overall effect of RK (400 mg/kg) compared with vehicle; *p* < 0.05.

**Figure 7.** Representative coronal micrographs from NTS and AP in mice dosed with vehicle or raspberry ketone (400 mg/kg). Immunopositive c-Fos cells are stained black. NTS subareas are caudal (cNTS; −7.92 mm from Bregma); medial (mNTS; −7.48 mm from Bregma); intermediate (iNTS; −7.08 mm from Bregma); rostral (rNTS; −6.84 mm from Bregma). AP are located mNTS. Black bar in each image represents 153 μm.

#### **4. Discussion**

RK is marketed in the United States and other countries as a nutraceutical for appetite control and weight loss [18,27,37]. This study sought to investigate the dose-dependent effects of RK on the meal patterns and hemodynamic alterations associated with an obesogenic diet. Our major finding was a dose-differential response in RK's effects on weight gain, meal patterns, and hemodynamic parameters.

RK has reported effects on weight gain due to an obesogenic diet in male mice when administered by admixture in the diet [16,17] or, in our previous studies, by oral gavage [18]. In these studies differences in weight gain were observed after three to five weeks of exposure to RK. Therefore, to understand the preceding changes that drive the reduced weight gain, we dosed animals for two weeks with a 200 mg/kg RK and high dose of RK 400 mg/kg. The dose of 400 mg/kg in a high-fat driven obesogenic environment has not been studied before. For weight gain, male mice that received 400 mg/kg RK had a reduced gain in body weight. The difference in body weight was observed on 2–5 days of the daily dosing. In females, there was a reduction in gain in body weight with RK 400 mg/kg compared with RK 200 mg/kg. These body weight differences between males and females are not surprising. Female mice are more resilient to the effect of a high-fat diet [38], and we expect to see greater differences in body weight due to RK in a longer exposure to daily dosing. Previously, we demonstrated that there was no difference in the acute oral dosed RK bioavailability in males compared with females [27]. However, metabolism of phenolic compounds can vary between males and females [39], a possibility that is currently being explored with RK.

Previously, we observed a reduction in weight gain with oral RK 200 mg/kg in males fed a high-fat diet for twice as long (i.e., 28 days) [18]. In that study, meal patterns were measured after 21 days of dosing and were not altered with RK 200 mg/kg [18]. In the present study, we analyzed meal patterns separately for dark and light cycles. We observed a dose effect with 200 mg/kg on nocturnal meal patterns in males. In males, the frequency of meals was reduced, but meal size and duration were increased. The 400-mg/kg dose increased the satiety ratio in males and females. Meal frequency was also reduced in females with 400 mg/kg compared with 200 mg/kg. Taken together, our results suggest that RK doses influence meal patterns, specifically meal frequency. Our study demonstrated an increase in meal number, size and duration associated with high-fat diet, which has been previously reported [23,40]. Similarly, diet-induced obesity susceptible rats consume more meals compared with control fed animals [22]. Therefore, nutraceuticals that mitigate the high-fat diet-induced changes in meal parameters can be useful to normalize aberrant feeding. Notably, the differences in RK-induced changes in body weight and meal patterns were not specific to the obesogenic high-fat diet condition.

Weight changes are often a secondary outcome of anxiogenic and antipsychotic agents [24, 25]. As such, we examined whether RK-induced weight loss and meal patterns alterations results in anxiety-like behaviors. The major findings were that RK 400 mg/kg decreased total distance traveled in the open field test. RK 200 mg/kg also increased time spent in the center, which is often noted with anxiolytic agents [41]. However, there were no significant findings in the elevated plus maze with 200 mg/kg. In addition, the reductions in entries into both open and closed arms are suggestive that RK 400 mg/kg reduces locomotor activity and does not promote anxiogenic behaviors. Alterations in pre-pulse inhibition have been noted with antipsychotic medications and amphetamines in rodents [34,42,43]. These compounds are also associated with adverse body weight and meal pattern alterations [44,45]. In this study, RK did not alter pre-pulse inhibition, suggesting the weight reduction was not secondary to sensorimotor gating impairments.

FDA-approved obesity medications and nutraceuticals that reduced body weight have been demonstrated to produce adverse cardiovascular outcomes [46–48]. Previous studies in rodents have suggested that RK has a cardioprotective role [19,20]. A 28-day treatment of oral RK (100–200 mg/kg) has been shown to prevent isoproterenol-induced cardiac tissue damage and dyslipidemia in rats [19,20]. That is, RK increased cardiac levels of peroxisome proliferator-activated receptor (PPAR)-alpha, and reduced markers of cell death suggesting a protective role [20]. RK shares structural similarity with synephrine [16], which is known to have an effect on hemodynamics [48,49]. As such, our study

demonstrated that RK 400 mg/kg reduced systolic blood pressure (BP), diastolic BP, and mean BP. These effects were demonstrated in males and females, regardless of diet. Future studies will examine the blood parameters, such as pro- and anti-inflammatory cytokines and lipid profiles related to these cardiometabolic parameters. Additionally, the present study demonstrated RK 400 mg/kg increased neural activation, as measured by c-Fos immunohistochemistry, of the nucleus of the solitary (NTS). The NTS receives sensory input from the vagal nerve and has been shown to be involved in meal patterns and cardiovascular control [29,30]. Future experiments will be conducted to elucidate the feeding and cardiovascular mechanisms of action and receptor activation of RK on the NTS and other brain regions.

#### **5. Conclusions**

Our study demonstrated that RK effectively reduced body weight, altered meal patterns, and reduced cardiovascular outcomes. These differences were differentially observed with dose, but they were not specific to the obesogenic diet. These studies suggest that RK might have limited use to *prevent* weight gain and metabolic signatures associated with high-fat diet.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6643/12/6/1754/s1, Figure S1: Experimental design for measurement of meal patterns and behavioral outcomes during exposure to respective diet and dose treatments. Figure S2: Experimental design for measurement of hemodynamic parameters after 14-day exposure to respective diet and dose treatments. Figure S3: Experimental design to study activation of caudal hindbrain in response to respective diet and dose treatments.

**Author Contributions:** Conceptualization and experimental design, N.T.B.; experimental execution, D.K., L.H., X.L.; writing—original draft preparation, D.K.; writing—review and editing, D.K., N.T.B., X.L., L.H.; funding acquisition, N.T.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** Research reported in this publication was supported by the NCCIH of the NIH under Award Number R01AT008933 and USDA (NIFA) NJ06280.

**Acknowledgments:** The authors would like to thank, Matthew Wereski, Nardine Nasr, and Gina M. Giunta for their technical assistance and Kathy Manger for her editorial assistance. The authors would especially like to thank Research Diets Inc., specifically Edward A. Ulman, Douglas Compton, and Juliet Gentile for generously providing the use of and the training with the BioDAQ system.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Allithiamine Alleviates Hyperglycaemia-Induced Endothelial Dysfunction**

#### **Attila Biró 1, Arnold Markovics 1, Mónika Éva Fazekas 1, Gábor Fidler 2, Gábor Szalóki 3, Melinda Paholcsek 2, János Lukács 4, László Stündl <sup>1</sup> and Judit Remenyik 1,\***


Received: 8 May 2020; Accepted: 29 May 2020; Published: 5 June 2020

**Abstract:** Diabetes mellitus-related morbidity and mortality is a rapidly growing healthcare problem, globally. Several nutraceuticals exhibit potency to target the pathogenesis of diabetes mellitus. The antidiabetic effects of compounds of garlic have been extensively studied, however, limited data are available on the biological effects of a certain garlic component, allithiamine. In this study, allithiamine was tested using human umbilical cord vein endothelial cells (HUVECs) as a hyperglycaemic model. HUVECs were isolated by enzymatic digestion and characterized by flow cytometric analysis using antibodies against specific marker proteins including CD31, CD45, CD54, and CD106. The non-cytotoxic concentration of allithiamine was determined based on MTT, apoptosis, and necrosis assays. Subsequently, cells were divided into three groups: incubating with M199 medium as the control; or with 30 mMol/L glucose; or with 30 mMol/L glucose plus allithiamine. The effect of allithiamine on the levels of advanced glycation end-products (AGEs), activation of NF-κB, release of pro-inflammatory cytokines including IL-6, IL-8, and TNF-α, and H2O2-induced oxidative stress was investigated. We found that in the hyperglycaemia-induced increase in the level of AGEs, pro-inflammatory changes were significantly suppressed by allithiamine. However, allithiamine could not enhance the activity of transketolase, but it exerts a potent antioxidant effect. Collectively, our data suggest that allithiamine could alleviate the hyperglycaemia-induced endothelial dysfunction due to its potent antioxidant and anti-inflammatory effect by a mechanism unrelated to the transketolase activity.

**Keywords:** allithiamine; garlic; hyperglycaemia; advanced glycation end-products; cytokines

#### **1. Introduction**

In recent decades, several research have focused on the pharmacologically active, plant-derived compounds [1]. Extensive research on physiological effects of nutraceuticals is expected to continue in the near future. A recently published review unequivocally declared that the ongoing discovery of naturally occurring drugs stands as a major contributor to cope with diseases, reaching high prevalence, globally [2].

Among plants having significant pharmacological activity, garlic (*Allium sativum* L.) is among the most studied ones [3]. Several studies have shown that garlic exerts antioxidant, antimicrobial [4], anti-inflammatory, immunomodulatory [5], antithrombotic [6], anti-atherosclerotic [7], antihypertensive [8], and anti-carcinogenic [9] effects. The biological effects of garlic are mainly attributed to its characteristic organosulfur compounds, including alliin, allicin, ajoene, S-allylmercaptocystein, diallyl disulfide, and S-allyl-cysteine, among others. [10]. Limited data in the scientific literature are available on the biological effects of another garlic component, allithiamine, which is a less polar thiamine (B1-vitamin) derivative and, similar to the molecules mentioned above, has a prop-2-en-1-yl disulfanyl moiety. According to a recent study, allithiamine is also accumulated in red sweet pepper (*Capsicum annuum* L.) seeds, implying that its occurrence is more frequent than as thought until now. Nevertheless, several studies revealed that numerous garlic compounds have beneficial effects on hyperglycaemia in diabetes mellitus [11].

Diabetes mellitus is a rapidly growing public health burden, particularly in developed countries [12]. Diabetes mellitus is a metabolic, endocrine disorder, which can cause an acute life-threatening homeostasis imbalance as well as chronically developing micro- and macrovascular complications (blindness, neuropathy, myocardial infarction, stroke, etc.) [13]. There is a common agreement that endothelial dysfunction precedes the development of micro- and macrovascular complications associated with diabetes mellitus [14]. These complications are caused—at least partially by the detrimental effects of hyperglycaemia, which affects endothelial cell biology by accelerating the formation of advanced glycation end-products (AGEs), thereby increasing pro-inflammatory signaling and resulting in oxidative stress [15].

Glucose reacts with an amino group of the circulating proteins during the formatting of AGEs. The level of AGEs elevates heavily in the presence of chronic hyperglycaemia to evoke both damaging biological functions of glycated molecules, resulting in altered intracellular signaling, gene expression, release of pro-inflammatory molecules, and enhanced oxidative stress by bonding to cell surface receptors (RAGE), and so consequently, AGEs play a major role in diabetic microvascular complications [16]. Hyperglycaemia, alone can trigger inflammation by activating the pro-inflammatory transcription factor nuclear κB (NF-κB), resulting in an increased inflammatory chemokine and cytokine release including interleukin-6 (IL-6), interleukin-8 (IL-8), and tumor necrosis factor-α (TNF-α), among others. [17]. A recent study reported that alleviating the release of pro-inflammatory cytokines has a beneficial effect in chronic hyperglycaemia [18]. In addition, a high level of glucose enhances oxidative stress, when the rate of oxidant production exceeds the rate of oxidant scavenging [19]. In the case of hyperglycaemia, there are both enhanced oxidant production and impaired antioxidant defenses by multiple interacting pathways [20]. Studies have demonstrated that compounds with a strong antioxidant property can potentially be effective in delaying diabetes-related complications.

To date, there is no preclinical evidence for the antidiabetic effect of allithiamine, therefore, the main objective of our current research was to study whether this compound is able to exert a beneficial effect on diabetes. Primary cultured human umbilical cord vein endothelial cells (HUVECs) were used as a unique hyperglycaemic model, which appeared to be ideally capable to investigate the level of AGEs, antioxidant status, and pro-inflammatory cytokines.

#### **2. Materials and Methods**

#### *2.1. Materials*

#### Chemicals

All reagents were obtained from the distributor of iBioTech Hungary Ltd. (Budapest, Hungary) and DIAGON Ltd. Hungary (Budapest, Hungary).

#### *2.2. Methods*

#### 2.2.1. Preparation and Purification of Allithiamine

Preparation and purification of allithiamine were carried out based on the method of our recent allithiamine-oriented study [21]. Briefly, allyl thiosulphate and thiamine hydrochloride with an opening thiazole ring were reacted. As a result of the reaction, many organosulfur compounds were formed, including allithiamine. Reaction products were separated and allithiamine was purified by reversed-phase chromatography using LaChrom HPLC equipped with a diode array detector. (Hitachi, Osaka, Japan). To confirm the accuracy and efficiency of the allithiamine synthesis and purification, matrix-assisted laser desorption/ionization mass spectrometric (MALDI-MS) analysis and HPLC-MS/MS fragmentation were performed applying a Bruker Biflex MALDI-TOF mass spectrometer (Bruker, Billerica, MA, USA) and Thermo Scientific Q Exactive Orbitrap mass spectrometer (Thermo Fisher Scientific Inc., Waltham, MA, USA).

#### 2.2.2. Isolation and Treatment of Primary HUVECs

The HUVECs were isolated from human umbilical cords and maintained according to the method previously described [22]. In our experiments, supplemented M199 medium was used as a control and had 5.6 mMol/L glucose. To create a hyperglycaemic model, glucose was added to M199 for a final concentration of 30 mMol/L.

#### 2.2.3. Characterization of HUVECs by Flow Cytometry

HUVECs were incubated with four fluorescent dye-labeled antibodies, involving fluoresceinisothiocyanate (FITC)-labeled mouse anti-human CD31, phycoerythrin (PE)-labeled anti-human CD54, allophycocyanin (APC)-labeled mouse anti-human CD106, and PerCP-Cy5.5-labeled mouse anti-human CD45 (BD Biosciences, Franklin Lakes, NJ, USA), and then characterized by using a FACSAriaIII Cell Sorter (Becton Dickinson, Mountain View, CA, USA) [22].

#### 2.2.4. Determination of Cellular Viability

Cell viability was determined by an MTT (3-[4,5-dimethylthiazol-2-yl]-2,5 diphenyltetrazolium bromide) assay. Cells were plated into 96-well plates at a density of 20,000 cells per well in quadruplicates and treated with allithiamine of different concentrations (0.1, 0.5, 1, 5, 10, and 50 μg/mL) and without allithiamine (control group) for 24, 48, and 72 h. The medium was removed and incubated with 100 μL of MTT solution (0.5 mg/ml dissolved in Dulbecco's modified Eagle's medium) for 3 h. Subsequently, the formazan crystals were dissolved in 100 μL solubilizing solution (81% (*v*/*v*) isopropyl alcohol, 9% (*v*/*v*), 1 M HCl, 10% (*v*/*v*) Triton X-100) and the absorbance was measured at 465 nm by using a Clariostar microplate reader (BMG Labtech, Ortenberg, Germany). Cell viability at different allithiamine concentrations was expressed relative to 100% of the control group.

#### 2.2.5. Determination of Apoptosis

The mitochondrial membrane potential of the HUVECs was determined by using 1,1 ,3,3,3 ,3 hexamethylindodicarbocyanin iodide (DilC1(5)) dye. The decrease in the DilC1(5) intensity indicated mitochondrial depolarization, i.e., the onset of early apoptotic processes of HUVECs [23,24]. Cells were seeded to 96-well plates at a density of 20,000 cells per well treated with allithiamine of different concentrations (0.1, 0.5, 1, 5, 10, and 50 μg/mL). After the removal of the medium, the cells were incubated for 30 min with 50 μL/well DilC1(5) working solution (50 nM dissolved in Dulbecco's modified Eagle's medium). After incubation, cells were washed twice with PBS and the fluorescence of DilC1(5) was measured at 630 nm excitation and 670 nm emission wavelengths by using a Clariostar microplate reader (BMG Labtech, Ortenberg, Germany). The results were expressed relative to 100% of the control group.

#### 2.2.6. Determination of Necrosis

Necrotic processes were evaluated by SYTOX Green staining. The dye is able to penetrate only necrotic cells with ruptured plasma membranes and then binds to the nucleic acids, whereas intact cells with unimpaired surface membranes show a negligible SYTOX Green staining intensity [23,24]. Cells were cultured in 96-well plates, and treated as indicated in Section 2.2.5. After the removal

of medium, cells were incubated for 30 min with 50 μL/well SYTOX Green dye (1 μM dissolved in Dulbecco's modified Eagle's medium) and then washed with PBS. The fluorescence of SYTOX Green was measured at 490 nm excitation and 520 nm emission wavelengths by using a Clariostar microplate reader (BMG Labtech, Ortenberg, Germany). The results were expressed relative to 100% of the control group.

#### 2.2.7. Performing ELISAs

Measurement of Advanced Glycation End-Products

The assay was performed according to the manufacturer' instructions by using an OxisSelectTM Advanced Glycation End Product (AGE) Competitive ELISA Kit (Cell Biolabs Inc., San Diego, CA, USA).

#### Measurement of NF-κB

The assay was performed according to the manufacturer' instructions by using a Human NF-κB p65 Sandwich ELISA Kit (Fine Biological Technology Ltd., Wuhan, China)

#### 2.2.8. Determination of Cytokine Release

HUVECs were seeded into a 6-well plate (500,000 cells/well), and were incubated in 5 mMol/L glucose and 30 mMol/L glucose with or without 5 μg/mL allithiamine for 6, 12, or 24 h. Supernatants were collected, centrifuged for 10 min 10,000 r·min−<sup>1</sup> and the released amount of IL-6, IL-8, and TNF-<sup>α</sup> was determined by using a MILLIPLEX MAP Human cytokine/chemokine Magnetic Bead Panel (EMD Millipore Corp., Billerica, MA, USA) based on the manufacturer's recommendation.

#### 2.2.9. Measurement of Transketolase Activity

Transketolase activity was measured by adding 100 μL cytosolic fraction to 200 μL reaction mixture containing 14.8 mMol/L ribose-5-phosphate, 253 μMol/L NADH, 185 U/mL triosephosphate isomerase, and 21.5 U/mL glycerol-3-phosphate dehydrogenase in Tris buffer (pH = 7.9). The optical density was measured at 340 nm immediately and then every 10 min for 2 h by using a Clariostar microplate reader (BMG Labtech, Ortenberg, Germany). The activity was calculated from the difference in the optical density readings at 10 and 80 by min using the extinction coefficient of NADH. Results are expressed in nMol/min/mg protein.

#### 2.2.10. Determination of Protein Content

The protein concentrations were determined in the cell lysate by using a Pierce BCA Protein assay (Pierce Biotechnology Inc., Rockford, IL, USA). Protease inhibitor cocktail (Pierce Biotechnology Inc., Rockford, IL, USA) was added to the cell lysate prior to its storage or measurement.

#### 2.2.11. Determination of ROS Production

The cells were seeded in a 24-well plate and exposed to 100 μMol/L 2 ,7 -dichlorofluorescin diacetate (DCFDA) for 1 h at 37 ◦C to label the intracellular ROS. After incubation, the cells were washed twice with PBS and divided into three groups: incubating with M199 medium as control; or with 100 μMol/L H2O2; orwith 100μMol/L H2O2 plus 5μg/mL allithiamine. Fluorescence (excitation = 485 nm; emission = 530 nm) was assessed by using a Clariostar microplate reader (BMG Labtech, Ortenberg, Germany).

#### 2.2.12. Statistical Analysis

For multiple comparisons, results were analyzed by an ANOVA followed by a modified *t*-test for repeated measures according to Bonferroni's method. Data were presented as mean ± SEM. Differences were considered statistically significant, when *p* < 0.05.

#### *2.3. Ethics*

The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of the University of Debrecen (registration number RKEB/IKEB 3712-2012).

#### **3. Results**

#### *3.1. Purification and Verification of Allithiamine*

Since allithiamine is not commercially available and that isolation from the plant does not work in large amounts, chemically prepared and purified allithiamine was applied for our experiments. The synthesis of allithiamine resulted in a wide variety of organosulfur compounds, which were separated by using the reversed-phase chromatographic method. After the chromatographic separation and purification of allithiamine, a fraction of allithiamine was verified by using the MALDI-MS and HPLC-MS/MS techniques. These methods clearly indicated that the chromatographic purification of allithiamine was accurate and efficient. The chromatogram (Supplementary Figure S1) of the synthetic mixture and MALDI-MS spectrum (Supplementary Figure S2) as well as the MS2 spectrum (Supplementary Figure S3) of purified allithiamine can be seen in the Supplemental Materials.

#### *3.2. Flow Cytometric Studies*

Isolated HUVECs were characterized to positive and negative marker expressions by applying flow cytometry and antibodies against specific marker proteins, a routinely used method in our recent HUVEC-oriented study [22]. CD106, CD45, CD31, and CD54 were applied to label the cells. As shown in Figure 1, the isolated HUVECs showed a high CD31 and CD54 positive marker expression, while approximately 93% of the cells did not express CD45 and CD106 (negative markers). These results suggest that the isolation of HUVECs was sufficiently accurate and efficient.

**Figure 1.** Flow cytometric analysis of human umbilical cord vein endothelial cells (HUVECs). Isolated HUVECs were verified using specific fluorescent-labeled antibodies. Forward- and side-scatter plot and dot-plots (**A**) of HUVEC positive (CD54, CD31) (**B**) and negative (CD45, CD106) (**C**) markers are shown. FSC: Forward scatter, SSC: Side scatter.

#### *3.3. Determination of Optimal Concentration of Allithiamine*

Up to 5 μg/mL, Allithiamine Treatment Has No Effect on Survival Rate of HUVECs

At first, the effect of allithiamine on cell viability was evaluated by using an MTT assay. HUVECs were exposed to allithiamine at different concentrations (0.1–50 μg/mL) for 24, 48, and 72 h. We found that up to 5 μg/mL allithiamine did not decrease the viability of HUVECs in this time window (Figure 2).

**Figure 2.** Viability of HUVECs was examined after 24 (**A**), 48 (**B**), or 72 (**C**) h. Results are expressed in the percentage of the control (0 μg/mL allithiamine). Data are expressed as the mean ± SEM of three individual experiments. Two additional experiments yielded similar results. \*, \*\*\*, and \*\*\*\* mark significant (*p* < 0.05, 0.0005, and 0.0001, respectively) differences compared with control.

In order to exclude the possibility of early apoptotic and necrotic events, which are not obvious alterations in the MTT assay, we further evaluated the effect of allithiamine whilst performing fluorescent labeling (DilC1(5) and SYTOX Green dyes). The results show that allithiamine, in line with the MTT data, did not induce a necrotic and apoptotic process and can be used without the risk of any biologically relevant cytotoxic actions in this concentration range (≤5 μg/mL; 24–72 h; Figure 3). Based on these preliminary experiments, a concentration of 5 μg/mL allithiamine was selected for further investigations.

#### *3.4. Allithiamine Can Reduce the Level of Advanced Glycation End-Products (AGEs)*

In cases of chronic hyperglycaemia, cells are exposed to prolonged elevated glucose concentrations leading to an excessive formation of AGEs [25]. Several studies have reported the positive effect of garlic sulfur compounds on the level of AGEs [26,27]. Therefore, we aimed to study the effects of allithiamine on the level of AGEs after one-day and one-week exposure to hyperglycaemia. As expected, 30 mMol/L glucose significantly increased the level of AGEs in HUVECs (Figure 4). We found that this hyperglycaemia-induced nearly 2-fold increase was significantly suppressed by the above-revealed non-cytotoxic concentration (5 μg/mL) of allithiamine.

**Figure 3.** Fluorescent DilC1(5) and SYTOX Green labeling. Effect of allithiamine on apoptosis and necrosis after 24 (**A**), 48 (**B**), or 72 (**C**) h. Results (intensity of fluorescence) are expressed in the percentage of the control (0 μg/mL allithiamine; 100% is represented by the solid lines). Apoptosis is indicated by a decrease in fluorescence of DilC1(5), and necrosis is indicated by an increase in fluorescence of SYTOX Green. Data are expressed as the mean ± SEM of three individual experiments. Two additional experiments yielded similar results. \*\*\*\* and #### mark significant (*p* < 0.0001 in both cases) differences compared with the control group (0 μg/mL allithiamine). CCCP: carbonyl cyanide m-chlorophenyl hydrazone (positive control for apoptosis); LB: lysis buffer (positive control for necrosis). SYTOX Green: non-permeable fluorescent nucleic acid dye; DilC1(5): 1,1 ,3,3,3 ,3 -hexamethylindodicarbocyanin iodide dye.

**Figure 4.** Effect of allithiamine on level of advanced glycation end-products (AGEs) in HUVECs after one day and one week. Data are expressed as the mean ± SEM of three individual experiments. \*\*\*\* marks a significant (*p* < 0.0001) difference between the control and HG, and between HG and HG+5 μg/mL allithiamine after one week. n.s.: not significant. HG: hyperglycaemia (30 mMol/L glucose).

#### *3.5. Allithiamine Can Alleviate the Hyperglycaemia-Induced Inflammatory Response in HUVECs*

Several studies reported the close relationship between hyperglycaemia and the inflammatory response of endothelium. In order to get a deeper insight into the effects of allithiamine on these inflammatory reactions, we assessed the activation of NF-κB in the cell lysate after 6 and 12 h and the release of TNF-α, IL-6, and IL-8 in the cell supernatant after 6, 12, and 24 h incubation in 5 mMol/L glucose and 30 mMol/L glucose with or without 5 μg/mL allithiamine. We found that the hyperglycaemic condition was able to trigger the inflammatory processes at an early time. The activation of NF-κB was significantly increased after the hyperglycaemic treatments in all the sampling times. Allithiamine was able to decrease this increment (Figure 5). The secretion of IL-6 and IL-8 was significantly increased after the hyperglycaemic treatments in all the sampling times. The TNF-α secretion significantly increased after the hyperglycaemic treatment at 6 and 12 h. At 24 h, a statistically not significant but marked biological change was observed. We found that allithiamine could alleviate the hyperglycaemia-induced inflammatory response by suppressing the pro-inflammatory cytokines mentioned above (Figure 6).

**Figure 5.** Effect of allithiamine on the activation of NF-κB in HUVECs after 6 and 12 h. Data are expressed as the mean ± SEM of three individual experiments \*, \*\*, and \*\*\* mark significant (*p* < 0.05, *p* < 0.005, and *p* < 0.0005) differences between the control and HG, and between HG and HG+5 μg/mL allithiamine, HG: hyperglycaemia (30 mMol/L glucose).

#### *3.6. Allithiamine Has No E*ff*ect on Transketolase Activity*

Considering the encouraging results presented above, we assayed to determine which mechanism may be responsible for the positive biological changes. First, we assumed that allithiamine, similar to thiamine derivatives, is able to enhance the activity of transketolase, a key enzyme in the pentose phosphate pathway. Benfothiamine is a potent transketolase activator [28], therefore, we applied it as a positive control in our experiments involving transketolase. The enzyme activity was evaluated in HUVECs after 6, 12, and 24 h of incubation. In the 6-h sample, we observed that 20 μg/mL benfothiamine increased significantly the transketolase activity in cells incubated in 30 mM glucose compared with cells incubated in 5 mM glucose and 30 mM glucose. However, 5 μg/mL allithiamine in cells incubated in 30 mM glucose did not change the transketolase activity significantly (Figure 7). Further incubations did not result in a significant increase in the transketolase activity (data not shown).

**Figure 6.** Effect of allithiamine on the level of pro-inflammatory cytokines including IL-6 (**A**), IL-8 (**B**), and TNF-α (**C**) in HUVECs after 6, 12, and 24 h. Data are expressed as the mean ± SEM of three individual experiments \*, \*\*, \*\*\*, and \*\*\*\* mark significant (*p* < 0.05, *p* < 0.005, *p* < 0.0005, and *p* < 0.0001) differences between the control and HG, between HG and HG+5 μg/mL allithiamine, and between control and HG+5 μg/mL allithiamine. HG: hyperglycaemia (30 mMol/L glucose).

**Figure 7.** Effect of allithiamine on the transketolase activity in HUVECs after 6 h of incubation. Data are expressed as the mean ± SEM of three individual experiments. Two additional experiments yielded similar results. \*\* marks a significant (*p* < 0.005) difference between the control and HG+20 μg/mL benfotiamin. HG: hyperglycaemia (30 mMol/L glucose).

#### *3.7. Allithiamine Exerts a Potent Antioxidant E*ff*ect*

As small molecule organosulfur compounds with a prop-2-en-1-yl disulfanyl moiety of garlic have strong antioxidant properties [29], we further intended to investigate the potential antioxidant effect of allithiamine. To assess the antioxidant capacity of allithiamine, hydrogen-peroxide (H2O2), a routinely used oxidative agent [30], was applied in our HUVEC model, ensuring an enhanced ROS

production and imbalance in the oxidative status of cells. As expected, the administration of H2O2 significantly increased the production of ROS. Allithiamine was able to eliminate a significant part of this increment, indicating the potent antioxidant effect of allithiamine (Figure 8).

**Figure 8.** Antioxidative effect of allithiamine in HUVECs incubated with or without 100 μMol/L H2O2. Fluorescent intensity was normalized to the baseline (**A**). H2O2 was administrated as indicated by the arrow. Statistical analysis was performed at the peak fluorescence (F/F0) values (**B**). Data are expressed as the mean ± SEM of three individual experiments. Two additional experiments yielded similar results. \*\*\*\* marks a significant (*p* < 0.0001) difference between the control and H2O2, and between H2O2 and H2O2+5 μg/mL allithiamine. H2O2: hydrogen-peroxide. – indicates the absence of treatment substance; + indicates the presence of treatment substance.

#### **4. Discussion**

In recent years, several research indicate that plant-derived compounds will be among the most important sources of new drugs [31]. Plants tend to produce several chemically highly diverse secondary metabolites, which may be suitable for exerting positive effects in human diseases [32]. Numerous studies have shown that garlic compounds (ajoene, alliin, allicin, diallyl disulfide S-allyl-cystein, etc.) are particularly valuable in this regard [33]. An experiment with mice demonstrated that hyperglycaemia was suppressed by ajoene treatment [34]. Another study revealed that allicin had a protective effect on hyperglycaemia-induced injury in aortic endothelial cells [35]. Findings of a study in rats suggest that S-allyl-Cystein treatment exerts a therapeutic protective effect on diabetes by decreasing oxidative stress [36]. Furthermore, comprehensive studies on garlic showed its therapeutic potential in various diseases accompanied by hyperglycaemia, including diabetes mellitus [37]. Our objective was to investigate the effect of allithiamine, a less-studied garlic component, on hyperglycaemia-induced endothelial pathologic changes (AGEs yielding, inflammatory processes, ROS production) in a HUVEC model.

Primarily, HUVECs were characterized by flow cytometric analysis using antibodies against specific marker proteins including CD31, CD45, CD54, and CD106. The results proved that the cells used in our experiments were endothelial cells, indeed. Subsequently, the non-cytotoxic concentration of allithiamine was determined by using an MTT assay and fluorescent labeling with DilC1(5) (examination of apoptosis) and SYTOX Green (investigation of necrosis) dyes. The viability tests of HUVECs after exposure to allithiamine at different concentrations (0.1–5 μg/mL) for up to 72 h indicate that cells were not significantly affected.

In parallel with the research on the garlic component which showed a suppressed formation of AGEs [27], excellent markers of tissue damage caused by persistent hyperglycaemia [38], we examined the effect of allithiamine on the level of AGEs. We found that a prolonged (one week) hyperglycaemiainduced increase in the level of AGEs was significantly suppressed by allithiamine.

In order to investigate the effects of allithiamine on inflammatory processes, we examined NF-κB which has been proven to be upregulated under hyperglycaemic conditions. NF-κB has a pivotal role in the inflammatory process as a transcriptional factor of several pro-inflammatory cytokines [39]. We found that allithiamine alleviated the expression of NF-κB caused by an elevated glucose level. We also assessed the release of several cytokines including IL-6, IL-8, and TNF-α, which faithfully reflect the inflammatory state of HUVECs [40]. To observe the early release of the preformed cytokine pool and secretion of de novo-synthetized cytokines, we measured the level of the abovementioned molecules after 6, 12, and 24 h of treatment. We also revealed that allithiamine was able to significantly decrease the level of the investigated cytokine at the time of sampling.

Seeking possible reasons behind the positive effect of allithiamine, two plausible explanations were raised. At first, we assumed that allithiamine is able to enhance the transketolase activity, resulting in a decreased flux of hyperglycaemia-induced pathologic pathways involving the polyol, AGEs, PKC, and hexosamine pathways, similar to thiamine [41] and benfothiamine [28]. However, we found that 5 μg/mL allithiamine did not change the transketolase activity significantly. We assumed that this non-toxic concentration may be rather negligible to increase the transketolase activity. At second, the widely researched garlic compounds having prop-2-en-1-yl moiety, as a two-edged sword, are able to attenuate the oxidative damage [29] and exert a beneficial anti-inflammatory effects as a slow H2S donor [42,43]. Allithiamine possesses a prop-2-en-1-yl moiety, as well. In order to get a deeper insight into the putative antioxidant effect of allithiamine, an experiment was performed to verify the eliminating ability of ROS using H2O2 as an oxidative agent. Allithiamine was able to eliminate the significant part of ROS after the administration of H2O2. Consequently, our results clearly indicate that allithiamine exhibits a potent antioxidant effect, which may be responsible for the improvement in hyperglycaemia-induced endothelial dysfunction.

#### **5. Conclusions**

Collectively, in this study, we observed that—without influencing the viability, necrosis, or apoptosis of HUVECs—allithiamine was able to attenuate particular negative effects of an elevated glucose level by its potent antioxidant effect and a mechanism unrelated to the transketolase activity. Further studies are needed to elucidate the mechanisms of action of allithiamine as well as its supposed beneficial role in the spectrum of plant-derived bioactive molecules. However, our results contribute to a better understanding of this relatively less-researched compound, with particular respect to its beneficial effects on hyperglycaemia.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6643/12/6/1690/s1, Figure S1: HPLC chromatogram of allithiamine at 250 nm, Figure S2: MALDI-TOF spectrum of purified allithiamine, Figure S3: MS2 spectrum of allithiamine.

**Author Contributions:** J.R., A.B., conceived and designed the experiments, A.B., A.M., M.É.F., M.P. and G.F. performed the experiments. A.B., A.M., G.S., L.S. and J.L. analyzed the data. A.B. and J.R. wrote the paper. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** The work is supported by the GINOP-2.3.2-15-2016-00042 project. This project is co-financed by the European Union and the European Social Fund. The authors are grateful for the generous help of the Hungarian Academy of Sciences-University of Debrecen (HAS-UD) Vascular Biology and Myocardial Pathophysiology Research Group, Faculty of Medicine, Nephrology Division, Debrecen, Hungary, for the human umbilical vein endothelial cell isolation. The authors would like to thank Ádám Biró for helpful advice and corrections.

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

### **Antiarthritic E**ff**ects of a Root Extract from** *Harpagophytum procumbens* **DC: Novel Insights into the Molecular Mechanisms and Possible Bioactive Phytochemicals**

### **Alessia Mariano 1, Antonella Di Sotto 2, Martina Leopizzi 3, Stefania Garzoli 4, Valeria Di Maio 3, Marco Gullì 2, Pietro Dalla Vedova 5, Sergio Ammendola <sup>6</sup> and Anna Scotto d'Abusco 1,\***


Received: 28 July 2020; Accepted: 21 August 2020; Published: 23 August 2020

**Abstract:** *Harpagophytum procumbens* (Burch.) DC. ex Meisn. is a traditional remedy for osteoarticular diseases, including osteoarthritis (OA), although the bioactive constituents and mechanisms involved are yet to be clarified. In the present study, an aqueous *H. procumbens*root extract (HPE; containing 1.2% harpagoside) was characterized for its effects on synoviocytes from OA patients and phytochemical composition in polyphenols, and volatile compounds were detected. HPE powder was dissolved in different solvents, including deionized water (HPEH2O), DMSO (HPEDMSO), 100% *v*/*v* ethanol (HPEEtOH100), and 50% *v*/*v* ethanol (HPEEtOH50). The highest polyphenol levels were found in HPEDMSO and HPEEtOH50, whereas different volatile compounds, mainly β-caryophyllene and eugenol, were detected in all the extracts except for HPEH2O. HPEH2O and HPEDMSO were able to enhance CB2 receptor expression and to downregulate PI-PLC β2 in synovial membranes; moreover, all the extracts inhibited FAAH activity. The present results highlight for the first time a multitarget modulation of the endocannabinoid system by HPE, likely ascribable to its hydrosoluble compounds, along with the presence of volatile compounds in *H. procumbens* root. Although hydrosoluble compounds seem to be mainly responsible for endocannabinoid modulation by HPE, a possible contribution of volatile compounds can be suggested, strengthening the hypothesis that the entire phytocomplex can contribute to the *H. procumbens* healing properties.

**Keywords:** osteoarthritis; nutraceuticals; polyphenols; volatile compounds; β-caryophyllene; eugenol; FAAH; cannabinoid receptors; phospholipases

#### **1. Introduction**

Osteoarthritis (OA) is a pathology of the whole joint structure, involving several cellular and molecular processes, in different types of cells, such as chondrocytes, osteoblasts, synoviocytes, and immune cells [1]. The clinical symptoms include pain as well as joint dysfunction and deformity, often leading to joint replacement surgery, with high costs for healthcare [2].

Recently, an association between inflammation and endocannabinoid receptors has been described [3]. In particular, the role of CB2 receptors seems very interesting in OA inflammation. CB2 is a peripheral cannabinoid receptor and has been found in immune system cells, raising the possibility that the endocannabinoid system could have a role in immunomodulatory processes [3]. In this respect, several studies reported that mice lacking CB2 receptors showed an exacerbated inflammatory phenotype [4]. Moreover, CB2 receptors are involved also in inhibiting nociceptive transmission [5]. CB2 receptor signaling has been associated with phospholipase C (PI-PLC) activation in calf pulmonary endothelial cells and in mast cells [6,7]. Moreover, the inhibition of inflammatory pathways, such as NF-κB nuclear translocation, has been obtained by inhibiting the PI-PLC β pathway in osteoblast-like cells [8]. The presence of CB2 in articular joints has been explored, mainly in animal models, and it has been highlighted in chondrocytes and in synoviocytes [9,10]. Thus, therapeutic strategies able to modulate CB2 signaling could be considered as a novel approach.

OA is treated with analgesic agents and anti-inflammatory and painkiller drugs, mainly non-steroidal anti-inflammatory drugs (NSAIDs), with the aim of alleviating symptoms [11]. Structure-modifying agents, such as nutraceuticals, are also administered to OA patients, with the aim of preventing or delaying cartilage degradation, even though further studies are required to confirm their effectiveness [12–14]. ARTRIT DOL, an Italian food supplement (Italian Minister of Health food supplement no. 71362 since 2013), is a composition containing glucosamine, chondroitin sulfate, extracts from *Harpagophytum procumbens* DC. and *Glycyrrhiza glabra* L., *Curcuma longa* L. roots, manganese, and copper in traces. Although this work does not focus on ARTRIT DOL, it is interesting to note that, as stated by the manufacturer, this nutritional supplement obtained good feedback from OA patients, in particular for the reduction of pain.

*Harpagophytum procumbens* (Burch.) DC. ex Meisn. (Fam. Pedaliaceae), commonly known as devil's claw, is a plant used worldwide as a traditional remedy for joint pain associated with OA and mild rheumatic ailments [15,16]. Moreover, it has been described to have analgesic effects on neuropathic pain in rats [17]. The harpagoside, one of the characteristic constituents of *H. procumbens* root, has been shown to be effective for osteoarthritis and low back pain [18].

The pharmacological activity of devil's claw root is attributed to the whole phytocomplex containing iridoid glucosides, such as harpagoside, phenolic glycosides (acteoside and isoacteoside), mono- and polysaccharides, triterpenes (mainly oleanolic acid, 3β-acetyloleanolic acid, and ursolic acid), phytosterols, phenolic acids (caffeic, cinnamic, and chlorogenic acids), flavonoids, and minor components such as volatile compounds [15]. A number of studies have been conducted in order to characterize the analgesic and anti-inflammatory activities of *H. procumbens* and its secondary metabolites. An in vitro study showed that *H. procumbens* was able to decrease the production of proinflammatory cytokines and inhibit metalloprotease activity in human monocytes [19]. Studies conducted in animals showed that an aqueous *H. procumbens* extract showed dose-dependent analgesic and anti-inflammatory activity, but the purified harpagoside did not effectively inhibit the inflammatory pathways, at least at the dosage used in that study [20]. However, human clinical studies showed that the administration of *H. procumbens* root extract was able to improve the clinical picture of OA patients, in terms of pain and limitation of movements [21–23], suggesting that the phytocomplex may contribute to observed effects [20]. The biological effects of volatile compounds have not been extensively studied. Recently, the combination of purified β-caryophyllene with curcumin was shown to reduce the inflammation pathway through inhibition of NF-κB in OA primary chondrocytes [24]. Purified β-caryophyllene has been described to mitigate pain in a mouse model of arthritis through a mechanism involving CB2 receptors [25]. Moreover, the volatile eugenol, a known dentistry analgesic, has been shown to be effective in a monoiodoacetate-induced rat model of osteoarthritis [26].

In the present study, an *H. procumbens* root extract was studied for its effects on human primary synoviocytes, with particular attention to the expression of CB2 receptors; furthermore, the extract was characterized for its polyphenol and volatile phytochemical content, in order to highlight possible novel bioactive compounds.

#### **2. Materials and Methods**

#### *2.1. Harpagophytum Procumbens Extract*

The dry aqueous extract from *H. procumbens* root (HPE) was provided by Ambiotec S.A.S. It was obtained starting from 300 g of root slices, which were repeatedly washed with cold ethanol 99% (*v*/*v*). After being washed, the slices were reduced by grinding with an electric miller for 10 min, and the powder was recovered by a sieve, with 4 mm holes. This powder was dissolved in ethanol/water solution (1:5 *w*/*v*) in the dark, shaking at 58 ◦C for 24 h, and then, it was filtered using 0.45 μm membrane. The retained fraction was air dried at 80 ◦C in a static dryer and in the presence of 5% maltodextrin (as carrier). The powder was again screened by a certified sieve (diameter 200 μm, mesh 100; net light 0.150 μm). The final yield of this preparation was 70% (drug-extract ratio, DER 1.4) of the initial root slices. The extract was a fine brown-colored powder with characteristic smell and taste, containing 1.2% (*w*/*w*) harpagoside. The powder was stored at room temperature in dry and dark conditions until use.

In order to evaluate antiarthritic activity and to highlight possible bioactive constituents, HPE was further dissolved in different solvents, including dimethyl sulfoxide (DMSO), 100% *v*/*v* ethanol (EtOH), 50% *v*/*v* EtOH, and deionized water. These solvents were chosen on the basis of their biocompatibility and ability to dissolve different classes of phytochemicals, mainly focusing on volatile compounds and polyphenols. Particularly, DMSO possesses high solubilizing properties for both polar and nonpolar compounds, thus, being able to dissolve the entire phytocomplex. Conversely, deionized water and ethanol recover mainly polar and nonpolar molecules, respectively, whereas 50% *v*/*v* ethanol is able to collect compounds dissolved by both solvents.

#### *2.2. Phytochemical Analysis*

#### 2.2.1. Determination of Total Polyphenols, Tannins, and Flavonoids

Total amounts of polyphenols and tannins in the tested extracts were determined spectrophotometrically by the Folin–Ciocalteu method, as previously reported [27]. For both polyphenols and tannins, the absorbance was measured at 765 nm, and the amount was calculated as tannic acid equivalent (TAE) per milligram of dry HPE extract. Furthermore, total flavonoids and its subclass flavonols were measured by applying the aluminum chloride method with minor changes [27]. Specifically, total flavonoids were measured after mixing equal volumes of each extract (2 mg/mL) and aluminum trichloride (2% *w*/*v* in methanol), whereas the content of flavonols was determined by mixing 50 μL extract (4 mg/mL), 20 μL aluminum trichloride (10% *w*/*v* in methanol), 60 μL sodium hydroxide (1 M), 10 μL sodium nitrite (5% *w*/*v* in deionized water), and 70 μL deionized water. After a 10-min incubation, the absorbance was measured at 415 nm, and the total contents of flavonoids and flavonols were determined and expressed as quercetin equivalent (QE) per milligram of dry HPE extract.

#### 2.2.2. Solid-Phase Microextraction (SPME)

For the extraction of volatile compounds, solid-phase microextraction (SPME) holders and coating fibers (Supelco; Bellefonte, PA, USA) were used. The sampling was performed with an SPME fiber (50/30 μm divinylbenzene/carboxen/polydimethylsiloxane—DVB/CAR/PDMS). Before sampling, the SPME fiber was conditioned by heating in the injector of a gas chromatograph at 270 ◦C for 30 min in order to remove traces of contaminants. Prior to analysis, a fiber blank was run to confirm the absence of contaminant peaks. To obtain a better extraction, SPME conditions, such as the most suitable temperature and equilibration time, were adjusted. Each sample (0.5 mL) was placed in a septum-sealed glass vial. The fiber was exposed to the headspace of the sample for 20 min at 40 ◦C. During this time, samples were stirred with a magnetic stirrer. After equilibration, the fiber was removed from the sample and immediately inserted into the GC injection port for the thermal desorption (2 min) at 270 ◦C.

#### 2.2.3. Gas Chromatography–Mass Spectrometry (GC–MS)

The extracts were analyzed using a GC–MS Perkin Elmer Clarus 500 instrument (Perkin Elmer, Waltham, MA, USA) equipped with a flame ionization detector (FID). Chromatographic separations were performed on a Varian FactorFour VF-1 fused-silica capillary column (length 60 m × 0.32 mm ID × 1.0 μm film thickness). The oven temperature program was as follows: 60 ◦C for two min, then, a gradient of 6 ◦C/min to 250 ◦C for 10 min, and an injector temperature of 270 ◦C. Helium was used as the carrier gas with a flow rate of 1.0 mL/min. Split injection with a split ratio of 1:20 was used. The electron-impact ionization mass spectrometer was operated as follows: ionization voltage, 70 eV; ion source temperature, 200 ◦C; scan mode, 30.0 to 500.0 mass range. The volatile compounds were identified by comparing mass spectra with those in the NIST02 and Wiley libraries. Furthermore, linear retention indices (LRIs) of each compound were calculated using a mixture of n-alkane hydrocarbons (C8-C30, Ultrasci, Ultra Scientific Italia, Bologna, Italy) injected directly into GC injector using the same temperature program reported above. Semiquantitative analysis was performed by normalizing the peak area generated in the FID (%) without using correction factors (relative response factors, RRFs). All analyses were repeated twice.

#### *2.3. Human Tissue*

Human synovial membranes were isolated from 6 OA patients and 5 non-OA (fractured) patients that underwent surgical treatment. Full ethical consent was obtained from all donors and the Research Ethics Committee, Sapienza University of Roma (#290/07, 29 March 2007), and ASL Lazio 2 (#005605/2019, 3 March 2019) approved the study. The tissues were fixed in 4% paraformaldehyde in 0.1 M phosphate buffer pH 7.2 immediately after removal from patients.

#### *2.4. Immunohistochemistry*

Histological sections were deparaffinized and rehydrated in graded ethanol. Endogenous peroxidase activity was blocked by 3% hydrogen peroxide for 10 min. Antigen retrieval was performed in 10 mM sodium citrate buffer (pH 6.0) for 15 min. The sections were then incubated with anti-CB1, anti-CB2, anti-PI PLC β2, and anti-PI PLC β3 (Santa Cruz Biotechnology, Inc., Dallas, TE, USA), all diluted 1:50, overnight at 4 ◦C. After incubation, specimens were washed and incubated with the secondary-biotinylated antibody and subsequently, with streptavidin–biotin–peroxidase (DAKOLSAB Kit peroxidase; DAKO, Carpinteria, CA, USA). The signals were developed by incubating with freshly prepared 3,3 -diaminobenzidine (DAB) substrate–chromogen buffer at room temperature. Slides were counterstained with hematoxylin and mounted with permanent mounting media. Negative controls were used in each experiment. The samples were scored semiquantitatively using a score based on the intensity and distribution: 0, undetectable; 1+, weak staining; 2+, medium staining; 3+, strong staining [28].

#### *2.5. Human Primary Cell Isolation*

Human primary synoviocytes (FLSs) were isolated from synovial membranes, obtained, as above described, from patients who underwent a total knee and hip arthroplasty. In brief, the synovial membrane fragments were minced and treated with 1 mg/mL collagenase type IV and 0.25% trypsin for 1 h, at 37 ◦C in agitation. Then, FLSs were grown to 80% confluence in DMEM (HyClone, Logan, UT, USA) supplemented with L-glutamine, penicillin/streptomycin (Sigma-Aldrich, St. Louis, MO, USA), and 10% fetal bovine serum (FBS) and cultured at 37 ◦C and 5% CO2. All experiments were carried out with synoviocytes at first passage (p1), isolated from at least 3 different donors.

#### *2.6. Immunofluorescence*

Vimentin, CB2, and PI-PLC β2 were visualized by immunofluorescence. Cells were plated at a density of 8 <sup>×</sup> <sup>10</sup>3/cm2 and cultured for 48 h and then, washed in PBS, fixed in 4% paraformaldehyde in PBS for 15 min at 4 ◦C, and permeabilized with 0.5% Triton-X 100 in PBS for 10 min at room temperature. After blocking with 3% bovine serum albumin (BSA) in PBS for 30 min at room temperature, cells were incubated at 1 h, at room temperature, with mouse monoclonal anti-vimentin antibody (Proteintech Group, Manchester, UK) 1:50, mouse monoclonal anti-CB2 antibody 1:150, and mouse monoclonal anti-PI-PLC β2 antibody (Santa Cruz Biotechnology) 1:50. Cells were washed with PBS and then, incubated for 1 h, at room temperature, with Alexa Fluor 488 donkey anti-rabbit antibody 1:300 (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA), to stain vimentin green; with Alexa Fluor 488 donkey anti-goat antibody 1:600 (Invitrogen, Thermo Fisher Scientific), to stain CB2 receptors green; and Alexa Fluor 595 donkey anti-rabbit antibody 1:300 (Invitrogen, Thermo Fisher Scientific), to stain PI-PLC β2 red. Slides were washed and then, stained with DAPI (Invitrogen, Thermo Fisher Scientific) to visualize the nuclei. The images were captured by a Leica DM IL LED optical microscope, using an AF6000 modular microscope (Leica Microsystem, Milan, Italy).

#### *2.7. Densitometric Analysis*

The free software ImageJ (https://imagej.nih.gov/ij/) was used to perform the densitometric analysis of protein production. For each cell culture condition, the integrated density values of fluorescence were considered.

#### *2.8. Cell Treatment*

Cells were left untreated (CTL) or treated, for the required time, with 0.1 mg/mL of *Harpagophytum procumbens* root extract (HPE) dissolved in deionized water (HPEH2O), DMSO (HPEDMSO), 100% *v*/*v* EtOH (HPEEtOH100), and 50% *v*/*v* EtOH (HPEEtOH50). Experiments were independently repeated at least three times.

#### *2.9. Cell Viability*

To assess a potential cytotoxic effect of *H. procumbens* extracts on FLSs at different concentrations and time points, an MTS (3-(4,5-dimethylthiazol-2–yl)-5-(3-carboxymethoxyphenyl)- 2-(4-sulfophenyl)-2*H*-tetrazolium)-based colorimetric assay was performed (Promega Corporation, Madison, WI, USA). Briefly, 8 <sup>×</sup> <sup>10</sup><sup>3</sup> cells per well were seeded in a 96-well plate. The day after seeding, cells were starved overnight in reduced serum medium, in order to align cell cycle progression. Cells were then left untreated (CTL) or treated with *H. procumbens* extracts for 24, 48, and 72 h. After each time point, 100 μL MTS solution was added to the wells. Spectrophotometric absorbance was directly measured at 492 nm after 3 h incubation.

#### *2.10. RNA Extraction and Reverse Transcription*

Total RNA was extracted with TRIZOL (Invitrogen, Thermo Fisher Scientific), purified using a micro RNeasy column (Qiagen, Valencia, CA, USA), and reverse transcribed by Improm II enzyme, (Promega Corporation, Madison, WI, USA), according to the manufacturers' instructions.

#### *2.11. Quantitative Real-Time PCR*

Quantitative real-time PCR analysis was performed using an ABI Prism 7300 (Applied Biosystems, Thermo Fisher Scientific). Amplification was carried out using SensimixPlus SYBR master mix (Bioline, London, UK). Primers were designed using Primer Express software (Applied Biosystems) and were synthesized by Biofab Research (Rome, Italy). Primers' sequences are reported in Table 1.


**Table 1.** Sequences of the primers used in RT-PCR analysis.

Data were analyzed by the 2−ΔΔCt method, which determines the transcript abundance relative to the 18S housekeeping gene [29].

#### *2.12. FAAH Inhibition*

The potential ability of the tested extract to inhibit fatty acid amide hydrolase (FAAH) was evaluated using a commercial fluorescence-based kit (Cayman's FAAH Inhibitor Screening Assay Kit, Vinci Biochem, Vinci (FI), Italy), according to the manufacturer's instructions. The fluorescence of the FAAH-catalyzed product was measured at an excitation wavelength of 340 to 360 nm and an emission wavelength of 450 to 465 nm by a BD Accuri™ C6 flow cytometer (BD Biosciences, Milan, Italy). Suitable control wells treated with vehicles (maximum FAAH activity) and with the known FAAH inhibitor JZL 195 (maximum FAAH inhibition) were included. Each treatment was assayed at least in triplicate and at least in two different experiments. The enzyme activity was evaluated as % inhibition with respect to the vehicle.

#### *2.13. Statistical Analysis*

All data were obtained from at least three independent experiments, each performed either in duplicate or in triplicate (n = 6 or n = 9). Data were statistically analyzed with two-way repeated measures analysis of variance (ANOVA) with Bonferroni's multiple comparison test using Prism 5.0 software (GraphPad Software, San Diego, CA, USA). The *p* value < 0.05 was considered significant.

The Hill equation E = Emax/(1 + (10LogEC50/A) HillSlope) (E, effect at a given concentration; Emax, maximum activity; EC50 or IC50, concentration giving a 50% inhibition; A, concentration of agonist; HillSlope, slope of the agonist curve) was applied to obtain a concentration–response curve.

#### **3. Results**

#### *3.1. Phytochemical Characterization*

The highest levels of total polyphenols were extracted by DMSO and 50% *v*/*v* EtOH, followed by deionized water and pure ethanol. Indeed, their levels in HPEDMSO and HPEEtOH50 were 1.2- to 1.4-fold higher than in HPEH2O and HPEEtOH100 (Table 2). Tannins were mainly recovered by H2O and DMSO, their amount in HPEH2O and HPEDMSO being 1.5- to 1.9-fold higher than that found in HPEEtOH100 and HPEEtOH50 (Table 2). Regarding flavonoids and flavonols, the highest extraction power was exhibited by DMSO; HPEDMSO exhibited about 1.5- to 3-fold and 1.5- to 1.7-fold higher concentrations of total flavonoids and flavonols than HPEEtOH50 or HPEEtOH100 and HPEH2O, respectively (Table 2).



TAE—tannic acid equivalent. QE—quercetin equivalent. \*\* *p* < 0.01 and \*\*\* *p* < 0.01 significantly higher than HPEEtOH100 (ANOVA followed by Bonferroni's multiple comparison post hoc test). § *p* < 0.05 and §§ *p* < 0.01 significantly higher than HPEH2O (ANOVA followed by Bonferroni's multiple comparison post hoc test).

Altogether, these results revealed that DMSO is the most suitable for recovering total polyphenols, including tannins, flavonoids, and flavonols, followed by 50% *v*/*v* EtOH, especially for total flavonoids and flavonols. Pure ethanol extract possesses similar features compared to 50% *v*/*v* EtOH regarding total flavonoids and flavonols, with a lower recovery of total polyphenols. Finally, pure deionized water was able to better extract tannins, with significantly lower flavonoid levels.

The SPME–GC–MS analysis highlighted the presence of different volatile compounds in all the samples except for the aqueous HPEH2O extract. Among them, eight compounds, listed in Table 3, were identified and their relative percentage amounts were calculated. Eugenol and β-caryophyllene were the main volatile phytochemicals in all extracts. Eugenol achieved a maximum 51.6% amount in HPEETOH50, while HPEDMSO and HPEETOH100 contained the highest percentages of β-caryophyllene (i.e., 77.4% and 77.1%, respectively). Moreover, β-pinene (5.1%) and isoeugenol (1.0%) were found in HPEETOH50, while thymol (0.8%) was found in HPEDMSO. α-copaene was also identified in ethanolic extracts HPEEtOH100 and HPEEtOH50 (6.3% and 3.7%, respectively). On the other hand, α-humulene (10.0%; 0.8%) and δ-cadinene (5.8%; 1.3%) were detected in HPEDMSO and HPEEtOH100, respectively.

**Table 3.** Volatile compounds (relative percentage in the volatile fraction) detected in *Harpagophytum procumbens* root extract (HPE) dissolved in DMSO (HPEDMSO), 100% *v*/*v* EtOH (HPEEtOH100), 50% *v*/*v* EtOH (HPEEtOH50), and deionized water (HPEH2O).


<sup>1</sup> Compound identification number; <sup>2</sup> compounds are reported according their elution order on column; <sup>3</sup> linear retention indices measured on apolar columns; <sup>4</sup> linear retention indices from literature; <sup>+</sup> normal alkane RI; <sup>5</sup> identification by MS spectra, tr <sup>&</sup>lt; 0.1%.

#### *3.2. Expression of Cannabinoid Receptors in OA Synovial Membrane*

The presence of CB1 and CB2 receptors and PI-PLC β2 and β3 was detected by immunohistochemistry in synovial membranes isolated both from non-OA and OA patients. CB1 receptors were expressed in both normal and pathological tissues but with different scores; they were moderately present in non-OA patients (score 1+), while strongly present in OA patients (score 3+) (Figure 1, left-upper panel). CB2 receptors were detected only in non-OA tissue (score 2+) (Figure 1, left-bottom panel). Regarding the presence of PI-PLC β2 and β3, we found that the expression of PI-PLC β2 showed a lesser staining in non-OA (score 2+) than OA tissue (score 3+), and it was localized in the cytoplasm of both tissues (Figure 1, right-upper panel). PI-PLC β3 was equally expressed in tissues from non-OA (score 3+) and OA patients (score 3+) (Figure 1, right-bottom panel). We also analyzed PI-PLC β1 and PI-PLC β4, finding that were both barely expressed both in non-OA and OA synovial membranes.

**Figure 1.** Immunohistochemical analysis of synovial membranes from OA (osteoarthritis) and non-OA patients. Left panel: Slices were stained with anti-CB1 and anti-CB2 receptor antibodies. Right panel: Slices were stained with anti-PI-PLC β2 and anti-PI-PLC β3 antibodies. Slides were counterstained with hematoxylin and mounted with permanent mounting media. This figure shows representative images of different experiments (n = 5 non-OA and n = 6 OA).

#### *3.3. E*ff*ects of Harpagophytum Extracts on Synoviocyte Cell Viability*

Synoviocytes were isolated by both OA and non-OA tissues. Two types of synoviocytes are present in the synovial membrane, A and B. The synoviocytes A are macrophage-like and the synoviocytes B are fibroblast-like synoviocytes (FLSs). The latter are characterized by their elongated shape and the expression of vimentin [30]; we verified the expression of this protein in our cells, finding that they were able to produce vimentin (Figure 2A).

The effects of *Harpagophytum procumbens* extracts on FLS cell viability were determined by the MTS colorimetric method. The extracts, tested at 1 mg/mL, 0.5 mg/mL, and 0.1 mg/mL for 24, 48, and 72 h, did not show detrimental effects at any analyzed concentration or time point (Figure 2B). We decided to use the 0.1 mg/mL concentration for further experiments.

#### *3.4. Mechanism of Action of Harpagophytum Extracts on FLSs*

In order to assay the effects of the different extracts on cannabinoid receptor expression, they were added to cell culture medium at a concentration of 0.1 mg/mL for 24 h, then, the mRNA expression level of CB1 and CB2 receptors was analyzed. The HPEH2O and HPEDMSO and to a lesser extent, HPEEtOH50, were able to increase the CB2 mRNA expression level, whereas HPEEtOH100 did not show any effect (Figure 3). CB1 receptor mRNA expression level was increased by HPEH2O and HPEDMSO and to a lesser extent by HPEEtOH100, whereas it was decreased by HPEEtOH50 (Figure 3).

**Figure 2.** Characterization of human primary synoviocytes (FLSs) and analysis of cell viability. (**A**) Human primary FLSs, isolated by synovial membranes and cultured in vitro, were stained with anti-vimentin primary antibody and with Alexa Fluor 488 (green) secondary antibody. (**B**) Cell viability was assessed by the MTS colorimetric method, and FLSs were treated with three concentrations, 1, 0.5, and 0.1 mg/mL of *Harpagophytum procumbens* root extract (HPE) dissolved in deionized water (HPEH2O), DMSO (HPEDMSO), 100% *v*/*v* EtOH (HPEEtOH100), and 50% *v*/*v* EtOH (HPEEtOH50), for 24, 48, and 72 h. Cell viability of treated samples was normalized to the untreated cells, which is reported as 100% and represented by a horizontal line.

**Figure 3.** Effects of all HPE extracts on CB1 and CB2 mRNA expression level in human primary FLSs. After 24 h treatment with 0.1 mg/mL of *Harpagophytum procumbens* root extract (HPE) dissolved in deionized water (HPEH2O), DMSO (HPEDMSO), 100% *v*/*v* EtOH (HPEEtOH100), and 50% *v*/*v* EtOH (HPEEtOH50), cells were harvested and mRNA was extracted and analyzed by RT-PCR. CB1 and CB2 receptor mRNA levels were reported as relative mRNA expression level with respect to 18S mRNA (2−ΔΔCt method). Results are expressed as mean <sup>±</sup> S.E.M. of data obtained by three different experiments. Statistical significance was \* *p* < 0.05; \*\* *p* < 0.01.

Considering that only CB2 receptors are associated with inflammation and pain in peripheral tissues, we verified whether the CB2 receptors also increased at the protein level by immunofluorescence staining. FLSs were treated with 0.1 mg/mL extracts for 24 and 48 h, then, the cells were stained with antibody anti-CB2. We found that HPEH2O and HPEDMSO extracts were able to stimulate the exposure in membrane of CB2 receptors, both at 24 and 48 h, whereas HPEEtOH50 and HPEEtOH100 did not stimulate CB2 receptor expression at any analyzed time (Figure 4 and Figure S1). Taking into account that the expression of PI-PLC β2 was increased in synovial membranes from OA patients, we verified whether the HPE extracts were able to reduce the production of this phospholipase. HPEH2O and HPEDMSO were able to inhibit the expression of PI-PLC β2, whereas HPEEtOH50 and HPEEtOH100 did not show effects (Figure 4).

**Figure 4.** Effects of all HPE extracts on CB2 receptor and PI-PLC β2 protein production. Upper panel: Cells were treated with 0.1 mg/mL of *Harpagophytum procumbens* root extract (HPE) dissolved in deionized water (HPEH2O), DMSO (HPEDMSO), 100% *v*/*v* EtOH (HPEEtOH100), and 50% *v*/*v* EtOH (HPEEtOH50), for 24 h and then, analyzed by immunofluorescence using anti-CB2 and anti-PI-PLC β2 primary antibodies and Alexa Fluor 488 (green, CB2) and Alexa Fluor 568 (red, PI-PLC β2) secondary antibodies, respectively. Nuclei were stained with DAPI (original magnification 40×). Lower panel: The pixel intensities in the region of interest were obtained by ImageJ. \* *p* < 0.05.

#### *3.5. Inhibition of Fatty Acid Anandamide Hydrolase*

In order to check whether the HPE extracts were able to affect the fatty acid amide hydrolase (FAAH), we analyzed both the FAAH mRNA expression level and enzymatic activity. HPEH2O was able to decrease the FAAH mRNA level even if the downregulation was not statistically significant, whereas all other extracts were ineffective (Figure 5A). Interestingly, under our experimental conditions, all the extracts were able to interfere with the FAAH activity, although with different efficacy and potency (Figure 5B). Particularly, HPEH2O was the least effective sample, achieving a maximum 58.5% enzyme inhibition at the highest concentration of 2500 μg/mL. Conversely, the other extracts could completely inhibit the FAAH enzyme, HPEEtOH100 being slightly more potent than HPEDMSO and HPEEtOH50, which displayed similar potencies. Indeed, the IC50 value of HPEEtOH100 was about 1.2- to 1.5-fold lower than those of HPEEtOH50 and HPEDMSO (Table 4). Under the same experimental conditions, 100 μg/mL harpagoside (corresponding to 200 μM) was found to be ineffective in the inhibition of the FAAH enzyme (about 9% inhibition compared the control). Conversely, the positive control JZL 195 (20 μM corresponding to 8.7 μg/mL) produced a maximum 90% enzyme inhibition (Table 4). As expected, the positive control was significantly more potent than the HPE extracts (Table 4).

**Figure 5.** Effects of *Harpagophytum procumbens* root extract (HPE) on fatty acid anandamide hydrolase (FAAH) expression and enzymatic activity. (**A**) Cells were treated with 0.1 mg/mL of *Harpagophytum procumbens* root extract (HPE) dissolved in deionized water (HPEH2O), DMSO (HPEDMSO), 100% *v*/*v* EtOH (HPEEtOH100), and 50% *v*/*v* EtOH (HPEEtOH50), for 24 h. Cells were then harvested, and mRNA was extracted and analyzed by RT-PCR. FAAH mRNA levels were reported as relative mRNA expression level with respect to 18S mRNA (2−ΔΔCt method). Results are expressed as mean <sup>±</sup> S.E.M. of data obtained by three different experiments. (**B**) Concentration–response curves showing the inhibitory effects on FAAH from HPEDMSO, 100% *v*/*v* HPEEtOH100, 50% *v*/*v* HPEEtOH50, and HPEH2O. Data are the mean ± SE of at least three independent experiments with two replicates for each experiment (n = 6).



Not evaluable being lower than 80% inhibition achieved. \* *p* < 0.05 significantly lower than HPEDMSO (ANOVA followed by Bonferroni's multiple comparison post hoc test).

#### **4. Discussion**

The aim of this study was to investigate the effects of *Harpagophytum procumbens* extract (HPE) on fibroblast-like synoviocytes (FLSs) from osteoarthritis patients, with particular attention on the endocannabinoid-mediated mechanisms. Osteoarthritis (OA) is characterized by chronic inflammation, and it is currently treated with anti-inflammatory drugs, which are only able to counteract the symptoms [2,31]. Several nutraceuticals, such as glucosamine, chondroitin sulfate, and curcumin, are administered with the aim of delaying the cartilage degradation, with inconsistent results [32]. Extracts from plants are also traditionally used to treat OA [16]. Among them, *H. procumbens* DC. has been studied for its chondroprotective activities, mainly for the ability to inhibit the production of proinflammatory mediators, such as TNFα and IL-1β, and enzymes able to hydrolyze the extracellular matrix components, metalloproteases, and elastase [19].

Iridoid glycosides have received major attention as possible bioactive compounds of *H. procumbens* secondary metabolites [33]. Harpagoside is the most investigated one, and it is considered a reference standard of *H. procumbens* for titration purposes [15]. The anti-inflammatory activity of harpagoside has been found to be mediated by the inhibition of COX-1 and COX-2 enzymes along with by a lowered cytokine and NO release [33]. It was also able to counteract inflammation in primary human osteoarthritic chondrocytes through the suppression of c-FOS/AP-1 signal and the inhibition of proinflammatory cytokine and fibrinogenic factor production [23]. Moreover, harpagoside requires hydrolysis to a bioactive metabolite to exert its anti-inflammatory activity [34].

Despite this promising evidence, the anti-inflammatory properties of harpagoside cannot fully explain those of the entire *H. procumbens* phytocomplex, thus, suggesting that other compounds can contribute to the activity of the plant [15,33]. In the phytocomplex, some phenylpropanoids were reported to contribute to the *H. procumbens* anti-inflammatory effects [15].

Some studies have highlighted that plant roots can release volatile compounds as a defense strategy to counteract pathogen and fungal infections and to mediate the interaction between plant and soil bacteria [35–37]. Accordingly, our SPME–GC–MS analysis of HPE extracts highlighted the presence of several volatile compounds, mainly recovered by HPEDMSO, 100% *v*/*v* HPEEtOH, and 50% *v*/*v* HPEEtOH. Among these volatile compounds, β-caryophyllene and eugenol were identified to be the major sesquiterpene and monoterpene present, followed by the sesquiterpenes α-humulene and α-copaene. Previously, 31 different volatile compounds, obtained by an heptanoic extraction, were characterized in *H. procumbens* root, although neither sesquiterpenes nor monoterpenes were detected [38]. In several preclinical models, β-caryophyllene, α-humulene, and eugenol have been shown to possess anti-inflammatory activities, affecting proinflammatory cytokine secretion and inhibiting inducible nitric oxide synthase (iNOS) and cyclooxygenase (COX-2) expression [39,40]. β-caryophyllene acts as an agonist of the endocannabinoid CB2 receptor, which is involved in the modulation of inflammation and the immune system and inhibits the fatty acid amide hydrolase enzyme [41,42]. Recent evidence highlighted the ability of this sesquiterpene to reduce articular and systemic inflammation in several animal models of arthritis [25,43,44]. Particularly, its antiarthritic effects in human articular chondrocytes were mediated by a crosstalk between CB2 and PPAR-γ receptors [25].

In articular joints, FLSs are involved both in supporting health chondrocytes and the immune system during inflammation [45]. In this study, we decided to use FLSs isolated from human synovial membranes as an in vitro model to study the effects of *H. procumbens* extracts. Preliminarily, the expression of CB2 receptors was checked in synovial membranes from OA and non-OA joints, and we found that it was expressed only in non-OA joints and was completely absent in OA. This finding agrees with the observations of Fukuda and coworkers, who found that CB2 receptors are expressed in rheumatoid arthritis (RA) synovial membranes and not in OA synovial membranes [46]. Moreover, they showed that CB2 plays an anti-inflammatory role in RA, and the treatment of RA synoviocytes with an agonist of CB2 blocked the production of proinflammatory mediators, through the inhibition of adenylyl cyclase, which, in turn, did not produce cAMP. Thus, protein kinase A was not activated, finally leading to failure of the activation of NF-κB [46]. Accordingly, the absence of CB2 in OA synovial membranes, which has been observed in our samples, can be associated with the activated inflammatory pathways present in joints of OA patients. Several studies showed the activation of NF-κB in OA joints [34,47]; thus, the activation of pathways or molecules that can downregulate NF-κB activity is very desirable [48,49]. Moreover, Sophocleous and coworkers showed that CB2-/- mice had a greater susceptibility to OA [4].

The phosphoinositide (PI)-dependent signal plays important roles in many cellular processes, among them proinflammatory pathways, the alteration of which is involved in the onset and progression of several diseases [50]. The PI-phospholipase C (PI-PLC) β1 isoform has been described as differently modulated in osteoblasts from OA and RA patients [51]. We analyzed the synovial membranes from OA and non-OA patients for the presence of PI-PLC β1–β4, finding that β1 and β4 were almost unexpressed both in OA and non-OA tissues, whereas β3 was equally expressed in both tissues. Interestingly, β2 was highly expressed in OA and poorly expressed in non-OA, suggesting that only β2 is related to injured OA tissue. These findings prompted us to evaluate whether the *H. procumbens* extracts were able to modulate both CB2 and PI-PLC β2 in isolated human primary FLSs, finding that HPEH2O and HPEDMSO stimulated the expression of CB2 receptors and decreased the expression of PI-PLC β2, whereas the HPEEtOH50 had a weak effect and HPEEtOH100 was completely ineffective on the expression of these two molecules. Thus, the ability of HPEH2O and HPEDMSO to increase the expression of CB2 receptors may explain the anti-inflammatory and antinociceptive activity of this plant. The role of PI-PLC β2 in inflammatory pathways has not been described so far; for this reason, the inhibition of its expression needs to be explored more in depth. We suppose that its expression in OA tissue could be associated with stimulation of the proinflammatory pathway through the activation of protein kinase C.

Moreover, further anti-inflammatory mechanisms are affected, as shown by the inhibition of FAAH in all the analyzed HPE extracts. The best inhibition was shown by HPEEtOH100 and HPEEtOH50, so we can hypothesize that volatile compounds, such as β-caryophyllene, eugenol, and α-humulene, can be involved in these HPE effects.

#### **5. Conclusions**

Harpagoside, harpagide, and procumbide are present in all extracts, and ethnopharmacology considers them primarily responsible for *H. procumbens* activity. However, several studies, performed at the molecular level, showed that the administration of these purified compounds cannot justify the analgesic activity of the whole phytocomplex. The present study showed that further extractions with deionized water or DMSO can recover some bioactive compounds able to increase the synthesis of CB2 receptors, which are unexpressed in the osteoarthritic tissues. Moreover, the deionized water and DMSO extracts decrease the expression of the PI-PLC β2 isoform, which, in turn, could inhibit the FAAH synthesis of endocannabinoids. Interestingly, DMSO, 50% ethanolic, and to a greater extent, 100% ethanolic extracts, were able to inhibit FAAH activity.

Further studies in in vitro cell models and in vivo animal models are needed to support the hypothesis that *H. procumbens* root can be effective in controlling osteoarthritic pain, through the modulation of endocannabinoid system. The specific contribution of iridoid glucosides, polyphenols, and terpenes deserve to be further investigated too.

**Supplementary Materials:** The following materials are available online at http://www.mdpi.com/2072-6643/12/ 9/2545/s1, Figure S1: Effects of all HPE extracts on CB2 receptor and PI-PLC β2 protein production after 48 h treatment.

**Author Contributions:** Conceptualization, A.S.d., A.D.S. and S.A.; methodology and investigation, A.M., A.D.S., M.L., S.G., V.D.M., M.G., P.D.V.; data curation, A.S.d., A.M., A.D.S.; writing—original draft preparation, A.S.d., S.A. and A.D.S. All authors have read and agree to the published version of the manuscript.

**Funding:** This research was partially funded by "Progetto di Facoltà 2019".

**Acknowledgments:** We would like to thank Ssa E. Bisicchia for helpful discussion. A.D.S. fellowship was funded by grants from Sapienza University, "Progetto di Ateneo 2019".

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **In Vivo Anti-Inflammatory E**ff**ects and Related Mechanisms of Processed Egg Yolk, a Potential Anti-Inflammaging Dietary Supplement**

**Joan Cunill 1,\*, Clara Babot 1, Liliana Santos 2, José C. E. Serrano 2, Mariona Jové 2, Meritxell Martin-Garí <sup>2</sup> and Manuel Portero-Otín 2,\***


Received: 28 July 2020; Accepted: 30 August 2020; Published: 4 September 2020

**Abstract:** Egg-yolk based supplements have demonstrated biological effects. We have developed a novel processed egg-yolk (PEY) complement, and we have tested whether it has inflammation modulatory properties. These were evaluated in a lipopolysaccharide (LPS)-challenge in 1-month male rats by in vivo circulating cytokine profiles measured by multiplexing techniques. Cell culture was used to explore ex vivo properties of derived serum samples. We explored growth factor composition, and mass-spectrometry metabolome and lipidome analyses of PEY to characterize it. PEY significantly prevented LPS-induced increase in IL-1 β, TNF-α, and MCP-1. Further, serum from PEY-treated animals abrogated LPS-induced iNOS build-up of the Raw 264.7 macrophage-like cell line. Immunochemical analyses demonstrated increased concentrations of insulin-like growth factor 1 (IGF-1), connective tissue growth factor (CTGF), and platelet-derived growth factor (PDGF) in the extract. PEY vs. egg-yolk comparative metabolomic analyses showed significative differences in the concentrations of at least 140 molecules, and in 357 in the lipidomic analyses, demonstrating the complexity of PEY. Globally, PEY acts as an orally-bioavailable immunomodulatory extract that may be of interest in those conditions associated with disarranged inflammation, such as inflammaging.

**Keywords:** fecundation; inflammation; cytokine; growth factors; metabolomics; lipidomics

#### **1. Introduction**

Inflammaging is a chronic increase in the body's pro-inflammatory status with advancing age in some tissues [1]. As inflammatory reactions increase, neurohormonal signaling (e.g., renin-angiotensin, adrenergic, insulin-IGF1 signaling) tends to be deregulated, immunosurveillance against pathogens and premalignant cells declines, and the composition of the peri- and extracellular environment changes, thereby affecting the mechanical and functional properties of tissues involved [2]. With the rise of this aging characteristic, a new concept of "anti-inflammaging" was also proposed, which influences progressive pathophysiological changes, as well as lifespan, and acts along with inflammaging [1]. In physiological conditions, the immune system helps to maintain homeostasis by mounting nonspecific innate and specific adaptive responses (inflammation) against potential aggressions. The inflammatory response is driven by a complex network of mediators and signaling pathways and is the net effect of interactions between pro-inflammatory and anti-inflammatory molecules (cytokines) that determines the immune response [3]. A properly functioning and reliable immune system is essential

for maintaining health. When the expression of cytokines is chronically altered, it can lead to chronic inflammation, tumorigenesis, and autoimmunity. Inflammaging comprises the phenomena explaining the age-related trend towards increasing pro-inflammatory cytokine concentrations, more pronounced for IL-6, IL-8, IL-2, IFN-γ, and TNF-α [4], which is an indicator of the inflammation. It is logical to think, thus, that adequating the wrong cytokine profile, can prevent or slow down over some of the harmful consequences of the aging process.

Among the measures addressed to this rearrangement, pharmacological treatments can be found—statins to improve cardiac health [5]; also, nutritional approaches like caloric restriction [5]. Also, dietary supplements have emerged as potential anti-aging treatments, like antioxidants (curcumin, polyphenols), vitamins, and probiotics. The NIH defines dietary supplements as "substances you might use to add nutrients to your diet or to lower your risk of health problems, like osteoporosis or arthritis" [6]. Raising health concerns, changing lifestyles, and dietary habits, are driving increased attention to these products. In this line, there is ample scientific evidence that eggs contain biologically active compounds that may have a role in the therapy and prevention of chronic and infectious diseases. Several dietary supplements derived from eggs can be found in the market, broadly divided into unfertilized (commercialized), and fertilized. Commercialized hen eggs are highly nutritious food; its macronutrient content includes low carbohydrates and about 12 g per 100 g of protein and lipids (most of which are monounsaturated) as well as several nutrients such as zinc, selenium, retinol, and tocopherols. It is known that regular commercialized egg yolk contains proteins and peptides with biological activity. For example, it has been seen that whole egg yolk in the native form as a potential source of angiotensin converting enzyme-inhibitory peptides, or that yolk lipoproteins are important for lipid-mediated antimicrobial activity; also, a protein called γ-livetin, also referred to as immunoglobulin Y, exerts the same immunomodulatory activity as immunoglobulin G, and has been shown to have immunoregulatory effects [7]. In fact, in vivo anti-inflammatory and analgesic effects of nonfertilized egg yolk have been reported [8]. Based on preclinical studies, egg phosphatidylcholine and sphingomyelin species appear to regulate cholesterol absorption and inflammation; in clinical studies, egg phospholipid intake is associated with beneficial changes in biomarkers related to HDL reverse cholesterol transport [9]. Fertilized eggs contain every nutrient essential to sustaining a new life, in a closed shell that only permits O2, CO2 and H2O gas exchange [10], and its composition changes as the embryo develops. The screening of the bioactive components in the different development stages has attracted researcher's attention, and important differences have been observed, showing very interesting potential applications. However, reports on the immune mechanism and the related bioactive components are limited.

Furthermore, to our knowledge, there are no studies regarding the effects of the intake of fertilized eggs *in vivo*. We have obtained, by a patented process [11], an extract of fertilized eggs, termed here patented egg yolk (PEY). To shed light over its effects in the inflammaging context, we explored its potential anti-inflammatory effect in comparison with the commercial egg yolk, and we characterized its composition in order to establish further screening protocols of the potential mechanisms of action, with the final goal of establishing the basis for potential interventional inflammaging studies.

#### **2. Materials and Methods**

#### *2.1. Chemicals*

Unless stated otherwise, all chemicals were obtained from indicated major commercial suppliers. PEY consists of an egg preparation obtained through a patented process [11]. It comprises a mixture of yolk and white extracted from a fertilized egg which has been incubated for a short period, which then is lyophilized to obtain a commercial product called Excelvit®.

#### *2.2. Animals*

One-month-old male Wistar rats (initial weight 75–100 g, *n* = 10 per group in the 5 day treatment regime, and *n* = 5 in the two or one day-treatment regime) obtained from Harlan Laboratories (Catalunya, Spain) were maintained at 23 ± 2 ◦C under a 12:12 h light-dark cycle (lights on from 07:00 to 19:00). All rats were allowed unlimited access to Harlan Teklad 14% protein rodent maintenance diet during the whole experimental period. After one week of acclimation to animal facilities, the animals were weighed and divided into two groups (PEY and egg yolk group) with equal bodyweight. Experimental treatment consisted of a 4-h fast (8:00–12:00 a.m.), where further a daily dose of 2000 mg/kg of PEY or egg yolk during one, two or five days was administered by oro-gastric gavage. Bodyweight and signs of toxicity were recorded daily during the treatment period, in which all treated animals showed adequate health. After the designated duration (one, two or five days), 2-h after the oro-gastric gavage of PEY or egg yolk, all animals were intraperitoneally injected with a 2.5 mg/kg dose of lipopolysaccharide (LPS, Sigma-Aldrich, Sant Louis, MO, USA). One hour after LPS injection, all animals were sacrificed by cervical displacement, and blood samples were collected by cardiac puncture. Serum from blood samples was collected within 30 min by centrifugation and immediately frozen with liquid nitrogen, and stored at −80 ◦C until further analyses. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the University of Lleida Institutional Animal Care Committee (Project Code CEEA 03/03-12).

#### *2.3. Cytokine Analysis*

Cytokines were determined by Milliplex RECYMAG65K27PMX magnetic bead immunology multiplex assay (Merck-Millipore, Burlington, MA, USA) following the manufacturer's instructions.

#### *2.4. Cell Culture and Treatments*

Mouse RAW 264.7 macrophages (ATCC TIB-71) were grown in a humidified incubator containing 5% CO2 and 95% air at 37 ◦C. Cells were grown in DMEM medium supplemented with 10% of heatinactivated fetal bovine serum, 100 UI/mL of penicillin, and 100 μg/mL of streptomycin. In all experiments, cells were grown until confluence and then transferred to 6-well plates. On the day of the assay, the medium was changed to serum-free medium for 4 h, and further cells were exposed to DMEM medium supplemented with 10% rat serum obtained from rats exposed to a dose of 2000 mg/day (for five days) of PEY or egg yolk, or fetal bovine serum. For iNOS activation, a dose of 0.1 microgram/mL of LPS was added to selected wells. After 24-h, cells were collected and lysed using RIPA buffer with protease and phosphatase inhibitors. Protein concentrations were measured using the Bradford assay (BioRad Laboratories, München, Germany) with bovine serum albumin as a standard, and further cell-lysates were frozen a −80 ◦C for further analyses.

#### *2.5. Western Blot Analysis*

Total protein (15–40 μg) was resolved by SDS-PAGE and electroblotted onto polyvinylidene difluoride membranes (Immobilon-P Millipore, Bedford, MA, USA). Immunodetection was performed using primary antibodies anti-iNOS (Cell Signaling Technologies #2977), PDGF (RD Systems, AB23NA), NGF (Abcam Ab6199), CTGF (Genetex, GTX37727), and IGF-1 (Abcam, Ab106838). A monoclonal antibody to β-actin (Sigma, Saint Louis, MO, USA) was used to control protein loading from cell culture samples. Protein bands were visualized with the chemiluminescence ECL® method (Millipore Corporation, Billerica, MA, USA). Luminescence was recorded and quantified in Lumi-Imager equipment (Boehringer, Mannheim, Germany), using the Quantity One 4.6.5. software (Bio-Rad, Hercules, CA, USA).

#### *2.6. Lipidomic and Metabolomic Analyses*

Metabolites and lipids in PEY and egg yolk were analyzed by liquid chromatography coupled to mass spectrometry as previously described [12]. For metabolomics, samples were depleted of proteins by methanol addition [13]. The resulting metabolites were separated using a reverse-phase column (Zorbax SB-Aq 1.8 μm 2.1 × 50 mm; Agilent Technologies, Barcelona, Spain). We employed a gradient of water to methanol (all with 0.2% acetic acid) in a chromatograph (LC Agilent 1290) coupled to a mass spectrometry system (time of flight (TOF) mass spectrometer Agilent 6520). Mass spectra were collected in both negative and positive ionization modes, scanning from *m*/*z* values from 50 to 1600, at 1.5 scans/s. For lipidomic analyses, lipids were obtained by chloroform:methanol extraction, internal standards added, and processed as described [14,15] employing the same system as above (Agilent Technologies, Santa Clara, CA, USA) with adequate solvents [14]. This method allows the orthogonal characterization (based on exact mass (<10 ppm) and on retention time) of lipids. When combined with internal standards, this strategy is useful for proposing potential identities with low uncertainty [14,16]. In this case, data were collected in both positive and negative electrospray ionization in full-scan mode at 100–3000 *m*/*z* in an extended dynamic range (2 GHz).

#### *2.7. Statistical Analyses and Data Annotation*

Statistic calculations were performed using SPSS (IBM SPSS v 25, Armonk, NY, USA) and GraphPad Prism 8 (GraphPad Software, San Diego, CA, USA). The Kolmogorov–Smirnov test checked the normality of the distribution of variables. For metabolomics and lipidomic analyses, only common features (found in ≥75% of the replicas of PEY and egg yolk samples) were taken into account, to correct for individual bias. Principal component analysis (PCA), partial least squares discrimination analysis (PLS-DA), and hierarchical clustering analysis were performed using MassHunter Mass Profiler Professional (Agilent Technologies, Barcelona, Spain) after the transformation of chromatographic results to the CEF® format. Analyses (Volcano plots, PCA and PLS-DA reported here) were performed employing the Metaboanalyst 3.0 platform [17].

Annotations of metabolites were produced based on an accurate mass-retention time algorithm (Agilent®, MassHunter Mass Profiler Professional) and for lipids on the comparison with the retention time of internal standards. Therefore, the level of evidence of annotation is 2 (i.e., putatively annotated compounds, without chemical reference standards, based upon physicochemical properties and spectral similarity with public/commercial spectral libraries), according to [18]. A level of *p* < 0.05 was selected as the point of minimal statistical significance in every comparison. Pathway analyses were produced according to these putatively annotated compounds by employing the ConsensusPathDataBase platform [19].

#### **3. Results**

As expected, LPS injection induced an increase in plasma levels of cytokines (Supplemental Figure S1A), increasing ca 43% circulating levels of TNF-α after 2 h. To study the systemic anti-inflammatory effects of PEY, plasmatic concentrations of cytokines of animals feed during five days with PEY were measured after 2 h of performing the inflammatory stimulus (LPS injection) (Figure 1A). Compared with egg-yolk, the PEY group presented a significant reduction in plasmatic concentrations of IL-1 β (Figure 1B), TNF-α (Figure 1C), and MCP-1 (Figure 1D). Interestingly, shorter times of treatment also sufficed to abrogate LPS-induced buildup of circulating cytokines (1 day, Supplemental Figure S1A; or 2 days, Supplemental Figure S1B).

Demonstrating the bioavailability of PEY and the solubility of PEY-induced effects, the antiinflammatory effect was confirmed ex-vivo using Raw 264.7 cells treated with the serum of the animals exposed during five days to PEY and egg-yolk (Figure 1E). RAW 264.7 cells were treated with LPS to stimulate their inflammatory response, and i-NOS content was measured. A significant difference in the iNOS immunoreactivity between the LPS group and the other groups, showing the different inflammatory responses of the cells, was observed (Figure 1E). iNOS production by the cells treated with the serum of the egg yolk group was significantly lower than control, and the serum of rats treated with PEY had the highest anti-inflammatory effect, significantly lower than the egg yolk.

**Figure 1.** Characterization of the anti-inflammatory effects of processed egg-yolk in vivo. (**A**). Oral administration of processed egg-yolk (PEY) during five days and application of inflammatory aggression the 5th day by lipopolysaccharide (LPS). (**B**–**D**). Plasmatic concentrations of cytokines IL-1β, TNF-α, and MCP-1 in the control group (egg yolk) and treated group (PEY) after 2 h of the administration of lipopolysaccharide (LPS) (\* *p* < 0.05, \*\*\* *p* < 0.001). (**E**). iNOS immunoreactivity of Raw 264.7 cell culture exposed to 0.1 μg/mL of LPS after the application of standard cell medium (no treatment), blood serum from rats treated with egg yolk (control) and serum from rats treated with PEY; a \*\*\*\* significant difference with cell medium LPS− *p* < 0.001; b \*\*\*\* significant difference with cell medium LPS+ *p* < 0.0001; c \* significant difference with serum from egg yolk LPS+ *p* < 0.05. One-way ANOVA or Student's *t*-test was used for statistical analyses. In vitro experiments were performed in triplicate, while as treatments were delivered to *n* = 10 animals per group.

The composition of PEY was studied to explore the origin of the anti-inflammatory effects observed. We evaluated the presence of growth factors (Figure 2), as well as the comparative metabolomics and lipidomics (Figure 3) of egg yolk and PEY.

It was observed that PEY contained 2.5 more connective tissue growth factor (CTGF), 1.7 more platelet-derived growth factor (PDGF), 1.3 more nerve growth factor (NGF) and 8.6 more insulin-like growth factor (IGF-1) than egg yolk (see Figure 2). All the increases were significant, but NGF. The presence of growth factors has been previously described [20]. The results of liquid chromatography coupled to mass spectrometry indicate (Figure 3) that PEY contains a non-negligible number of methanol-soluble (metabolites) and organic solvent-soluble (lipids) differing from egg yolk. Interestingly, lipidomic profiles allow a more thorough differentiation in comparison to metabolomics (Figure 3A,E). Even applying a Benjamini–Hochberg correction for false discovery rate, a total of 357 differential lipids were found (231 increased in PEY and 126 increased in PEY, FDR = 0.05; Supplemental Dataset). A total number of 140 metabolites differentially present in the extract were also found (16 increased in egg yolk and 124 increased in PEY, FDR = 0.05; Supplemental Dataset). Collectively, regarding putative identifications, these lipids clustered among different sphingolipid-related pathways, as well as some immune-related pathways (Figure 3D). Interestingly, differential metabolites were also present in pathways related to omega-3 fatty acids, among many other metabolic nodes (Figure 3H). All differential metabolites and lipids, including exact mass, chromatographic behavior (retention

time in respective systems), composite spectrum, fold-change and *p*-values (both raw and adjusted for false-discovery rate) are available as a Supplemental Dataset in Excel® file in Supplemental materials, found online. Similarly, the individual levels of metabolites and lipids employed for pathway analyses (Figure 3D,H) can be found in an ad-hoc designed web page (https://excelv.herokuapp.com/).

**Figure 2.** PEY contains a significant amount of growth factors. (**A**). Connective tissue growth factor (CTGF) immunoreactivity in egg yolk and PEY, showing a difference between them with a *p* < 0.0001 significance. (**B**). Platelet-derived growth factor (PDGF) immunoreactivity in egg yolk and PEY, showing the difference between them with a *p* < 0.01 significance. (**C**). Nerve growth factor (NGF) immunoreactivity in egg yolk and PEY, showing no significant difference between them. (**D**). Insulin-like growth factor 1 (IGF-1) immunoreactivity in egg yolk and PEY, showing a significant difference between them (\* *p* < 0.05, \*\* *p* < 0.01, \*\*\*\* *p* < 0.0001).

**Figure 3.** *Cont*.

**Figure 3.** PEY exhibits a differential lipidomic and metabolomic profile. Lipidome (**A**–**D**) and metabolome (**F**–**H**) profiles of PEY and egg yolk were analyzed by liquid chromatography coupled to mass spectrometry. As shown by principal component analyses graphs, PEY and egg yolk show different lipidome (**A**) and metabolome (**E**) signatures. This is reinforced by hierarchical clustering analyses which show a perfect clustering in lipidomics (**B**), but not in metabolomics (**F**). Univariate statistics such as a Volcano plot are shown, indicating that there are marked differences in a high number of molecules both in the lipidome (**C**) and metabolome (**G**). Pathway analyses, shown in (**D**) for lipidomics and in H for metabolomics, indicate the pathways where putatively annotated molecules are located—network neighborhood-based entity sets of lipidomic and metabolomics differences between PEY and egg yolk. Differential lipids (**D**) and metabolites (**H**) (see main text) were entered into the ConsensusPathDB platform, and nodes, representing neighborhood-based entity sets (whose size is proportional to the number of metabolites/lipids of the set, and color intensity denote *p*-value for hypergeometric tests) are linked by interactions consisting of the number of metabolites shared by nodes. The type of network chosen was 2-next neighbors, with a minimum number of 1 metabolite overlap with members of the entity set (and a *p* < 0.05 for lipidomics and a *p* < 0.01 for metabolomics as the cutoff). The sets were obtained, considering only Wikipathway based ones. Lipids included were C21480, C21481, C00350, C02737, C13883, C00195, C12126, C00550, C01190, and C02686. Metabolites used were C01179, C06104, C06425, C00239, C00364, C00984, C14214, C02043, C06429, C08491, C00256, C05332, C00410, C05441, and C02477 (KEGG nomenclature). Individual levels of metabolites and lipids employed for pathway analyses are available in https://excelv.herokuapp.com/.

#### **4. Discussion**

It is known that nutrition can affect the functioning of various immune parameters, and the immunomodulation through dietary supplements is not new; e.g., linoleic acid promotes the production of leukotrienes and prostaglandins, or that arginine and glutamine enhance macrophage phagocytosis [21]. Also, epidemiological studies show that both overall diet or specific dietary components like polyphenols can reduce inflammatory cytokines in animal and cell culture models [22]. Then, it was expectable that a dietary component so rich in biomolecules as egg yolk would lead to some effects in the immunity. What it was not predictable, is the fact that PEY would significantly improve these effects.

In our view, PEY optimizes the immune response. This effect means that, more than down-regulating hyperactive immune functions or up-regulating suppressive immune functions, PEY strengthens the resilience of immune functions to respond to external 'stressors' [21]. The results of this work agree with PEY intake induced changes at the systemic level that led to a less aggressive immune response in rodents since a lower induction of pro-inflammatory cytokines was observed after an inflammatory stimulus. PEY, like any biological extract, is a complex matrix containing thousands of molecules with biological activity, in a vast range of concentrations; thus, elucidating its mechanism of action is not an easy task. The individual effects of single compounds can be easily studied in vitro. However, synergisms and antagonisms, as well as metabolic reactions, take place when a food, as a complex matrix, is ingested, digested, and processed by a living being. Accounting for cytokine blood levels, it was clear that some differences in the composition between PEY and egg yolk must exist, and screening these differences could give some clues.

The increased presence of growth factors (CTGF, PDGF, IGF-I and NGF) on PEY, when compared with egg yolk, is not surprising from a biological point of view. Growth factors play a vital role in the evolution and resolution of inflammatory reactions [22]. Thus the presence of these proteins leads to thinking that they could be related to the different immune responses observed in vivo. CTGF is a critical player in connective tissue homeostasis since it helps to maintain extracellular matrix remodeling in normal physiological processes such as wound healing, and it has also been shown to possess apoptotic and nonmitogenic properties [23]. PDGF has a vital role in the early differentiation of hematopoietic/endothelial precursors; it has been used in clinical trials as a topical treatment for healing chronic neuropathy, as well as to improve periodontal regeneration in severe periodontal disease [24]. Insulin-like growth factor I (IGF-I) is a polypeptide hormone secreted by multiple tissues in response to growth hormone (GH). It is partly responsible for GH activity, and also has anabolic effects. NGF is a neurotrophic factor that promotes the growth and survival of peripheral sensory and sympathetic nerve cells. It is a pleiotropic factor, since it produced and utilized by several cell types, including structural (epithelial cells, fibroblasts/myofibroblasts, endothelial cells, smooth muscle cells and hepatocytes), accessory (glial cells, astrocytes and Muller cells) and immune (antigen presenting cells, lymphocytes, granulocytes, mast cells and eosinophils) cells [25], having neuroprotective and tissue repairing properties. Strikingly, since PEY is administered orally, it must be assumed that there exists bioavailability of these molecules. PEY effects could be explained through absorption in the digestive tract by indirect interaction with the host's microbiota. Previous studies showed how the Platelet-Rich Plasma rich in growth factors (PDGF, IGF, VEGF, TGFb) promoted the regenerative processes inhibiting the macrophage activation and the release of cytokines (TNFa, MCP-1, and RANTES) [26] in vitro. However, another study showed that orally administered IGF-I mainly acts at the intestine, a portion of ingested IGF-I is absorbed into the general circulation to enhance the growth of selective organs/tissue [27]. As indicated, interaction with microbiota cannot be excluded—for instance, Padlyia et al. [28] analyzed fertilized eggs, finding that the proteins that increased in abundance play a role in angiogenesis (pleiotrophin, histidine-rich glycoprotein), in defending the developing embryo against microbial pathogens (avian β-defensin 11, polymeric immunoglobulin receptor, serum amyloid P-component, ovostatin and mannose-binding ligand) and in augmenting the structural integrity of the egg shell (ovo-calyxin-32), necessary to provide a substantial barrier against microbial infection.

The action of PEY is not only due to the presence of growth factors. The lipids and metabolites that were found increased in PEY clustered among different sphingolipid-related pathways, as well as some immune-related pathways. A plethora of cell biological processes are critically modulated by bioactive sphingolipids, including growth regulation, cell migration, adhesion, apoptosis, senescence, and inflammatory responses [29]. Similarly, Duan et al. [30] reported that fertilized eggs exhibited higher essential fatty acids (EFAA) and monounsaturated (MUFA) fatty acids levels than unfertilized eggs, and lower cholesterol concentrations, having the potential of being utilized as an EFAA/MUFA-rich, low-cholesterol dietary supplement for the aged and people with special dietary requirements. There are also in vitro experiments demonstrating the pharmacological effects of these bioactive molecules; Xi Li et al. [31] observed that 12-day chicken embryo extracts enhanced spleen lymphocyte proliferation (and IL-2 production), and macrophage function (phagocytosis and NO production). Accounting the high bioavailability of lipids, we cannot discard a role of lipids in PEY in the immune optimization.

Globally, the observed decrease in LPS-induced increase in circulating TNFa, IL1b, and MCP-1 could be related with the complex PEY composition. Growth factors, together with a vibrant matrix of bioactive lipids and metabolites, are promoting a less aggressive immune response, which means better maintenance of homeostasis. Further, these responses can be ascribed to a change in the M1-M2 macrophage status, favoring the resolutive (M2) phase, a thread that will be the focus of future studies.

As for the limitations of our work, we acknowledge that a single nutrient cannot explain observed effects; that the oral bioavailability of the compounds present in PEY can be limited; and that perhaps the system is limited to preclinical studies. Further, we think that measurement of systemic cytokine data induced by PEY treatment, as well as its surrogate metabolome and lipidome changes, the focus of future studies, could offer more light on the potential mechanisms behind this. However, we think that the demonstration that PEY is a complex of bioactive molecules (growth factors, lipids, metabolites) that encloses the molecules that an embryo needs to develop a whole organism affects the host's immunity is clear. It is reasonable to think that these active compounds are bioavailable even if orally administered since a systemic effect has been observed. These active molecules probably act together; this differentiates it from a single active ingredient since possibly this set of molecules produces a series of simultaneous effects that include synergies and antagonisms, promoting an anti-inflammatory microenvironment and leading to a state of homeostasis. Since chronic inflammation contributes to the development of chronic diseases (cancer, cardiovascular disease, and diabetes) and aging, consumption of PEY could effectively reduce the incidence and the progression of these processes.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6643/12/9/2699/s1, Supplemental Dataset, which is an Excel® file containing differential molecules between PEY and egg yolk in the metabolomics and lipidomics analyses, and Supplemental Figure S1. Individual levels of metabolites and lipids employed for pathway analyses are present in https://excelv.herokuapp.com/.

**Author Contributions:** Conceptualization, J.C., C.B. and J.C.E.S.; Methodology, J.C.E.S., M.M.-G., L.S., M.J. and C.B.; Validation, M.M.-G., J.C.E.S., L.S., M.J. and M.P.-O.; Formal analysis, M.M.-G., J.C.E.S., C.B. and J.C.; Investigation, J.C. and M.P.-O.; Resources, J.C.; Data curation, M.M.-G. and J.C.E.S.; Writing—original draft preparation, C.B. and J.C.E.S.; Writing—review and editing, M.P.-O., C.B. and J.C.; Visualization and supervision, J.C.E.S.; Project administration and funding acquisition, J.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** Support from Generalitat de Catalunya, 2017SGR696.

**Conflicts of Interest:** Authors J.C.E.S., M.J., L.S. and M.P.-O. declare no conflict of interest. J.C. is the inventor of the patent protecting PEY obtention. J.C. and C.B. are employees of Ovovity s.l., which is the owner of the patent protecting PEY obtention. The funders (Ovovity s.l.) had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

### **Protocatechuic Acid Extends Survival, Improves Motor Function, Diminishes Gliosis, and Sustains Neuromuscular Junctions in the hSOD1G93A Mouse Model of Amyotrophic Lateral Sclerosis**

**Lilia A. Koza 1,2, Aimee N. Winter 1, Jessica Holsopple 1, Angela N. Baybayon-Grandgeorge 1, Claudia Pena 1,2, Je**ff**rey R. Olson 1,2, Randall C. Mazzarino 1,2, David Patterson 1,2,3 and Daniel A. Linseman 1,2,3,\***


Received: 4 May 2020; Accepted: 15 June 2020; Published: 18 June 2020

**Abstract:** Amyotrophic lateral sclerosis (ALS) is a devastating disorder characterized by motor neuron apoptosis and subsequent skeletal muscle atrophy caused by oxidative and nitrosative stress, mitochondrial dysfunction, and neuroinflammation. Anthocyanins are polyphenolic compounds found in berries that possess neuroprotective and anti-inflammatory properties. Protocatechuic acid (PCA) is a phenolic acid metabolite of the parent anthocyanin, kuromanin, found in blackberries and bilberries. We explored the therapeutic effects of PCA in a transgenic mouse model of ALS that expresses mutant human Cu, Zn-superoxide dismutase 1 with a glycine to alanine substitution at position 93. These mice display skeletal muscle atrophy, hindlimb weakness, and weight loss. Disease onset occurs at approximately 90 days old and end stage is reached at approximately 120 days old. Daily treatment with PCA (100 mg/kg) by oral gavage beginning at disease onset significantly extended survival (121 days old in untreated vs. 133 days old in PCA-treated) and preserved skeletal muscle strength and endurance as assessed by grip strength testing and rotarod performance. Furthermore, PCA reduced astrogliosis and microgliosis in spinal cord, protected spinal motor neurons from apoptosis, and maintained neuromuscular junction integrity in transgenic mice. PCA lengthens survival, lessens the severity of pathological symptoms, and slows disease progression in this mouse model of ALS. Given its significant preclinical therapeutic effects, PCA should be further investigated as a treatment option for patients with ALS.

**Keywords:** amyotrophic lateral sclerosis; anti-inflammatory; antioxidant; phenolic acid; neuroprotective; neurodegeneration

#### **1. Introduction**

Amyotrophic lateral sclerosis (ALS), also known as Lou Gehrig's disease, is a devastating, progressive, and fatal neurodegenerative disease that affects motor neurons of the central nervous system. ALS patients exhibit a median survival of only 2–3 years following diagnosis, with death typically caused by respiratory failure [1]. ALS presents as either a sporadic or familial disease. Sporadic ALS cases account for approximately 90% of all patients and do not have an obvious genetic

cause. Familial ALS accounts for the remaining 10% of patients and has been linked to mutations in genes such as Cu, Zn-superoxide dismutase 1 (SOD1), chromosome 9 open reading frame 72 (C9orf72), fused in sarcoma, and TAR DNA-binding protein 43 (TDP-43) [2]. Although ALS is classified as a rare disease, with a prevalence of 5 in 100,000 people living in the United States, the effects of the disease are calamitous for those who are afflicted [1]. ALS is characterized pathologically by the death of motor neurons, axonal retraction away from the neuromuscular junctions (NMJs), skeletal muscle atrophy, and ultimately, death.

The pathogenesis underlying both familial and sporadic forms of ALS has been extensively studied but is still not completely understood. Protein aggregation, disrupted axonal transport, perturbed RNA metabolism, excitotoxicity, neuroinflammation, mitochondrial dysfunction, and oxidative stress have all been identified as underlying mechanisms and contributing factors in ALS. In the context of motor neuron degeneration, neuroinflammation and oxidative stress appear to be major pathogenic mechanisms. Both astrocytes and microglia can adopt distinct anti- or pro-inflammatory phenotypes, depending on signals from the surrounding environment. These anti- or pro-inflammatory phenotypes are neuroprotective or neurotoxic to motor neurons, respectively. Astrocytes, although beneficial to neurons in their resting state, become reactive and contribute to motor neuron death in various models of ALS [3–6]. Reactive astrocytes have also been shown to impair the process of autophagy in motor neurons, resulting in increased protein aggregation and reduced motor neuron health in in vivo and in vitro models of ALS [7–9]. Microglia also contribute significantly to the neuroinflammation in ALS. Microglia have been shown to induce astrocyte reactivity by releasing pro-inflammatory cytokines, resulting in an inability of astrocytes to protect motor neurons [10,11]. Microglia also become pro-inflammatory and have been found to display an ALS-specific phenotype that contributes to rapid disease progression and increased motor neuron loss [12–15].

Oxidative stress is another mechanism that has been identified as a causative factor in ALS. Increased oxidative stress burden correlates positively with disease severity [16,17]. Superoxide and nitric oxide have been found to be elevated in ALS [18,19]. Oxidative stress markers such as glutathione peroxidase, malondialdehyde, glutathione status, and 8-oxodeoxyguanosine demonstrate significant alterations in ALS patients [20,21]. Furthermore, mitochondrial dysfunction and mutations in genes that affect mitochondrial processes have been linked to ALS [22–26]. Specifically, in the case of ALS caused by mutations in SOD1, the mutant SOD1 protein has been shown to aggregate within mitochondria, resulting in mitochondrial dysfunction and mitochondrial oxidative stress [24,27]. Given the above findings, a beneficial therapeutic approach for ALS may be to reduce both neuroinflammation and oxidative stress.

Currently, only two drugs, Riluzole and Edaravone, have been approved by the U.S. Food and Drug Administration (FDA) to treat ALS. Riluzole is administered orally, only has a modest effect on slowing disease progression, and shows an extension of life of only a few months [2]. Edaravone has been shown to delay disease progression by only 33% when compared to a placebo, is costly, and must be administered intravenously [28]. Unfortunately, both Riluzole and Edaravone have only modest effects on disease progression and survival. Furthermore, both drugs are expensive and display substantial side effects. Many other therapeutics have failed in the clinic or are still undergoing clinical trials. However, no new pharmacological therapies are immediately on the horizon for ALS patients.

Nutraceuticals, natural bioactive compounds found in foods, may be safe, easy to administer, and cost-effective therapeutic treatments for ALS. More specifically, anthocyanins, a type of flavonoid, may be beneficial to ALS patients due to their substantial antioxidant and anti-inflammatory properties [29]. Anthocyanins vary in color from red to blue and are responsible for the vibrant coloring of many fruits and vegetables [30]. We have previously demonstrated a therapeutic effect of an anthocyanin-enriched strawberry extract in a transgenic mouse model of ALS that expresses mutant human SOD1 with a glycine to alanine substitution at position 93 (hSOD1G93A) [31]. This strawberry extract is enriched in callistephin, an anthocyanin derived from pomegranates and strawberries [29,32]. We have also shown that callistephin suppresses apoptosis induced by mitochondrial oxidative stress

and protects neurons from glutamate excitotoxicity in vitro [29,32]. In vivo, we found that hSOD1G93A mice treated with strawberry anthocyanin extract beginning at 60 days old showed delayed disease onset, improved grip strength throughout the disease, and significantly extended survival when compared to untreated hSOD1G93A littermate mice. Furthermore, treated mice displayed significantly decreased astrogliosis in the spinal cord and preserved NMJs in gastrocnemius muscle when compared to untreated littermate mice [31]. These findings indicate that anthocyanin compounds may have therapeutic potential in ALS. However, despite their potential benefits, parent anthocyanins suffer from poor bioavailability. In contrast, their phenolic acid metabolites typically display much higher bioavailability and most can readily cross the blood brain barrier [33].

Here, we examined the therapeutic potential of protocatechuic acid (PCA) in the hSOD1G93A preclinical mouse model of ALS. PCA is a phenolic acid metabolite of kuromanin, the parent anthocyanin found in blackberries, bilberries, and black rice. We have previously shown that kuromanin protects neurons from oxidative stress induced by glutamate excitotoxicity, nitrosative stress induced by nitric oxide, and it also suppresses mitochondrial oxidative stress and the consequent apoptosis by preserving mitochondrial glutathione [29,32]. In a similar manner, PCA protects neurons against nitrosative and oxidative stress and reduces nitric oxide production in microglial cells treated with lipopolysaccharide, demonstrating both antioxidant and anti-inflammatory activities [34]. Based on our previous findings, we tested the therapeutic effects of PCA in the hSOD1G93A mouse model of ALS.

#### **2. Materials and Methods**

#### *2.1. The hSOD1G93A Mouse Model of ALS*

FVB/NJ mice harboring a human transgene coding for a mutated form of SOD1 with a glycine to alanine substitution at position 93 were obtained from The Jackson Laboratory (Bar Harbor, ME, USA). Mice were bred and maintained at the University of Denver animal facility under a standard 12 h light/dark cycle with food and water provided ad libitum. Genotyping to identify transgenic mice was carried out by a third-party company, Transnetyx Inc. (Cordova, TN, USA). All procedures were performed in accordance with two protocols approved by the institutional animal care and use committee at the University of Denver. The initial protocol (927091) was approved on 21 July 2016 and the second protocol (1454889) was approved on 12 July 2019.

#### *2.2. Survival Data*

For the survival study, mice were divided into four groups consisting of 15 mice each. The first group consisted of non-transgenic wild-type (WT) age- and sex-matched littermate controls. The other three groups consisted of age- and sex-matched transgenic hSOD1G93A littermate mice, either untreated or treated with either 50 or 100 mg/kg PCA. PCA was dissolved in sterile deionized water and was administered once per day as a 0.25 mL dose by oral gavage 5 days/week. Oral gavage treatment of PCA began at disease onset (90 days of age) and continued until mice reached end stage, defined as the point at which a mouse no longer had the ability to right itself within 15 s after being placed on its side. PCA-treated and untreated hSOD1G93A littermate mice were euthanized at end stage by an overdose of inhaled isoflurane (Vet One, Boise, ID, USA). The WT littermate control mouse was euthanized at end stage of whichever hSOD1G93A littermate mouse lived the longest, regardless of treatment.

#### *2.3. Paw Grip Endurance and Rotarod Testing*

Hind limb strength was assessed by paw grip endurance (PaGE) testing twice per week beginning at disease onset. Briefly, mice were placed on top of a standard wire cage lid which was suspended a few inches above the bench top. Mice were briefly allowed to acclimate before the cage lid was smoothly inverted to prompt the mouse to grip the wire with both its fore and hind limbs. A stopwatch was started as soon as the cage lid was inverted and the time was measured to determine how long the mouse could hold on before its hind legs released their grip from the cage lid, expressed as latency to fall. Care was taken not to jostle the lid during this time. The stopwatch was stopped at a maximum of 30 s. Mice were given five scored attempts and the highest and lowest scores were excluded from the final score. Final scores are reported as an average of the three remaining scored attempts ± standard error of the mean (SEM) for each time point. Time points correspond to the age of the animal at the time that testing was performed and are reported as a range of several days since multiple sets of animals having slightly different ages were tested concomitantly.

Motor function and endurance were assessed by accelerating rotarod testing once per week beginning at disease onset. Mice were placed on a rod, 30 mm in diameter, rotating at 4 rpm. Each animal was placed in one lane and subjected to three trials. The lane was cleaned before the next mouse was tested to prevent interference. Once the mice were acclimated to the initial speed of 4 rpm, the rod was accelerated from 4 to 40 rpm over the course of 5 min. The time was stopped when the mouse fell off the rotating rod, expressed as latency to fall. Mice were given three scored attempts reported as an average ± SEM for each time point. Time points correspond to the age of the animal at the time that testing was performed and are reported as a range of several days since multiple sets of animals having slightly different ages were tested concomitantly.

#### *2.4. Analysis of Mice at End Stage of the Untreated hSOD1G93A Littermate Mouse*

For assessment of Nissl-stained motor neuron counts, glial fibrillary acidic protein (GFAP) and ionized calcium-binding adapter molecule 1 (Iba-1) staining, NMJ area, perimeter, and Sholl analysis, mice were divided into three groups consisting of approximately 10 mice per group. The first group consisted of non-transgenic WT age- and sex-matched littermate controls. The other two groups consisted of age- and sex-matched transgenic hSOD1G93A littermate mice, either untreated or treated with 100 mg/kg PCA orally as described above. PCA treatment continued until the untreated hSOD1G93A littermate mouse reached end stage. At that point, mice from all 3 groups were euthanized and spinal cord and gastrocnemius muscles were collected.

#### *2.5. Analysis of Mice at 105 Days of Age*

We observed the greatest improvements in motor function as assessed by rotarod and PaGE testing between 97 and 114 days of age in PCA-treated hSOD1G93A mice. Therefore, we analyzed the gastrocnemius muscle wet weight, vesicular acetylcholine transporter (VAChT) and alpha-bungarotoxin (BTx) co-stained gastrocnemius muscle, and 4-hydroxynonenal (4-HNE)-stained spinal cord ventral horn from a separate cohort of mice at 105 days of age. This cohort consisted of both WT and hSOD1G93A mice with or without PCA treatment, with each treatment group containing approximately 10 mice. Mice receiving PCA were given a daily dose of 100 mg/kg beginning at 90 days of age and continuing until the mice reached 105 days of age. At 105 days of age, mice from all 3 groups were euthanized and gastrocnemius muscles and lumbar spinal cord were collected.

#### *2.6. Tissue Preparation and Cryosectioning*

Following euthanasia, the thoracolumbar portion of the spinal column, containing the lumbar spinal cord, and gastrocnemius muscle were removed. Each specimen was washed with 1× phosphate-buffered saline (PBS, pH 7.4), and placed in 4% paraformaldehyde at 4 ◦C overnight. Each specimen was then washed again and allowed to sit in 1× PBS for 20 min. Gastrocnemius muscle was placed in 30% sucrose in 1× PBS and allowed to sink for cryoprotection. The lumbar spinal cord was placed in 6% trichloroacetic acid (Sigma-Aldrich, St. Louis, MO, USA) in deionized water for 6 days for decalcification. Following decalcification, the lumbar spinal cord was placed in 30% sucrose until saturated. Both gastrocnemius muscle and spinal cord were frozen rapidly in optimal cutting temperature (OCT) compound with liquid nitrogen and stored at −80 ◦C until sectioning took place. Prior to sectioning, tissue was allowed to acclimate in the microtome cryostat for at least 20 min. For gastrocnemius muscle and spinal cords, sections of 30 μm in length were cut and

every viable tissue section was collected onto the surface of Fisherbrand Superfrost Colorfrost Plus coated slides (Fisher Scientific, Pittsburgh, PA, USA). Slides were stored at −20 ◦C until subjected to immunohistochemistry.

#### *2.7. Immunohistochemistry of Spinal Cord Sections*

Prior to staining, slides were allowed to equilibrate at room temperature for at least 30 min. Tissue sections on the slides were then outlined with a hydrophobic pen (Liquid Blocker Super PAP Pen; Daido Sangyo Co., Tokyo, Japan) and washed twice with 1× PBS to remove any residual OCT. Tissue was then incubated at room temperature in blocking buffer, containing 5% (*w*/*v*) bovine serum albumin (BSA) and 1× PBS containing 0.2% triton-X 100 for 90 min. For astrocyte staining, primary antibody to GFAP (Abcam, Cambridge, MA, USA) was then prepared as a 1:500 dilution in 1× PBS containing 0.2% triton-X 100 and 2% BSA (*w*/*v*). For microglial staining, primary antibody to Iba-1 (Abcam, Cambridge, MA, USA) was prepared as a 1:167 dilution in PBS containing 0.2% triton-X 100 and 2% BSA (*w*/*v*). For 4-HNE staining, primary antibody to 4-HNE (Alpha Diagnostic Intl. Inc., San Antonio, TX, USA) was prepared as a 1:500 dilution in PBS containing 0.2% triton-X 100 and 2% BSA (*w*/*v*). Tissue was incubated in primary antibody overnight at 4 ◦C. Tissue was washed 3–4 times with 1× PBS to remove any unbound primary antibody. FITC-conjugated donkey anti-rabbit antibody (Jackson Immunoresearch Laboratories, West Grove, PA, USA) or Alexa Fluor 488-conjugated donkey anti-goat antibody (Jackson Immunoresearch Laboratories, West Grove, PA, USA) were then prepared at 1:500 dilutions (*v*/*v*) in 1× PBS containing 0.2% triton-X100 and 2% BSA (*w*/*v*) and Hoechst nuclear stain (Sigma-Aldrich, St. Louis, MO, USA) at a 1:500 dilution to detect GFAP and Iba-1, and to label nuclei, respectively. For 4-HNE staining, FITC-conjugated donkey anti-rabbit antibody (Jackson Immunoresearch Laboratories, West Grove, PA, USA) was prepared at 1:500 dilution (*v*/*v*) in 1× PBS containing 0.2% triton-X100 and 2% BSA (*w*/*v*) and Hoechst nuclear stain (Sigma-Aldrich, St. Louis, MO, USA) at a 1:500 dilution to detect 4-HNE and to label nuclei, respectively. Sections were incubated with secondary antibodies at room temperature for 90 min, then washed 3–4 times with 1× PBS. ProLong Gold anti-fade reagent (Thermo Fisher Scientific, Eugene, OR, USA) was used as mounting medium and slides were sealed with coverslips. Stained slides were stored in the dark at −20 ◦C until imaging took place. Twelve sections of spinal cord were stained for Iba-1, GFAP, and 4-HNE per mouse.

#### *2.8. Imaging and Quantification of Spinal Cord Sections Stained for Iba-1 and GFAP at End Stage of the Untreated hSOD1G93A Littermate Mouse*

Tissue was imaged using a Zeiss Axio Observer epifluorescence microscope to capture a single image of each ventral horn on the 20× objective. Imaging was performed by blinded researchers. For Iba-1 and GFAP, images were captured on the Alexa Fluor 488 channel and the exposure time was set appropriately for the untreated hSOD1G93A littermate and kept constant when imaging the PCA-treated hSOD1G93A and WT littermate mice. At least 6 ventral horns per animal were imaged and analyzed for fluorescence intensity using Adobe Photoshop CC software for both Iba-1 and GFAP staining. For both Iba-1 and GFAP quantification, the ventral horn image of the untreated hSOD1G93A littermate control mouse with the most background was chosen, and the green channel input level was adjusted so that background staining was best eliminated. This value was recorded and used for each subsequent image, including those taken from the WT control littermate and the PCA-treated hSOD1G93A littermate mice such that all images were adjusted by an equivalent amount. After the channel levels were adjusted, the ventral horn was outlined using the lasso tool and green channel values for mean pixel intensity and pixel area were recorded for each ventral horn image. Total pixel intensity of the green channel was obtained by multiplying the pixel intensity and pixel area. An average of the total pixel intensity was taken for each mouse. Quantification was performed twice by two different and blinded researchers. An average of the pixel intensities for each mouse was obtained. For analysis, the average total pixel intensity for the WT littermate control mouse

for each group was set at 100% and the average total pixel intensities for the untreated hSOD1G93A littermate control and PCA-treated hSOD1G93A littermate mice were calculated as percentages relative to the WT littermate control mouse. Mean raw GFAP and Iba-1 fluorescence intensity in the lumbar spinal cord ventral horn in the untreated hSOD1G93A littermate mice and WT littermate mice were also counted and statistically compared to verify a significant disease effect.

#### *2.9. Nissl Staining of Spinal Cord Sections*

Prior to staining, slides were allowed to equilibrate at room temperature for at least 30 min. Slides were then washed twice with 1× PBS to remove any residual OCT. Tissue was then incubated in 70% ethanol and then 95% ethanol for 3 min each. Slides were then incubated in 100% ethanol for 3 min and then 5 min, changing the ethanol between each incubation. Slides were then washed in deionized water 3–4 times. A Cresyl Violet Counterstain Solution (Bioenno Tech, Santa Anna, CA, USA) was applied to each slide for 3 min. De-staining was then performed by washing briefly in 70 mM acetic acid solution. Slides were rinsed in deionized water 3 times. Lastly, slides were incubated sequentially in solutions of 70%, 95%, and 100% ethanol as described above. Slides were mounted with ProLong Gold anti-fade reagent and were sealed with coverslips. Stained slides were stored in the dark at −20 ◦C until imaging took place. Two slides from each mouse spinal cord were stained for Nissl with 6 sections per slide.

#### *2.10. Imaging and Quantification of Nissl-Stained Spinal Cord Sections at End Stage of the Untreated hSOD1G93A Littermate*

A single image of each ventral horn was captured on the 20× objective using bright field by blinded researchers. At least six ventral horns for each mouse were imaged. Any neuron greater than 20 μm in length along its longest axis was considered to be a viable alpha motor neuron. Using Adobe Photoshop CC, the contrast was set to the WT littermate and kept the same within the group to ensure consistency in staining. Next, the ruler tool was used to measure all motor neurons greater than 20 μm. For each mouse, the average number of alpha motor neurons was taken. Quantification was performed twice by two different and blinded researchers. Each average of the untreated hSOD1G93A littermate and 100 mg/kg PCA-treated hSOD1G93A littermate mouse were calculated as a percent of the average of the WT littermate control mouse (set at 100%). The mean numbers of alpha motor neurons in the lumbar spinal cord ventral horn in the untreated hSOD1G93A littermate mice and WT littermate mice were also statistically compared to verify a significant disease effect.

#### *2.11. Alpha-Bungarotoxin and VAChT Staining of Gastrocnemius Sections*

Prior to staining, slides were allowed to equilibrate at room temperature for at least 30 min. Tissue sections on the slides were then outlined with a hydrophobic pen and washed twice with 1× PBS to remove any OCT. Tissue was then incubated at room temperature in blocking buffer, containing 5% (*w*/*v*) BSA and 0.2% triton-X 100 in 1× PBS for 90 min. For analysis of NMJ area, perimeter, and complexity, the slides were then incubated for 90 min with alpha-BTx conjugated to Alexa Fluor® 594 (ThermoFisher Scientific Inc., Rockford, IL, USA) at a 1:200 dilution in blocking buffer containing Hoechst nuclear stain at a dilution of 1:500. For analysis of NMJ innervation, slides were stained with primary antibody to VAChT (C-terminal) (Sigma-Aldrich, St. Louis, MO, USA) which was prepared as a 1:500 dilution in PBS containing 0.2% triton-X 100 and 2% BSA (*w*/*v*). Tissue was incubated in primary antibody overnight at 4 ◦C. FITC-conjugated donkey anti-rabbit antibody (Jackson Immunoresearch Laboratories, West Grove, PA, USA) was then prepared at a 1:500 dilution (*v*/*v*) and alpha-BTx conjugated to Alexa Fluor® 594 (ThermoFisher Scientific Inc., Rockford, IL, USA) was prepared at a 1:200 dilution in 1× PBS containing 0.2% triton-X100 and 2% BSA (*w*/*v*) and Hoechst nuclear stain (Sigma-Aldrich, St. Louis, MO, USA) at a 1:500 dilution to detect VAChT on the presynaptic axon terminal, NMJs, and to label nuclei, respectively. Slides were washed with 1× PBS 3–4 times and then

mounted using ProLong Gold anti-fade reagent and sealed with coverslips. Stained slides were stored in the dark at −20 ◦C until imaging took place.

#### *2.12. Imaging and Area, Perimeter, and Sholl Analysis Quantification of Gastrocnemius Sections Stained with Alpha-Bungarotoxin at End Stage of the Untreated hSOD1G93A Littermate Mouse*

Neuromuscular junctions (20–25) were imaged for each mouse by blinded researchers on the 40<sup>X</sup> objective for all mice that were euthanized at end stage of the untreated hSOD1G93A littermate control mice. Using the magic wand tool in Adobe Photoshop CC, the area was measured, in pixels, for each NMJ. This was achieved by outlining the outside of each NMJ, pressing "Record Measurements", and taking the provided "Area" value. The perimeter, in pixels, of each NMJ was also measured by tracing the NMJ inside and outside, pressing "Record Measurements", and taking the provided "Perimeter" value. An average perimeter and area were calculated for each mouse from all the values recorded for the mouse. Quantification was performed twice by two different and blinded researchers. The untreated hSOD1G93A and 100 mg/kg PCA-treated hSOD1G93A littermate mice averages were calculated as a percent of the WT littermate mouse, which was set at 100%. Mean NMJ pixel area and perimeter in the untreated hSOD1G93A littermate mice and WT littermate mice were also statistically compared to verify a significant disease effect. The Sholl analysis was also performed on the same NMJ images that were used to analyze area and perimeter. Briefly, using ImageJ, the red (BTx) channel was separated from all channels and the NMJ was isolated using the plugins "Despeckle" and "Find Edges". The edges of the NMJ were traced with "Simple Neurite Tracer" and "Analyze Skeleton" plugins. Each NMJ was circled by hand and pasted into a new image. The plugin "Skeletonize" was run to draw a line around the border of the NMJ. The "Sholl Analysis" plugin was run to decide a point in the center of the NMJ and to draw concentric circles around that point at a radius increasing by 10 pixels. Mean Sholl analysis values of NMJ complexity, or the number of intersections formed between the concentric circles and the skeleton of the NMJ, were reported for each untreated hSOD1G93A and 100 mg/kg PCA-treated hSOD1G93A littermate mouse to be taken as a percent of the WT littermate control mouse (set at 100%). Mean Sholl analysis values from untreated hSOD1G93A littermate mice and WT littermate mice were also statistically compared to verify a significant disease effect.

#### *2.13. Analysis of Gastrocnemius Muscle Wet Weight at 105 Days of Age*

Gastrocnemius muscle from all 105-day-old mice were placed into a tared weigh boat on a standard analytical balance to obtain muscle wet weight. Weights of gastrocnemius muscle were taken from WT, 100 mg/kg PCA-treated hSOD1G93A, and untreated hSOD1G93A littermate mice. Each gastrocnemius muscle wet weight from the untreated hSOD1G93A littermate mouse and 100 mg/kg PCA-treated hSOD1G93A littermate mouse was calculated as a percent of the average of the WT littermate control mouse (set at 100%).

#### *2.14. Imaging of Gastrocnemius Sections Stained with Alpha-BTx and VACht at 105 Days of Age*

Neuromuscular junctions (20–25) were captured on the Rhodamine channel for each mouse by blinded researchers on the 20× objective for all mice that were euthanized at 105 days of age. VACht was captured on the Alexa Fluor 488 channel for each NMJ. Representative images are shown from a WT, untreated hSOD1G93A mouse, and a 100 mg/kg PCA-treated hSOD1G93A mouse.

#### *2.15. Imaging of Quantification of 4-HNE-Stained Spinal Cord Sections at 105 Days of Age*

A single image of each ventral horn on the 20× objective was captured for each mouse. Imaging was performed by blinded researchers. Images were captured on the Alexa Fluor 488 channel (pseudo-colored red) and the exposure time was set appropriately for the untreated hSOD1G93A littermate and kept constant when imaging the PCA-treated hSOD1G93A and WT littermate mice. Two littermate groups were analyzed and 6 ventral horns per animal were imaged and analyzed for

fluorescence intensity using Adobe Photoshop CC software. The ventral horn image of the untreated hSOD1G93A littermate control mouse with the most background was chosen, and the red channel input level was adjusted so that background staining was best eliminated. This value was recorded and used for each subsequent image, including those taken from the WT control littermate and the PCA-treated hSOD1G93A littermate mice such that all images were adjusted by an equivalent amount. After the channel levels were adjusted, the ventral horn was outlined using the lasso tool and red channel values for mean pixel intensity and pixel area were recorded for each ventral horn image. Total pixel intensity of the red channel was obtained by multiplying the pixel intensity and pixel area. An average of the total pixel intensity was taken for each mouse. Quantification was performed once by a blinded researcher. An average of the pixel intensities for each mouse was obtained. For analysis, the average total pixel intensity for the WT littermate control mouse for each group was set at 100% and each individual total pixel intensity for each ventral horn for the untreated hSOD1G93A littermate control and PCA-treated hSOD1G93A littermate mice were calculated as percentages relative to the average total pixel intensity of the WT littermate control mouse. Mean raw 4-HNE fluorescence intensity in the lumbar spinal cord ventral horn in the untreated hSOD1G93A littermate mice and WT littermate mice were also counted and statistically compared to verify a significant disease effect.

#### *2.16. Statistical Analysis*

Histological analyses were performed on at least 6 mice per treatment group. Differences between untreated and PCA-treated hSOD1G93A littermate mice for Iba-1, GFAP, and 4-HNE intensity, NMJ area and perimeter, Nissl-stained counts, and gastrocnemius muscle wet weight were analyzed using a paired *t*-test. PaGE and rotarod were analyzed using an unpaired *t*-test at each time point. Correlation analysis of PaGE data and survival were analyzed using a Pearson correlation. Mean NMJ Sholl analysis values were analyzed using a one-way analysis of variance (ANOVA) with post-hoc Tukey's test. Survival Kaplan–Meier curves and body weight data were analyzed using a log-rank test. For all analyses, differences were statistically significant when *p* < 0.05.

#### **3. Results**

#### *3.1. PCA Orally Administered Beginning at Disease Onset Results in a Significant Extension of Survival but Does Not Preserve Body Weight in the hSOD1G93A Mouse Model of ALS*

In order to determine the therapeutic benefit of PCA in an ALS mouse model, we first evaluated the ability of PCA to extend the lifespan of hSOD1G93A mice. Mice were dosed by oral gavage with either 50 or 100 mg/kg PCA beginning at 90 days of age. This time point corresponds with average disease onset. At this age, mice typically display gait disturbances, decreased weight, and lower limb tremors [35]. Mice were dosed by oral gavage with PCA until end stage, assessed by the ability of the mouse to right itself to sternum when placed on its side. Administration of both 50 and 100 mg/kg PCA significantly extended median survival in hSOD1G93A mice to 129 and 133 days, respectively, when compared to untreated hSOD1G93A mice, which exhibited a median survival of 121 days (*p* = 0.0025; Figure 1A). This impressive extension of survival indicates that PCA is slowing the disease progression in hSOD1G93A ALS mice. Body weight of the hSOD1G93A mice was assessed twice per week and is expressed as a percentage of peak body weight at each time point. Despite significantly extended survival, administration of PCA had no significant effect on the decline in body weight in the hSOD1G93A mouse model of ALS (Figure 1B).

**Figure 1.** PCA treatment extends survival in the hSOD1G93A mouse model of ALS. (**A**) Survival of hSOD1G93A mice (untreated or treated with 50 or 100 mg/kg PCA) and wildtype mice (WT). Oral administration of either 50 or 100 mg/kg PCA beginning at disease onset (90 days of age) significantly extended median survival in hSOD1G93A mice to 129 and 133 days, respectively, when compared to untreated hSOD1G93A mice, which exhibited a median survival of 121 days. Curves are significantly different as determined by log-rank (Mantel–Cox) test (*p* = 0.0025; *n* = 15 mice per group). (**B**) Body weight of hSOD1G93A mice (untreated or treated with 100 mg/kg PCA) and WT mice. Body weight was assessed twice per week and is expressed as the percent of peak body weight at each time point. Data are displayed as the mean ± SEM with *n* = 15 mice for each group.

#### *3.2. PCA Treatment Improves Grip Strength and Motor Performance in the hSOD1G93A Mouse Model of ALS*

Since the dose of 100 mg/kg PCA had a more pronounced effect on survival than the 50 mg/kg PCA dose, grip strength and motor function were assessed in mice dosed with 100 mg/kg PCA. hSOD1G93A mice treated with 100 mg/kg PCA were further evaluated using PaGE testing and rotarod testing in order to assess motor function [36]. PaGE testing was performed twice a week beginning at disease onset until end stage of disease to assess grip strength. Administration of 100 mg/kg PCA beginning at disease onset significantly increased the latency to fall as assessed by PaGE testing at 100 to 114 days of

age when compared to the untreated hSOD1G93A littermate controls (*p* < 0.05, Figure 2A). PaGE data were also analyzed for differences between male and female mice. PCA-treated male hSOD1G93A mice did not show a significantly increased latency to fall at any time points when compared to the untreated male hSOD1G93A littermate controls (Figure 2B). However, PCA-treated male hSOD1G93A mice did trend towards a significant increase in latency to fall at 95–99 (*p* = 0.118, Figure 2B) and 100–104 days of age (*p* = 0.115). PCA-treated female hSOD1G93A mice displayed a significantly increased latency to fall as assessed by PaGE testing at 105–109 (*p* < 0.001, Figure 2C) and 110–119 (*p* < 0.05, Figure 2C) days of age when compared to untreated female hSOD1G93A littermate controls. Lastly, we sought to understand the relationship between PaGE testing and survival in PCA-treated hSOD1G93A mice. We averaged latency to fall as assessed by PaGE testing at 100–114 days of age from male and female hSOD1G93A mice treated with 100 mg/kg PCA beginning at disease onset. The latency to fall at these time points was chosen to be averaged because PCA-treated hSOD1G93A displayed an increased latency to fall as assessed by PaGE testing at these time points when compared to untreated hSOD1G93A littermate mice. The average latency to fall over these time points was correlated with the days lived for each mouse. Average latency to fall at 100–114 days and days lived was positively and significantly correlated in PCA-treated hSOD1G93A mice (r = 0.558, *p* < 0.05, Figure 2D).

**Figure 2.** PCA treatment improves grip strength as assessed by PaGE in the hSOD1G93A mouse model of ALS. (**A**) PaGE testing of hSOD1G93A mice (untreated or treated with 100 mg/kg PCA) beginning at disease onset. PaGE testing was performed twice a week beginning at 90 days of age and is expressed as latency to fall. PaGE data are expressed as the mean ± SEM for each time point; *n* = 15 mice per group. \* indicates *p* < 0.05 in comparison to untreated hSOD1G93A littermate controls. All data were analyzed using an unpaired *t*-test at each time point. (**B**) PaGE testing of male hSOD1G93A mice (untreated or treated with 100 mg/kg PCA) beginning at disease onset. PaGE data are expressed as the mean ± SEM for each time point; *n* = 9 mice per group. All data were analyzed using an unpaired *t*-test at each time point. (**C**) PaGE testing of female hSOD1G93A mice (untreated or treated with 100 mg/kg PCA)

beginning at disease onset. PaGE data are expressed as the mean ± SEM for each time point; *n* = 6 mice per group. All data were analyzed using an unpaired *t*-test at each time point. \* indicates *p* < 0.05 in comparison to untreated hSOD1G93A littermate controls. \*\*\* indicates *p* < 0.001 in comparison to untreated hSOD1G93A littermate controls. (**D**) Increased average PaGE latency to fall between 100 and 114 days of age is significantly correlated with longer survival of 100 mg/kg PCA-treated hSOD1G93A mice. PaGE data from 100–114 days of age are expressed as the mean for each mouse; *n* = 14 mice. All data were analyzed using a Pearson correlation (*r* = 0.558, *p* < 0.05).

For further motor function analysis, rotarod testing was performed once a week beginning at disease onset until end stage of disease. This behavioral assay also showed an impressive enhancement in the motor function of PCA-treated hSOD1G93A mice when compared to untreated hSOD1G93A littermates. Administration of 100 mg/kg PCA beginning at disease onset significantly but transiently increased the latency to fall as measured by rotarod testing at 97 and 104 days of age when compared to the untreated hSOD1G93A littermate control mice (*p* < 0.001, Figure 3A). The results of these behavioral tests indicate that oral administration of PCA beginning at disease onset significantly improves balance, grip strength, and motor coordination in the hSOD1G93A mouse model of ALS.

**Figure 3.** PCA treatment improves motor function as assessed by rotarod and also preserves gastrocnemius muscle weight and neuromuscular junction (NMJ) innervation at 105 days of age in the hSOD1G93A mouse model of ALS. (**A**) Rotarod testing of hSOD1G93A mice (untreated or treated with 100 mg/kg PCA beginning at disease onset). Rotarod testing was performed beginning at 90 days of age and extending through end stage and is expressed as latency to fall. Rotarod data are represented as the mean ± SEM for each time point; *n* = 10 mice per group. \*\*\* indicates *p* < 0.001 in

comparison to untreated hSOD1G93A littermate controls. All data were analyzed using an unpaired *t*-test at each time point. (**B**) Quantification of gastrocnemius muscle weights. Data are expressed as a percent of the wildtype (WT) littermate mouse muscle weight and are shown as the mean ± SEM; *n* = 7 mice per group. \* indicates *p* < 0.05 compared to untreated hSOD1G93A control mice (paired *t*-test). Mean gastrocnemius wet weight for the untreated hSOD1G93A mouse (0.0912 <sup>±</sup> 0.0070) is significantly decreased when compared to the WT littermate control mouse (0.1433 ± 0.0055) (*p* = 0.0002; *n* = 7 mice per group). (**C**) Representative images of gastrocnemius muscle from wildtype control mice (WT), untreated hSOD1G93A mice (G93A), and hSOD1G93A mice treated orally with 100 mg/kg PCA beginning at disease onset (G93A+PCA). Mice were euthanized at 105 days of age and gastrocnemius muscles were stained with alpha-BTx (red) and VAChT (green) to label NMJs and innervation of NMJs, respectively. Scale bar = 20 μm. Arrowheads point to NMJs stained positively with alpha-BTx and VAChT.

#### *3.3. PCA Treatment Preserves Gastrocnemius Muscle Wet Weight, Protects NMJ Innervation, and Reduces Oxidative Stress in the hSOD1G93A Mouse Model of ALS at 105 Days of Age*

We observed the greatest improvements in motor function as assessed by rotarod and PaGE testing between 97 and 114 days of age in PCA-treated hSOD1G93A mice. Therefore, we next analyzed gastrocnemius muscle weight at 105 days of age, the time point at which we saw the greatest behavioral therapeutic effect of PCA. Wet muscle weight isolated from untreated hSOD1G93A mice (0.091 g ± 0.007) was significantly decreased when compared to WT littermate mice (0.143 g ± 0.006) (*p* < 0.01, Figure 3B). PCA-treated mice exhibited a significant preservation of gastrocnemius muscle wet weight when compared to untreated hSOD1G93A littermate mice. PCA-treated hSOD1G93A mice had a mean muscle wet weight of 80% of the WT littermate mice, while untreated hSOD1G93A mice had a mean muscle weight of only 63% of WT littermate mice (*p* < 0.05, Figure 3B). These results show that PCA treatment significantly delayed atrophy of the gastrocnemius muscle in the hSOD1G93A mouse model of ALS. To support these findings, we stained gastrocnemius muscle isolated at 105 days of age with alpha-BTx and VACht in order to visualize innervated NMJs. Although these findings are preliminary, they indicate that PCA-treated mice exhibit a protection of NMJ innervation when compared to untreated hSOD1G93A littermate mice, as evidenced by retention of the overlapping staining of alpha-BTx and VACht in the treated mouse NMJs (Figure 3C).

Oxidative stress is an underlying pathology of ALS and contributes to motor neuron death and subsequent skeletal muscle atrophy and deficits in motor function [16,17]. Therefore, we also analyzed the effect of PCA on the production of 4-HNE in the lumbar spinal cord ventral horn of the hSOD1G93A mouse model of ALS. Lumbar spinal cord isolated at 105 days of age from two littermate groups was stained with antibodies against 4-HNE to measure lipid peroxidation. Untreated hSOD1G93A mice in both groups exhibit significantly higher raw 4-HNE fluorescence units (((11.70 <sup>±</sup> 0.78) <sup>×</sup> 106) and ((6.56 <sup>±</sup> 0.88) <sup>×</sup> <sup>10</sup>6))) when compared to their WT littermate control mice (((7.83 <sup>±</sup> 0.16) <sup>×</sup> 106) and ((4.17 <sup>±</sup> 0.24) <sup>×</sup> <sup>10</sup>6))), respectively. These data indicate profound lipid peroxidation in the ventral horn (*p* < 0.01 and *p* < 0.05, Figure 4A and C, respectively). Administration of 100 mg/kg PCA beginning at disease onset significantly reduced lipid peroxidation in the ventral horn of the lumbar spinal cord in the hSOD1G93A mouse model of ALS relative to the untreated littermate control (*p* < 0.01 and *p* < 0.05, Figure 4B and D, respectively).

**Figure 4.** PCA treatment significantly reduces lipid peroxidation in the ventral horn of the spinal cord in the hSOD1G93A mouse model of ALS. (**A**,**C**) Representative images of lumbar spinal cord ventral horns stained for 4-HNE from two littermate groups of wildtype control mice (WT), untreated hSOD1G93A mice (G93A), and hSOD1G93A mice treated orally with 100 mg/kg PCA beginning at disease onset (G93A+PCA). Mice were euthanized at 105 days of age and ventral horns were stained with an antibody to 4-HNE to measure lipid peroxidation. Scale bar = 70 μm. (**B**) Quantification of spinal cord ventral horns stained with 4-HNE as described and shown in A. The 4-HNE fluorescence intensity of untreated and PCA-treated hSOD1G93A littermate mice were normalized and expressed as a percentage of mean 4-HNE fluorescence measured in the WT littermate control mouse. Data are expressed as the mean ± SEM; 6 ventral horns were imaged per mouse. \*\* indicates *p* < 0.01 compared to the untreated hSOD1G93A control littermate (paired *t*-test). Raw mean 4-HNE fluorescence units for the untreated hSOD1G93A mouse (11.70 <sup>±</sup> 0.78) <sup>×</sup> 106) are significantly higher than the WT littermate mouse (7.83 <sup>±</sup> 0.16) <sup>×</sup> <sup>10</sup>6) (*p* < 0.01) (**D**) Quantification of spinal cord ventral horns stained with 4-HNE as described and shown in C. The 4-HNE fluorescence intensity of untreated and PCA-treated hSOD1G93A littermate mice were normalized and expressed as a percentage of mean 4-HNE fluorescence measured in the WT littermate control mouse. Data are expressed as the mean ± SEM; 6 ventral horns were imaged per mouse. \* indicates *p* < 0.05 compared to the untreated hSOD1G93A control littermate (paired *t*-test). Raw mean 4-HNE fluorescence units for the untreated hSOD1G93A mouse (6.56 <sup>±</sup> 0.88) <sup>×</sup> <sup>10</sup>6) are significantly higher than the WT (4.17 <sup>±</sup> 0.24) <sup>×</sup> 106) (*<sup>p</sup>* <sup>&</sup>lt; 0.05).

#### *3.4. PCA Treatment Significantly Preserves Motor Neurons in the Ventral Horn of the Spinal Cord in the hSOD1G93A Mouse Model of ALS*

It was evident that PCA had a beneficial therapeutic effect, as evidenced by the extension of survival and motor function improvements in the hSOD1G93A mouse model of ALS. However, the effects of PCA on inflammation, motor neuron preservation, and neuromuscular junction integrity needed to be studied in order to support the survival and behavioral assay results. Lumbar spinal cord was isolated at end stage of the untreated hSOD1G93A littermate control mouse and Nissl staining was performed so that neuronal cell bodies could be identified. The number of alpha motor neurons, with somas typically greater than 20 μm along the longest axis, were counted in the ventral horn of the lumbar spinal cord of untreated and PCA-treated hSOD1G93A littermate mice. These values were normalized and expressed as a percentage of the number of alpha motor neurons counted in the WT littermate control mouse. The cell bodies of alpha motor neurons are located in the ventral horn of the lumbar spinal cord and their axons project to innervate skeletal muscle fibers of the leg muscles including the gastrocnemius muscle. The hSOD1G93A mouse model of ALS exhibits a rapidly progressive lower limb muscular atrophy and subsequent paralysis as a result of alpha motor neuron death in the ventral horn of the spinal cord and retraction of motor axons away from the NMJs [35]. As anticipated, untreated hSOD1G93A mice exhibited nearly 60% fewer alpha motor neurons in the lumbar spinal cord ventral horn than their healthy WT littermates (Figure 5A). However, when 100 mg/kg PCA was administered beginning at disease onset, the average alpha motor neuron count in the ventral horn of the lumbar spinal cord was significantly increased in comparison to the untreated hSOD1G93A littermate mouse (*p* < 0.05, Figure 5B). These findings may contribute to the observed improvements in motor function as assessed by rotarod and PaGE testing.

**Figure 5.** PCA treatment significantly preserves motor neurons in the ventral horn of the spinal cord in the hSOD1G93A mouse model of ALS. (**A**) Representative images of lumbar spinal cord ventral horns from wildtype control mice (WT), untreated hSOD1G93A control mice (G93A) and hSOD1G93A mice treated orally with 100 mg/kg PCA beginning at disease onset (G93A+PCA). Mice were euthanized at end stage of the untreated hSOD1G93A littermate control mouse and ventral horns were Nissl stained to label neuronal cell bodies. Stained soma were measured along the longest axis and cells were considered alpha motor neurons if the length was greater than 20 μm. Scale bar = 20 μm. (**B**) Quantification of Nissl-stained alpha motor neurons as described in A. The number of alpha motor neurons in the ventral horns of lumbar spinal cord of untreated and PCA-treated hSOD1G93A littermate mice were normalized and expressed as a percentage of the number of alpha motor neurons measured in the WT littermate control mouse. Data are expressed as the mean ± SEM; *n* = 7 mice per group; 4–6 ventral

horns were imaged per mouse. \* indicates *p* < 0.05 compared to untreated hSOD1G93A littermate controls (paired *t*-test). Mean number of alpha motor neurons present in the ventral horn lumbar spinal cord for the untreated hSOD1G93A littermate mouse (1.92 <sup>±</sup> 0.47) is significantly less than that of the WT (5.55 ± 0.57) (*p* = 0.004; *n* = 7 mice per group). Abbreviations used: MNs = motor neurons.

*3.5. PCA Treatment Significantly Reduces Astrogliosis and Microgliosis in the Ventral Horn of the Spinal Cord in the hSOD1G93A Mouse Model of ALS*

A robust neuroinflammatory response in the central nervous system of the hSOD1G93A mouse model contributes to the motor neuron death and muscle atrophy producing the ALS-like phenotype [35]. Therefore, we next analyzed the effect of PCA on astrogliosis and microgliosis in the lumbar spinal cord ventral horn of the hSOD1G93A mouse model of ALS. Lumbar spinal cord isolated at end stage of the untreated hSOD1G93A littermate mouse was stained with antibodies against GFAP and Iba-1 to identify astrocytes and microglia, respectively. Untreated hSOD1G93A mice exhibit significantly higher raw mean GFAP fluorescence units ((3.21 <sup>±</sup> 0.46) <sup>×</sup> 106) when compared to WT littermate control mice ((0.97 <sup>±</sup> 0.16) <sup>×</sup> 106) indicating profound astrogliosis in the ventral horn (*p* = 0.001, Figure 6A). Microgliosis is also present in the lumbar spinal cord ventral horn of untreated hSOD1G93A mice, as these mice exhibited significantly higher raw mean Iba-1 fluorescence units ((3.41 <sup>±</sup> 0.52) <sup>×</sup> 106) when compared to healthy WT littermate control mice ((1.56 <sup>±</sup> 0.29) <sup>×</sup> 106) (*p* < 0.05, Figure 7A). Administration of 100 mg/kg PCA beginning at disease onset significantly reduced astrogliosis in the ventral horn of the lumbar spinal cord in the hSOD1G93A mouse model of ALS from a mean 394% increase to a mean 258% increase in GFAP fluorescence units relative to the WT littermate control (*p* < 0.05, Figure 6B). Administration of PCA also significantly reduced microgliosis in the ventral horn lumbar spinal cord in the hSOD1G93A mouse model of ALS from a mean 248% increase to a mean 156% increase in Iba-1 fluorescence units relative to the WT littermate control (*p* < 0.05, Figure 7B). From these data, we conclude that PCA significantly reduces gliosis which may ultimately protect alpha motor neurons in the ventral horn of the lumbar spinal cord in the hSOD1G93A mouse model of ALS.

**Figure 6.** PCA treatment significantly reduces astrogliosis in the ventral horn of the spinal cord in the hSOD1G93A mouse model of ALS. (**A**) Representative images of lumbar spinal cord ventral horns

stained for GFAP from wildtype control mice (WT), untreated hSOD1G93A mice (G93A), and hSOD1G93A mice treated orally with 100 mg/kg PCA beginning at disease onset (G93A+PCA). Mice were euthanized at end stage of the untreated hSOD1G93A littermate control mouse and ventral horns were stained with an antibody to GFAP to label astrocytes. Scale bar = 20 μm. (**B**) Quantification of spinal cord ventral horns stained with GFAP as described in A. GFAP fluorescence intensity of untreated and PCA-treated hSOD1G93A littermate mice were normalized and expressed as a percentage of GFAP fluorescence measured in the WT littermate control mouse. Data are expressed as the mean ± SEM; n = 10 mice per group; 4–6 ventral horns were imaged per mouse. \* indicates *p* < 0.05 compared to untreated hSOD1G93A controls (paired *t*-test). Raw mean GFAP fluorescence units for the untreated hSOD1G93A mice ((3.21 <sup>±</sup> 0.46) <sup>×</sup> <sup>10</sup>6) are significantly higher than the WT ((0.97 <sup>±</sup> 0.16) <sup>×</sup> <sup>10</sup>6) (*<sup>p</sup>* <sup>=</sup> 0.001; *n* = 10 mice per group).

**Figure 7.** PCA treatment significantly reduces microgliosis in the ventral horn of the spinal cord in the hSOD1G93A mouse model of ALS. (**A**) Representative images of lumbar spinal cord ventral horns stained for Iba-1 from wildtype control mice (WT), untreated hSOD1G93A mice (G93A), and hSOD1G93A mice treated orally with 100 mg/kg PCA beginning at disease onset (G93A+PCA). Mice were euthanized at end stage of the untreated hSOD1G93A littermate mouse and ventral horns were stained with an antibody to Iba-1 to label microglia. Scale bar = 20 μm. (**B**) Quantification of spinal cord ventral horns stained with Iba-1 as described in A. Iba-1 fluorescence intensity of untreated and PCA-treated hSOD1G93A littermate mice were normalized and expressed as a percentage of Iba-1 fluorescence measured in the WT littermate control mouse. Data are expressed as the mean ± SEM; *n* = 6 mice per group; 4–6 ventral horns were imaged per mouse. \* indicates *p* = 0.05 compared to untreated hSOD1G93A controls (paired *t*-test). Raw mean Iba-1 fluorescence units for the untreated hSOD1G93A mouse ((3.41 <sup>±</sup> 0.52) <sup>×</sup> <sup>10</sup>6) are significantly higher than the WT ((1.56 <sup>±</sup> 0.29) <sup>×</sup> 106) (*<sup>p</sup>* <sup>&</sup>lt; 0.05; *<sup>n</sup>* <sup>=</sup> 6 mice per group).

#### *3.6. PCA Treatment Significantly Preserves Neuromuscular Junctions of the Gastrocnemius Muscle in the hSOD1G93A Mouse Model of ALS*

After determining that PCA has neuroprotective and anti-inflammatory effects in the ventral horn of the lumbar spinal cord isolated from hSOD1G93A mice, we sought to analyze NMJs in gastrocnemius muscle in order to further elucidate how administration of PCA improved motor deficits. Neuromuscular junctions represent the synapse between the alpha motor neuron axon terminal and the gastrocnemius muscle fiber. In the hSOD1G93A mouse model of ALS, NMJs become weakened and break down as the alpha motor neuron cell body dies and the axon retracts away from the muscle. This results in skeletal muscle atrophy and paralysis characteristically seen in this mouse model [35]. To analyze the NMJs, gastrocnemius muscle was isolated at end stage of the untreated hSOD1G93A littermate mouse and stained with alpha-bungarotoxin. Alpha-bungarotoxin binds to nicotinic acetylcholine receptors found on the gastrocnemius muscle. Mean NMJ area, perimeter, and Sholl analysis values of untreated and PCA-treated hSOD1G93A littermate mice were normalized and expressed as a percentage of the corresponding NMJ values measured in the WT littermate control mouse. Untreated hSOD1G93A mice exhibited an overall decreased NMJ pixel area (11,842 <sup>±</sup> 1204) when compared to WT littermate control mice (17,966 ± 818) (*p* = 0.01, Figure 8A). Furthermore, untreated hSOD1G93A mice also exhibited an overall decreased NMJ pixel perimeter (731 <sup>±</sup> 40) when compared to the healthy WT littermate control mice (1141 ± 95) (*p* = 0.0007, Figure 8A). Administration of 100 mg/kg PCA beginning at disease onset significantly preserved NMJ area yielding an average NMJ pixel area of 79% of the WT littermate mice compared to the untreated hSOD1G93A littermate mice, which had an average area of 61% of the WT littermate mice (*p* < 0.05, Figure 8B). Furthermore, PCA significantly preserved NMJ perimeter in the hSOD1G93A mouse model of ALS. PCA-treated hSOD1G93A mice exhibited a mean NMJ pixel perimeter of 81% of WT littermate mice versus untreated hSOD1G93A littermate mice, which exhibited a mean of 64% of WT littermate mice (*p* < 0.05, Figure 8C). Untreated hSOD1G93A mice also exhibited an overall decreased NMJ mean Sholl analysis value (62 <sup>±</sup> 1.6) when compared to WT littermate mice (97 ± 4.3) (*p* < 0.001, Figure 8D). More importantly, PCA-treated hSOD1G93A littermates exhibited a significantly greater mean Sholl analysis value (79 <sup>±</sup> 6.8) when compared to untreated hSOD1G93A littermate mice (*p* < 0.05, Figure 8D). These findings indicate that PCA protects the synapse between the alpha motor neuron axon and the gastrocnemius muscle, ultimately allowing for improved motor function and mobility.

**Figure 8.** PCA treatment significantly preserves neuromuscular junctions (NMJ) in the gastrocnemius muscle in the hSOD1G93A mouse model of ALS. (**A**) Representative images of gastrocnemius muscle from wildtype control mice (WT), untreated hSOD1G93A mice (G93A), and hSOD1G93A mice treated orally with 100 mg/kg PCA beginning at disease onset (G93A+PCA). Mice were euthanized at end stage of the untreated hSOD1G93A littermate mouse and gastrocnemius muscles were stained with alpha-BTx (red) and Hoechst (blue) to label NMJs and nuclei, respectively. Scale bar = 40 μm. (**B**) Quantification of gastrocnemius NMJ area stained with alpha bungarotoxin as described in A. The NMJ area, measured in pixels by tracing the outside of the neuromuscular junction, of untreated and PCA-treated hSOD1G93A littermate mice were normalized and expressed as a percentage of NMJ area measured in the WT littermate control mouse. Data are expressed as the mean ± SEM; *n* = 8 mice per group; 20–25 NMJs were imaged per mouse. \* indicates *p* < 0.05 compared to untreated hSOD1G93A control mice (paired *<sup>t</sup>*-test). Mean NMJ pixel area for the untreated hSOD1G93A littermate mouse (11,842 <sup>±</sup> 1204) is significantly less than the WT littermate control mouse (17,966 ± 818) (*p* = 0.01; *n* =8 mice per group). (**C**) Quantification of gastrocnemius NMJ perimeter stained with alpha bungarotoxin as described in A. The NMJ perimeter, measured in pixels by tracing the inside and outside of the NMJ, of untreated and PCA-treated hSOD1G93A littermate mice were normalized and expressed as a percentage of NMJ perimeter measured in the WT littermate control mouse. Data are expressed as the mean ± SEM; *n* = 9 mice per group; 20–25 neuromuscular junctions were imaged per mouse. \*\* indicates *p* < 0.01 compared to untreated hSOD1G93A controls (paired *t*-test). Mean NMJ pixel perimeter for the untreated hSOD1G93A littermate mouse (730.9 <sup>±</sup> 39.37) is significantly less than the WT littermate control mouse (1141 ± 94.52) (*p* = 0.0007; *n* = 9 mice per group). (**D**) Quantification of gastrocnemius NMJ Sholl analysis stained with alpha bungarotoxin as described in A. WT, untreated, and PCA-treated hSOD1G93A littermate mice were given a mean Sholl analysis value, which represents the number of intersections that the NMJ makes with concentric circles every 10 pixels from a center point. Data are expressed as the mean ± SEM; *n* = 8 mice per group; 20–25 neuromuscular junctions were imaged per mouse. \* indicates *p* < 0.05 and \*\*\* indicates *p* < 0.001 compared to WT littermate control mice (one-way ANOVA with post-hoc Tukey's test).

#### **4. Discussion**

Plant-derived polyphenolic compounds exhibit anti-inflammatory, antioxidant, and neuroprotective capabilities; therefore, they have the potential to be safe, cost-effective, and successful agents in treating neurodegenerative diseases such as ALS. The catechol, protocatechuic acid (PCA), is a phenolic acid metabolite of kuromanin, an anthocyanin found in foods such as blackberries, bilberries, and black rice. Here, we demonstrate that PCA extends survival, improves motor function, reduces gliosis, protects motor neurons, and preserves NMJs in the hSOD1G93A mouse model of ALS.

To ensure that PCA was non-toxic and had therapeutic potential, we first studied the effects of PCA on survival. We found that daily administration of 100 mg/kg PCA by oral gavage beginning at disease onset significantly extended survival of hSOD1G93A mice when compared to untreated hSOD1G93A littermate control mice. PCA-treated hSOD1G93A mice also exhibited significantly improved motor function as assessed by rotarod and PaGE testing when compared to untreated hSOD1G93A littermate control mice. To supplement these findings, we also analyzed gastrocnemius muscle weight at 105 days of age, a time point at which we observed significant peak motor performance in PCA-treated hSOD1G93A mice. We found a preservation of muscle wet weight in PCA-treated hSOD1G93A littermate mice when compared to untreated littermate controls at 105 days of age. At this time point, we also observed that PCA-treated hSOD1G93A mice seemed to display a preservation of NMJ innervation in gastrocnemius muscle when compared to untreated hSOD1G93A mice. Indeed, PCA administration preserved motor function and muscle weight well into the disease course, indicating that this compound was able to slow the progression of motor symptoms, which could indicate improved quality of life.

We next sought to determine how PCA was able to elicit a significant improvement in survival and motor function. We analyzed the ventral horn of the lumbar spinal cord at end stage of untreated hSOD1G93A littermate mice and found that PCA-treated hSOD1G93A mice had significantly reduced astrogliosis, microgliosis, and increased motor neuron count when compared to untreated hSOD1G93A littermate control mice. Furthermore, we analyzed NMJs within the gastrocnemius muscle isolated from PCA-treated hSOD1G93A mice at end stage of the untreated hSOD1G93A littermate control mice and measured them in terms of overall area, perimeter, and complexity (via Sholl analysis). We found that oral treatment with PCA significantly preserved NMJ area, perimeter, and complexity when compared to untreated hSOD1G93A littermate control mice. Furthermore, at 105 days of age, we analyzed gastrocnemius muscle stained with alpha-BTx along with antibodies against VAChT. We wanted to visualize the NMJs and their innervation by presynaptic cholinergic neurons. Representative images indicate that administration of PCA was able to protect the innervation of the NMJ at this time point. Although we did not quantitatively assess NMJ innervation, the combined results of preserved NMJ size and complexity at end stage of the untreated hSOD1G93A littermate control mice, along with a preservation of gastrocnemius muscle wet weight and improved motor performance at 105 days of age, indicate that PCA has a beneficial effect on preserving skeletal muscle performance. Taken together, these findings indicate that PCA exhibits neuroprotective properties while also reducing gliosis in vivo and therefore, should be further explored as a therapeutic for ALS.

In future studies, we plan to further examine the mechanism of action of PCA in mitigating the deleterious effects of ALS. We have previously shown that PCA exhibits antioxidant activity due to its catechol structure. PCA has the ability to chelate metal ions, act as a reducing agent, and scavenge free radical species including nitric oxide [29,32]. In the current study, we performed a preliminary analysis and explored the effect of PCA on lipid peroxidation in the ventral horn of lumbar spinal cord isolated at 105 days of age from two groups of WT, untreated hSOD1G93A, and PCA-treated hSOD1G93A littermate mice. Ventral horn was stained with 4-HNE, a byproduct of lipid peroxidation, and fluorescence intensity was measured. These data show that untreated hSOD1G93A mice had significantly higher levels of 4-HNE fluorescence intensity in the ventral horn when compared to their WT littermates, a result that has previously been shown by others [37]. Furthermore, PCA-treated hSOD1G93A mice exhibited significantly lower levels of 4-HNE fluorescence intensity when compared to their untreated hSOD1G93A littermate controls. These data are further supported by previous

research indicating that PCA has the ability to protect cells from mitochondrial dysfunction and apoptosis in vitro and in vivo [38,39]. Furthermore, PCA is able to increase glutathione and superoxide dismutase activity and decrease lipid peroxidation in vitro [40,41]. An increase in free radical species, release of pro-inflammatory cytokines by microglia and astrocytes, and mitochondrial dysfunction all contribute to the oxidative stress burden seen in ALS patients [18,23,24,42]. Consequently, markers such as glutathione peroxidase and malondialdehyde have been found to be elevated in the serum, plasma, and urine of ALS patients [20,21]. Research has demonstrated that oxidative stress heavily contributes to motor neuron death in ALS and PCA has been studied for its antioxidant properties. Therefore, it is possible that PCA is aiding in the preservation of motor neuron viability by reducing the oxidative stress burden through free radical scavenging or by an upregulation of endogenous antioxidant activity. Each of these potential mechanisms should be evaluated in future studies.

Acting in parallel with oxidative stress in the pathogenesis of ALS is neuroinflammation. For example, mutant SOD1 contributes to the death of motor neurons and promotes microgliosis and astrogliosis in the spinal cord. In the hSOD1G93A mouse model of ALS, glial cells, such as astrocytes and microglia, overexpress mutant SOD1. This is toxic to motor neurons and causes an accelerated disease progression. It is theorized that ALS disease onset induced by expression of mutant SOD1 is non-cell autonomous and that glial cells play a central role in motor neuron death [43,44]. Microglial cells expressing mutant SOD1 become activated and release pro-inflammatory cytokines and free radical species [10,45]. Furthermore, mutant SOD1-expressing microglia release increased nitric oxide, superoxide, and decreased insulin-like growth factor-1 when compared to WT microglia in the presence of lipopolysaccharide [12]. Neuroinflammatory microglia also contribute to the activation of astrocytes. Activated astrocytes lose the ability to promote motor neuron survival, phagocytosis, and synaptogenesis [11]. Furthermore, activated astrocytes expressing mutant SOD1 contribute to the death of primary spinal motor neurons [46]. In vitro, PCA has the ability to reduce pro-inflammatory cytokines and nitric oxide production in lipopolysaccharide treated microglial cultures [34,47]. In vivo, PCA treatment reduces cyclooxygenase-2, interleukin-1β, interleukin-6, tumor necrosis factor-α, and prostaglandin E2 expression in inflammatory models in rats and mice [40,47,48]. In the hSOD1G93A mouse model of ALS, we have found that PCA significantly reduces both astrogliosis and microgliosis in the ventral horn of the lumbar spinal cord. While further study is required to determine whether reducing the presence of these cells also reduces the release of pro-inflammatory factors, our data suggest that the reduction in the presence of reactive astrocytes and microglia could be a principal mechanism by which PCA preserves motor neuron survival and overall motor function in hSOD1G93A mice.

In the hSOD1G93A mouse model of ALS, NMJs become weakened and break down as the alpha motor neuron cell body dies and the axon retracts away from the muscle. This results in skeletal muscle atrophy and paralysis characteristically seen in this mouse model. Since PCA-treated hSOD1G93A mice exhibited neuroprotection in the ventral horn of the lumbar spinal cord, we aimed to explore the preservation of the NMJ which represents the synapse between the alpha motor neuron axon terminal and the gastrocnemius muscle fiber. In the hSOD1G93A mouse model of ALS, detachment of nerve terminals from the neuromuscular junction can be seen as early as 10 weeks of age [49]. Although no previous research has explored the effect of PCA treatment on NMJs in mice, we found that PCA was able to preserve the size and complexity of NMJs in hSOD1G93A mice. These findings help further explain the improved motor function measured by rotarod and PaGE testing in PCA-treated hSOD1G93A mice. However, the precise mechanism by which PCA protects the NMJs is currently unknown.

This study is valuable in that it highlights the ability of PCA to significantly reduce neuronal death and gliosis in a mouse model of ALS. However, the therapeutic benefits of PCA are not limited to ALS. Oxidative stress and neuroinflammation contribute to the pathology and subsequent neuronal death observed in many neurodegenerative diseases such as Alzheimer's disease (AD) and Parkinson's disease (PD). PCA has been previously studied in mouse models of AD and was able to improve spatial learning, decrease inflammatory cytokine expression, and increase expression of brain-derived neurotrophic factor in the APP/PS1 mouse model of familial AD [50]. Furthermore, a diet high in date palm fruits, which contain high amounts of phenolic compounds, including PCA, improved spatial learning deficits, and resulted in a reduction in lipid peroxidation and restoration of antioxidant enzymes in the [APPsw]/Tg2576 mouse model of AD [51,52]. In cell models of PD, PCA treatment resulted in a significant upregulation of antioxidant enzymes and inhibited the activation of nuclear factor-κB and expression of inducible nitric oxide synthase [53]. PCA was also able to prevent apoptosis, reduce reactive oxygen species (ROS) production, decrease activation of caspase-3, and enhance SOD activity in in vitro models of PD [54,55]. In vivo, PCA was able to improve motor function and ameliorate PD pathology in the substantia nigra in both MPTP and 6-hydroxydopamine mouse models of PD [53,56]. Interestingly, PCA has been shown in vitro to inhibit aggregation of pathogenic proteins including amyloid beta peptide and alpha-synuclein [57]. Therefore, it will be of interest in future studies to determine whether PCA attenuates aggregation of mutant SOD1 or TDP-43 in models of ALS.

#### **5. Conclusions**

This preclinical study is the first to explore the therapeutic benefits of PCA in a mouse model of ALS. Our findings of the neuroprotective and anti-inflammatory effects of PCA in the hSOD1G93A mouse model of ALS are supported by previous studies showing similar effects of PCA in other models of neurodegeneration. Although PCA has been well studied in other disease models (e.g., AD and PD), it should be further studied in additional models of ALS, such as C9orf72 and TDP-43 ALS, to further elucidate its benefit for treating ALS in a diverse patient population. A thorough analysis of the effect of PCA treatment on levels of biomarkers relating to oxidative stress and neuroinflammation should also be performed. It would be important to measure the effect of PCA treatment on levels of pro-inflammatory cytokines and intracellular levels of ROS (superoxide anion, nitric oxide) in lumbar spinal cord and gastrocnemius muscle in the hSOD1G93A model and other mouse models of ALS, and to analyze these levels over the time course of the disease. Such an analysis would allow for identification of predictive biomarkers for the efficacy of PCA treatment and ALS disease progression. Furthermore, other phenolic acid metabolites of anthocyanin compounds (e.g., 4-hydroxybenzoic acid, gallic acid, and syringic acid) should be studied for similar benefits in ALS and other neurodegenerative diseases. Our findings indicate that nutraceutical phenolic compounds, such as PCA, have the potential to help treat patients with ALS and should be investigated as possible therapeutics for this devastating disorder.

**Author Contributions:** D.P. and D.A.L. conceptualization; L.A.K. and D.A.L. data curation; L.A.K., J.R.O. and D.A.L. formal analysis; D.A.L. funding acquisition; L.A.K., A.N.W., J.H., A.N.B.-G. and C.P. investigation; D.P. and D.A.L. methodology; L.A.K., A.N.W. and D.A.L. project administration; R.C.M., D.P. and D.A.L. resources; L.A.K., A.N.W., R.C.M. and D.A.L. supervision; L.A.K., A.N.W., J.H., A.N.B.-G., C.P. and J.R.O. validation; L.A.K. and D.A.L. visualization; L.A.K. writing—original draft; A.N.W. and D.A.L. writing—review and editing. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received funding from the Ralph L. Smith Foundation.

**Acknowledgments:** We would like to thank the Ralph L. Smith Foundation for the funds and support for this project.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

### **Phytoplankton Supplementation Lowers Muscle Damage and Sustains Performance across Repeated Exercise Bouts in Humans and Improves Antioxidant Capacity in a Mechanistic Animal**

**Matthew Sharp 1,\*, Kazim Sahin 2, Matthew Stefan 1, Cemal Orhan 2, Raad Gheith 1, Dallen Reber 1, Nurhan Sahin 2, Mehmet Tuzcu 2, Ryan Lowery 1, Shane Durkee <sup>3</sup> and Jacob Wilson <sup>1</sup>**


Received: 15 May 2020; Accepted: 1 July 2020; Published: 4 July 2020

**Abstract:** The purpose of this study was to investigate the impact of antioxidant-rich marine phytoplankton supplementation (Oceanix, OCX) on performance and muscle damage following a cross-training event in endurance-trained subjects. Additionally, an animal model was carried out to assess the effects of varying dosages of OCX, with exercise, on intramuscular antioxidant capacity. Methods: In the human trial, endurance-trained subjects (average running distance <sup>=</sup> 29.5 <sup>±</sup> 2.6 miles <sup>×</sup> week<sup>−</sup>1) were randomly divided into placebo (PLA) and OCX (25 mg) conditions for 14 days. The subjects were pre-tested on a one-mile uphill run, maximal isometric strength, countermovement jump (CMJ) and squat jump (SJ) power, and for muscle damage (creatine kinase (CK)). On Day 12, the subjects underwent a strenuous cross-training event. Measures were reassessed on Day 13 and 14 (24 h and 48 h Post event). In the animal model, Wistar rats were divided into four groups (*n* = 7): (i) Control (no exercise and placebo (CON)), (ii) Exercise (E), (iii) Exercise + OCX 1 (Oceanix, 2.55 mg/day, (iv) Exercise + OCX 2 (5.1 mg/day). The rats performed treadmill exercise five days a week for 6 weeks. Intramuscular antioxidant capacity (superoxide dismutase (SOD), catalase (CAT), glutathione peroxidase (GSH-Px)) and muscle damage (CK and myoglobin (MYOB) were collected. The data were analyzed using repeated measures ANOVA and *t*-test for select variables. The alpha value was set at *p* < 0.05. Results: For the human trial, SJ power lowered in PLA relative to OCX at 24 h Post (−15%, *p* < 0.05). Decrements in isometric strength from Pre to 48 h Post were greater in the PLA group (−12%, *p* < 0.05) than in the OCX. Serum CK levels were greater in the PLA compared to the OCX (+14%, *p* < 0.05). For the animal trial, the intramuscular antioxidant capacity was increased in a general dose-dependent manner (E + Oc2 > E + Oc1 > E > CON). Additionally, CK and MYOB were lower in supplemented compared to E alone. Conclusions: Phytoplankton supplementation (Oceanix) sustains performance and lowers muscle damage across repeated exercise bouts. The ingredient appears to operate through an elevating oxidative capacity in skeletal muscle.

**Keywords:** phytoplankton; antioxidants; muscle damage; muscle recovery; muscle soreness

#### **1. Introduction**

Sport performance depends on the ability of an athlete to produce and sustain high levels of physical, technical, decision-making, and psychological skills throughout competition. The deterioration of any of these skills could appear as a symptom of fatigue. The phenomenon of fatigue is complex, with the underlying processes developing as exercise proceeds to ultimately manifest as a decline of performance. The issue of fatigue is compounded when there is an imbalance between recovery and exertion. Recent research has demonstrated that endurance and cross-training athletes partake in multiple high-intensity competition events, which limits the recovery opportunities. The result is extreme stress to the body which is not counterbalanced by proper rest [1]. The lack of rest between intra-competition bouts impairs regeneration and leads to decrements in strength, power, and endurance-based performance [2,3]. Previous research has demonstrated that functional impairments are strongly associated with an increase in oxidative stress [4,5], defined as disturbances in the homeostatic balance where oxidant capacity exceeds the antioxidant capacity, causing the redox state to be more pro-oxidizing [6]. Mechanistically, extreme exercise can elevate oxygen consumption by up to 20-fold over resting levels [7]. The heightened use of oxygen may result in a leakage of reactive oxygen species (ROS) such as superoxide dismutase (SOD). If not properly regulated, these changes can alter cellular structure and function, leading to performance declines [4,5].

While athletes find it challenging to change the demands of competition, they may improve their recovery in response to exercise through the consumption of nutraceuticals that are rich in antioxidants. In fact, it is commonplace for athletes to supplement with antioxidants under the conviction that they will enhance the recovery of muscle function and performance [8,9]. As a result, a fair amount of attention has been given to antioxidant supplements for exercise recovery, which is largely due to their capacity to enhance the endogenous support in diminishing oxidative damage by discarding ROS [10,11]. The majority of studies investigating the impact of antioxidant supplementation on exercise report that antioxidants can reduce oxidative stress [12]; however, the physiological implications of this effect are not well known in more practical competition environments. Furthermore, strong evidence supporting antioxidant supplementation as a protectant against muscle damage is incomplete, as most investigations do not consider oxidative stress makers and intramuscular enzymes in conjunction with the functional indices of muscle damage (e.g., losses in force and power) [12,13].

Dieticians have made a push for natural and sustainable super foods, which contain an array of performance aids [14]. In response, scientists have isolated a unique source of marine phytoplankton, microalga *Tetraselmis chuii* (Oceanix (OCX), Lonza Consumer Health, Morristown, NJ, USA). This microalga contains highly active antioxidant enzymes, particularly superoxide dismutase (SOD), which speeds the reaction that converts superoxide into ordinary molecular oxygen, thereby protecting cells from oxidative damage [15]. The ingredient was also found to upregulate glutathione peroxidase and catalase enzymes in human skeletal muscle myoblasts in vitro [15]. When isolated, antioxidants have been shown to aid in recovery [16–19]. However, the effects on recovery of a marine-derived, SOD-rich ingredient remains to be investigated on high-intensity, cross-training events. Therefore, the purpose of this study was to investigate the targeted marine phytoplankton supplementation on recovery across repeated bouts of activity encompassing endurance, strength, and power and to investigate its effects on intramuscular antioxidant capacity using human and animal models. We hypothesized that marine phytoplankton would better reduce muscle damage while sustaining endurance and strength, which collectively indicates improved recovery. We also hypothesized that the ingredient would improve antioxidant capacity and lower oxidative stress in our mechanistic animal model.

#### **2. Materials and Methods**

#### *2.1. Human Modal*

#### 2.1.1. Subjects

Subjects were recruited by word of mouth, email contact, and direct contact with the local runner's clubs. Subjects were excluded from the study if they: had a body mass index (BMI) <sup>≥</sup>30 kg/m2, had allergies to fish, shellfish, algae, or seaweed; had any cardiovascular, metabolic, or endocrine disease; had undergone surgery that affects digestion and absorption, smoke, drink heavily (>7 and >14 drinks per week for women and men, respectively), were pregnant or planning to be pregnant, were on medication to regulate blood glucose, lipids, and/or blood pressure; had used anabolic–androgenic steroids, were currently using antioxidant supplements, non-steroidal anti-inflammatory drugs, or nutritional supplements known to stimulate recovery or muscle mass gains. A total of 20 subjects volunteered for participation in the study. Eighteen of the enrolled subjects completed the study and two subjects withdrew from the study due to work or family requirements. Prior to engaging in any study procedures, the subjects signed a written informed consent form that was approved by an Institutional Review Board (IntegReview, Austin, TX; Protocol # 1219) for study participation. Subjects in the study were considered active athletes who could withstand multiple endurance bouts (i.e., at least ran 20 miles <sup>×</sup> week−1). The self-reported weekly running mileage ranged from 20–60 miles (mean <sup>±</sup> SD = 29.5 <sup>±</sup> 10.9 miles <sup>×</sup> week−1). The subject characteristics are reported in Table 1 as the mean ± standard error.

**Table 1.** Subject characteristics.


#### 2.1.2. Study Protocol

This study was carried out in a randomized, double-blind, placebo-controlled and parallel manner. Prior to allocation into conditions, the subjects were assessed for maximal oxygen uptake (VO2max) on a graded treadmill test. The subjects were then classified into quartiles according to the VO2max values, and then subjects forming each quartile were randomly assigned to conditions. Following the condition allocation, the subjects underwent baseline testing on day 0 (Pre) which included: salivary cortisol, serum creatine kinase, maximal isometric strength, maximal muscle power, one-mile timed run, and perceptual measures using visual analog scales (VAS). Immediately following pre-testing, the subjects were given their respective condition (Oceanix (OCX) or microcrystalline cellulose-based placebo (PLA)). The OCX ingredient was independently examined by Brunswick Laboratories (Southborough, MA, USA) for oxygen radical absorbance capacity (ORAC) expressed in micromole Trolox equivalency (μmole TE) per gram. The results indicated that the values were high for hydroxyl radicals at 178.71 μmole TE/gram and super oxide anions at 348.11 μmole TE/gram, moderate in peroxynitrite and peroxyl radicals at 8.65 and 29.65 μmole TE/gram, respectively, and not detectable in singlet oxygen and hypochlorite. The ORAC values for the super oxide anion corresponded with high total values (38,000 IU per 100 g) of SOD in the raw powder measured.

The conditions were stored in visually identical capsules and containers. The subjects were required to consume one serving (25 mg) a day, either 30 min prior to exercise or with the first meal of the day on non-exercise days. Supplement compliance was assessed by supplement logs and the collection of supplement containers. The subjects were instructed to refrain from consuming any nutritional supplements for the duration of the study. Additionally, the subjects were informed to follow their habitual routine regarding exercise and diet for days 1–9 of the supplement period; however, the subjects were instructed to refrain from exercise on days 10 and 11 to avoid a potential interference effect from the cross-training bout on day 12. On the twelfth day of supplementation, the subjects returned to the laboratory to complete a supervised cross-training, muscle damage protocol. The subjects continued to supplement for two days after completing the supervised training protocol. Approximately 24 h and 48 h post training (days 13 and 14), the subjects were retested in a manner identical to the Pre to assess the changes in endurance, strength, power, and muscle damage. Study procedures are further described below.

#### 2.1.3. Maximal Oxygen Uptake (VO2max)

Subjects began the assessment at a velocity of 3 mph with a grade of 0%. The velocity was increased by 1 mph at the top of each min, and the rating of perceived exertion (RPE) was recorded using the Borg Scale [20]. Once an RPE of 12 was reached, the velocity increments ceased, and the grade was increased by 1% at the top of each min. This process continued until volitional fatigue was reached or if two of the three following criteria was reached: (1) leveling off (plateau) of oxygen uptake with an increase in work rate [21], (2) respiratory exchange ratio (VCO2/VO2) greater than 1.10 [22], or (3) 90% of the age-predicted maximum heart [23]. The Cardio Coach CO2 metabolic analyzer (Korr Medical Technologies Inc., Salt Lake City, UT, USA) was used for the collection and analysis of respiratory gases. This device has been validated for assessing maximal and submaximal VO2 [24]. The samples of respiratory gases were sampled every 15 s and measured usinga5L mixing chamber technique. A Hans Rudolph one-way valve and silicone face mask was used for the gas collection. A 6-foot breathing tube connected the non-breathing valve to the mixing chamber inlet. Ventilatory oxygen was calculated using modified Haldane equations while the CO2 was measured directly by the analyzer.

#### 2.1.4. Salivary Cortisol (sCT)

Salivary cortisol samples were collected using IPRO Oral Fluid Collector (OFC) Kits (Soma Bioscience; Wallingford, UK). The OFC kits collect 0.5 mL of oral fluid and contain a color-changing volume adequacy indicator within the swab, giving collection times typically in the range of 20–50 s [25]. All samples were collected following a 10 h overnight fast. The samples were analyzed using an IPRO POC Lateral Flow Device (LFD), specific for cortisol, in an IPRO LFD Reader. Two drops of saliva/buffer mix from the OFC were added to the sample window of the LFD. The liquid runs the length of the test strip via lateral flow, creating a control and test line visible in the test window. Ten min after the sample was added, the test line intensity was measured in an IPRO LFD Reader and converted into a quantitative value.

#### 2.1.5. Serum Creatine Kinase (CK)

Venous blood was extracted by the venipuncture of the antecubital vein using a 21-gauge syringe and collected into a 10 mL ethylenediaminetetraacetic acid (EDTA) vacutainer tube (BD Vacutainer, Becton, Dickinson and Company, Franklin Lakes, NJ, USA) by a certified phlebotomist. Afterward, the blood samples were centrifuged at 2500 rpm for 10 min at 4 ◦C. The resulting serum were then aliquoted and stored at −80 ◦C until further analysis. The serum creatine kinase was assayed via commercially available ELISA kits. The samples were thawed once and analyzed in duplicate in the same assay for each analyte to avoid compounded inter-assay variance.

#### 2.1.6. Maximal Isometric Muscle Strength

Each subject was tested for maximal isometric strength using isometric mid-thigh pull (IMTP) performed in an Olympic style half rack to allow the fixation of the bar at any height. Subjects were

secured to the bar using lifting straps and athletic tape. Utilizing a pronated clean grip, the subjects were instructed to assume a body position similar to the second pull of the snatch and clean. Knee angle was confirmed between 125◦ and 135◦ using a hand-held goniometer and the hip angle was approximately set at 175◦. Once the body positioning was stabilized, the subject was given a countdown. Minimal pre-tension was allowed to eliminate slack prior to the initiation of the IMTP. Each subject performed two warm-up reps, one at 50% and one at 75% of the perceived maximum effort. Thereafter, subjects completed 2 maximal IMTPs separated by 2–3 min rest. Subjects were instructed to pull fast and hard and were given strong verbal encouragement during the assessment. Peak isometric force production was recorded using a linear position transducer [26].

#### 2.1.7. Maximal Muscle Power

Maximal muscle power was assessed via a countermovement jump (CMJ) and squat jump (SJ) on a dual force plate platform (Leonardo Mechanograph GRFP XL; Novotec Medical GmbH, Pforzheim, Germany). The platform was composed of two symmetrical force plates that separated the platform into a left and a right half. The resonance frequency of each plate was at 150 Hz. Each plate contained four strain gauge force sensors (the whole platform thus had eight force sensors). The sensors were connected to a laptop computer via a USB 2.0 connection. The signal from the force sensors was sampled at a frequency of 800 Hz and was analyzed using the Leonardo Mechanography GRFP Research Edition software (in this study version 4.2-b05.53-RES was used).

Prior to the test, the subjects completed a warm-up of 10 body weight squats and two submaximal effort CMJs. The subjects were instructed to stand in a comfortable and upright position with the feet about shoulder width apart and parallel to each other. The subjects then performed a countermovement by flexing the hips and knees. Once the subjects reached a preferred countermovement depth, they explosively extended their hip, knee, and ankle joints to perform a maximal vertical jump. Subjects performed 3 hands-free, maximal effort CMJs with 30 s rest between jumps. After completing 3 CMJs, the subjects rested for 2 min and performed 3 hands-free, maximal effort SJs separated by 30 s rest. The SJ started from a similar upright position as the CMJ and subjects were instructed to lower into a squat position (90◦ knee angle). The subjects held their squat position for approximately 3 s until a "jump" command was announced by the researcher. Immediately following the command, the subjects jumped as high as possible from the squat position without reloading or further descent.

#### 2.1.8. One-Mile Timed Run

As a warm-up, the subjects walked on a treadmill for 3 min at 4.8 km/h. Afterward, the subjects ran on the treadmill for 1 min at 9.7 km/h. Immediately following the warm-up, the subjects were allotted a 2 min period to rest or stretch. Thereafter, the treadmill grade and velocity settings were adjusted to a 1% incline and 4.8 km/h, respectively. The subjects maintained this velocity until the distance covered reached 0.05 miles. Upon reaching this distance, the subjects could control the velocity on the treadmill to complete one mile (marked as 1.05 miles) as fast as possible. Control panel placements displaying speed and time were covered to block the subject view. Time was kept by a research technician using a stopwatch. One-mile completion time was recorded to the nearest whole second.

#### 2.1.9. Perceptual Measures

The perceptual measures collected for the study were the perceived recovery status (PRS) scale, rating of perceived soreness (RPS), and the rating of perceived exertion (RPE). PRS was collected in a manner describe by Laurent et al. [27]. The PRS and RPS scales consist of a scalar representation numbering from 0–10. On the PRS scale, the visual descriptors of "very poorly recovered", "adequately recovered" and "very well recovered" for perceived recovery are presented at numbers 0, 5, and 10, respectively. On the RPS scale, the visual descriptors of "no soreness at all", "moderate soreness", and "extreme soreness" are presented at numbers 0, 5, and 10, respectively. The subjects provided PRS at baseline (Pre), immediately before the training protocol (Pre2), and prior to the performance testing at

24 h and 48 h post-workout (24 h Post and 48 h Post) and the RPS was collected at Pre2, 24 h Post, and 48 h Post. The standard 6–20 Borg Scale was used to assess the RPE [20], and the RPE was collected four times: immediately after the 1 mile timed run at baseline (Pre), immediately after the training protocol, and immediately after the 1-mile timed run at 24 h Post and 48 h Post training protocol. Each perceptual measure was recorded as an arbitrary unit (au).

#### 2.1.10. Cross-Training Muscle Damaging Event

Prior to the muscle-damaging session, the subjects completed a dynamic warm-up. Following the warm-up, the subjects completed a cross-training obstacle style program, which was anticipated to induce fatigue and muscle damage. The protocol consisted of 2 separate "blocks". Each block contained 2 exercise stacks, with each stack containing 3 movements (e.g., A1, A2, and A3). The total amount of exercises performed during the training session was 12. Each exercise consisted of 10 reps per set, except for a sled push (45 meters push) and a rowing machine (300 m meters row). Each exercise stack was performed 3 times before moving on to the second stack of each block. Once Block 1 was completed, subjects moved to Block 2. Up to 2 min of rest was allowed between stacks, and then between blocks. However, it was at the discretion of the subject whether they used the full 2 min to rest. Minimal rest was given between exercises within a superset. The subjects were instructed to attempt to finish each stack in as little time as possible with compromising exercise form. An overview is provided in Table 2.



m = meters; cm = centimeters; e = each leg; BW = bodyweight; DB = dumbbell; KB = kettlebell; W = prone shoulder internal/external rotation; T = prone horizontal shoulder abduction; Y = prone shoulder extension.

#### 2.1.11. Statistical Analysis

Prior to carrying out inferential statistics, normality was assessed via Shapiro–Wilk testing and the visual inspection of box blots. All data passed normality testing except for CK, which was logged transformed (y = log(y)) and reported as an arbitrary unit (au). Dependent variables were scrutinized using a two-way mixed analysis of variance (ANOVA) with the condition as the "between-group" factor, time as the "within-group" factor, and the subjects as a random factor. Whenever a significant *F*-value was obtained, post-hoc testing was performed with a Bonferroni correction for multiple comparisons. Additionally, the absolute mean differences (Time2–Time1) were analyzed using a two-tailed, unpaired *t*-test. The alpha level was set at *p* < 0.05 for statistical significance. For significant within- and between group differences, the mean difference (meandiff), 95% confidence interval (95% CI), and *p*-value were reported in text. The data are presented as the mean ± standard error.

#### *2.2. Animal Modal*

#### 2.2.1. Animals and Protocol

Male Wistar albino rats (*n* = 28, 8 weeks old) were provided from the Laboratory Animal Research Center, Firat University (Elazig, Turkey). The animals were kept in a room with standard conditions (22 ± 2 ◦C temperature, 55 ± 5% humidity, a 12 h light–12 h dark cycle). The ethical permission of the experiment was obtained from the Animal Experimentation Ethics Committee of Firat University (2019/139–206) according to the relevant laws, guidelines, and restrictions.

Rats were randomly divided into four groups (*n* = 7): (i) Control (no exercise and placebo), (ii) Exercise (E), (iii) Exercise + Oceanix 1 (2.55 mg/day, (E + Oc1)), and (iv) Exercise + Oceanix 2 (5.1 mg/day (E + Oc2)). Oceanix and placebo (physiological saline) were administered orally via gavage every day before exercise during the experiment period (6 weeks).

The rats were subjected to treadmill exercise on a motorized rodent treadmill (Commat Limited, Ankara, Turkey). The treadmill contained a stimulus grid at the back end of the treadmill giving an electric shock when the animal placed its paw on the grid. The apparatus had a 5-lane animal exerciser utilizing a single belt unit divided with walls suspended over the tread surface. In order to eliminate the diurnal variations, all the exercise tests were applied during the same time of the day. A week of adaptation was provided as pre-training practice for the animals in order to get familiar with the treadmill equipment and handling. In doing so, the rats in the exercise training groups were accustomed to treadmill exercise over a 5 day period such that: (i) first day, 10 m/min, 10 min, (ii) second day, 20 m/min, 10 min, (iii) third day, 25 m/min, 10 min, (iv) fourth day, 25 m/min, 20 min and (v) fifth day, 25 m/min, 30 min. Upon the adaptation of a week to the treadmill system for the novel and stress impacts, the rats in the treadmill exercise groups ran on the treadmill 25 m/min, 45 min/day and five days per week for 6 weeks according to the protocol described by Liu et al. [28]. The exercise model was chosen because it is the most common procedure to carry out animal exercise training from weeks to months at 45 min–1 h/day and 5 days/week. In addition, the exercise model provides adaptations to the cardiovascular system including the physiological remodeling of the heart representative with the increased O2 consumption, the improvement of cardiac contractile function, and calcium handling [29]. The chosen long-term animal exercise model fits an effective program to benefit both healthy and individuals at cardiovascular risk [30].

#### 2.2.2. Biochemical Analysis

Serum samples were obtained by taking blood samples to gel biochemical tubes after centrifugation (5000 rpm at 4 ◦C for 10 min). For the assays, the muscle tissue samples were homogenized within 10 min in 10 volumes of cold Tris 10 mM (pH 7.4). The serum concentrations of creatine kinase (CK) were assayed using a portable automated chemistry analyzer (Samsung LABGEO PT10, Samsung Electronics Co., Suwon, Korea). ELISA was also used in measuring the serum myoglobin (MyBioSource, San Diego, CA, USA). Tissue homogenates (10%, *w*/*v*) were prepared in 10 mM phosphate buffer and centrifuged at 13.000× *g* for 10 min at 4 ◦C. The muscle activities of superoxide dismutase (SOD), catalase (CAT) and glutathione peroxidase (GSH-Px) were determined using the commercially available kits (Cayman Chemical, Ann Arbor, MI, USA) according to the manufacturer's procedure.

#### 2.2.3. Statistical Analysis

Data were presented as the mean ± standard error. All the tests were performed with the SPSS software program (IBM SPSS, Version 21.0; Chicago, IL, USA). Significance was determined by one-way ANOVA. Whenever a significant *F*-value was obtained, the Tukey HSD post-hoc test was applied for comparisons. Statistical significance for the data was defined as *p* < 0.05.

#### **3. Results**

#### *3.1. Human Trial*

#### 3.1.1. Salivary Cortisol and Serum Creatine Kinase

There was no significant between-group difference for any dependent variable at Pre (*p* > 0.05). No significant differences were detected for sCT (*p* > 0.05). A significant group-by-time interaction was demonstrated for CK (*p* < 0.05). The post-hoc analysis indicated that both groups had higher CK levels at 24 h Post (OCX: meandiff = 0.395 au, 95% CI: 0.153 to 0.637 au, *p* < 0.001; PLA: meandiff = 0.759 au, 95% CI: 0.517 to 1.00 au, *p* < 0.0001) and 48 h Post (OCX: meandiff = 0.364 au, 95% CI: 0.122 to 0.606

au, *p* < 0.005; PLA: meandiff = 0.554 au, 95% CI: 0.312 to 0.796 au, *p* < 0.0001). However, 24 h Post levels were significantly higher in the PLA compared to the OCX (meandiff = 0.354 au, 95% CI: 0.026 to 0.715 au, *p* < 0.05; Figure 1).

**Figure 1.** Bar charts displaying the mean and standard error for the log transformation of the plasma creatine kinase (CK) in the Oceanix (OCX) and placebo (PLA) conditions collected prior to the supplement period (Pre), one day, and two days following the supervised cross-training bout (24 h Post and 48 h Post, respectively). **\*** = significantly greater than Pre (*p* < 0.05). ˆ = significantly greater than OCX (*p* < 0.05).

#### 3.1.2. Muscle Strength and Power

The absolute mean difference in IMTP strength from Pre to 48 h Post was significantly lower in the PLA group (meandiff = −9.01 kg, 95% CI: −17.6 to −0.4 kg, *p* < 0.05; Figure 2). A significant main effect of time was detected for the CMJ relative power (*p* < 0.05) in which levels at 24 h Post (meandiff <sup>=</sup> <sup>−</sup>1.63 W kg<sup>−</sup>1, 95% CI: 0.27 to 2.99 W kg<sup>−</sup>1, *<sup>p</sup>* <sup>&</sup>lt; 0.05 and 48 h Post (meandiff <sup>=</sup> <sup>−</sup>2.07 W kg<sup>−</sup>1, 95% CI: 0.71 to 3.43 W kg<sup>−</sup>1, *p* < 0.01) were significantly lower than Pre. A significant group-by-time interaction was demonstrated for the SJ relative power (*p* < 0.05). The post-hoc analysis revealed that the PLA was significantly lower than OCX at 24 h Post (meandiff = <sup>−</sup>5.93 W kg−1, 95% CI: <sup>−</sup>14.32 to <sup>−</sup>0.31 W kg<sup>−</sup>1, *<sup>p</sup>* <sup>&</sup>lt; 0.05; Figure 3).

**Figure 2.** Bar charts displaying the mean and standard error for the absolute mean difference in maximal strength assessed by the isometric mid-thigh pull (IMTP) in the Oceanix (OCX) and placebo (PLA) conditions from Pre to 1 day following the cross-training bout (24 h Pre), Pre to 2 days following the cross-training bout (48 h Pre), and 1 day to 2 days following the cross-training bout (48 h–24 h). \* = significantly lower than OCX (*p* < 0.05).

**Figure 3.** Bar charts displaying the mean and standard error for the relative power output during the squat jump (SJ) in the Oceanix (OCX) and placebo (PLA) conditions collected prior to the supplement period (Pre), one day, and two days following the supervised cross-training bout (24 h Post and 48 h Post, respectively). \* = significantly lower than OCX (*p* < 0.05).

#### 3.1.3. One Mile Timed Run

A significant main effect for time was detected for the one mile timed run (*p* < 0.0001) whereby times at 24 h Post (meandiff = 79 s, 95% CI: 45 to 113 s, *p* < 0.0001) and 48 h post (meandiff = 70 s, 95% CI: 36 to 104 s, *p* < 0.0001) were greater than Pre.

#### 3.1.4. Perceptual Measures

A significant main effect of time was detected for Soreness (*p* < 0.0001). The post-hoc analysis indicated that Soreness levels were elevated at 24 h Post (meandiff = 2.3 au, 95% CI: 0.9 to 3.8 au, *p* < 0.01) and 48 h Post (meandiff = 3.8 au, 95% CI: 2.4 to 5.3, *p* < 0.0001) compared to Pre. Additionally, the levels were elevated at 48 h Post compared to 24 h Post (meandiff = 1.5 au, 95% CI: 0.1 to 3.0 au, *p* < 0.05). A significant main effect of time was detected for the PRS (*p* < 0.0001). The post-hoc analysis revealed that the PRS was lower at 24 h Post (meandiff = −2.3 au, 95% CI: −3.9 to −0.7 au, *p* < 0.01) and 48 h Post (meandiff = −3.5 au, 95% CI: −5.2 to −1.9 au, *p* < 0.0001) compared to Pre. Moreover, 48 h Post was significantly lower than Pre2 (meandiff = −2.7 au, 95% CI: −4.3 to −1.0 au, *p* < 0.001). No significant differences were detected for RPE.

#### *3.2. Animal Modal*

#### 3.2.1. Muscle Damage

For the exercise arm and both supplement arms, the CK was significantly greater than the control (Figure 4a). However, the CK in both supplement arms were lower than in the exercise arm, and E + Oc2 was lower than E + Oc1 (*p* < 0.05). The myoglobin concentration was significantly greater in the exercise and supplement arms compared to the control (Figure 4b). Additionally, both supplement arms were lower than the exercise arm (*p* < 0.05).

**Figure 4.** Bar charts displaying the mean and standard error for serum creatine kinase (**a**) and myoglobin (**b**) following the 6-week motorized rodent treadmill exercise protocol described in Section 3.1. Conditions are: Control (CON), Exercise (E), Exercise + Oceanix 1 (2.55 mg day/rat, (E + Oc1)), and Exercise + Oceanix 2 (5.1 mg day/rat (E + Oc2)). Conditions without a common letter differ significantly (*p* < 0.05).

#### 3.2.2. Intramuscular Antioxidant Enzymes

Serum malonaldehyde (MDA) was lowered by exercise and exercise plus supplementation compared to control, and levels in E + Oc2 was significantly lower than E + Oc1 and E (*p* < 0.05). The exercise and supplement arms demonstrated higher SOD and GSH-Px relative to the control (*p* < 0.05). Furthermore, the supplement arms demonstrated higher levels compared to the exercise arm in a dose-dependent manner (i.e., E + Oc2 > E + Oc1 > E). Catalase was significantly greater in the exercise and supplement arms compared to the control, and both supplement arms were greater than the exercise arm (*p* < 0.05). The raw mean and standard error data for these biochemical measures are provided in Table 3.


**Table 3.** Intramuscular activity of MDA, SOD, CAT, and GSH-Px across conditions.

Data are presented as mean and standard error a–d: Means in the same line without a common superscript differ significantly (*p* < 0.05). MDA = malonaldehyde; SOD = superoxide dismutase; CAT = catalase; GSH-Px = glutathione peroxidase.

#### **4. Discussion**

The purpose of this study was to investigate targeted marine phytoplankton supplementation (Oceanix, OCX) on recovery across repeated bouts of activity encompassing endurance, strength, and power, and to investigate its effects on intramuscular antioxidant capacity using human and animal models. We hypothesized that OCX would better reduce muscle damage and sustain endurance and strength, which collectively indicates improved recovery. We also hypothesized that the ingredient would improve the antioxidant capacity and lower oxidative stress in our mechanistic animal model. The primary findings of this study supported four out of our five major hypotheses. Specifically, OCX was able to lower muscle damage, sustain power and prevent declines in strength across repeated endurance and cross-training bouts. Mechanistically, the ingredient appears to operate through elevating oxidative capacity in skeletal muscle, which led to decreased oxidative stress and muscle damage.

Excessive exercise during competitions may contribute to disturbances in biological function through a physiological deregulation that leads to impairments in performance and recovery [31]. While athletes find it challenging to change the demands of a competition, they may alter their responses through the consumption of supplements that are rich in ingredients capable of influencing recovery. Previous investigations have taken an isolated approach to supplement research by determining their effects on recovery from either aerobic or anaerobic activities. However, many sports (e.g., soccer, basketball, and hockey) require individuals to be exposed to concurrent training stimuli [32]. Moreover, the emergence of CrossFit as a sport has driven athletes to excel across multiple physiological domains [33]. The uniqueness of the present study was that subjects participated in two endurance uphill runs, one cross-training event, and six maximal strength and power bouts (three maximal attempts on two occasions). We feel that this protocol allows researchers to truly determine the ecological validity of the present nutraceutical-based intervention.

Strength and power are two of the most critical attributes underlying success in competitions [32,34]. These variables are intimately related and allow athletes to be successful in their respective sport [35]. Moreover, the ability to sustain repeated outputs over distances is commonly referred to as endurance [32]. In the short term, these opposite spectrum attributes appear to compete with one another [32]. For example, just 25 min of endurance exercise ranging from 40 to 100% VO2 max acutely decreased power and strength by 19–36% [36]. From another perspective, resistance training acutely impairs endurance in a volume and intensity-dependent manner [37]. The present study found that repeated concurrent bouts of strength, power, and endurance decreased the majority of performance metrics examined in this study and resulted in noticeable muscle damage.

When observing CMJ power, it was found that performance dropped equally in both groups. Intriguingly, however, SJ power declined more in the PLA compared to the OCX group (−15% at 24 h Post). These differences were paralleled by greater increases in muscle damage in the PLA (+14% at 24 h Post) compared to the OCX. The ability of OCX to improve power in the SJ but not the CMJ may be due to the differences in mechanics between the two exercises [38]. Specifically, athletes may be able to better use passive elastic components of connective tissue and tendons to overcome muscle damage in a CMJ relative to a SJ, which removes much of the ability to store and release elastic energy [39]. Intriguingly, the ability to use elastic energy increases as the stretch-shorten cycle time decreases [39]. Correspondingly, the high intensity run in our study would demonstrate the shortest stretch shortening cycle in all the measures examined. Thus, the lack of differences in endurance between conditions despite greater muscle damage in the PLA may be explained by similar reasons to the CMJ. In support, we found that decrements in our isometric measure of strength from Pre to 48 h Post were greater in the PLA group (−12%) than OCX.

Since improving recovery through lowering muscle damage is of great interest to athletes, it is worth exploring how OCX may influence these outcomes. To begin to answer this question, it is important to understand the contribution of reactive oxygen species (ROS) to exercise-induced muscle damage. This topic has been addressed in impactful reviews [40]. Briefly, exercise elevates metabolism, and the use of oxygen is heightened [41]. The result is a leakage of ROS from the mitochondria [41]. Superoxide is a major ROS produced as a by-product of oxygen metabolism and, if not regulated, causes many types of cell damage [42]. As such, ROS alters cell structure and function, and contributes to muscle damage causing performance declines [42].

The ingredient administered in this study was derived from the microalga *Tetraselmis chuii*. This microalga contains highly active antioxidant enzymes, particularly superoxide dismutase (SOD), which speeds the reaction that converts superoxide into ordinary molecular oxygen, thereby protecting cells from oxidative damage [15]. The ingredient was also found to upregulate glutathione peroxidase and catalase enzymes in human skeletal muscle myoblasts in vitro [15]. Research has shown that antioxidant supplementation can protect against exercise-induced muscle damage [43,44] and as a result, reduce muscle performance loss and fatigue [45,46]. Our human and animal trials demonstrated that exercise increased muscle damage (e.g., elevated CK and myoglobin levels). Moreover, in both studies, muscle damage was lowered by supplementation. An independent analysis of the present ingredient demonstrated high antioxidant capacity via high concentrations of SOD (38,000 IU per 100 g)

with robust corresponding ORAC values for the enzymes' corresponding anion. These properties led us to explore multiple intramuscular antioxidant enzymes in exercise alone and exercise with OCX supplementation. We found that the ingredient improved measures of antioxidant capacity in a generally dose-dependent fashion.

The results of our study agreed with the past literature demonstrating that exercise alone increases antioxidant capacity, likely as an adaptation to increased oxidative stress [47]. Our animal study also found that exercise combined with the administration of the SOD rich ingredient further decreased oxidative stress. Our findings agreed with a number of antioxidant supplement studies, which demonstrated decreased oxidative stress. These studies include positive effects with vitamin E [48–50], vitamin C [10,51], polyphenolic compounds [52], β-carotene [53], and different antioxidant combinations [54]. These changes occurred as result of elevations in antioxidant capacity as indicated by greater levels of the three intramuscular enzymes measured. The increased capacity found agreed with the past research demonstrating that antioxidant supplementation can increase antioxidant enzyme activity both at rest [55] and in combination with exercise [51,54].

While the mechanism of action still needs exploring, previous in vitro research with the ingredient in human skeletal muscle myoblasts observed that the changes in the proteins studied in our research occurred via the transcriptional responses of genes encoding the antioxidant enzymes and the further regulation of the polypeptide translation downstream of these events [15]. Specifically, a parallel and positive response in enzyme activities and transcripts for SOD, CAT and GSH-Px encoding genes in myoblasts, as a consequence of microalga *Tetraselmis chuii* treatment, was found [15]. This previous finding appears to unravel the potential molecular basis of the cytoprotective effect of the studied ingredient in relation to the primary antioxidant enzymes investigated.

#### **5. Conclusions**

We can conclude that the training program was effective at inducing high levels of neuromuscular fatigue. However, marine phytoplankton supplementation (Oceanix, OCX) was able to improve recovery, sustain power, and prevent declines in strength across repeated endurance and cross-training bouts. Since recovery can be defined as, "returning what was lost due to exercise," [56], we demonstrated that OCX supplementation can improve recovery from intense competitions. Moreover, we present for the first time mechanistic data, which support the OCX role in improving intramuscular antioxidant capacity when combined with exercise.

These findings may allow practitioners to better recommend supplementation for athletes competing under high stress scenarios. While no person should be expected to maximally display every skill of athleticism on a daily basis, OCX supplementation may assist in improving recovery during demanding training or competition periods.

**Author Contributions:** Conceptualization, M.S., K.S., R.Y., S.D., and J.W.; data curation, M.S., M.S., C.O., R.G., D.R., N.S., and M.T.; formal analysis, M.S., C.O., N.S., M.T., J.W.; investigation, M.S., M.S., C.O., R.G., D.R., N.S., M.T.; methodology, M.S., K.S., M.S., C.O., R.G., D.R., N.S., M.T.; supervision, M.S. and J.W.; writing—original draft, M.S., K.S., J.W.; writing—review and editing, M.S., K.S., R.L., J.W. All authors have read and agreed to the published version of the manuscript.

**Funding:** All projects reported in this manuscript was supported by Lonza Consumer Health Inc. (Morrison, NJ, USA).

**Conflicts of Interest:** S.D. is an employee of Lonza Consumer Health Inc. Other authors have no other relevant affiliations or competing financial involvement with the subject matter or materials in this manuscript.

#### **Abbreviations**


#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **The E**ff**ect of Resveratrol or Curcumin on Head and Neck Cancer Cells Sensitivity to the Cytotoxic E**ff**ects of Cisplatin**

**Marinela Bostan 1,2,**†**, Georgiana Gabriela Petrică-Matei 3,**†**, Nicoleta Radu 4,5, Razvan Hainarosie 6, Cristian Dragos Stefanescu 6, Carmen Cristina Diaconu 7,\* and Viviana Roman 1,\***


Received: 31 July 2020; Accepted: 21 August 2020; Published: 26 August 2020

**Abstract:** Natural compounds can modulate all three major phases of carcinogenesis. The role of the natural compounds such as resveratrol (RSV) and curcumin (CRM) in modulation of anticancer potential of platinum-based drugs (CisPt) is still a topic of considerable debate. In order to enhance head and neck cancer (HNSCC) cells' sensitivity to the cytotoxic effects of CisPt combined treatments with RSV or CRM were used. The study aim was to evaluate how the RSV or CRM associated to CisPt treatment modulated some cellular processes such as proliferation, P21 gene expression, apoptotic process, and cell cycle development in HNSCC tumor cell line (PE/CA-PJ49) compared to a normal cell line (HUVEC). The results showed that RSV or CRM treatment affected the viability of tumor cells more than normal cells. These natural compounds act against proliferation and sustain the effects of cisplatin by cell cycle arrest, induction of apoptosis and amplification of P21 expression in tumor cells. In conclusion, using RSV or CRM as adjuvants in CisPt therapy might have a beneficial effect by supporting the effects induced by CisPt.

**Keywords:** resveratrol; curcumin; cisplatin; head and neck cancer; cell cycle; apoptosis

#### **1. Introduction**

The heterogeneous nature, at the molecular level, of many types of cancer has hindered both the identification of specific targets and development of efficient therapeutic strategies against tumours. The most common types of treatment for cancer include surgery, radiation and chemotherapy, which can be used either alone or in combination. But these approaches are associated with significant morbidity and a dramatic reduction in quality of life. Head and neck squamous cell carcinomas (HNSCCs) affect different regions of the head and neck, including the tongue, pharynx, larynx, the nasal

cavity and salivary glands. The HNSCC is one of the most aggressive cancers, with a complex etiology and pathophysiology [1,2] and these aspects make finding an optimal treatment strategy relatively difficult. Moreover, regardless of the specificity and efficiency of monotherapy, it is difficult to obtain optimal cytotoxic effects on cancer cells because of their molecular adaptability [3]. The poor prognosis of HNSC is due to the occurrence of recurrence or metastasis in many patients after radiation or chemotherapy. Cisplatin (*cis*-diamminedichloroplatinum; CisPt) is the most effective and widely spread chemotherapy choice of treatment available for patients with recurrent and metastatic HNSCC [4]. Unfortunately, there is a high rate of clinical failure of CisPt due to intrinsic or acquired resistance which may lead to therapy discontinuation [5,6]. Tumor cells' CisPt resistance could occur by reduction of intracellular CisPt accumulation, changes in DNA repair or in the apoptotic cell death pathways [7,8].

In the past few decades, efforts to improve efficacy of cancer treatment have been largely without success, highlighting the need for finding new solutions such as complementary treatment approaches in order to improve the response to chemotherapy [9]. Many natural herbal compounds have attracted the attention of both researchers and clinicians for their use in preventing or improving the current treatment of various chronic diseases, including different types of cancer [10–13]. Natural compounds are able to interact with multiple cellular targets associated with tumor growth, drug resistance, and metastasis, simultaneously making them a potential source of synergistic cancer treatments. It is possible that, through their multi-targeting activity, natural compounds have the potential to enhance the efficacy of current cancer treatments or to reduce the treatment resistance. The main aim of cancer treatment is to remove or destroy the tumor cells without killing normal cells. Most of the natural compounds have less toxicity, a low cost, are associated with limited side effects and many studies described their effects on the process of carcinogenesis. Due to side effects and drug resistance appearance during conventional therapy, it is becoming obvious that natural compounds have potential to demonstrate anticancer activities or can be used as adjuvants in chemotherapy [14–16].

Curcumin (CRM) is a natural compound extracted from the rhizome of turmeric (*Curcuma longa* L.) with reported antiproliferative, antitumoral, antioxidant, anti-inflammatory and chemo-preventive properties and no apparent side effects. In some clinical trials [17–19] curcumin use showed low toxicity and good tolerability. CRM exerts anti-carcinogenic activity against a wide variety of human cancers by regulation of various signaling pathways involved in tumorigenesis, gene expression, cell cycle regulation and apoptosis. Curcumin can influence the expression of various protein kinases, transcription factors, inflammatory cytokines and other oncogenic proteins [20–23].

Resveratrol (3,5,4 -trihydroxystilbene,RSV) is a phytoalexin produced by a wide variety of plants, such as grapes, peanuts and mulberries. This natural compound is one of the most studied componds for its anti-cancer properties besides its other biological properties such as anti-diabetic, anti-platelet, cardioprotective, neuroprotective, anti-aging, antioxidant and anti-inflammatory activity [24–26]. Resveratrol appears to be an important player in the fight against cancer, as it may influence the mechanisms responsible for inducing the suppression of tumor cell proliferation, as well as the mechanisms involved in sensitization to chemotherapy [27–29]. Rigorous control of cell proliferation and differentiation are necessary to ensure the normal growth and development. Any disorder of the cell division pathways leads to the amplification of the cell division process, the formation of tumors and the appearance of the carcinogenesis process. The carcinogenesis of HNSCC is characterized by multiple events such as activation or suppression of tumour suppressor genes, cell cycle phases disruption, increasing of cell proliferation associated with the decreasing of the apoptotic process [30]. Tumor cells are able to bypass the control point of cell cycle in G1, do not respond to internal regulation and continue to proliferate. It is possible that there are changes in the other phases of the cell cycle, which could be responsible for generating an exaggerated cell proliferation. The balance between cell growth and cell death is influenced by the various molecule regulators like cyclins, cyclin dependent kinases, oncogenes and tumour suppressor genes [31]. One of gene known as a key regulator of the cell cycle as well as cell death and DNA repair is P21 (WAF1/CIP1) a tumor suppressor gene located on chromosome 6 [32,33]. P21 is a cyclin-dependent kinase inhibitor, which is active in response to

cellular and environmental signals to develop tumor suppressor activity. In addition, P21 may act as a key mediator of cell cycle arrest after DNA damage [34]. Many studies suggest that P21 gene by direct association with the promoter region of individual genes or by binding to specific transcription factors/coactivators, contribute to modulation of their activity [35,36]. P21 can exert either positive or negative activities toward a specific cellular response in a context-dependent manner, including the cell type and the source of stress signals. Although abnormal expression of P21 gene has been found in various types of malignancy, current views on the role of P21 as a tumor suppressor or tumor-promoting protein remain ambiguous [37–41]. Our study aimed to define the role of P21 on cell control of the cell cycle progression, programed cell death and response to cisplatin in tumor line PE/CA-PJ49 comparatively with normal cell line HUVEC. Despite invasive treatment protocols that comprise surgical resection of the tumor, radiotherapy, chemotherapy and often in combination, the 5-years survival rate of HNSCC patients remain around 40–50% [42]. New therapy approaches are awaited to reduce toxicities, improve survival rates, and quality of life. Natural compounds could be used as adjuvants in HNSCC therapy, due to their good tolerability and low toxicity, as well as their acceptance as dietary supplements [43]. Moreover, numerous studies have displayed the potential utility of natural compounds against HNSCC [44,45]. Currently, there is a great concern about finding natural compounds to support the effects of conventional therapy used in the treatment of HNSCC. The results of this study will provide additional information about P21 gene or protein expression in response to cisplatin mediated by natural compounds (CRM or RSV). Extensive knowledge regarding the molecular mechanisms of natural compounds induced apoptosis, cell cycle regulation and influence on cisplatin response is indispensable for the development of improved therapeutic strategies. In this study we have analyzed the influence of CRM or RSV treatment on protein (p21) or gene expression in HNSCC line (PE/CA-PJ49) treated or not with CisPt, compared to a normal cell line (HUVEC). In addition, we tried to analyze how the CRM or RSV treatment influences the apoptotic processes and the tumor cells arrest in different phases of the cell cycle and the proliferation activity of tumor cells in response to cisplatin therapy. Data analysis was performed in order to investigate possible correlation between the expression level of p21 and the apoptotic process or cell cycle progression in PE/CA-PJ49 tumor cells treated with CisPt in the presence or absence of CRM or RSV compared to the normal HUVEC cell line. The results of this study suggest the use of natural compounds CRM or RSV, as adjuvants in order to improve the response to CisPt therapy.

#### **2. Materials**

Dulbecco's modified Eagle's medium (DMEM), fetal bovine serum (FBS), glutamine, cisplatin (*cis*-diammineplatinum (II) dichloride, DDP), resveratrol (3,5,4 -trihydroxystilbene), curcumin, 0.9% sodium chloride (NaCl), dimethyl sulfoxide (DMSO), trypsin-EDTA (0.25% trypsin, 0.03% etylenediaminetetraacetic acid), phosphate buffer solution (PBS) and 3-(4,5 dimethylthiazol-2-yl)-2,5 di-phenyltetrazolium bromide (MTT), propidium iodide (PI) stock solution (4 mg/mL PI in PBS); and RNase A stock solution (10 mg/mL RNase A) were purchased from Sigma-Aldrich (St. Louis, MO, USA). CellTiter 96® AQueous One Solution Cell Proliferation Assay (G3580) was from Promega (Madison, WI, USA). RIPA buffer was obtained from Pierce (Rockford, lL, USA), protease inhibitor cocktail and phosphatase inhibitor were from Roche (Rockford, lL, USA); BCA protein assay kit was from Thermo Scientific (Waltham, MA, USA); DuoSet® IC ELISA Intracellular Human Total p21/CIP1/CDKN1A (DYC1047-2) and stop solution–2 N H2SO4 were purchased from R&D Systems Inc. (Minneapolis, MI, USA); the Annexin V-FITC kit was from BD Pharmingen (San Jose, CA, USA); TRIzol reagent was from Invitrogen (Carlsbad, CA, USA); 96-well plates, micropipettes, special tips, Eppendorf tubes; High-capacity cDNA Reverse Transcription Kits (Applied Biosystems, Foster City, CA, USA) contained the following components: RT-Buffer 10×, 1 mL; RT-10× random primers, 1 mL; 25× dNTP mix (100 mM); MultiScribe ™ Reverse Transcriptase, 50 U/μL; RNase inhibitor 100 uL; ultrapure water; TagMan validate Hs00355782\_m1 gene CDKN1A and Hs02800695\_m1 gene HPRT1; QPCR master mix; plates (MicroAmp Fast Optical 96-Well Reaction Plate, 0.1 mL).

#### **3. Methods**

#### *3.1. Cell Lines and Treatment*

The cell lines were obtained from European Collection of Authenticated Cell Cultures (ECACC, Salisbury, UK). The squamous carcinoma cell line PE/CA-PJ49 (ECACC cat. no. 0060606) was obtained from a 57-year old male patient with tongue carcinoma. The PE/CA-PJ49 cell line was grown and maintained in DMEM supplemented with 10% FBS, 2 mM glutamine, at 37 ◦C in 5% CO2.

Human Umbilical Vein Endothelial Cells line (HUVEC, ECACC cat. no.06090720) was maintained at 37 ◦C and 5% CO2 in complete endothelial cell growth medium (ECACC cat. no.06091509). The sub-confluent cultures of both cell lines (70–80%) were split 1:4–1:8 (i.e., seeding at 1–3 × 10,000 cells/cm2) using trypsin-EDTA (0.25% trypsin, 0.03% EDTA) according to the manufacturer's instructions [46].

#### 3.1.1. Preparation of Stock Solution

CRM and RSV were dissolved in DMSO to make a stock solution of 10 mM. Cisplatin was prepared as a 10 mM stock in 0.9% sodium chloride (NaCl). The solutions stock were filtered through a 0.22 μm membrane, aliquoted and stored at −20 ◦C until further use. The cisplatin, curcumin or resveratrol stock was used to obtain the working concentrations related to the treatment by performing dilutions in the culture medium [47].

#### 3.1.2. Cell Cultures Treatment

Cells were cultured in specific cell line medium supplemented with 10% inactivated FBS, 2 mM L-glutamine and incubated at 37 ◦C and in 5% CO2 humid atmosphere for 24 h to achieve around 60–70% confluence, and then treated with various concentrations of CisPt, CRM or RSV for different periods of time (6, 24, 48 h). After treatments, adherent cells were detached from flasks with a trypsin-EDTA solution, washed twice in PBS. PE/CA=PJ49 and HUVEC cells were used for the evaluation of cytotoxicity capacity, apoptosis, distribution of cell cycle phases or for storing as cell pellets at −80 ◦C for preparation of cell lysates for further use in ELISA or PCR assays. In all experiments described in this study, all untreated cells were designated as control cells. For treated cells, the conditions (dose and treatment duration) and reagents that were used together at specific dose, were indicated in the figure legends. Non-treated cells were used as controls throughout the experiments.

#### *3.2. Drug Sensitivity Assay (MTT)*

The MTT assay is based on the ability of viable cells to reduce the reagent 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide to colored formazan compounds [48]. Because the transformation is only possible in viable cells, the amount of blue formazan is proportional to the number of viable cells and thus, a linear dependence exists between cell activity and absorbance. Cells were seeded in triplicate in 96-well plates at a density of 5000 cells/well, incubated at 37 ◦C for 24 h. Then cells were treated with or without different concentrations of CisPt (1–80 μM) and CRM or RSV (1–160 μM) for the different time period (6, 24, and 48 h). After the incubation period, 20 μL MTT (5 mg/mL in PBS) was added and further incubated for 4 h. The supernatant was removed, and the formazan product was analyzed spectrophotometrically (570 nm) after dissolution in DMSO. Wells without cells serve as a blank, and their absorbance was subtracted from the other results. Untreated cells were used as control and their viability was assumed as 100%. Results were expressed as mean values of three determinations ± standard deviation (SD). Data are presented as percent of cell viability and compared to untreated [Viability% = (T − B)/(U − B) × 100, {where T = absorbance of treated cells; U = absorbance of untreated cells, B = absorbance of blank}]. Once percent viability was obtained, the drug response curve was generated, and inhibitory concentration (IC50 and IC25) was calculated using GraphPad Prism version 5.01 [49].

In the present study, the degrees of selectivity of the CisPt, RSV or CRM are expressed as a selectivity index (SI): SI=IC50 of compound in a normal cell line/IC50 of the same compound in cancer cell line, where IC50 is the concentration required to kill 50% of the cell population. SI values of more than 2 were considered as indicative of high selectivity [50,51].

#### *3.3. Cell Proliferation Assay*

In order to determine the effect of CRM or RSV on proliferative activity of cisplatin treated cells was used the Promega CellTiter 96® AQueous One Solution Cell Proliferation Assay, a test that is based on the reduction of yellow MTS tetrazolium salt by the viable cells and generation of colored formazan soluble in the culture medium. The product was spectrophotometrically quantified by measuring the absorbance at λ = 490 nm using a Dynex plate reader (DYNEX Technologies-MRS, Chantilly, VA, USA) [52,53]. Results were expressed as mean values of three determinations ± standard deviation (SD). Untreated cells served as control and considered to have a proliferation index (PI = absorbance of treated cells/absorbance of untreated cells) equal to 1.

#### *3.4. Cell Lysates*

Cells were treated for 24 h with CisPt, RSV and CRM at the indicated concentrations. After incubation, cells were washed twice with ice-cold PBS, scraped, pelleted and lysed in RIPA buffer (Pierce) supplemented with protease inhibitor cocktail (Roche) and phosphatase inhibitor (Roche) for 30 min on ice. The lysates were transfered into microcentrifuge tubes and centrifuged at 14,000× *g* for 5 min at 4 ◦C followed by sonication (10–15 s × 3). The supernatants were collected into clean microcentrifuge tubes and stored at −80 ◦C. Lysates protein concentration were quantified by BCA protein assay kit (Thermo Scientific) and aliquots of 50 μg of total cell lysate were used for ELISA assay [54].

#### *3.5. ELISA Assay*

DuoSetR IC ELISA-Human Total p21/CIP1/CDKN1A(: DYC1047-2) was purchased from R&D Systems Inc. and contains the basic components required for the development of sandwich ELISAs to measure human p21 protein also known as CIP1 and CDKN1A in cell lysates. An immobilized capture antibody specifically binds human p21. After washing away unbound material, a biotinylated detection antibody specific for human p21 is used to detect the captured protein, utilizing a standard Streptavidin-HRP format [55]. All experiments were performed in triplicates and absorbance was measured at λ = 450 nm using a Dynex plate reader.

#### *3.6. Apoptosis Analysis*

The apoptosis assay was carried out with the Annexin V-FITC kit using the manufacturer's (BD Pharmingen) protocol. Treated and untreated 1 <sup>×</sup> <sup>10</sup><sup>6</sup> cells/mL were resuspended in cold binding buffer and staining simultaneously with 5 μL FITC-Annexin V (green fluorescence) and 5 μL propidium iodide (PI) in the dark, at room temperature for 15 min. Then, 400μL of Annexin V binding buffer was added and 10,000 cells/sample were acquired using a BD Canto II flow cytometer. The analysis was performed using DIVA 6.2 software in order to discriminate viable cells (FITC−PI−) from necrotic cells (FITC+PI+) and early apoptosis (FITC+PI−) from late apoptosis [56,57].

#### *3.7. Cell Cycle Analysis by Flow Cytometry*

For analysis of cell-cycle distribution, the cells were collected after treatment, fixed with 70% ethanol for at least one hour at 4 ◦C. The cells were stained with propidium iodide (PI) an agent which intercalates into the major groove of double-stranded DNA and produces a highly fluorescent adduct that can be excited at 488 nm with a broad emission around 600 nm. Since PI can also bind to double-stranded RNA, it is necessary to treat the cells with RNase for optimal DNA resolution. Cells

(1×10<sup>6</sup> cells/mL) were washed in PBS and centrifuge at 300<sup>×</sup> *g*, 5 min at 4 ◦C. The pellet of the cells were incubated for 10 min at 37 ◦C with 0.5 mg/sample RNase A and then was added 10 μg/sample of PI staining solution to cell pellet, mix well and incubated 10 min at 37 ◦C. The samples stored at 4 ◦C until analyzed by flow cytometry. A minimum of 20,000 events for each sample were collected using a FACS CantoII flow cytometer and ModFIT software (BectonDickinson) and used to determine the cell cycle phase distribution after debris exclusion [58,59].

#### *3.8. RT-PCR [60,61]*

#### 3.8.1. RNA Isolation with TRIzol

In order to isolate of total RNA from treated or untreated cells was used TRIzol reagent (Invitrogen)-a monophasic solution of phenol and guanidine isothiocyanate. During the isolation phases, the TRIzol reagent maintains RNA integrity, while disrupting cells and dissolving cell components. The addition of chloroform followed by centrifugation ensures the separation phase of the solution into an aqueous phase and an organic phase. RNA remains in the aqueous phase and is recovered by isopropanol precipitation. The RNA pellet washed once with 75% ice-cold ethanol and entrifuged at 10,000× *g* for 5 min at 4 ◦C. The RNApellet is redissolved in nuclease-free water and incubating in a 60 ◦C water bath for 10 min. The concentration of isolated RNA was assessed using a NanoDrop spectrophotometer (NanoDropTechnologies, Montchanin, DE, USA). RNA purity was determined by A260/A280 ratios (an A260/A230 ratio as close as possible to 2 indicates the presence of highly purified RNA).

Reverse transcription of messenger RNA molecules was performed using the High-capacity cDNA Reverse Transcription Kit from Applied Biosystems, using non-specific, randomic primers, following the manufacturer's instructions (Table 1). To perform cDNA synthesis 1μg of total RNA was used. The obtained cDNA was stored at 4 ◦C and further used in the amplification reaction Real-Time PCR (RT-PCR) [60,61].


**Table 1.** Preparation of the reaction mixture/reaction.

#### 3.8.2. Real-Time PCR

The analysis of the CDKN1A gene (P21) expression level was performed by real time PCR using a ViiA ™ 7 Real-Time PCR System by setting the ABI 7500 Fast program (Applied Biosystems). The reference gene used in the experiments was hypoxanthine ribosyltransferase (HPRT1) because this gene is found in all cell types and has a stable, relatively constant expression regardless of experimental conditions. The reference gene is useful in calibrating and interpreting qRT-PCR. The products obtained after reverse transcription reaction were diluted to a final concentration of 100 ng/uL with DEPC-treated water. The samples were prepared according to the standard protocol kit, in a reaction volume of 20 μL (Table 2):


**Table 2.** Real-Time PCR reaction components.

Each sample was performed in duplicate. Thermal cycling conditions of PCR were as follows: 95 ◦C for 10 min for amplification activation and 40 cycles at 95 ◦C for 12 s and 60 ◦C for 15 s.

The samples were analyzed using the formula 2−ΔΔCt. The value obtained indicates how many times the expression of the gene has increased or decreased compared to the control sample (untreated cells):

ΔCt1 = Ct (gene of interest) − Ct (reference gene) (treated cells)

ΔCt2 = Ct (gene of interest) − Ct (reference gene) (control cells)

ΔΔCt = ΔCt1 − ΔCt2

gene expression = 2 − ΔΔCt

#### *3.9. Statistical Analysis*

Data analyses were performed using GraphPad Prism 7 (GraphPad Software Inc., La Jolla, CA, USA). The differences between the treatment and control groups were statistically analysed using unpaired two tailed t-test and one-way ANOVA. Statistical significance was considered at *p* < 0.05.

#### **4. Results**

#### *4.1. E*ff*ects of CisPt and*/*or Natural Compounds RSV, CRM on Cell Viability*

The anticancer cytotoxic activity in vitro was evaluated by the MTT assay. The tumor cells (PE/CA-PJ49) and normal cells (HUVEC) were treated for 24 h with different concentrations of CisPt (1, 2.5, 5, 10, 20, 30, 40, 80 and 160 μM) and/or natural compounds CRM or RSV (1, 5, 10, 20, 30, 40, 80 and 160 μM). The most widely used and informative measure of a drug's efficacy is IC50-the half-maximal inhibitory concentration. IC50 was determined at half of the difference between the maximum (plateau) and minimum absorbance values, by plotting the absorbance value at 570 nm (Y axis) versus the concentration of the compound analysed (X axis). The IC50 values were reported as a mean of three independent experiments ± standard deviation (S.D.). In addition, we also determined IC 25 for both CisPt and CRM and ESV in order to choose the optimal working concentrations for these compounds (Table 3). Results showed that in comparison to the control, both CisPt and CRM or RSV caused dose-dependent toxicity. The IC50 values (±SEM) for CisPt were 9.72 ± 1.7 μM on PE/CA-PJ49 cells and 20.93 ± 2.1 μM on HUVEC cells (Table 3a). RSV, under our experimental conditions, had an IC50 of 46.8 ± 2.6 μM on PE/CA-PJ49 cells and 110.4 ± 8.6 μM on HUVEC cells (Table 3a). CRM treatment for 24 h reduced the cellular viability to an IC50 = 16.3 ± 3.4 μM on PE/CA-PJ49 tumor cells and IC50 = 59.3 ± 6.1 μM on normal cells HUVEC (Table 3a).

**Table 3.** Inhibitory concentration (IC25 and IC50) values of the CisPt, RSV and CRM were performed using a linear regression equation for the cytotoxicity curve for PE/CA-PJ49 tumor cells and for normal cells HUVEC. IC25 and IC50 values are presented as mean ± SEM according to two independent assays, each done in triplicate (**a**). The selectivity index (SI), which indicates the cytotoxic selectivity for CisPt, RSV or CRM against cancer cells versus normal cells, and. SI values over 2 were considered as high selectivity (**b**).


Selectivity index (SI) values were also calculated for RSV or CRM on both cell lines and compared to those calculated for CisPt. The results are presented in Table 3b. The SI value calculated for RSV (2.36) was close to the SI value for CisPt (2.15), but he highest SI value was obtained for CRM (3.66). Knowing that the greater the SI value is, the more selective it is and SI values less than 2 indicate general toxicity [62], we concluded that compared to CisPt, the common chemotherapy drug, RSV (SI = 2.36), works in a similar manner as CisPt, while CRM exhibits a high degree of cytotoxic selectivity (SI = 3.66).

The aim of our study was to use natural compounds capable of potentiating the cytostatic effects of CisPt without having a toxic effect on their own on normal cells. For this purpose, we also determined IC25 for CisPt, CRM or RSV treatments in order to select the optimal concentrations used in the combined treatment. The optimal time for cell-treatment was 24 h and the used concentrations for the natural compounds were 15 μM for CRM and 40 uM for RSV for treatment of both cell lines as shown in Figure 1 For CisPt treatment, the choice to use 2 concentrations was made (5 μM and 20 μM) in order to determine if the modulatory effects induced by the two natural compounds were dependent by the CisPt dose (Figure 1).

The obtained results showed that the cisplatin treatment affects the viability of both normal HUVEC cells and PE/CA-PJ49 tumor cells. As shown in Figure 1, the viability of tumor cells PE/CA-PJ49 is significantly affected by CisPt treatment 5 μM (\*\* *p* = 0.002) or 20 μM (\*\*\* *p* < 0.0001), compared to untreated cells. Analysis of the effect of CisPt treatment on the viability of PE/CA-PJ49 tumor cells showed that the difference between the effect induced by 20 μM CisPt treatment is significantly higher than the one induced by 5 μM CisPt treatment (\*\* *p* = 0.0058).

CisPt treatment of normal human HUVEC cells showed that regardless of the chosen concentration CisPt 5 μM (\*\* *p* = 0.0055) and CisPt 20 μM (\*\*\* *p* = 0.0002) the cell viability was significantly reduced compared to untreated cells. In addition, CisPt 20 μM treatment of HUVEC cells led to a decrease cell viability compared to the effect induced by treatment with 5μMCisPt (\*\* *p* = 0.0081). Comparative analysis of the effect of CisPt treatment on the two cell lines led to the following observation-both the low-concentration 5μM CisPt (non-significant, *p* = 0.287) and the high-concentration 20μMCisPt (non-significant, *p* = 0.105) treatment acted similarly on the viability of tumor and normal cells.

The viability of tumor cells PE/CA-PJ49 (\*\*\* *p* = 0.0004) or normal HUVEC cells (\*\* *p* = 0.0077) is significantly affected by RSV 40 μM treatment compared to untreated cells (Figure 1). The effect of the 40 μM RSV treatment on cell viability does not significantly differ from the effect induced by 5 μM CisPt on both normal cells (non-significant, *p* = 0.578) and tumor cells viability (non-significant, *p* = 0.0891).

**C.**

When analyzing the effect of the 40 μM RSV treatment on PE/CA-PJ49 tumor cells, compared to the induced effect on normal HUVEC cells, it was observed that RSV reduced the viability of tumor cells more than in the case of normal cells (\* *p* < 0.013) (Figure 1).

Treatment with 15 μM CRM reduced tumor cell viability compared to normal cells HUVEC (\* *p* < 0.015). The effect induced by CRM treatment on tumor cells PE/CA-PJ49 was not different from the effect induced by treatment with 5 μM CisPt (non-significant, *p* = 0.578), but was different compared to the effects induce by the 20 μM CisPt treatment (\*\* *p* = 0.0077), (Figure 1).

**A. HUVEC B. PE/CA-PJ49** 

**Figure 1.** Cisplatin or natural compounds effects on cells viability. (**A**) HUVEC and (**B**) PE/CA-PJ49 cells were either left untreated or treated 24 h with different concentrations of CisPt, CRM. Data shown are representative of three independent experiments and are expressed as mean of three replicates ± SD (*n* = 3). Untreated cells were considered to have 100% viability. Viability% = (T − B)/(U − B) × 100, (where T, absorbance of treated cells; U, absorbance of untreated cells; and B, absorbance of blank). (**C**) Tumor cell viability compared to normal HUVEC cell viability after treatment with CisPt and / or RSV,CRM; (\* *p* < 0.05, \*\* *p* < 0.005; \*\*\* *p* < 0.0005).

#### *4.2. E*ff*ects of CisPt and*/*or Natural Compounds RSVor CRM on the Cell Proliferation Process*

PE/CA-PJ49 and HUVEC cells were treated with 5 μM and 20 μM CisPt and/or 40 μM RSV, 15 μM CRM for 24 h. Analysis of the effect induced by the individual treatment with CisPt, RSV or CRM on the proliferation process of the PE/CA-PJ49 tumor cells and of the normal HUVEC cells are shown in Figure 2A,B. Treatment with 20 μM CisPt for 24 h determined a significant decrease of the proliferation process compared to the 5 μM CisPt treatment, both in the case of tumor cells PE/CA-PJ49 (\*\* *p* = 0.0065) and that of normal cells HUVEC (\*\* *p* = 0.0011). Treatment with 40 μM RSV for 24 h affected the

proliferation of tumor cells PE/CA-PJ49 much more than the one of normal cells HUVEC (\*\* *p* = 0.0063) (Figure 2A).

**A. CisPt and/or RSV B. CisPt and/or CRM** 

**Figure 2.** The index proliferation (IP) of HUVEC and PE/CA-PJ49 cells treated with CisPt and/or (**A**) RSV (**B**) CRM was calculated as IP = absorbance of treated cells/absorbance of untreated cells. Results are expressed as DO mean values of three determinations ± standard deviation (SD). Untreated cells were considered to have IP equal 1. (\*\* *p* < 0.005; \*\*\* *p* < 0.0005).

Analysis of the effect induced by the combined treatment of 5 μM CisPt + 40 μM RSV on PE/CA-PJ49 tumor cells showed that RSV can amplify the effect induced by lower CisPt concentration by recording a decrease of the IP (5 μM CisPt + 40 μM RSV vs. 5 μM CisPt; \* *p* = 0.0145). In case of applying the same treatment on the normal HUVEC cells no significant changes of IP were registered (5 μM CisPt + 40 μM RSV vs. 5 μM CisPt; non-significant, *p* = 0.105). Using a high concentration of CisPt, the proliferative response of tumor cells to the combined treatment of 20 μM CisPt + 40 μM RSV recorded a significant decrease in IP (20 μM CisPt + 40 μM RSV vs. 40 μM RSV; \*\* *p* = 0.0093). Treatment of normal cells with 20 μM CisPt + 40 μM RSV led to a response similar to that induced by treatment with 20 μM CisPt (non-significant, *p* = 0.53), but there was a significant decrease in IP compared to the response induced by treatment with 40 μM RSV alone (\*\* *p* = 0.005) (Figure 2A). These results show that the effect of treatment with 20 μM CisPt is dominant in the treatment of normal or tumor cells. Treatment with 15 μM CRM for 24 h on tumor cells PE/CA-PJ49 led to a much more intense decrease of the proliferative process compared to the effect induced on normal HUVEC cell proliferation (\*\* *p* = 0.0065) (Figure 2B). Applying the combined 5 μM CisPt + 15 μM CRM treatment to normal cells did not alter the proliferative process compared to the effect induced by individual treatment with 5 μM CisPt (non-significant, *p* = 0.784) or 15 μM CRM (non-significant, *p* = 0.209). IP of tumor cells PE/CA-PJ49 treated with 5 μM CisPt + 15 μM CRM suggested a significant inhibition of the proliferative process compared to the effect induced by treatment with 5 μM CisPt (\*\* *p* = 0.008) or 15 μM CRM (\*\* *p* = 0.0042). The combined treatment of 20 μM CisPt + 15 μM CRM very strongly inhibited the proliferation of tumor cells compared to the effect induced by treatment only with CRM (\*\*\* *p* = 0.0006) (Figure 2B). The combined treatment applied to normal cells HUVEC led to a decrease in IP, but the dominant effect on cell proliferation appears to be due to the high concentration of CisPt (20 μM CisPt + 15 μM CRM vs. 15 μM CRM; \* *p* = 0.0424). Comparative analysis of the proliferative response of PE/CA-PJ49 tumor cells versus the normal HUVEC cells to the combined CisPt + CRM treatment shown a significantly different cellular behavior (\*\* *p* = 0.004 –5 μM CisPt + 15 μM CRM; \*\* *p* = 0.0051 – 20 μM CisPt + 15 μM CRM) (Figure 2B).

#### *4.3. Modulation of p21 Protein Expression by Natural Compounds and*/*or Cisplatin Treatment*

In cancer development and evolution p21 protein might acts as an oncogene or tumor suppressor and for this reason it could be an important player in processes such as the cancer aggressiveness or the response to chemotherapy [63,64]. In this study, using ELISA assay we analyzed the expression level of p21 in the tumor cells (PE-CA/PJ49) compared to a normal cells (HUVEC). In addition, we looked at how treatment with CisPt, RSV or CRM applied individually or in combination could affect cell p21 protein expression. The data showed that the p21 protein is expressed more (3.4×) in the untreated tumor cells of the PE-CA/PJ49 vs. normal cells of the HUVEC line (625/184 pg/mL). Individually applied treatment with CisPt, 40 μM RSV or 15 μM CRM did not significantly affect the expression of p21 protein (*p* = non-significant) in normal HUVEC cells. Treatment of normal cells with 20 μM CisPt + 40 μM RSV induced an increase of p21 protein expression compared to untreated cells (*p* < 0.029) or to cells treated only with RSV (\*\* *p* < 0.0085). A significant increase in p21 protein expression of HUVEC cells was recorded in the case of the combined treatment 5 μM CisPt + 15 μM CRM compared to untreated cells (\*\* *p* < 0.0048) (Figure 3 and Table 4).

**Figure 3.** The total p21 protein expression (pg/mL) in normal cell line-HUVEC and in tumor cell line-PE/CA-PJ49 cells treated 24 h with CisPt and/or RSV, CRM. The experiments were performed in triplicates. Results are expressed as mean values of three determinations ± standard deviation (SD). (\*\* *p* < 0.005, \*\*\* *p* < 0.0005; \*\*\*\* *p* < 0.00005).

PE/CA-PJ49 tumour cells were treated 24 h with 5 μM CisPt (\*\*\* *p* = 0.0009) or 20 μM CisPt (\*\* *p* = 0.0039) showed a significant increase of total p21 protein expression compared to untreated cells (control). 40 μM RSV (\*\*\*\* *p* < 0.0001) applied alone, determined a significant increase of the total p21 protein expression in PE/CA-PJ49 cells versus control (Figure 3). On the contrary, after combined treatment with 5 μM CisPt + 40 μM RSV (\* *p* < 0.014) versus treatment with 5 μM CisPt alone the results showed a slight increase in p21 expression of PE/CA-PJ49 cells. As shown in Figure 3 and Table 4 the simultaneous treatment with 20μMCisPt+40μMRSV induced in tumor cells a significant decrease of p21 protein expression comparatively with the effect induced by RSV alone (\*\* *p* = 0.0058). These results support the notion that the dominant effect is attributed to the high concentration of CisPt. Thus we can say that RSV can potentiate the induced effect of 5 μM CisPt but cannot counteract

the induced effects of treatment with a much higher concentration of CisPT on p21 protein expression in tumor cells.

**Table 4.** The effects of CisPt, RSV, CRM treatment applied alone or in combination for 24 h on the level of p21 protein expression (n-fold p21protein expression) in normal cell line-Huvec and in tumor cell line-PE/CA-PJ49. The n-fold p21expression was calculated using the formula: n-fold p21expression = p21 pg/mL treatment/p21 pg/mL control.


PE/CA-PJ49 tumor cells treatment with 15 μM CRM led to a significant increase of the p21 protein expression compared to control (\*\* *p* = 0,0022). Combined treatment 15 μM CRM and 5 μM CisPt (\*\*\*\* *p* = 0.0001) or 20 μM CisPt (\*\*\*\* *p* = 0.0001) led to a much higher amplification of p21 protein expression compared to CisPt only-treated cells (Figure 3 and Table 4). Our data showed that in PE/CA-PJ49 tumor cells the effect induced by the CisPt and CRM was cumulative. In addition, the analysis reveal the potential of CRM to sustain and amplify the effect induced by CisPt treatment, regardless of the CisPt concentration used.

#### *4.4. Modulation of P21 Gene Expression by Natural Compounds and*/*or Cisplatin Treatment*

Changes of the P21 gene expression have been found in various types of malignancy including head and neck [65–67] but the impact of the P21 level on the disease progression and prognosis remains controversial. Therefore, we performed in vitro experiments to assess how natural compounds (RSV or CRM) modulated the P21 expression in head and neck tumor cells PE/CA-PJ49 versus normal cells HUVEC. Analysis of the expression level of P21 was performed by real-time qPCR assay in order to identify their role in response to chemotherapeutic agent-CisPt. Obtained data from this study may help define how P21 acts in modulating the DNA repair processes of tumor cells, in hope of finding new effective strategies in the treatment of head and neck cancer. Using real-time qPCR assay, we evaluated the P21 gene expression induced by CisPt and/or RSV, CRM treatment of HUVEC and PE/CA-PJ49 tumor cells. The data showed that RSV treatment acts differently on the two cell lines analyzed. RSV induces a slight decrease in P21 gene expression in HUVEC cells and an increase in the PE/CA-PJ49 tumor cells. Treatment with 5 μM CisPt (\* *p* = 0.027) or 20 μM CisPt (\*\* *p* = 0.016) induces an antagonist response compared to the effect induced by RSV treatment alone in P21 gene expression of normal HUVEC cells. The results obtained from the combined CisPt + RSV treatment show that RSV did not significantly modify the P21 gene expression generated by CisPt treatment in normal cells. CRM treatment applied alone did not affected P21 expression in HUVEC cells. The combined 5 μM CisPt + 15 μM CRM treatment showed a significant increase in P21 expression in normal cells (\*\* *p* = 0.019 versus 15 μM CRM). On the contrary, 20 μM CisPt + 15 μM CRM treatment of HUVEC cells induced a decrease of P21 gene expression compared with the effect induced by CRM alone (\* *p* = 0.021) or 20 μM CisPt (\* *p* = 0.013) (Figure 4 and Table 5).

**Figure 4.** The effect of treatment with CisPt, RSV, CRM applied independently or in combination on P21 gene expression in tumor cells PE/CA-PJ49 compared to normal cells HUVEC. Each sample was performed in duplicate. The samples were analyzed using the formula 2-ΔΔCt = gene expression. (\* *p* < 0.05, \*\* *p* < 0.005; \*\*\* *p* < 0.0005).

**Table 5.** n-fold P21gene expression in PE/CA-PJ49 tumor cells versus normal cells HUVEC treated 24 h with CisPt and/or RSV, CisPt and/or CRM. The results were obtained using gene reference HPRT.


RSV alone induced a significant increase of P21 expression (\*\* *p* = 0.006) while 5 μM CisPt (*p* = non-significant) applied individually induced a slight increase of the P21 gene expression compared to untreated PE/CA-PJ49 tumor cells. The application of the combined treatment 5 μM CisPt + 40 μM RSV in tumor cells led to a significant increase in P21 gene expression compared to untreated cells (\*\* *p* = 0.002). When using a higher concentration of CisPt together with RSV, a significant increase in P21 gene expression was observed in tumor cells versus control (\*\* *p* = 0.009) or versus RSV (\* *p* = 0.046) (Figure 4 and Table 5). These data support the modulatory role of RSV in tumor cells on the effect induced by 20 μM CisPt on the gene expression of P21 without influencing the effect induced by 5 μM CisPt.

Although CRM appears to induce a slight increase in P21 gene on the PE/CA-PJ49 tumor cells, this is not significantly different from P21 expression in untreated cells. The use of combination treatment with 5 μM CisPt + 15 μM CRM showed a significant increase in P21 in tumor cells (\*\* *p* = 0.0014) compared to the effect induced by individual treatment with CRM. Analysis of the induced effect of 20 μM CisPt +15 μM CRM in PE/CA-PJ49 tumor cells also showed a significant increase in P21 gene expression compared to the induced effect of CRM alone (\*\*\* *p* = 0.0009) or compared to the 20 μM CisPt effect (\*\* *p* = 0.0018) (Figure 4 and Table 5). These data demonstrate the ability of CRM to support the effect induced by CisPt treatment on P21 gene expression levels.

In conclusion, CisPt treatment led to an increase of the P21 gene expression level of in both lines. RSV treatment for 24 h had a different effect in PE-CA/PJ49 tumor cells (inducing an increase) versus the effect induced in normal HUVEC cells (induced a decrease) in P21 gene expression. CRM

alone amplified slightly P21 gene expression in both lines. Applied in combination with CisPt, CRM amplified the effect induced by 5 μM CisPt in both lines, the effect being twice as high in tumor cells. When the high concentration of 20 μM CisPt was used, the effects induced by CRM in tumor cells are the same as those recorded at CRM + 5 μM CisPt. However, in the case of normal cells, 20 μM CisPt + CRM treatment caused a drastic decrease in P21 gene expression, probably due to the toxic effect of CisPt, which killed a large number of normal HUVEC cells.

#### *4.5. E*ff*ects of the Natural Compounds (RSV,CRM) on Cell Cycle Phases Distribution in Cisplatin Treated Cells*

Cell cycle phases distribution was evaluated to further characterise the cytotoxic effect of CisPt and/or RSV, CRM on tumor cells PE-CA/PJ49 compared with HUVEC normal cells.

PE/CA-PJ49 tumor cells and HUVEC normal cells were treated with 5 μM or 20 μM CisPt in the presence or absence of 40 μM RSV or 15 μM CRM for 24 h. The progression through the cell cycle phases induced by treatment were evaluated by flow cytometry using a FACSCanto II flow cytometer. Data analysis were performed using the ModFit 3.2 program which offers the possibility of evaluating the cell cycle phases distribution (G0/G1, S, G2 + M).

The results obtained in the case of the tumor cells PE-CA/PJ49 revealed that single 5 μM or 20 μM CisPt applied treatment decreased of the phase G0/G1 (24.2% or 22.84% versus 41.24 in untreated cells) associated with a slight increase of the synthesis phase (50.5% or 49.6% versus 43.2% in control cells) and followed by the increase of G2+M phase (25.3% or 27.5% versus 15.5% in untreated cells) (Figure 5A,B). The effects induced by the 40 μM RSV treatment in PE/CA-PJ49 tumor cells shown a small decrease of G0/G1 phase (41.24 to 39.05%) associated with an increase in phase S (38.1 to 41.52%) and phase G2 + M (17.56 to 19.41%) compared to untreated cells (Control). Combined CisPt treatment with RSV applied to PE/CA-PJ49 induced an increase of G0/G1 phase (48.4% compared with the untreated cells 41.2%) and a decrease of the synthesis phase (34.1% versus 43.2% in control). Cell cycle phases distribution did not differ significantly when use different concentration of CisPt (5 or 20 μM) and RSV 40 μM are combined. Treatment of PE/CA-PJ49 tumor cells with 15 μM CRM alone or in combination with CisPt induced a decrease in synthesis phase (CRM = 38.6%; CisPt 5 μM + CRM = 33.6%; CisPt 20 μM + CRM = 35.14%; versus control = 43.2%) accompanied by an increase of the G2 +M phase (CRM=33.6%; CisPt 5uM + CRM = 32.2%; CisPt 20uM + CRM = 34.02%; versus control = 15.5%) (Figure 5A,B). These results suggested that the combined CisPt treatment with CRM induced a tendency to block the cell cycle in G2+M phase in tumor cells.

The HUVEC cells used as the normal line showed a distribution in the synthesis phase of less than 20% (untreated cells). The treatments applied individually 5 μM or 20 μM CisPt caused an increase of the synthesis phase to 29.2% and 39.65%, respectively, compared to 19.1% recorded in untreated cells (Control). These data show that treatment with CisPt affects normal HUVEC cells especially at higher concentrations. 40 μM RSV and 15 μM CRM did not appear to significantly affect the cell cycle when applied individually, and cannot influence the effect induced by CisPt treatment when used in combination (Figure 6A,B).

#### *4.6. E*ff*ects of the Natural Compounds (RSV,CRM) on the Apoptotic Process in Cisplatin-Treated Cells*

Apoptosis can be seen as an efficient method of preventing malignant transformation, as it ensures the removal of cells which present with genetic alterations. Deficient apoptosis can promote the development of cancer, both through the accumulation of cells found during division and the blocking of the removal of cells with high malignant potential genetic variations. The factors which determine the cells to follow one of three possible pathways are not yet known – the start of apoptosis after cellular injury, the repair of the lesion or the continuation of the cellular cycle. Many therapeutical agents have anti-tumoral effects generated by their capacity to activate the apoptotic process [68]. Tumoral cells PE/CA-PJ49 and normal cells HUVEC were treated for 24 h with CisPt, RSV, CRM in order to detect the effect induced by treatment with CisPt on tumor cells in the presence or absence of natural compounds (RSV or CRM). The flow cytometric analysis was performed using DIVA 6.2

software and allowed to discriminate viable cells (FITC−PI<sup>−</sup> = Q3) from necrotic cells (FITC−PI<sup>+</sup> = Q1) and early apoptosis (FITC+PI<sup>−</sup> = Q4) from late apoptosis (FITC+PI<sup>+</sup> = Q2).

**Figure 5.** The effect of treatments with CisPt, RSV, CRM applied independently or in combination on tumor cells PE/CA-PJ49 on cell cycle phases distribution. (**A**) flow cytometry histograms; (**B**) cell cycle phases distribution (%).

Analysis the RSV or CRM effects on the apoptotic process of tumor cells PE/CA-PJ49 is shown in the Table 6 and Figure 7. Our results show that the untreated tumor cells PE/CA-PJ49 (Control) had low total apoptosis (Q2 + Q4 = 2%). Treatment with 40 μM RSV induced a raise of total apoptosis to 27.2%, much higher than the total apoptosis induced by 5 μM CisPt (12.5%) or 20 μM CisPt (17.4%). In the case of simultaneous treatment with CisPt and RSV a raise of the total apoptosis was registered (5 μM CisPt + RSV = 31.1%; 20 μM CisPt + RSV = 34.9%) and it seems to have been generated by the presence of RSV.

**Figure 6.** The effect of treatments with CisPt, RSV, CRM applied independently or in combination on normal cells HUVEC on the cell cycle phases distribution. (**A**) flow cytometry histograms; (**B**) cell cycle phases distribution (%).



**Figure 7.** Natural compounds (RSV or CRM) effects on the apoptotic process of PE/CA-PJ49 tumor cells treated 24 h with CisPt and/or RSV or CRM (Dot-plot analysis).

Treatment for 24 h with 15μMCRM (29.6%) has determined an increase of the total apoptosis in the tumor cells PE/CA-PJ49. Combined treatment CRM with CisPt applied to the PE/CA-PJ49 cells for 24 h has determined a raise of total apoptosis to 28,7% when using 5 μM CisPt or to 40,3% when using 20 μM CisPt (Table 6 and Figure 7).

Effects of the natural compounds RSV or CRM on apoptosis process in normal HUVEC cells were analyzed after applying the same treatment scheme used in the case of PE/CA-PJ49 tumor cells. As shown in Table 7 and Figure 8, treatment with 5 μM or 20 μM CisPt, RSV or CRM did not significantly modify the apoptosis, compared to untreated HUVEC cells (Control). However, combined treatment led to an increase of the apoptotic process. Thus, 5 μM and 20 μM CisPt combined with RSV have determined apoptosis in 14% and 21.6% of the analysed cellular population, a value that was double the one obtained when using individual stimuli (Table 7 and Figure 8). Combined treatment CRM with 5 μM or 20 μM CisPt has induced a raise of apoptosis in 19,7% or 23.9% of the analysed cellular

population (Table 7 and Figure 8). Comparative analysis of the obtained data shows that both RSV and CRM have the capacity to stimulate the apoptosis of PE/CA-PJ49 tumor cells without significantly affecting the apoptotic process of normal HUVEC cells (RSV-PE/CA-PJ49 versus RSV HUVEC = 27.2% versus 5.1%; CRM-PE/CA-PJ49 versus CRM-HUVEC = 28.2% versus 7.2%). In conclusion, the two analyzed natural compounds (RSV and CRM) have the ability to amplify the apoptotic process induced by CisPt treatment in PE/CA-PJ49 tumor cells.

**Table 7.** Apoptosis of HUVEC cells induced by 24 h treatment with CisPt and/or RSV, CRM.



**Figure 8.** Natural compounds (RSV or CRM) effects on the apoptotic process on normal HUVEC cells treated 24 h with CisPt (Dot-plot analysis.).

#### **5. Discussion**

The toxic effects of CisPt are dose-dependent and affect the kidneys and bone marrow, which are accompanied by an increase in transaminases and serum creatinine. In addition, CisPt therapy alone has not been found to be effective in treating patients with HNSCC. Cisplatin efficacy appear to be more increased in combination with other chemotherapeutic agents and/or radiation therapy, but side effects or resistance to these drugs are often mentioned [69,70]. This observation suggests that the combination of cisplatin with other anti-tumor agents could be much more effective in the treatment and evolution of head and neck cancer. Some natural compounds due to their chemosensitizing potential, antitumoral activity and ability to reduce the side effects of drugs used in conventional cancer therapy protocols promise to be one of the important components of combinatorial therapy [71,72]. Natural compounds must be well tolerated and have long-lasting benefits. They can also act at the level of different signaling pathways responsible for the tumorigenesis process in HNSCC, which recommends them as agents with multiple molecular targets [73,74]. The experimental approch of this study focused on the enhancing chemotherapy response while lowering side effects and the incidence of drug resistance. To design novel combinatorial therapeutic strategy for improving the patients' prognosis and response to chemotherapy is needed. To streamline the response to chemotherapy, we analyzed how the use of CisPt in combination with CRM and RSV may influence some cellular processes such as proliferation, P21 gene expression, apoptotic process, and cell cycle development in HNSCC cell line (PE/CA-PJ49) compared to a normal cell line (HUVEC). The results showed that the viability of the tumor and normal cells was affected by CisPt treatment in the same way in both cell lines in a concentration-dependent manner. RSV or CRM treatment affected the viability of tumor cells PE/CA-PJ49 more than of normal HUVEC cells. RSV and CRM have the ability to amplify the inhibitory effect of 5μM CisPt induced on PE-CA/PJ49 tumor cells without significantly affecting the 5 μM CisPt-induced effect in normal cells. The use of a high concentration of CisPt (20 μM) together with a natural compound (CRM or RSV) led to the observation that the effect induced by CisPt on cell proliferation is dominant in both tumor cells and normal cells. Because p21 can act as a tumor suppressor or a tumor-promoting protein [75], and this makes more difficult to establish its function in the evolution of cancer we analysed the role of P21 in response to chemotherapy. Using ELISA and RT-PCR were evaluated protein and gene P21 expression in tumor PE/CA-PJ49 cells and normal HUVEC cells treated with CisPt in presence of RSV or CRM. A significant increase of total p21 protein expression was recorded in PE/CA-PJ49 tumor cells treated with 5 μM CisPt and 20 μM CisPt compared with the effect induced by RSV applied alone. However, it was observed that RSV can potentiate the induced effect of 5 μM CisPt but cannot influence the effects induced by 20μMCisPt treatment on p21 protein expression in tumor cells. In PE/CA-PJ49 the CRM treatment amplified the increase of p21 protein expression induced by CisPt treatment, regardless the CisPt concentration. Increased p21 expression was correlated with a descreased of proliferative activity of tumor cells and a good response to CisPt therapy.

Analysis of P21 gene expression level showed that CisPt treatment induced an increase of P21 expression in both lines. RSV treatment influenced in a different manner P21 expression, inducing an increase in PE/CA-PJ49 and a decrease in normal HUVEC cells. In combination with CisPt, RSV did not influence the CisPt-induced effect in HUVEC cells. On the contrary in PE/CA-PJ49 tumor cells, using the combination CisPt+RSV induced an increase in P21 gene expression and this way, RSV can modulate the effect induced by 20 μM CisPt. CRM alone did not modify the P21 gene expression in both lines. In combination with CisPt, CRM amplified the effect induced by 5 μM CisPt in both lines, the effect being higher in tumor cells. Not significant differences were observed in P21 expression in tumor cells when used the high concentrations of CisPt (20 μM). In normal cells was recorded a drastic decrease in P21 gene expression upon treatment with 20 μM CisPt + CRM. These data support the toxic effect of CisPt used in high concentration on normal cells. The treatment of tumor cells with RSV or CRM in the presence of CisPt induced an increase in protein and gene expression levels of p21. These results may be associated with a more effective response of tumor cells to treatment with lower concentrations of CisPt using RSV or CRM as therapeutic adjuvants.

The cell cycle ensures the cell can proliferate and grow by passing through the G1, S, G2 and M phases. Cancer cells reveal disorders in the cell cycle progression which contributes to exacerbate cell proliferation and to loss of genomic integrity. Treatment with 5 μM or 20 μM CisPt applied alone on tumor cells induced a slight transition from G1 to S phases accompanied by the increase of the G2+M phase (25.3% or 27.5% versus 15.5% in untreated cells). Cell cycle phases distribution did not differ significantly when different concentrations of CisPt (5 or 20 μM) were used. Treatment of PE/CA-PJ49 tumor cells with 40 μM RSV did not affect significantly the cell cycle phases compared to untreated cells. PE/CA-PJ49 cells treated with CisPt + RSV showed an increase of the G0/G1 phase and a decrease in the S phase. Treatment of PE/CA-PJ49 tumor cells with 15 μM CRM alone or in combination with CisPt induced a decrease the number of cells in S phase which was accompanied the massive blockage of cells in G2/M phase. In addition, the results provide evidence that RSV and CRM induce an increase in the apoptotic process of tumor cells. Interaction of RSV or CRM with CisPt in triggering tumor cells apoptosis indicated the amplifier role of the used natural compounds with respect to cisplatin effect. Therefore, associated chemotherapy of CisPt with natural compounds (RSV or CRM) as adjuvants might have a beneficial effect in decreasing the CisPt doses and in reducing its adverse reactions.

#### **6. Conclusions**

In conclusion, the results suggest that RSV and CRM act against proliferation and sustain the effects of cisplatin by the induction of cell cycle arrest, amplification of DNA damage and contributing to cancer cell destruction by increasing the apoptotic process. In addition, RSV and CRM amplify the expression of P21 in tumor cells and might contribute to increasing the sensitivity to cisplatin.

**Author Contributions:** G.G.P.-M. had equal contribution with M.B. in designed the research, data acquisition, analysis and interpretation of data, and wrote the manuscript. V.R., R.H. and C.D.S. performed the experiments, data acquisition, analysis and interpretation of data and manuscript drafting. N.R. and C.C.D. contributed to statistical analysis, critical revision of the manuscript for important intellectual content. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Grant PN-III-P1-1.2-PCCDI-2017-0341/2018 and Competitiveness Operational Programme (COP) A1.1.4. Grant ID: P\_37\_798 MyeloAL-EDiaProT, Contract 149/26.10.2016, (SMIS: 106774) and ERAPerMed\_Joint Transnational Call for Proposal (2019) for Personalised Medicine: Multidisciplinary Research Towards Implementation, BronchoBOC, Contract 140/2020.

**Acknowledgments:** Competitiveness Operational Programme (COP) A1.1.4. ID: P\_37\_798 MyeloAL-EDiaProT, Contract 149/26.10.2016, (SMIS: 106774), MyeloAL

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Dietary Supplements for Male Infertility: A Critical Evaluation of Their Composition**

**Andrea Garolla 1,\*, Gabriel Cosmin Petre 1,**†**, Francesco Francini-Pesenti 2,**†**, Luca De Toni 1, Amerigo Vitagliano 3,4, Andrea Di Nisio <sup>1</sup> and Carlo Foresta <sup>1</sup>**


Received: 17 April 2020; Accepted: 18 May 2020; Published: 19 May 2020

**Abstract:** Dietary supplements (DS) represent a possible approach to improve sperm parameters and male fertility. A wide range of DS containing different nutrients is now available. Although many authors demonstrated benefits from some nutrients in the improvement of sperm parameters, their real effectiveness is still under debate. The aim of this study was to critically review the composition of DS using the Italian market as a sample. Active ingredients and their minimal effective daily dose (mED) on sperm parameters were identified through a literature search. Thereafter, we created a formula to classify the expected efficacy of each DS. Considering active ingredients, their concentration and the recommended daily dose, DS were scored into three classes of expected efficacy: higher, lower and none. Twenty-one DS were identified. Most of them had a large number of ingredients, frequently at doses below mED or with undemonstrated efficacy. Zinc was the most common ingredient of DS (70% of products), followed by selenium, arginine, coenzyme Q and folic acid. By applying our scoring system, 9.5% of DS fell in a higher class, 71.4% in a lower class and 19.1% in the class with no expected efficacy. DS marketed in Italy for male infertility frequently includes effective ingredients but also a large number of substances at insufficient doses or with no reported efficacy. Manufacturers and physicians should better consider the scientific evidence on effective ingredients and their doses before formulating and prescribing these products.

**Keywords:** fertility; ingredients; male reproduction; semen parameters; supplements

#### **1. Introduction**

Infertility is a pathological condition defined as the inability of a sexually active, non-contracepting couple to achieve pregnancy in one year [1]. Both male and female factors can lead to infertility. In particular, according to the causes, it has been reported that 29.3% is due to a male factor, 37.1% to a female factor, 17.6% to both male and female factors, with the remaining percentage considered as idiopathic [2].

It is estimated that around 10%–15% of all couples are affected by infertility, thus representing a global concern in most developed countries [3].

Among male infertility causes, many recent studies have emphasized the role of genital tract inflammation, incorrect lifestyles and malnutrition [4]. On this regard, weight excess and other conditions such as metabolic syndrome, alcohol abuse, cigarette smoking, exposure to environmental pollutants etc. have been strongly related to a decrease in sperm quality and fertility. A major driving hypothesis is that these conditions, by inducing an elevation of reactive oxygen species (ROS) and nitrogen species (RONS), are able to alter the balance of the redox status of both the steroidogenic cell population and the germ line cell populations, leading to the impairment of the hypothalamic–pituitary–testicular axis and the reduction of sperm quality [5].

A large number of recent studies have focused on the ability of many substances, generally termed as *nutraceuticals*, to improve the hormonal status and sperm parameters by different mechanisms [6]. Nutraceuticals are used as ingredients of dietary supplements (DS), widely marketed for the prevention or treatment of the most disparate pathological conditions. From a legislative point of view, the European Food Safety Agency (EFSA) defines that DS are not intended for the treatment or prevention of disease in humans, but only to support specific physiological function [7]. Currently, DS are widely prescribed to improve physiological aspects related to male fertility.

Many DS are available on the market with various formulations, containing both nutrients and botanicals at different doses. Despite many authors demonstrating positive effects of some ingredients on semen parameters and fertility outcomes [8], many others have also shown a lack of efficacy and even potentially harmful side effects [9]. In a recent position statement, the Italian Society of Andrology and Sexual Medicine (SIAMS) summarized the state of the art on each single ingredient currently used in the andrological field. In this paper, authors concluded that there was still limited scientific evidence on the possible role of any nutraceutical in andrology and the use of antioxidants could be suggested in patients with idiopathic infertility in the presence of documented abnormal sperm parameters only after a specific diagnostic workup. However, to date, no regulation or guidelines are available for the use of these products, generating confusion for both prescribers and patients [10]. Moreover, several factors make it difficult to empirically address the right ingredient for the right patient. In particular, it is difficult to identify the correct DS since each product contains different ingredients at different doses.

The purpose of this study was to critically evaluate the composition of DS employed in male infertility, using the Italian market as a sample.

#### **2. Materials and Methods**

In order to evaluate the potential efficacy of DS, a systematic literature review on substances used to improve sperm parameters was preliminarily performed. The literature search was conducted in MEDLINE, Scopus, EMBASE, and Cochrane Library registers until 31 March 2020. Only randomized clinical trials (RCTs) and systematic reviews or meta-analysis of RCTs were considered eligible. With the aim to rule out possible interactions between ingredients, only studies that used active substances alone or in combination with at most the other three ingredients were considered. The key terms used for the search were: fertility or male reproduction or semen parameters and supplements or ingredients. Figure 1 displays the flow diagram of the selection of eligible papers.

To establish the efficacy of each ingredient we considered only those having at least one RCT or systematic review or meta-analysis of RCTs, demonstrating a significant effect on any sperm parameters involved in male fertility. Significance was set at *p*-value < 0.05. When evaluating the findings of meta-analyses, we verified whether statistical methods incorporated substantial heterogeneity (Higgins I2 > 30%) into a random-effects model, as appropriate. Regarding the daily dose of each active ingredient with nutrient characteristics, we referred to the tolerable Upper intake Levels (UL) as reported in Dietary Reference Intake (DRI) [11].

**Figure 1.** Flow diagram of the selection of eligible papers.

Based on the results of available articles, we were able to identify, for each active ingredient, the minimal effective daily dose (mED) able to improve sperm parameters. To define mED we used the lowest effective dose reported in RCTs, systematic review or meta-analysis of RCTs. Therefore, we classified the ingredients contained in each supplement and suggested daily dose into three categories: reported efficacy with a dose achieving the mED (A), reported efficacy but with a dose below mED (B) and unreported data of efficacy (C). To classify DS, we created a formula taking into consideration the three classes of ingredients and their number:

$$Score = \frac{(2A + B - C)}{2N} \times (A + B/2)^2$$

In particular, the above formula was conceived based on the following sequential steps:


efficacy) the number of moderate efficacy ingredients (*A* + *B*/*2*), finally obtaining a corrected score for each supplement.

(5) Given the distribution of the scores resulted in three main clusters, we classified DS into three categories, resembling the efficacy of the ingredients: higher expected efficacy (corrected score ≥ 4), lower expected efficacy (4 < corrected score > 1) and no expected efficacy (corrected score ≤ 1).

We collected the names and formulations of the DS registered in Italy by referring to the register of the Italian Ministry of Health [12].

#### **3. Results**

The literature search on active ingredients allowed us to identify 41 studies (RCTs or meta-analyses) reporting their efficacy on sperm parameters (Figure 1). By this analysis we found that 18 of these ingredients had a reported efficacy. The complete list of ingredients with clinical evidence of efficacy, the respective references, evaluated sperm parameters and employed daily doses, are summarized in Table 1. In the right column, the mED of each ingredient is reported. In some studies, marked with an asterisk, the employed dose exceeded the reported UL. In particular, all the studies involving zinc evaluated the effect of this ingredient at a dose exceeding UL. For each active ingredient, the evidence of efficacy was supported by at least two RCTs or meta-analysis, excluding astaxanthin, D-aspartic acid and L-citrulline, which had only one reference.

**Table 1.** Active ingredients with evidence of efficacy, references, evaluated sperm parameters, employed daily doses and minimal effective dose (mED).



**Table 1.** *Cont.*

\* The employed dose exceeded/reach UL. LC: L-Carnitine; LAC: Acetyl L-Carnitine; EPA: Eicosapentaenoic acid; DE: Dry Extract; DHA: docosahexaenoic acid. Rev: Review; Met: Meta-analysis.

Ingredients without clinical evidence in the improvement of sperm parameters (no RCT or meta-analyses) are listed in Table 2.

> Astragalus DE Damiana DE Nettle DE Catuba DE Ecklonia bicyclis DE L-Taurine Glutathione Glucosamine SOD Vitamin D3 Vitamin B1 Riboflavin Niacin Vitamin B5 Vitamin B6 Biotin Manganese

**Table 2.** Ingredients without clinical evidence of efficacy.

DE: Dry Extract; SOD: super oxide dismutase.

We found 21 DS marketed in Italy for male infertility. Their composition and the daily doses of their active ingredients are summarized in Table 3. Moreover, for each supplement, the scores of expected efficacy and the symbols summarizing the efficacy of their ingredients are reported.



#### *Nutrients* **2020** , *12*, 1472




**Table 3.** *Cont.*


**Table3.***Cont.*

score of supplement's potential efficacy; EV: efficacy value of active ingredients; evidence of ingredients and dose efficacy: (A) reported efficacy and achievement of mED, (B) reportedefficacy but below mED and (C) unreported efficacy. DE: Dry Extract; SOD: super oxide dismutase; DHA: docosahexaenoic acid.

#### *Nutrients* **2020** , *12*, 1472

A detailed analysis of this table raised the following considerations: (i) all supplements were mixtures of active ingredients; (ii) in each supplement the number of ingredients ranged from 2 up to 17, with a mean number higher than 7; (iii) 13 of 21 supplements contained at least one ingredient without reported efficacy; (iv) 19 supplements had ingredients below mED; (v) indeed, 1 supplement contained seven ingredients dosed below mED; (vi) 1 supplement contained only active ingredients satisfying mED; (vii) the product number 9 had a nutrient reaching UL (zinc 40 mg/day); (viii) zinc was the most used ingredient, followed by selenium, arginine, coenzyme Q, folic acid and carnitine. These substances were present in more than 50% of DS, whereas all the remaining ingredients were represented in 10% or less of products.

The distribution of DS into the three classes of efficacy is reported in Figure 2. Two DS out of 21 (9.5%) were included in the higher expected efficacy group. The majority of remaining products (71.4%) fell in the lower expected efficacy group, and four (19.1%) in the group with no efficacy.


**Figure 2.** Distribution of supplements in classes of expected efficacy. \* This supplement has a content of Zinc reaching the UL. Numbers refer to a specific supplement.

#### **4. Discussion**

This critical review aimed to evaluate the formulation of supplements for male infertility using the Italian market as a sample. In general, there is still poor evidence in terms of large well-designed randomized and placebo-controlled trials availability, supporting the efficacy of nutraceutical products in the field of male reproductive health [54,55]. Nevertheless, these products are commonly administered to infertile patients [8,56]. Since a medical prescription is not necessary to purchase dietary supplements, subjects seeking fertility may have easy access to these products [10,57]. As a proof of concept, the Italian market of supplements generated 3.3 billion euros in 2019, with an increase of 4.3% compared to 2018 [58].

Whilst a rational use of supplements may be potentially beneficial for the improvement of sperm parameters, we need to stress that their uncontrolled use is potentially harmful for patients' health due to direct toxic effects and interaction with drugs or nutrients [59]. In this respect, we were surprised to point out that all RCTs and meta-analyses on zinc for male infertility relied on doses always exceeding the UL. Over this background, in the near future it would be desirable to better define thoughtful criteria for each supplement in use.

Our analysis found that beside the gap of literature, the market of food supplements is still supported by poor scientific evidence. The majority of DS contained a huge number of ingredients, up to 17. The mixture of such a high number of ingredients may generate different issues, including a low concentration of each substance (i.e., necessitating of two or more administrations to reach the daily effective dose), a large volume of pills and a high risk of interactions. What is more, we found that some ingredients included in many DS had no scientific evidence of efficacy (i.e., astragalus, vitamin D3, taurine and riboflavin). The formulation of pills with a large number of ingredients, some of which cause uncertain benefits, denotes a gap of knowledge of potential biologic targets by manufacturers. Moreover, it has been reported that some plant extracts, present in many of these supplements, are likely to interact with drug metabolism [60,61]. This aspect raises further concerns on the safety of these products.

Very frequently, nutrients were present in DS at a dosage below mED. This situation was more common among products with a high number of ingredients. The administration of any active substance with a dose below mED appears as scientifically unjustified due to uncertainties in the therapeutic results. Differently, when the number of ingredients was small, the dose often satisfied mED. Another major aspect in the evaluation of supplements concerns safety. Some ingredients, particularly when administrated in high doses, are not free from risks when used as dietary supplements. For example, folates can mask the B12 deficiency favoring the progression of neurological damage [62]. The combination of these two vitamins could have a synergic effect in improving homocysteine metabolism hence the sperm quality. It should be noted that vitamin B12, when present, was rarely associated to folic acid [63,64]. Furthermore, zinc reduces the copper intestinal absorption interfering with its carrier [65]. With respect to this, we want to stress that one supplement on the market contained a dose of zinc reaching the UL.

On a positive note, our analysis revealed that some active ingredients with reported efficacy are frequently present in analyzed supplements. Previous studies demonstrated that some ingredients are particularly effective in specific patients' conditions. Substances with antioxidant properties are indicated in inflammation of the male accessory glands, both related to microbial and non-microbial origin. Several studies performed in asthenozoospermic infertile patients, showed that the positive effect of selenium supplementation is dependent on the correct structure of the mitochondrial capsule [66,67]. Carnitine supplementation induced a significant increase in sperm motility in cases of asthenozoospermia with preserved mitochondrial function [68,69]. Due to the key role of zinc in the processes of DNA compaction, administration of this micronutrient was successful in improving sperm morphology and DNA integrity in patients with prostate abnormalities [70,71].

Based on active ingredients reaching mED we created a grading scale of supplements distinguishing three classes of expected efficacy. Three products were present in the higher class, some of which contained ineffective or underdosed ingredients. Most of the supplements fell in the lower group of expected efficacy. In this class, a large number of ineffective or underdosed products were also present. For an adequate evaluation of these classes, we considered the number of the effective ingredients as the most important criterion of efficacy. A relevant aspect was the use of ineffective or underdosed ingredients that should be absent or less than possible. Another parameter to evaluate a product was the presence of a lower number of ingredients.

We acknowledge the application of a non-validated statistical method to calculate scores for each DS may represent a point of weakness in this study. Very recently, a validated formula to score supplements was suggested by Kuchakulla et al. [72], based on the Budoff's score, previously conceived by cardiologists to evaluate their procedures [73]. However, this scoring system when applied to DS, does not take into account the effective dose of ingredients, a crucial point in the evaluation of their efficacy. For example, using this approach, a DS containing ingredients at ineffective or toxic doses would be considered useful. As a point of strength, our scoring system relied on high quality evidence coming from RCTs or a systematic review and meta-analyses of RCTs, which represents a reliable approach to critically weighing the expected efficacy of dietary supplements. The same approach could be applied to evaluate products used in other clinical conditions.

In conclusion, this study showed that most DS marketed in Italy for male infertility contain ingredients with reported efficacy in the improvement of sperm parameters. Nevertheless, a non-negligible number of DS are mixtures of substances with uncertain or unreported benefits, whose administration may be unhelpful or even harmful for infertile patients. On that basis, we believe manufacturers should carefully scrutinize scientific evidence before delivering each supplements' formulation. Accordingly, physicians should evaluate the composition of DS and the dose of each single constituent before considering their clinical use. Finally, the choice for DS should be tailored to the specific patient's fertility problem.

**Author Contributions:** A.G. and G.C.P. contributed to the conception/design of the research and acquisition/analysis of the literature data; A.G., G.C.P. and F.F.-P. equally contributed and drafted the manuscript; A.D.N. concepted and performed data analyses; L.D.T., A.V. and C.F. critically revised the paper for important intellectual content. All authors revised and approved the final manuscript, and agreed to be fully accountable for ensuring the integrity and accuracy of the work. All authors had full access to all the data in the study and are able to take responsibility for the integrity of them and the accuracy of the analysis.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors thank Marco Ghezzi for helpful discussion.

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* ω**-3 and** ω**-6 Polyunsaturated Fatty Acids, Obesity and Cancer**

#### **Stefania D'Angelo, Maria Letizia Motti and Rosaria Meccariello \***

Dipartimento di Scienze Motorie e del Benessere, Università di Napoli Parthenope, via Medina 40, 80133 Napoli, Italy; stefania.dangelo@uniparthenope.it (S.D.); motti@uniparthenope.it (M.L.M.)

**\*** Correspondence: rosaria.meccariello@uniparthenope.it; Tel.: +39-081-5474668

Received: 5 August 2020; Accepted: 7 September 2020; Published: 10 September 2020

**Abstract:** Recently, nutraceutical bioactive compounds in foods have been discovered for their potential health benefits regarding the prevention of chronic disorders, such as cancer, and inflammatory, cardiovascular, and metabolic diseases. Dietary omega-3 polyunsaturated fatty acids (ω-3PUFAs), including alpha-linolenic acid, docosapentaenoic acid, and eicosapentaenoic acid, are mostly attractive. They are available for the customers worldwide from commonly used foods and/or as components of commercial food supplements. The anti-inflammatory and hypotriglyceridemic effects of these fatty acids are well known, whereas pro-inflammatory properties have been recognized in their dietary counterparts, the ω-6PUFAs. Both ω-3 and ω-6PUFAs contribute to the production of lipid mediators such as endocannabinoids that are notably involved in control of food intake, energy sensing, and food–related disorders. In this review, we present ω-3 and ω-6PUFAs and their derivatives, endocannabinoids; discuss the anti-obesity effects of ω-3PUFAs; their roles in inflammation and colorectal cancer development; and how their action can be co-preventative and co-therapeutic.

**Keywords:** obesity; ω-3PUFAs; ω-6PUFAs; endocannabinoids; CRC; fatty acids

#### **1. Introduction**

The prevalence of obesity has increased worldwide. Obesity represents a major health challenge because it substantially increases the risk of comorbidities, including cardiovascular disease, hypertension, type 2 diabetes, dyslipidemia, nonalcoholic fatty liver disease, obstructive sleep apnea, and some cancers, thereby contributing to a decline in both quality of life and life expectancy [1]. This syndrome is a complex condition involving social, biological, and psychosocial factors. The genesis of obesity is multifactorial: it is characterized by chronic low-grade inflammation, primarily due to an imbalance between the production/secretion of pro-inflammatory cytokines vs. anti-inflammatory cytokines [2]; deregulated lipid and glucose metabolism in metabolic organs is thought to be a critical factor. High-calorie diets and sedentary lifestyles are the most important factors in the development of obesity. As a consequence, global anti-obesity strategies focus on dietary and lifestyle modifications. In fact, weight loss, energy restriction, and nutrient dense diets can restore this imbalance, at least in part [3]. Therefore, the most used approaches aim at suppressing appetite, normalizing lipid metabolism, and increasing energy expenditure [4], through limitation of sugar and fat consumption, promotion of physical activity, consumption of fruits and vegetables, and pharmacological approaches.

Dietary interventions using natural bioactive food compounds have emerged as promising therapeutic tools for metabolic diseases, with limited deleterious side effects. Composition of the diet may affect metabolic and endocrine functions and overall energy balance [5]. Studies conducted in both animal models and humans support the assertion that dietary bioactive compounds can increase energy expenditure and thermogenesis, providing benefits in preventing/limiting obesity. Most health recommendations emphasize diets rich in fruits and vegetables, which have lower caloric density and higher nutrient density [6]. Such diets would provide significant amounts of phytochemicals, bioactive components with nutraceutical effects due in part to their anti-oxidant and anti-inflammatory properties [7]. Natural bioactive compounds, for example, the polyphenols (epigallocatechin, resveratrol, curcumin, quercetin, oleuropein, anthocyanins, ellagic acid, and others), have been studied as factors with possible indirect or direct impacts on specific molecular pathways, associated with the pathophysiologies of different syndromes [8,9] due to their well-documented anti-oxidant [10–12], anti-proliferative [13–16], anti-inflammatory, anti-cancer, anti-aging, and anti-obesity effects [17–21]. In addition to polyphenols, other nutraceuticals with an anti-obesity effects are the dietary ω-3 polyunsaturated fatty acids (PUFAs), which can act on adipose tissue inflammation, in contrast to omega-6 (ω-6) PUFAs, which exhibit pro-inflammatory properties. Both ω-3 and ω-6PUFAs contribute to the production of lipid mediators such as endocannabinoids, which are notably involved in control of food intake, energy sensing, and food–related disorders; reproduction; inflammation; the stress response; and cancer, among other things [22–28].

Therefore, in this review article we present ω-3 and ω-6PUFAs and their derivatives, endocannabinoids; discuss the anti-obesity effects of ω-3PUFAs; their role in inflammation and colorectal cancer (CRC) development; and co-preventative and co-therapeutic applications.

#### **2.** ω**-3. and** ω**-6PUFAs**

The amount and type of dietary fat in the diet are important factors influencing adipose tissue function and whole-body metabolism, with important health ramifications. Fatty acids (FA) are hydrocarbon chains with a carboxyl group at one end and a methyl group at the other. FA species are classified by their varying degrees of saturation into three main classes: saturated fatty acids (SFA), monounsaturated fatty acids (MUFA), and polyunsaturated fatty acids (PUFA) [29].

SFAs have a simple carbon chain containing no double bonds; MUFAs contain one double bond; and PUFAs are classified as carbon chains containing two or more double bonds. The variations in the chemical structures of these diverse classes can lead to dissimilar physiological activities. For example, SFA has been linked to the development of metabolic dysfunction; contrariwise, MUFAs and PUFAs have helpful activities in metabolism [30].

PUFAs are additionally classified into ω-3 and ω-6 groups, based on the position of the first double bond from the methyl end of the fatty acid. The structural dissimilarities of these FAs also give rise to functional differences, in terms of their actions on inflammation and metabolism [5].

ω-3 FAs are PUFAs with more than one carbon-carbon double bond in their backbones. They are polyunsaturated because their chains have numerous double bonds. One way in which a FA is named derives from the position of the first double bond, counted from the tail, that is, the omega (ω-) or the n-end. Thus, in ω-3 FAs, the first double bond is between the third and fourth carbon atoms from the tail end; then, they have a double bond at the third carbon from the methyl end of the carbon chain.

Humans do not own the essential ω-3 desaturase to add a double bond at the 15th carbon of a long chain FAs, and are, therefore, unable to endogenously synthesize α-linoleic acid (ALA 18:3n-3) and linoleic acid (LA 18:2n-6), making them vital FAs.

In addition to ALA, ω-3 PUFAs can be defined as a heterogeneous mix of FAs, among which eicosapentaenoic acid (EPA, 20:5n-3) and docosapentaenoic acid (DHA, 22:6n-3) are presently thought to be the most bioactive of the ω-3 species; however, docosapentaenoic acid (DPA, 22:5), an intermediate of EPA and DHA, may also have positive health properties [30,31].

ω-6PUFAs are also essential fatty acids and normally have metabolically distinct properties to ω-3PUFAs. While the human body cannot synthesize ω-3 and ω-6PUFAs, it does have the capability to further metabolize these FAs through stages of elongation and desaturation. ALA can be metabolized to EPA and DHA by Δ6 desaturase and Δ5 desaturase correspondingly, while LA is transformed to arachidonic acid (AA 22:4n-6). However, the change of ALA to DHA is very inefficient with <10% transformation in females and <3% in males [30,32]. While ALA is the favorite substrate for

Δ6 desaturase, plenty of dietary linoleic acid has been observed to suppress the change of ALA to DHA, which may be a confounding factor in these data. Evidence suggests that supplementing with stearidonic acid (18:3n-3) may increase the efficiency of transformation to DHA, demonstrating Δ6 desaturase as a rate limiting step. There is also a degree of individual change in the lipidome following ω-3 integration in humans, which may be a factor in the equivocal metabolic conversion measured in many human integration trials [30,33].

The structural differences of the FAs give rise to functional dissimilarities, in terms of their actions on inflammation and metabolism. For example, due to the pro-inflammatory actions of saturated FAs, intake of these molecules is associated with an increase in cardiovascular disease risk. In contrast, ω-3PUFAs have anti-inflammatory capability, and their intake is connected to a decrease of in cardiovascular disease risk. Dietary FAs are involved in glucose-insulin homeostasis and modulating adipose tissue properties [5]. High saturated-fat intake causes adipose tissue inflammation and obesity in mice; those effects can be partially reduced when these high-fat diets are energy-restricted [29]. Instead, dietary EPA supplementation ameliorates adipose tissue inflammation, regardless of adipose tissue mass. These data, taken as a whole, demonstrate the importance of AF in modulating the adipose tissue properties [34].

Saturated FA, generally, contributes to adipose tissue inflammation, probably due to TLR-2 and TLR-4 activation, and switching of downstream pro-inflammatory signaling pathways comprising the nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) pathway [29]. In contrast, ω-3 FA, primarily DHA and EPA, mitigate adipose tissue inflammation in diverse animal models of obesity [34]. Compared to ω-6PUFAs such as AA, ω-3PUFAs generate less eicosanoids with inflammatory properties. Additionally, ω-3PUFAs competitively decrease AA-mediated inflammatory eicosanoid prostaglandin E2 (PGE2) synthesis. As a whole, food sources rich in healthy fats can provide health benefits in both the long and short term.

Figure 1 summarizes the essential FAs and their dietary sources. In general, the primary source of ω-3PUFAs in the human diet is marine products, in particular phytoplankton that enters at multiple levels in the food chain [34], fatty fish, and cod liver oil, food rich in DHA and EPA [35,36].

**Figure 1.** Essential fatty acids and dietary source (AA: arachidonic acid; ALA: α-linolenic acid; DHA: docosahexaenoic acid; EPA: eicosapentaenoic acid; LA: linoleic acid; PUFA: polyunsaturated fatty acid).

In plants, ALA can be extracted from seeds such as flaxseed (linseed), green leafy vegetables, legumes, and nuts. Vegetable oils such as sunflower, corn, perilla, canola, and soybean are the principal sources of LA, and they provide a small amount of ALA [31].

#### *2.1. Endocannabinoids: PUFAs Derivatives Involved in the Central and Peripheral Control of Food Intake*

The endocannabinoid system (ECS) comprises lipid mediators capable of binding to and activating cannabinoid receptors—traditionally, the central and the peripheral cannabinoid receptor, CB1 and CB2 respectively, membrane G-coupled receptors originally found to be mainly expressed in the brain and peripheral tissues. The system also includes a large set of biosynthetic and metabolizing enzymes and transporters [37]. Since the discovery in the 1990s of the first endocannabinoids, anandamide (AEA) and 2-arachinonoylglycerol (2-AG), the relevance of ECS has been progressively widening due to the inclusion in the system of non-canonical cannabinoid receptors (i.e., the transient receptor potential cation channel subfamily V member 1 (TRPV1); the orphan G-coupled receptors GPR18, GPR119, and GPR55; and also peroxisome proliferator-activated receptors (PPARs) such as PPARα and γ), the discovery of several "endocannabinoid-like" compounds, and the large spectrum of biological functions [38]. Currently ECS represents a conserved, widely-expressed signaling system involved in the control of most biological activities within the brain and peripheral tissues, from food intake to reproduction, the immune response, and cancer, among others [22–28,39], and it is susceptible to epigenetic modulation by diet [40].

Traditionally, AEA and 2-AG are the N-ethanolamide and the glyceryl ester of ω-6PUFA AA respectively, and represent the main endogenous ligands of CB1 and CB2, with AEA having low CB2 affinity, and 2-AG capable of binding both receptors [41]. The endocannabinoids N-docosahexaenoyl ethanolamine (DHEA) and N-eicosapentanoyl ethanolamine (EPEA) are ω-3 DHA and ω-3 EPA derivatives respectively; docosahexaenoylglycerol (DHG) and eicosapentaenoyl glycerol (EPG) are the glycerol esters of ω-3 DHA and EPA derivatives and have been discovered following the investigations on 2-AG analogues. While ω-6 AA derivatives AEA and 2-AG bind the canonical receptors [41], ω-3 EPA/DHA derived endocannabinoids, and endocannabinoid-like compounds exhibit lower affinity binding to CB1/CB2 than AEA/2-AG, and in some cases bind the aforementioned non canonical cannabinoid receptors [42]. Currently, ω-3 derived endocannabinoids are known at lesser extent than canonical ones, and their basic characterizations support possible involvement in inflammation, neuroprotection, and cancer [42]. Details concerning the metabolic/hydrolyzing pathways linking ω-3 and ω-6, endocannabinoids, and inflammatory mediators are summarized in Figure 2.

Endocannabinoids, via CB1, have a recognized role as orexigenic factors, in that they stimulate food intake and body fat deposition [26,43]; at the periphery ECS activity is largely reported in the gastrointestinal tract, and liver and adipose tissue, along with possible involvement in the microbiota–gut–brain axis and obesity onset, as excellently reviewed in [23]. Obese subjects display high endocannabinoid tone in the plasma and brain; furthermore, altered expression of CB1 and higher endocannabinoid levels in the muscle, adipose tissue, pancreas, and liver have been reported ([23] for a recent review). Consistently, CB1 activation increases food intake [44], whereas its pharmacological and genetic impairment reduces food assumption, protecting against the development of obesity, liver steatosis, and related inflammation [45,46]. Traditionally endocannabinoid signaling via CB1 is involved in the central control of food intake exerting its activity within the hypothalamus. This brain region catches and integrates the environmental cues, including fuel availability, in order to maintain energy homeostasis [47], projects neuronal networks towards different nuclei within the hypothalamus or in the brain stem controlling both the homeostatic regulation of energy balance, and biological functions deeply related to energy homeostasis and availability, such as reproduction [26]. Appetite inhibiting neuropeptides, such as proopiomelanocortin (POMC) precursor and cocaine-amphetamine regulated transcript (CART), and appetite stimulating neuropeptides such as neuropeptide Y (NPY), melanin-concentrating hormone (MCH), and agouti related protein (AgRP), are produced within the hypothalamic arcuate nucleus (ARC) to mediate the adaptive changes in food intake and energy expenditure in response to nutrient availability and peripherally produced "metabolic sensors" (i.e., glucose from liver, interleukin-6 (IL-6) from muscle, leptin from white adipose tissue, insulin, amylin and pancreatic polypeptide from pancreas, glucagon-like peptide-1, gherlin, cholecystokinin and peptide YY from the gastro-intestinal tract, and lastly gut microbiome-derived signals) [23,26,47]. ECS activity may be affected by metabolic sensors and may modulate neuronal population producing orexigenic/anorexigenic-peptides [26]. Among "metabolic sensors," leptin is the major peripheral indicators of body metabolic reserves acting as satiety signal [48]. In the natural mutant mice for *Leptin* gene, the Ob/Ob mice, over activation of ECS has been reported in the hypothalamus whereas leptin inhibits the hypothalamic activity of ECS [49]. Additionally leptin resistance, a condition causing food intake alteration and consequent obesity development, has been linked to the over-activation of the ECS [49], with a sex-specific epigenetic modulation of *cnr1*, the gene encoding CB1, following maternal high fat diet (HFD) [50]. Therefore, CB1 antagonist-based therapy has been used for the treatment of obesity, but, in spite the promising anti-obesity effects, the treatment caused severe psychiatric side effects and has been discontinued in patients [51].

**Figure 2.** A schematic representation of the metabolic/hydrolyzing pathways linking ω-3 and ω-6, endocannabinoids, and inflammatory mediators. The syntheses of DHEA and EPEA, and AEA from ω-3 and ω-6 PUFAs respectively, requires the activity of N-acetyltransferase (NAT) followed by N-acyl phosphatidylethanolamine-specific phospholipase D (NAPEPLD). The synthesis of 2-AG from ω-6 PUFAs requires the subsequent activity of phospholipase Cβ (PLCâ) and diacylglycerol lipase (DAGL). The hydrolysis of the endocannabinoids requires the activity of the fatty acid amide hydrolase (FAAH) and monoacylglycerol lipase (MAGL). The activity of lipoxygenases (LOX), cyclooxygenase (COX), and cytochrome P450 enzymes (CyP450) drives the production of inflammation mediators. AA: arachidonic acid; AEA: anandamide; 2-AG: 2-arachinonoylglycerol; DHA: docosahexaenoic acid; DHEA: N-docosahexaenoyl ethanolamine; EPA: eicosapentaenoic acid; EPEA: N-eicosapentanoyl ethanolamine; Mar1: maresin 1; NPD1: neuroprotectin D1; PUFA: polyunsaturated fatty acid; RvD1: resolvin D1.

A deep interplay between the synthesis and actions of the AA, DHA, and EPA-derived endocannabinoids and ECS occurred, as recently reviewed [52]. In this respect, many authors suggest that dietary PUFAs and in particular the ω-6/ω-3 ratio may affect the endogenous tone of endocannabinoids with consequences on health [26,53]. For example, in animal models fed for three or four months with a ω-3 deficient diet, low DHA levels in the brain with effects on ECS and synaptic plasticity have been discovered [54,55]; similarly, long-term EPA and DHA supplementation reduces AEA and 2-AG levels, with reciprocal increases in levels of the corresponding endocannabinoid-like EPA- and DHA-derived molecules [52].

Lastly, dietary intervention may epigenetically affect ECS in animal models, and cell lines [40]; and at present, a recurrent process potentially influencing the development of eating disorders such as binge-eating has been related to the epigenetic modulation of FAAH [56], the gene encoding for the main endocannabinoid hydrolyzing enzyme.

#### **3. The Antiobesity E**ff**ects of** ω**-PUFAs**

DHA and EPA exert numerous beneficial effects, and they act as natural hypolipidemics, decrease risk of cardiovascular syndromes and could prevent the progress of insulin resistance and obesity [57].

Human dietary intervention trials suggest that fish oil (EPA and DHA) supplementation might decrease waist circumference. ω-3PUFAs decrease adiposity by numerous effects. For example, DHA and EPA start AMP-activated protein kinase (AMPK), which in turn actives FA β-oxidation in adipose tissue [5]. DHA and EPA are also promoting mitochondrial biogenesis, which can increase energy metabolism [58].

In rodents, DHA and EPA also raise FA oxidation in the small intestine and liver in in vivo experiments. DHA and EPA prevent hepatic lipogenesis in an AMPK and PPARα-dependent manner [5]. PPARs are transcription factors that form heterodimers with retinoid X receptors in the promoter regions of different genes implicated in glucose and lipid e metabolism [59]. PPARγ works as a main regulator of adipogenesis and checks numerous genes and adipokines in glucose and lipid metabolism. ω-3PUFAs work as ligands for PPARγ, and it has been observed that PPARγ plays a evident role in the capability of ω-3PUFAs, particularly DHA, to prompt M2 macrophage polarization and thereby decrease inflammation since these data are not detected in PPARγ-knockdown RAW264.7 cells [60]. The EPA- and DHA-mediated increment in FA oxidation and decrease in lipogenesis could be accountable for their anti-obesity actions [61].

In Figure 3, some signaling mechanism-mediating effects of ω-3PUFAs are summarized.

DHA and EPA may act as anti-inflammatory agents directly. EPA improves adipose tissue inflammation and decreases insulin resistance. The ameliorations in adipokines' profiles are characterized by rises in anti-inflammatory adipokines, such as adiponectin, and reductions in pro-inflammatory cytokines, such as IL-6, tumor necrosis factor-alpha (TNF-α), monocyte chemo attractant protein 1 (MCP-1), and plasminogen activator inhibitor 1 (PAI-1). EPA and DHA's action of normalizing plasma adiponectin concentrations appears to be largely responsible for their insulin sensitizing action. This favorable activity on adiponectin secretion seems to be PPARγ-dependent, because fish oil fails to raise plasma adiponectin in PPARγ-null mice [62].

Prostaglandins, eicosanoids with pro-inflammatory action, are secreted by adipocytes. AA-originated eicosanoids such as thromboxane A2 and PGE2 possess stronger inflammatory action than EPA-originated ones. Since EPA contends with AA for incorporation into cell membranes, it is possible that enhancement dietary EPA intake decreases synthesis of AA-originated eicosanoids. Indeed, EPA hinders AA-induced secretion of PGE2 from 3T3-L1 adipocytes in vitro [63].

**Figure 3.** Some signaling mechanisms mediating effects of ω-3PUFAs (AMPK: AMP-activated protein kinase; FFAR: free fatty acid receptor; IL-: Interleukin-; MCP-1: Monocyte chemo attractant protein 1; NF-kB: nuclear factor-κB; PGE2: Prostaglandin E2; PPAR: Peroxisome proliferator-activated receptor; PUFA: polyunsaturated fatty acid; TNF-α: Tumor necrosis factor-alpha).

A close link was observed between inflammatory markers, BMI, and body fat percentage. NF-κB, a key transcription factor in gene expression and cytokine inflammation is inhibited by ω-3 PUFAs. Studies in humans and in vitro have shown that ω-3PUFAs are involved in the reduction of cytokines, such as IL-1, IL-6, and TNF-α, whose concentrations are high in cases of obesity [64]. The ω-3PUFAs behave as agonists for numerous free fatty acid receptors (FFARs) typical of different cell types, involved in both the inflammatory response and energy homeostasis. Some unsaturated and saturated long-chain fatty acids can activate FFAR1 and FFAR4 [65]. For example, FFAR4 stimulation prevents lipopolysaccharide (LPS)-mediated release of inflammatory cytokines, such as TNF-α and IL-6 in the macrophage-type RAW264.7 cell line [64].

Current data show that reduction of inflammation is an active process. EPA and DHA-derived resolvins and protectins are key examples of inflammation resolution agonists. Experiments involving treatment with resolvins or transgenic restoration of protectins have shown a slowdown of the adipose tissue macrophage infiltration, and enhanced insulin resistance in rodents. Secretion of these mediators could be another mechanism by which DHA and EPA ameliorate the inflammation in adipose tissue [5].

Human clinic trials have been organized to evaluate the effects of the intake of ω-3PUFAs (using as food different types of fish with different contents of DHA and EPA) on the variation of composition and body weight, and also on the evaluation of the caloric content of food intake. Fish oil and fishes have also been used in dietetic interventions of different duration and with or without associated physical activity, to evaluate a possible weight loss.

Participant-reported diet diaries show evident decreases in fat, carbohydrate, and total caloric intake with ω-3PUFAs integration [66], but others reported no variation in energy intake [29,67]. Since most trials only indicated total caloric intake, the action of ω-3PUFAs integration on macronutrient and energy intake should be repeated in larger trials to conclusively establish the action of these PUFAs on weight reduction in humans. Weight reduction data appear more encouraging when ω-3PUFA

integration and calorie restriction are combined, but it is problematic to draw deductions due to the diversity of calorie limit programs in dissimilar trials. Combined ω-3PUFAs supplementation and calorie restriction compared to calorie restriction alone or replacement of saturated fatty acids determined a major amelioration in insulin resistance and reduction of TGs [68,69]. ω-3PUFAs could decrease body weight, thereby improving the metabolic profile through various mechanisms: alteration in adipokines release; modification of gene expression in adipose tissue; adipokine-mediated or adipokine-connected pathways; variation in carbohydrate metabolism; appetite suppression; rise in fat oxidation; intensification in energy expenditure (probably by thermogenesis); initiation of mechanisms related to muscle anabolism; and, lastly, epigenetic actions.

The adipose tissue increase in obesity happens via hyperplasia (augmentation in adipocyte number due to adipogenesis) and adipocyte hypertrophy (growth of adipocytes). Both ω-3 and ω-6 PUFAs can bind and/or control transcriptional factors that regulate genes implicated in pre-adipocyte differentiation. Principally, AA and its derivatives act as ligands for PPARγ and PPARδ to cause fat cell differentiation and quicken maturation by increasing lipoprotein lipase expression in vitro [70]. Importantly, concentrations of ω-6 and ω-3PUFAs in human subcutaneous tissue are associated with less adipocyte size; improved saturated FA concentrations lead to amplified fat cell size. Considering all these data, it is possible to propose that ω-3PUFAs stimulate adipogenesis and a healthy expansion of adipose tissue during positive energy balance, stimulating a metabolically healthy phenotype [70].

Numerous trials have observed that ω-3PUFAs modulate adipokine secretion. Obese individuals have high plasma leptin values indicative of leptin resistance. Weight-loss-connected to decrease in leptin could act on hunger and a lower metabolic rate and ultimately lead to weight regaining [71]. EPA integration reduces the decrease in blood leptin values, which happens during weight loss in obese women, proposing a potentially prominent role of EPA in weight loss conservation [72]. ω-3PUFA-mediated consequences on leptin are related to a various factors, such as energy balance and kind of diet, which could determine incompatible data anyway [29]. It has been suggested that the anti-inflammatory capacities of ω-3PUFAs integration cause a rise in adipocyte adiponectin synthesis and get better leptin sensitivity. This interaction could have a substantial influence on body weight control.

A study described evident sensations of fullness in the attendees who took higher ω-3 PUFA content food compared to those who took lower ω-3PUFA content meals both immediately and 2 h after eating the meal. [73]. Consequently, it is possible that an increase in the feeling of satiety after a meal rich in ω-3PUFA content can help weight loss by reducing the next food intake. Additionally, FFAR4 could mediate appetite reduction. ω-3PUFAs are agonists for FFAR4, which provokes the secretion of cholecystokinin, a hormone that is synthesized, is freed from the small intestine, and is related to appetite suppression [74].

Brown adipose tissue (BAT) is a particular fat that disperses excess energy into heat (non- shivering thermogenesis) through mitochondrial uncoupling protein 1 (UCP1). Current studies confirm the metabolic activity of BAT by revealing BAT as a crucial regulator in ensuring energy balance by rising thermogenic energy consumption. An important quantity of BAT is dose in healthy adults and most children and adolescents, but not in the obese adults, indicating that loss of operative BAT depots is a contributing obesity factor. Cold- and diet-induced thermogenesis mediated by UCPs in the presence of ω-3PUFA have been analyzed in some studies [75]. UCPs are inner mitochondrial proteins moving hydrogen ions across the mitochondrial inner membrane [76]. ω-3PUFAs enhance mitochondrial oxidative capability in skeletal muscle and WAT, probably through UCP-3 up-regulation, but not in liver or BAT. Nevertheless, because most trials were carried out at 20 ◦C, it is uncertain whether increment in mitochondrial oxidative capability is ω-3PUFA-mediated or cold caused. Mechanisms relating the role of ω-3PUFAs in probable induction of energy expense and reduction of body fat should be investigated further at different temperatures since thermogenic markers act even at 22 ◦C [76].

Control of lipid metabolism may change by ω-3PUFA type, and by fat store. For example, EPA is preferably aimed to β-oxidation, while DHA and DPA avoid catabolism and are stored

in tissues. Furthermore, hormone-sensitive lipase, and gene expressions of fatty acid synthase, phosphoenol-pyruvate carboxykinase, and lipoprotein lipase, in retroperitoneal fat, are reduced with DHA and mixed EPA/DHA integration but not with EPA integration alone [77]. Additionally, ultimately, ω-3 PUFAs control lipid metabolism, encouraging fatty acid oxidation and repression of lipogenesis, causing a positive lipid profile and adipocyte metabolism.

The determination of ω-6 and ω-3PUFAs composition in the cell membrane of red blood cells (RBC) represents a biomarker of dietary intake and endogenous metabolism; in addition, it is a precise way to perform estimated studies and clinical trials in order to value their effects in weight increase and obesity. Harris et al. led a prospective study to observe the link between baseline RBC membrane phospholipids of ω-3PUFAs, ω-6PUFAs, ω-6/ω-3 ratio, and *trans* FA with the variations in body weight and the risk of becoming obese or overweight during a mean of 10.5 years follow up in the NIH Women's Health Initiative Study. This prospective analysis provided a strongly suggestive sign that ω-3PUFAs in RBC membrane phospholipids are reversely connected, while *cis* ω-6, ω-6/ω-3 ratio and *trans* fatty acids are favorably related with weight gain [78].

#### *Recommendations for PUFAs Intake*

The recommended intake for ω-3PUFAs is based on governing body. The Dietary Guidelines for Americans recommend consuming about 230 g/week of fish, corresponding to approximately 250 mg/day of EPA and DHA [29]. This recommended intake corresponds to consuming fish twice weekly, including one serving of oily fish. The U.S. Food and Drug Administration claimed that levels up to 3 g/day are considered as safe, while other authorities suggested at up to 5–6 g/day [79].

However, in intervention studies reporting a favorable health effect, the intake of fish oils or their derivatives resulted in long chain ω-3PUFAs daily intakes well above those "suggested" 200 mg/day and ranged from 0.5 to 9 g/day. Consequently, this justifies readjustments of nutritional guidelines to an upper level. Governments (UK, Belgium, The Netherlands, France, New Zeeland, and Australia) and health organizations (American Heart Association, FAO/WHO, American Dietetic Association,) now advise dietary consumption for total ω-3 PUFA of 1.4 to 2.5 g/day, with EPA and DHA ranging from 140 to 600 mg/day depending on the authority issuing guidelines, FOA/WHO making a relatively low recommendation of 250 mg/day, the average being around 500 mg/day [80]. This means minimum of 2 intake of fish/week (30–40 g/day), including one of oily fish (tuna salmon, sardine, and mackerel). In the light of the literature and inter individual changeability in PUFA metabolism and requirement, perhaps the minimal EPA+DHA supplies for healthy adults should reach 0.5–1 g/day (2–4 servings per week of fish, half of oily fish); that is minimal intake proved to reduce obesity and, in general, metabolic syndrome [81], with a total serving of ω-3 PUFA of 5–6 g/day as found in ancestral nutrition to which our metabolism is best fit [82]. Such levels are met in the traditional Japanese diet as it contains 80–100 g fish and shellfish/d/capita [81,83].

Numerous studies have discussed the actions of ω-3 PUFAs integration on obesity, in both animal and human models, highlighting possible mechanisms for ω-3PUFAs in decreasing body weight, improving body composition and counteracting the contrary metabolic effects of obesity [28,84]. However, clearly, findings of prospective studies concerning the favorable actions of ω-3 intake on obesity are far from agreement. The manifest discrepancies may have arisen due to differing or inadequate methods of data collection on food intake (food frequency questionnaires), changes in cooking procedures and other unaccounted for lifestyle behaviors (exercise, etc.) from study to study and among diverse study populations [29]. Therefore it is not possible at present to decide the ω3 dose to hinder obesity, neither through food, nor through integration. The advice is to act through nutritional interventions. In general, it is recommendable to replace the intake of SFAs with PUFAs. The World Health Organization (WHO) prompts eating at least two servings of oily fish per week, which is rich in the ω-3PUFAs (DHA and EPA) [85]. International and national guidelines on healthy eating agree in advising the intake of ω-3, both marine and vegetable, with a diverse and balanced diet, containing foods in which they are naturally present. In subjects at cardiovascular risk and on a

diet low in these fatty acids, or in patients in secondary prevention, integration at diverse levels should be estimated with the specialized doctor [81]. Human intervention trials indicate potential benefits of ω-3PUFAs supplementation, especially when combined with energy-restricted diets or exercise, but more well-controlled and long-term trials are needed to confirm these effects and identify doses for antiobesity-action.

ω-6PUFAs are pro-inflammatory and commonly occur in poultry, eggs, corn, and most vegetable oils, and also in processed and fast foods. ω-6PUFAs have pro-inflammatory capability and are considered to be the counterpart of ω-3PUFAs, which are anti-inflammatory [18]. A high-fat diet and the Western dietary pattern feature high quantities of ω-6PUFAs and low amounts of ω-3PUFAs. Instead, a low-fat diet (e.g., traditional Japanese diet) is low in ω-6PUFAs and high in ω-3PUFAs [86]. Data show that a higher ratio of ω-6/ω-3PUFAs increases inflammation and the probability of chronic inflammatory syndromes, including cardiovascular disease, obesity, and nonalcoholic fatty liver syndrome [18]. Preclinical studies show that ω-6 PUFAs have a tumor-enhancing effect. In a recent Japanese cohort study, incorporating 38,200 women, ω-6PUFA intake was positively associated with breast cancer risk [81,87].

Many data discussed the importance of preserving a low omega–6/omega–3 ratio for decreasing inflammation. Decreasing the ω-6/ω-3 ratio seems to reduce the inflammatory response to a high-fat meal [88]. A stable ω-6/ω-3 is one of the most significant dietary factors in the inhibition of obesity: a lower ω-6/ω-3 ratio should be reputed in the management of obesity [89]. A high omega-6 fatty acid intake and a high ω-6/ω-3 ratio are connected with weight gain in both animal and human investigations, whereas a high ω-3 FAs intake reductions the risk for weight gain [89].

Several sources of information recommend that human beings evolved on a diet that had a ratio of ω-6 to ω-3PUFA of about 1/1; whereas today, Western diets have a ratio of 10/1 to 20–25/1, demonstrating that Western diets are lacking in ω-3PUFA related with the diet on which humans evolved and their genetic patterns were established [90,91].

Due to agribusiness and modern agriculture western diets enclose unnecessary levels of omega-6 PUFAs but very low levels of ω-3PUFAs, leading to an unhealthy ω-6/ω-3 proportion, instead of 1:1 that was during evolution [90]. It is thought that hominids' foods during the Paleolithic era were high in seafood and low in seeds and vegetable oils, which led to a ω-6/ω-3 proportion of about 1:1 [30,92,93]. ω-6PUFAs are related to the synthesis of pro-inflammatory mediators while omega-3 PUFAs produce less powerful inflammatory mediators and inflammatory resolving proteins, so manipulating this proportion may bring about helpful health outcomes.

A balanced ω-6/ω-3 FA ratio (1:1 to 2:1 is optimal) is vital for homeostasis and regular development throughout the lifespan [92,94,95]. However, there are significant genetic variables in fatty acid biosynthesis including desaturase 1 and desaturase 2, which encode rate-limiting enzymes for FA metabolism. Data connected to genotyping of the desaturase region analyzed in human populations show that present-day humans vary dramatically in their capability to produce long-chain PUFAs [89,96].

In clinical investigations and intervention trials it is indispensable that the background diet is precisely determined in terms of the ω-6 and ω-3 FAs content. Because the concluding concentrations of ω-6 and ω-3PUFAs are defined by both dietary intake and endogenous metabolism, it is important that in all clinical investigations and intervention trials the ω-6 and ω-3 FAs are precisely defined in the red blood cell membrane phospholipids [89].

Mice fed the lowest ω-6/ω-3 ratio had the lowest non-HDL (i.e., atherogenic lipoporteins) and inflammation (IL-6). Mice fed lower ω-6/ω-3 ratio diets also had less macrophage cholesterol increase and less aortic atherosclerotic lesions. The lowest ω-6/ω-3 ratio (1:1) diet led to the least atherosclerotic formation and the severity of atherosclerosis augmented as the ω-6/ω-3 proportion increased [97].

Using long-chain ω-3PUFAs to suppress low-grade inflammation may advantage numerous chronic syndromes such as atherosclerosis, rheumatoid arthritis, diabetes, dyslipidaemia, obesity and heart failure. The ingesting of ω-6 seed oils may have the contrary action [98].

The consequences of extreme ω-6PUFAs remain controversial: ω-6PUFAs have intrinsic cardiovascular protective actions, justifying the latest FAO/WHO recommendations on maintaining high ω-6PUFAs consumptions if ω-3PUFA ones are fulfilled [99]. However, ω-6PUFAs compete with ω-3PUFAs for processing to eicosanoids, thereby limiting synthesis of anti-inflammatory ω-3PUFA derived mediators [100]. Moreover, there are convincing proofs that a low ω-6/ω-3PUFA ratio is determinant for the inhibition of pathologies connected to the metabolic syndrome, as colorectal cancer [101].

Deduced from ancestral nutrition, in an ideal balanced diet, fat should represent no more than 20–30% of total energy intake amongst which 5–6 g/day of ω-3PUFAs with a great percentage of EPA+DHA and the ω-6-to-ω-3 proportion should average 1 [102,103]. To keep in with a developmental approach and with the epigenetic consequences of the diet, a proportion of ω-6/ω-3 around 1 in breast milk should serve as a bench mark to decide the correct dietary requirements during pregnancy, lactation, and infant feeding [81,104].

#### **4. Role of** ω**-3PUFAs in Inflammation and Colorectal Cancer Development**

A large body of literature highlights the importance of dietary intake for the risk and progression of chronic disorders including inflammatory and neoplastic disease [105]. Nowadays it is well known that inflammation is a predisposing factor for cancer capable to promote the insurgence of several types of tumors [106], and that an inflammatory microenvironment is an essential component of all tumors. During metastasis development it has been shown that microenvironment modulates the capability of tumor cells and cancer stem cells to evade the innate immune response and survive. The metastatic niche is a complex system including several cell types, as vascular, stromal, and above all inflammatory and immune cells, in addition to many other molecules which provide survival, immune surveillance protection and metabolic requirements. The interaction among all these factors determines metastatic dissemination [107].

Only a few of all cancers depend on germline mutations, while the majority is determined by somatic mutations and environmental factors. Most often cancer is caused by chronic inflammation such as chronic infections, tobacco smoke, inhalation pollutants (such as silica and asbestos), and dietary factors (some forms of cancer are linked to obesity) [108]. Furthermore, in some cases exogenous diet-derived miRNAs might substantially contribute to the pool of circulating miRNAs, regulating tissue homeostasis and interfering with human health [109].

CRC is a multifactorial disease caused by multiple genetic and environmental factors. These include the type of diet, the lifestyle, the intake of alcoholic beverages, smoking, obesity, genomic abnormalities, alterations in the signaling pathways, chronic activation of the inflammatory response, oxidative stress, dysbiosis, etc. These factors work by altering intestinal homeostasis [110]. In fact, despite still lacking extensive epidemiological studies to date, most cases of early-onset colorectal cancer (EO-CRC) arise sporadically and are attributable to environmental factors [111]. In recent years, due to the activation of preventive screening activated in the population aged ≥50 years, the incidence of CRC in Western countries has stabilized and even decreased. In contrast, the incidence of CRC among people under the age of 50 has increased in both Europe and the United States, thereby representing a major public health problem [112,113]. However, CRC remains the fourth leading cause of cancer death in the world dependent on a close relationship between inflammation and environmental factors [114].

The gastrointestinal tract is not only responsible for digestion and absorption of nutrients but also represents a powerful barrier against pathogens and toxins harmful to the individual. It also has an endocrine function responsible for maintaining the metabolic homeostasis of the whole organism. Since the intestine comes into direct contact with food, it is very sensitive to dietary factors which directly influence both its structure and function [115]. In fact, the crypts and intestinal villi are structurally influenced by external factors by changing their size in response to changes in the diet [116].

Since chronic inflammation has been shown to promote the onset of CRC in humans, ω-3PUFAs, due to their strong anti-inflammatory function, have been shown to be protective against colon cancer [117,118].

Some studies have been conducted on the efficacy of ω-3PUFAs in the prevention of CRC through the integration of purified EPA with DHA or fish oil (FO). These studies have shown the importance of ω-3PUFAs in inhibiting the uncontrolled proliferation of CRC cells both when administered in large quantities for short times (8–9 g of EPA+DHA/day for 2 weeks) and in smaller quantities for longer times (2.5–4 g of EPA+DHA/day for 3–6 months). However, the effect on the control of intestinal cell proliferation was not seen in patients with the same supplementation but with a high-fat basal diet and a low ω-3/ω-6PUFAs ratio. For this reason, the effectiveness of ω-3PUFAs depends on both the total lipid content and the ω-3/ω-6PUFAs ratio [119].

Many studies have been conducted in CRC models to explain the molecular mechanisms to the base of the anti-inflammatory and anti-neoplastic activity of ω-3PUFAs. First of all, ω-3PUFAs are incorporated in phospholipid membrane inducing an alteration in structure, fluidity and function of lipid rafts. These membrane changes influence the activity of membrane receptors leading at the inhibition of signaling pathways involved in the activation of pro-inflammatory molecules of cell survival and apoptosis [120–122]. Moreover, in CRC, ω-3PUFAs modulate inflammatory pathways, generating lipid mediators implicated in the resolution of inflammation including resolvine, protectin, and maresins [123].

The ω-3PUFAs exert their antitumor actions through different mechanisms, involving proliferation, apoptosis, and migration. Their affects involve COX-dependent or COX-independent mechanisms, and they act on different pathways such as Wnt/β-catenin and Hippo or by regulation of oxidative stress and the expression of Granzyme B. To date, there are numerous papers in the literature that describe different mechanisms of action of ω-3PUFAs in CRC, as summarized in Table 1.


**Table 1.** ω-3PUFAs target different molecular pathways acting on classical hallmarks of cancer, i.e., proliferation, apoptosis, and migration.

Abbreviations: COX: cyclooxygenase, GPR: G-coupled receptor, Gαs: G-protein alpha subunit, PKA: protein kinase A, EMT: epithelial mesenchymal transition.

Multiple molecular mechanisms causing an increased apoptosis of CRC cells depend on the action of ω-3PUFAs. First of all, ω-3PUFAs influence the redox state of the cells: indeed, there is a link between anti-tumor effects of ω-3PUFAs and oxidative stress. PUFAs may induce an increase in apoptotic potential of CRC cells by enhancing the concentration of intracellular reactive oxygen species (ROS), inducing an elevated cancer cells apoptosis by the loss of mitochondrial membrane potential, ROS generation, activation of caspase 3 and 9, and by an increase of Bax/Bcl2 ratio [124].

Another important anti-inflammatory mechanism involves COX, a major player in inflammation. COX hyperactivation in the CRC, induces in turn the production of PGE2, a powerful pro-inflammatory and pro-carcinogenic agent [125]. ω-3PUFAs exert their anti-inflammatory role by modulating COX activity. In this respect, EPA, acting as an alternative substrate for COX-2, induces a switch in production from pro-tumorigenic PGE2 to three series PGs (PGE3) that abrogate the antiapoptotic activity of PGE2 in CRC cells [126]. However, the anti-cancer mechanism of ω-3PUFAs in CRC could also be explained

by a COX2-indipendent mechanism. Indeed, DHA and EPA inhibit the proliferation and induce the apoptosis of CRC cells in vitro and in animal models. At molecular level, involvement of the Hippo pathway, cytoplasmic retention of phosphorylated YAP by GPRs (GPR40 and GPR120)-Gαs-PKA cascade has been reported [127]. Moreover, the GPR120 is expressed on macrophages and regulates their polarization reducing inflammation [128].

In addition to apoptosis, ω-3PUFAs can also influence proliferation and migration capability of CRC cells. In colorectal cancer, the Wnt-β-catenin signaling pathway is the key regulator of tumor development, and alterations in this cellular signaling pathway can be found in most patients [129]. It has been shown that dietary ω-3PUFAs are able to inhibit significantly intestinal polyp growth in mice, correlating with the ECS described in Section 2.1. In fact, CB1 up-regulation reduces β-catenin and its transcriptional target c-myc, both involved in regulation of cell proliferation. In CRC patients, cancer tissue shows a significant inhibition of CB1 expression levels, compared to adjacent normal tissue, demonstrating that the "protective" action of endocannabinoids via CB1 is lost in the tumor [130]. Moreover, D'Eliseo and colleagues have studied the effect of DHA on migration of CRC cells and demonstrated that DHA inhibits Granzyme B expression, reducing CRC cells capacity to undergo epithelial mesenchimal transition (EMT) and invade matrigel [131].

Finally, ω-3PUFAs regulate the expressions of genes involved in inflammation and colon cancer development also through epigenetic modifications [132–134]. In fact, more recently, ω-3PUFAs have been attributed the ability to influence the epigenetic regulation of genes involved in the polarization of macrophages, negatively regulating the colorectal carcinogenesis; however, the interesting topic is not fully understood [135].

Taken together, although the many different mechanisms, ω-3PUFAs play anti-inflammatory and anticancer effects acting on the classic hallmarks of cancer, i.e., cell proliferation, apoptosis and migration.

To date, one of the most important problems in the treatment of tumors, including sporadic colorectal cancer, is the development of resistance to anti-tumor treatments and tumor relapse that can be related to self-renewing of cancer stem/stem-like cells (CSC/CSLC) within a tumor mass. Many groups have studied the effects exerted by ω-3PUFAs on cancer stem-like cells.

With the immunophenotyping of CSLC, the anti-CD133 antibody was found to be effective for isolating a population of colon cancer cells that retained the properties of stem cells (CSLC), while anti-cytokeratin 20 (CK20) and anti-Mucin-2 (MUC2) were specific epithelium colonic differenziation markers. EPA treatment induces an increase in of CK2 and MUC2 and an inhibition of CD133 expression. This means that the EPA could induce a more differentiated state of most cancer cells and could trigger the reduction of the stem state of the CSLC, as demonstrated by the reduction of the expression of the CD133 marker [136].

Moreover, Yang and colleagues showed an antiproliferative and proapoptotic effect on the dedifferentiated SW620 colon cell line, treated with DHA and EPA [137].

Additionally Sam et al., evaluating the effects of DHA and EPA treatments on LS174T cells, a model for colorectal cancer initiating cells with stem cell-like properties, demonstrated that ω-3PUFAs induce cell growth inhibition and promote cell death by down-regulating survivin expression and activating caspase-3 [138].

The effect of ω-3PUFAs on CSLC may be an important goal for cancer therapy and will constitute an interesting challenge for future studies. Anyway, the anti-tumor activity of ω-3PUFAs, shown through multiple mechanisms, suggests that they could have an important therapeutic role in the management of CRC.

#### **5. Conclusions**

Obesity is a preventable disease that can be treated through proper diet and exercise. A balanced ω-6/ω-3 ratio 1–2/1 is an important dietary factor in the prevention of obesity, along with physical activity. Different pro- and anti-inflammatory properties are exerted by ω-6 and ω-3PUFAs themselves and by their derivatives, such as endocannabinoids, lipid mediators deeply involved in the control of many biological functions, including the inflammatory response and the central and local control of food intake and energy homeostasis. Therefore, appropriate dietary intervention has primarily relevance in the prevention and the treatment of obesity in that it maintains the efficiency of key signaling pathways and avoids long term/chronic inflammatory states.

Inflammation is a predisposing factor for cancer, CRC included, with ω-3PUFAs exhibiting anti-cancer properties, once again confirming the need for a balanced ω-6/ω-3 ratio for health preservation.

The discovery of cancer stem cells offers a new perspective in cancer therapy. Since CSCs contribute to cancer onset and relapse after conventional therapy, they can represent a unique fundamental therapeutic target to completely cure cancer. Thus, the effect of ω-3s on CSLC may be an important goal for cancer therapy and will constitute an interesting challenge for future studies. Anyway, the anti-tumor activity of ω-3s, performed through multiple mechanisms, suggests that they could have an important therapeutic role in the management of CRC.

**Author Contributions:** Conceptualization, M.L.M., S.D., and R.M.; writing—original draft preparation, M.L.M., S.D., and R.M; supervision, R.M.; funding acquisition, R.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Prin-Miur 2017 to R.M. project code 20175MT5EM.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Abbreviations**



#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Review*

### **The Crosstalk between Prostate Cancer and Microbiota Inflammation: Nutraceutical Products Are Useful to Balance This Interplay?**

**Felice Crocetto 1,**†**, Mariarosaria Boccellino 2,**†**, Biagio Barone 1, Erika Di Zazzo 3,\*, Antonella Sciarra 4, Giovanni Galasso 2, Giuliana Settembre 2, Lucio Quagliuolo 2, Ciro Imbimbo 1, Silvia Bo**ff**o 5, Italo Francesco Angelillo <sup>6</sup> and Marina Di Domenico 2,5**


Received: 31 July 2020; Accepted: 26 August 2020; Published: 31 August 2020

**Abstract:** The human microbiota shows pivotal roles in urologic health and disease. Emerging studies indicate that gut and urinary microbiomes can impact several urological diseases, both benignant and malignant, acting particularly on prostate inflammation and prostate cancer. Indeed, the microbiota exerts its influence on prostate cancer initiation and/or progression mechanisms through the regulation of chronic inflammation, apoptotic processes, cytokines, and hormonal production in response to different pathogenic noxae. Additionally, therapies' and drugs' responses are influenced in their efficacy and tolerability by microbiota composition. Due to this complex potential interconnection between prostate cancer and microbiota, exploration and understanding of the involved relationships is pivotal to evaluate a potential therapeutic application in clinical practice. Several natural compounds, moreover, seem to have relevant effects, directly or mediated by microbiota, on urologic health, posing the human microbiota at the crossroad between prostatic inflammation and prostate cancer development. Here, we aim to analyze the most recent evidence regarding the possible crosstalk between prostate, microbiome, and inflammation.

**Keywords:** prostate cancer; microbiota; nutraceutical compounds

#### **1. Introduction**

Prostate cancer (PCa) is the second most commonly diagnosed malignancy in men and the fifth leading cause of tumor-associated death worldwide [1].

Global estimations are approximating 800,000 new PCa cases and 300,000 deaths per year [2], and this condition poses a significant health concern in the future due to the gradual aging of the

population. Genetics, family history, African descent, advanced age, diet, and environment are well-established risk factors for PCa development. However, the relevant pathways accounting for PCa development are not fully clarified [3–5]. The role of androgenic stimulation and the deficit of apoptosis of prostate cells are well-known explanations regarding the incidence and progression of PCa. Recent studies have also hypothesized a crucial role of microenvironment, infections, inflammation, and cytoskeletal changes induced by steroid integrating signals [6,7], influencing patients' outcomes and the rationale for the immunological treatment of PCa [8–12].

Chronic inflammation is a prominent contributing factor to the benign and malignant prostatic growth; however, the potential stimulus that induces or maintains this chronic inflammation remains poorly characterized [13]. Inflammation, sex hormones, and many other factors (e.g., infections, diet, physical activity, drugs), are known to affect the microbiota. The microbiota is a complex community composed of fungi, parasites, bacteria and viruses living within the human body. Microbiota components interact with each other and with the host, impacting, eventually, the overall human health. The purpose of this study is to summarize and analyze the most recent evidence regarding the possible crosstalk among prostate, microbiota, and inflammation.

#### **2. Prostate and Chronic Inflammation**

The role of inflammation in the carcinogenesis of a solid tumor is an accustomed aspect [13]. In fact, two key inflammatory cytokines, IL-6 and IL-2, have been convincingly implicated in prostate cancer pathogenesis. Inflammation may also contribute to impairing immune surveillance mechanisms, which are partially mediated by NK cells [14,15]. Repeated tissue damage and regeneration produce highly reactive nitrogen species (RNS) and oxygen species (ROS), which are responsible of cancer development and progression [16–19]. The underlying biological mechanism relies on DNA modifications of cells caused by this continuous process of damage and repair [20–22]. However, if there is a strong and proven connection between solid tumors and inflammation, the role of this condition in PCa development is still debatable and under revision. Different studies have suggested how chronic prostatitis could induce proliferation of stromal and glandular cells in response to ROS production, eliciting general tissue damage, and vascular injury [23,24]. ROS, moreover stimulate NF-kβ and TNF-α pathways by activating their proper kinases [25]. The morphological modification in the prostate tissue, associated with chronic and acute inflammation, is a glandular atrophy with hyperplasia called proliferative inflammatory atrophy (PIA) [26]. Up to 40% of PIA lead to the transition to a high-grade prostatic intraepithelial neoplasia (PIN), a precursor of PCa. Although some evidence of molecular changes has been observed in PIA, no certain clonal genetic alterations have been found in this condition [27]. However, genes such as NKX3.1 and CDKN1B have been shown to be downregulated in PIA, as in PIN and PCa, while the increased transcription of Hsp27 and PRDX6 could promote processes leading to tumorigenesis [28]. Several PCa susceptibility genes, such as MIC1, RNASEL, MSR1, PON1, TLR4, OGG, BRCA2 and CHEK2, are involved in prostate carcinogenesis and in other critical processes, as a host response to steroids, infection, inflammation and oxidative stress [29,30]. Furthermore, despite significant changes in inflammatory cellular infiltration between prostatitis, benign prostatic hyperplasia (BPH) and PCa have been found; the role of innate and adaptive immunity has not been completely cleared [13] (Figure 1). Chronic inflammation could have, moreover, a significant effect on cancer progression and metastatic invasion due to neo-angiogenesis and activation of epithelial–mesenchymal transitions (EMTs) [31,32]. These biologic and pathogenic processes are correlated to various molecules defined as biomarkers/indicators of normal, or pharmacologic, responses to a therapeutic intervention [33]. The control of these phenomena triggers pathways, as migration, proliferation, cell growth, apoptosis, and adhesion through various downstream effectors. The first key element that regulates cell proliferation, migration, and invasion in PCa is p85αPI 3Kinase [34–38]. Evidence from the literature supports the role of angiogenesis in human cancer progression, including PCa. The vascular endothelial growth factor (VEGF) is a potent angiogenic factor [39,40]. Several miRNAs, functioning as tumor suppressors or oncogenes

are deregulated in prostate tumorigenesis. miRNA dysregulation progress has a key role in prostate cancer [41]. Anti-VEGF therapy and combined chemotherapy treatments trigger apoptosis in cancer and, in particular, in prostate cancer [42–50].

**Figure 1.** Chronic prostatitis and immune cell infiltration. Outlined in blue are aggregates of lymphocytes, plasmacells and istiocites, which surround damaged glands (black arrows). In red circles, multinucleated giant cells are outlined.

Different cancer types (i.e., lung cancer, pancreatic cancer, glioblastoma, meningioma, myeloma, and myeloma) are characterized by distinct patterns revealed by corona composition, constituting a "fingerprint" for each cancer type [51–54].

Classically, the peripheral zone of the prostate gland is a common site of PCa development, while the transitional zone is mostly affected by benign prostatic hyperplasia [55]. However, in about 20% of cases, the two conditions subsist in the same zone and, despite different pathogenic pathways, several well-established epidemiologic studies confirm that both conditions are hormone-dependent and could be associated with a previous chronic prostatic inflammation [56]. A certain degree of inflammation is almost always present when prostate specimens are sampled: the REDUCE trial demonstrated on 8224 men that, indeed, 77.6% of biopsies are positive for some grade of inflammation, with the majority (>80%) showing a mild chronic inflammation [57]. To further support these findings, men diagnosed with prostatitis have an increased risk of developing, in the future, PCa compared to those without any grade of prostate inflammation. Specifically, 18% of those patients will develop PCa [58]. Chronic and acute inflammation is also frequently found in prostate tumor specimens obtained from prostatectomies and transurethral resections [59]. A study conducted by Daniels et al. on 5821 men >65 years old reported a positive association between a previous history of prostatitis and PCa (OR 5.4, 95% CI = 4.4–6.6) [60]. Similarly, Cheng et al. showed that protracted prostatitis symptoms could significantly increase the odds of PCa in 68,675 men (RR 1.3, 95% CI = 1.10–1.54) [61]. In addition, Dennis et al. reported, in a meta-analysis of 11 case-control studies, the evidence of a statistically significant risk of developing PCa in patients with a previous history of prostatitis (OR 1.6, 95% CI = 1–2.4) [62] and analogous results were found by a similar meta-analysis on 20 case-control studies (OR 1.50, 95% CI 1.39–1.62) [63]. Finally, a recent and wide meta-analysis by Perletti et al. reported, in 422,943 patients, a significant association between PCa and previous prostatitis (OR 1.83, 95% CI = 1.43–2.35) [4]. However, despite those data, the real impact of chronic inflammation on prostate carcinogenesis has been challenging to define. In particular, it is not easy to estimate the real

incidence of prostatitis due to the asymptomatic majority of cases (5–10%) [64]. Moreover, evidence that seems to show an increased risk for acute prostatitis rather than for chronic prostatitis, is influenced by the same potential detection bias.

#### *Etiology of Prostate Chronic Inflammation*

The etiology of chronic inflammation preceding PCa development remains unknown, however, infections and chemical trauma are often correlated to chronic inflammation.

Several putative etiological agents have been identified, from the xenotropic murine leukemia virus related-virus XMRV to different strains of bacteria [58,65–67]. Several studies support the potential role of infectious agents in PCa etiology with evidence that up to 87% of PCa patients show microbial DNA in their prostate [68,69]. However, if no clear association had been shown with HPV or other sexually transmitted viruses, men with previous gonorrhea or syphilis infections had a 60% increased risk of developing PCa [70]. A study based on animal models reported a mutagenic activity of inflammation caused by *Escherichia coli* in the prostatic gland, with the induction of epithelial hyperplasia, an increased tendency to apoptosis, and somatic mutations [71]. Moreover, the presence of an induced prostatic infection with *Escherichia coli*, in addition to the consumption of a diet enriched with a cyclic amine, the 2-amino-1-methyl-6-phenylimidazo [4,5-b]pyridine (PhIP) (a well-known prostatic carcinogen in rodents), further increased the risk of PCa development in mice with a marked drop in survival rate compared with PhIP-alone-treated animals, thus suggesting chronic inflammation as an enabling characteristic of PCa [72]. Cai et al., reported a significant increase in Gram-positive strains in patients with chronic prostatitis and a successively diagnosed PCa [73], while other significant associations between cancer development and infection were shown also for *Mycoplasma hominis* [74] and *Trichomonas vaginalis* [75,76]. In particular, previous *Trichomonas vaginalis* infection could create a favorable microenvironment, promoting PCa cell proliferation and invasiveness (activating the epithelial–mesenchymal transition), in addition to an increased overall inflammatory state of the gland [77]. Twu et al. reported, in fact, how *Trichomonas vaginalis* secretes a protein (TvMIF), which is 47% similar to the human macrophage migration inhibitory factor (HuMIF), which is reported to be elevated in PCa [78]. *Propionibacterium acnes*, which is frequently isolated in prostate tissue, has also been thought to have an influence on the development of PCa due to the association with reported histological inflammation in prostate-derived tissue models and prostatectomy specimens [79]. To further outline the role of *Propionibacterium acnes* in prostate carcinogenesis, Ugge et al. retrospectively analyzed the association between the presence of acne vulgaris during adolescence and the occurrence of PCa in 243,187 men for a median follow up of 36.7 years; 1633 of those patients developed PCa, reporting an adjusted OR of 1.43 (95% CI = 1.06–1.92) [80]. However, a recent meta-analysis by Zhang et al. did not find a significant association between acne and PCa, questioning this relation [81]. An EPICAP study reported instead the association between sexually transmitted and urinary tract infections and PCa, with an increased risk of developing this malignancy in patients with a previous history of prostatitis (OR 2.95, 95% CI = 1.26–6.92) and in patients who did not assume non-steroidal anti-inflammatory drugs (OR 2.00, 95% CI = 1.37–2.91) [82]. To further support the role of chronic inflammation in increased PCa risk, St. Hill et al. showed how EBV, HIV, HBV, HCV, or HSV chronic infections were associated with an increased risk of occurrence of PCA. Similarly, the risk was also increased in men with other chronic inflammatory diseases or conditions such as osteoporosis, diabetes mellitus, arthritis, or cardiovascular disease; however, currently, no inflammation marker could be associated with a higher risk of PCa development [83].

#### **3. Microbiota in Urological Disease**

#### *3.1. Urinary and Prostate Microbiota*

Commensal microorganisms colonize barrier surfaces of all multicellular organisms, coevolving and adapting with the host for more than 500 million years. As result, the commensal microbiota affects many processes of their hosts via biologically active molecules, playing critical roles in human diseases, in particular cancers and autoimmune conditions, influencing the innate and adaptive immune response [84]. The discovery of communities of bacteria in the genitourinary tract and their role in urologic diseases has introduced novel factors and implications in the pathophysiology of these conditions. The advent of such molecular-based methods as the quantitative real-time PCR and amplification of 16S rRNA for the identification and characterization of microbial populations has permitted the discovery of previously unrevealed microbial populations. Historically, the bladder, and generally the urinary tract, has always been considered sterile, however, recent studies have revealed important evidence of the presence of microbes in bladders of patients without clinical infection [85,86]. Human urinary microbiota characteristics depend on the age, gender, and disease status of individuals [87–92], and understanding its role in urological diseases is of particular interest. Moreover, novel molecular methods have made it possible to characterize the bladder microbiota formed by *Burkholderia cenocepacia* and different strains of *Lactobacilli* in urologic chronic pelvic pain syndrome (UCPSS), which was considered to be defined as "the absence of identifiable bacterial infection" [93]. Particularly interesting is that, in the same condition, an increased rate of *Lactobacilli*, compared to the remaining flora, was instead revealed in the urine of patients with Interstitial Cystitis (IC) [94]. Furthermore, *Lactobacillus casei* and *Lactobacillus rhamnosus* could also have interesting applications in the treatment of bladder cancer, demonstrating a decreasing effect on rates of metastasis and recurrence due to an enhanced recruiting of natural killer cells, both in vitro and in vivo [95]. Accordingly, different studies have hypothesized a link between prostate microbiota and pro-inflammatory bacterial species. In 2016, Mandar et al. reported a lower rate of *Lactobacilli* in patients with chronic prostatitis, while Shoskes et al. reported, for the same condition, higher rates of *Clostridia* and *Bacteroides* compared with controls [96,97]. In 2015, Yu et al. described how bacterial strains present in prostatic secretions, seminal fluid and voided urine are different among patients with BPH and PCa. Specifically, there are lower rates of *Eubacterium* and *Defluviicoccus* and higher rates of *Bacteroidetes* in patients with PCa, hypothesizing the role of certain bacteria in the induction of chronic inflammatory states, with enhanced production of factors favoring tumorigenesis [98]. An analogous study conducted on 135 PCa patients by Shrestha et al. in 2018, reported an increased presence of *Anaerococcus lactolyticus* and *obesiensis*, *Streptococcus anginosus*, *Varibaculum cambriense*, *Actinobaculum schaalii*, and *Propionimicrobium lymphophilum*. All patients were previously diagnosed with urinary tract infection caused by *Enterobacteriaceae*, which were instead more abundant in patients with BPH [99]. Similar conclusions were reported by Alanee et al., confirming the possible association between urinary and fecal microbiota with PCa after examination of prostate biopsies, which were characterized by higher rates of *Streptococcus anginosus*, *Anaerococcus lactolyticus*, and *Varibaculum cambriense* [100]. Analogously, Bhudia et al., reported increased rates of *Staphylococcus epidermidis*, *Streptococci*, *Corynebacterium amycolatum*, *Peptoniphilus harei*, and *Fusobacterium nucleatum* in prostate secretions of PCa patients [101]. Cavarretta et al. reported an abundance of *Propionibacterium* spp. and *Staphylococcus* spp. in 16 tumoral and peritumoral prostatectomy specimens [102]. Similarly, Feng et al. examined 65 radical prostatectomy specimens, reporting an increased rate of *Escherichia coli*, *Propionibacterium acnes*, *Pseudomonas* spp. and *Acinetobacter*; in particular, *Pseudomonas* has a gene expression profile that strongly correlates with human small RNA's profile, and that could be also related to metastasis [68]. Moreover, the same authors identified increased bacterial content (especially *Escherichia* spp. and *Acidovorax* spp.) in prostate specimens of African men, which were also associated with elevated tumor hypermutation, suggesting the possibility of a bacterially driven oncogenic transformation [69].

#### *3.2. Gut Microbiota*

The role of microbiota in urological diseases and PCa is, however, not limited to bacteria related to the urinary tract. Modification of the gut microbiota could modify the risk of incurring PCa and be influenced by the tumorigenesis process itself [103] (Table 1). Liss et al. reported significant differences in bacteria obtained via rectal swab between PCa and healthy patients, with an increase in certain genera such as *Bacteroides* and *Streptococci* and impoverishment of bacteria related to folate and biotin production [104]. Golombos et al., similarly, confirmed the abundance of *Bacteroides* in PCa patients and reported an increased presence of *Faecalibacterium prausnitzii* and *Eubacterium rectalie* in BPH patients [105]. Potential alterations of gut microbiota could influence, both directly and indirectly, prostate health via bacterial metabolites, and influence the enteric endocrine system [106]. Multiple studies have shown that gut microbiota also modulates the response to chemotherapy acting on the translocation, immunomodulation, metabolism, and enzymatic degradation of drugs [107]. This consideration, moreover, is valid also for androgen axis-targeted therapy in PCa treatment, which is influenced in its clinical response and antitumoral efficiency by gut microbiota. Conversely, androgen axis-targeted therapy enhances *Bacteroides* and *Streptococci* rates in the gastrointestinal tract while lowering overall bacterial diversity [108]. Besides, an analysis of the fecal microbiota of healthy volunteers and PCa patients by 16S rDNAsequencing, showed a greater abundance of *Akkermansia muciniphila* and *Ruminococcaceae* spp. in the microbioma of patients treated with oral androgen receptor axis-targeted therapies such as enzalutamide, bicalutamide and abiraterone acetate [109]. Finally, there are suggestions that butyrate, an anti-inflammatory micronutrient produced by *Faecalibacterium prausnitzii* and *Eubacterium rectale*, could be implicated in one of the pathways for the prevention of PCa, although further studies are required [105].




#### **Table 1.** *Cont.*

Abbreviations: PCa (prostate cancer), BPH (benign prostatic hyperplasia), ADT (androgen deprivation therapy), UCPSS (urologic chronic pelvic pain syndrome), IC (interstitial cystitis).

#### **4. Nutraceutical Aspects in the Interplay between Prostate and Microbiota**

#### *4.1. Unsaturated Fatty Acids*

Olive oil and unsaturated fats, high vegetable consumption, fruit intake, and allium vegetables, typical aspects of the Mediterranean diet, were related to a decreased risk of several cancer types. In particular, countries following the Mediterranean Diet have lower PCa incidence and mortality compared to other European regions. However, there are few studies that have assessed the effect of the Mediterranean diet on PCa incidence. Further large-scale studies are required to clarify the effect of the Mediterranean diet in order to establish the role of this diet in the PCa prevention [110,111]. PCa has a well-known association with food and, in particular, with fat intake; moreover, there is a relationship between PCa and gut microbiota that changes based on the diet [112]. A low-fat diet and/or intensive exercise involves changes in serum hormones and growth factors in vivo, which could reduce growth and induce apoptosis of LNCaP prostate tumor cells in vitro [113]. Low-fat diet-fed mice show significantly lower levels of prostate-specific serum antigen (PSA), insulin and Igf1 mRNA levels compared to mice with a high-fat diet, as well as a delayed tumor-growth rate in LAPC4 xenografts [114]. A high-fat diet induces, in fact, lipid accumulation in PCa and promotes metastasis via abnormal sterol regulatory element-binding protein (SREBP)-dependent lipid metabolism [115]. Several epidemiological studies suggest that an increased intake of saturated fatty acids and a sedentary lifestyle decreases the survival rate of PCa patients, whilst unsaturated fatty acids and physical activity reduce the risk of PCa [116,117]. In recent years, *n*-3 fatty acids, eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), present in fish oil, have been found to influence cancer cell proliferation. EPA and DHA were, moreover, effective in decreasing the proliferation, invasion, and migration of prostate PC3 cancer cells as well [118]. As known, sex hormones also play an important role in the development and progression of PCa. In prostate-specific Pten-/-mice, the reduction in serum cholesterol lowers intraprostatic androgens and suppresses tumor progression, although it does not change the incidence of PCa [119]. In transgenic mice, the consumption of high amounts of unsaturated fatty acid ω-3, produces a significant slow-down of prostate tumorigenesis by affecting estradiol, testosterone, and androgen receptor levels, suggesting a specific role of unsaturated fatty acids in the regulation of sex hormones, which may be the basis of fat-induced PCa progression [120].

#### *4.2. Carnitine*

Carnitine, and in particular its acetylated derivative, Acetyl-l-Carnitine (ALCAR) is involved in mitochondrial membrane trafficking in catabolic and anabolic pathways. Several studies have documented the antioxidant and scavenger activity of this compound, utilized in clinical settings related to disorders where the oxidative stress acts as a promoting factor (e.g., diabetes, Alzheimer's disease, and other neurometabolic disorders) [121,122]. ALCAR reduces PCa cell viability and induces apoptosis; moreover, ALCAR impairs the adhesion, invasion and migration of PC3, DU145, LNCaP, and BPH cells, eliciting a decreasing effect on TNF-α and other proinflammatory cytokines, such as

IL-6, CCL2 and CXCL12 [123]. Besides, ALCAR was able to limit inflammatory angiogenesis, in vitro and in vivo, downregulating the VEGF/VEGFR2, CXCL12/CXCR4, and FAK pathways [124].

#### *4.3. N-acetylcysteine (NAC)*

*N*-acetylcysteine (NAC) is an exogenous antioxidant primarily used as a mucolytic agent and as an antidote of acetaminophen toxicity. Its effects on increasing glutathione levels and scavenging free radicals pose NAC as a powerful antioxidant. The association of NAC with phenethyl isothiocyanate (PEITC) and sulforaphane (SFN), two compounds present in cruciferous vegetables (cauliflower, cabbage, and broccoli) inhibit LNCaP and DU145 cell growth in a dose-dependent manner, increasing p21, a potent inhibitor of cyclin-dependent kinases mediating cell replication, up to apoptosis. Besides, SFN-NAC reduces PSA and the expression of the androgen receptor [125,126]. NAC alone inhibits the growth of PC3 cells suppressing the transcription of nuclear factor (NF)-κB, while increasing Cyr61 levels and activating the Erk pathway [127]. Finally, NAC shows a significant anti-migration and anti-invasion activity on DU145 and PC3 cells, limiting the metastatic ability of those cells [128].

#### *4.4. Monoterpenes*

Terpenoids are natural constituents of plants and animals. The most common form occurs as monoterpenes, components of essential oils of herbs and spices. D-Limonene, the most abundant monoterpene present in orange, lemon, and peppermint essential oil, has been shown to inhibit PCa cell growth via Erk pathway activation and the induction of WAF1 and p21 [129]. Geraniol, another monoterpene found in geranium and citronella plants, inhibits tumor cell growth via the induction of apoptosis in PC3 cells, activating caspase-3, reducing Bcl-2 expression and increasing Bax and BNIP3 levels. Besides, geraniol has been found to inhibit AKT-mTOR signaling without influencing mitogen-activated protein kinase (MAPK) activity [130]. A thyme honey component, the trihydroxy ketone E-4-(1,2,4-trihydroxy-2,6,6-trimethylcyclohexyl)-but-3-en-2-one exerted significant apoptotic activity in PC3 cells, through a reduction in NF-κB activity and IL-6 secretion [131].

#### *4.5. Polyphenols*

Polyphenols are widely studied for their beneficial effects on human health, particularly in cancer prevention. Several studies associate, in particular, catechin and isoflavone with beneficial effects on PCa. The epigallocatechin-3-gallate (EGCG), the most common catechin in green tea (>50% of the total polyphenol content), shows a great physiological activity: EGCG arrests cell growth in the G0/G1-phase and induces apoptosis in both androgen-sensitive and insensitive human PCa cells [132]. Moreover, EGCG, in both androgen-sensitive and insensitive human PCa cells, attenuated the effects of arachidonic acid (AA) in increasing cell growth and prostaglandin E2 levels by reducing the concentration of the enzyme cyclooxygenase 2 (COX-2) [133]. EGCG also acts through different mechanisms in order to arrest cell cycle and induce apoptosis, in fact in 12-week-old TRAMP mice, contrary to 28-week-old mice, it suppressed PCa development at an early stage after oral intake of EGCG by regulating IGF-1-related signaling and COX-2 levels [134]. Green tea has, therefore, an inhibitory effect on PCa tumorigenesis when assumed in large quantities. Kurahashi et al. examined the relationship among green tea consumption and PCa risk, in a large-scale prospective study of 49,920 Japanese men, reporting how subjects who drank five or more cups of green tea each day had a lower risk of advanced PCa than those who drank less than one cup per day (RR 0.52, 95% CI = 0.28–0.96) [135]. More recently, a meta-analysis on ten large studies on the incidence of green tea and PCa has shown how the risk of PCa decreases in a dose-dependent manner, with a significant reduction in the risk for subjects who drank more than seven cups a day (RR 0.81, 95% CI = 0.67–0.97 for 7 cups/day; RR 0.74, 95% CI = 0.59–0.93 for 9 cups/day; RR 0.56, 95% CI = 0.35–0.92 for 15 cups/day) [136]. Isoflavones also play an important role in the prevention of PCa, with a reduction in PCa risk related to the intake of soy isoflavone [137]. Soy isoflavones, having a structure similar to 17β-estradiol, can bind to the estrogen receptor (ER), behaving as phytoestrogens with a binding affinity and transcriptional activity stronger

on ER-β than on ER-α and thus having more likely estrogenic effects in prostate tissue, which expresses higher levels of ER-β. Genistein, another isoflavone contained in fava beans, soy, and coffee, induces apoptosis of PC3 cells by suppressing NF-κB via the AKT signaling pathway [138]. In DU145 cells, genistein, EGCG, and Silymarin, a flavonolignan contained in Cardus marianus, induced the inhibition of erbB1 membrane receptor activation caused by TGFα, provoking a dose-dependent inhibition of cell growth [139]. In addition, EGCG could induce apoptosis in LNCaP cells by two pathways: the first acted on the stabilization of tumor suppressor gene p53 and on the reduction in MDM2 protein expression; the second was related to the negative regulation of NF-κB activity, leading to a decreased expression of the anti-apoptotic protein Bcl-2 [140]. In TRAMP mice, food genistein reduced PCa development in a dose-dependent manner [141]. Parallel studies in TRAMP-FVB mice showed that a low-dose genistein diet (250 mg/kg) promoted PCa growth and metastasis compared to control and a high-dose genistein diet (1000 mg/kg), showing a biphasic effect of isoflavones on PCa [142]. Paller et al. found that an increase in quercetin intake, another well-known isoflavone contained in capers, leads to a reduced risk of PCa, in African-Americans with vitamin D deficiency, while Sun et al. showed that its use, associated with metformin, inhibits the growth, migration, and invasion on PC3 and LNCaP cells by inhibiting the VEGF/AKT/PI3K signaling pathway [143,144]. Similarly, fisetin has been suggested to act as a dual inhibitor on PI3K/AKT and mTOR metabolic pathways in PCa cell lines. In addition, this compound could be used, alone or as an adjunctive drug in the chemotherapeutic treatment of PCa [145]. In two different prostate cancer cell lines, androgen-sensitive (LNCaP) and androgen-independent (DU145), cyanidin-3-O-beta-glucopyranoside (C3G), the most abundant anthocyanin in the diet, produced anti-proliferative effects through the activation of caspase-3 and the induction of p21 protein expression. Besides, treatment with C3G increased the levels of tumor suppressor P75 NGFR, indicating a possible role of C3G in the acquisition of a normal-like cell phenotype. C3G may, therefore, be considered a new therapeutic agent with both anti-proliferative and pro-differentiation properties [146]. The DU-145 cells treatment with anthocyanins extracted from black soybean provoked a significant increase in apoptosis and a significant decrease in p53, Bcl-2 and AR expressions with, in addition, a further decrease in PSA levels. Moreover, the anthocyanin treatment showed a significant inhibition of tumor growth in xenograft models [147]. Gallic acid (GA) induced apoptosis in DU145 and 22Rv1 cell lines, demonstrating, in nude mice fed with GA, inhibition of tumor growth [148]. In addition, GA reduces survival, proliferation, and invasion in PC3 cells [149]. Gallotannins, polymers formed by the esterification of GA, produce an apoptotic effect in DU145 and PC3 cell lines by decreasing the expression of different genes, such as Mcl-1, and inhibiting caspase activation [150]. Similarly, the ellagitannins of the pomegranate, named punicalagin (PN), have elicited the induction of apoptosis in PC-3 and LNCaP cells [151]. In the pomegranate, as well as juice, extract, or oil, in addition to the ellagitannins, there are also large quantities of anthocyanins that have powerful antioxidant and anticancer activities in different tumors, including PCa [152]. Caffeic acid and its natural ester-caffeic acid phenethyl ester (CAPE) are potent inhibitors of the androgen-dependent PCa lines [153]. Caffeic acid and CAPE from bee propolis showed a synergistic effect with chemotherapeutics and radiotherapy, repressing, moreover, tumor growth and AKT signals in human PCa cells [154]. Esters of cinnamic acid induce apoptosis and inhibit the growth of prostate and breast cancer [155]. Chlorogenic acid inhibits benign prostatic hyperplasia growth, probably via the inhibition of 5αR, in the animal model [156]. Ferulic acid induced the arrest of cell cycle in PC3 cells while, in LNCaP cells, it provoked apoptosis [157]. Resveratrol treatment of LNCaP cells led to the phosphorylation and the nuclear translocation of ERK1/2 (mitogen-activated protein kinase) and the accumulation of nuclear COX-2, and subsequently to the complex formation with pERK1/2 and p53 [158]. In addition, curcumin, a polyphenolic molecule extracted from the rhizome of the plant *Curcuma longa*, inhibits the proliferation of androgen-dependent and androgen-independent prostate cell lines [159]. Curcumin increases, in fact, the sensitivity of PCa cell cultures to gamma-radiation, reduces the trans-activation and the expression of AR (acting also as its antagonist), reduces the expression of EGF receptors, induces the degradation of HER2, reduces angiogenesis in vivo and the expression of VEGF [160]. Curcumin acts,

moreover, as an inhibitor of the tumor necrosis factor (TNF-α) and prostaglandin E2 (PGE2) production, but increases the caspase activity (3, 8, 9) in HL-60 PCa [161]. A recent study by Chen et al. examined the anti-carcinoma potential of curcumin, treating PC3 and DU145 cells with a series of curcumin analogs of the second generation, in concentrations of 0–10 μM, founding the ability of curcumin to decrease the expression of NF-kB, mTOR (mammalian target of rapamycin), AKT and p-AKT [162]. Colonic metabolites may participate in the chemoprevention of PCa by varied polyphenol-rich diet or composite polyphenol preparations. The gut microbiota-derived metabolites of ellagitannins and green tea catechins, urolithin A (uroA) and 5-(3 ,4 ,5 -trihydroxyphenyl)-γ-valerolactone (M4), respectively, are, in fact, the main compounds absorbed by the human system and derived from the metabolism of these polyphenols. Stanisławska et al. established the effects of M4, uroA, and their combinations on LNCaP cells: M4 showed modest antiproliferative activity in LNCaP cells (IC50 = 117 μM; CI: 81–154), while uroA decreased proliferation (IC50 = 32.7 μM; CI: 24.3–41.1) and induced apoptosis in the same line of cells with, furthermore, a synergistic antiproliferative activity of M4 plus uroA. Besides, M4 potentiated the inhibition of PSA secretion and enhanced AR retention in cytoplasm caused by uroA [163]. Urolithins induced apoptosis in LNCaP cells, negatively influencing the levels of Bcl-2 protein and probably decreasing the expression of AR and the PSA synthesis [164]. Moreover, the gut microbiota itself is influenced by those colonic metabolites, eliciting beneficial effects on intestinal probiotic bacteria [165]. The dietary pattern has, indeed, an important and direct influence on gut bacteria composition [166]. The western diet, consisting of high-fat content and high sugar content, reduces the diversity of the gut microbiota in mice, increasing in *Bacteroides* spp. and *Ruminococcus torques* [167] while, in humans, increasing *Enterobacteriaceae* rates and significantly decreasing short-chain fatty acids in feces, one of the metabolites generated by bacteria [168].

#### **5. Conclusions**

Although the prostate is not an organ directly affected by gut microbiota, a wealth of evidence suggests an indirect influence of cytokines and immune changes derived by different bacterial metabolites and gut microbiota modifications. The previously reported studies support a potential role of diet and nutrition in PCa pathogenesis, partially mediated by the gut microbiota itself. The gut microbiota could be targeted to improve therapies while attenuating adverse reactions. The influence of diet and nutrients on PCa pathogenesis and progression have received increasing attention. Several animal studies have reported how certain nutrients, including fat and polyphenols, are indeed involved through a variety of mechanisms, which include inflammation, antioxidant activity, and influence on sex hormones (Table 2). Generally, a healthy dietary pattern (e.g., low in meat and high in vegetables) could help in the prevention of PCa and lifestyle-related diseases. Due to such considerations, the close relationship between gut microbiota and cancer is a research area that is receiving considerable attention. Based on recent findings, gut microbiota alterations, which are caused by various external factors such as dietary composition, are involved in all stages of cancer, including initiation, progression, treatment outcomes, and adverse reactions [169]. The mechanism by which gut microbiota may influence PCa has not been elucidated. Therefore, it is challenging to understand how microbiota and host influence each other. It could be speculated that while microbiota could affect the natural cancer history, cancer itself could change the microbiota composition. However, it is undeniable that colonic metabolites may contribute to the chemoprevention of PCa by varied polyphenol-rich diet or composite polyphenol preparations. Understanding the specifics of gut microbiota in the context of PCa is needed in the era of precision medicine for the development of personalized treatments. However, a further investigation and understanding of the relationships between microbiota and PCa pathogenesis, development, and progression are warranted.


#### **Table 2.** Summary of the effects of several natural compounds.


#### **Table 2.** *Cont.*

**Author Contributions:** Conceptualization, F.C., B.B., M.B.; writing—original draft preparation, F.C., B.B., M.B., G.G., E.D.Z.; writing—review and editing, F.C., B.B., M.B., E.D.Z., M.D.D., C.I., I.F.A., A.S., G.G., G.S., L.Q., S.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **Lactoferrin from Bovine Milk: A Protective Companion for Life**

#### **Fabiana Superti**

National Centre for Innovative Technologies in Public Health, National Institute of Health, Viale Regina Elena 299, 00161 Rome, Italy; fabiana.superti@iss.it; Tel.: +39-06-4990-3149

Received: 30 July 2020; Accepted: 20 August 2020; Published: 24 August 2020

**Abstract:** Lactoferrin (Lf), an iron-binding multifunctional glycoprotein belonging to the transferrin family, is present in most biological secretions and reaches particularly high concentrations in colostrum and breast milk. A key function of lactoferrin is non-immune defence and it is considered to be a mediator linking innate and adaptive immune responses. Lf from bovine milk (bLf), the main Lf used in human medicine because of its easy availability, has been designated by the United States Food and Drug Administration as a food additive that is generally recognized as safe (GRAS). Among the numerous protective activities exercised by this nutraceutical protein, the most important ones demonstrated after its oral administration are: Antianemic, anti-inflammatory, antimicrobial, immunomodulatory, antioxidant and anticancer activities. All these activities underline the significance in host defence of bLf, which represents an ideal nutraceutical product both for its economic production and for its tolerance after ingestion. The purpose of this review is to summarize the most important beneficial activities demonstrated following the oral administration of bLf, trying to identify potential perspectives on its prophylactic and therapeutic applications in the future.

**Keywords:** lactoferrin; bovine milk; nutraceutical; human health

#### **1. Introduction**

Lactoferrin, an 80 kDa iron-binding glycoprotein belonging to the family of transferrin proteins, was first isolated in 1939 from cow's milk [1] and in 1960 was shown to be the main iron-binding protein in human milk [2]. Lactoferrin is also found in mucosal secretions such as tears, saliva, vaginal mucus, seminal plasma, nasal and bronchial secretions, bile, gastrointestinal fluids and urine [3]. It is present in plasma in relatively low concentrations, where it is predominantly neutrophil derived [4].

Bovine lactoferrin (bLf) has been extensively studied in the past 60 years, as research on this protein actually started around the 1960s, when technological progress had allowed its correct extraction from milk and its complete characterization [5].

Its role in numerous and varied biological functions is now accepted by the scientific community. Indeed, it has been shown that bLf is involved in various physiological and protective actions, among which some of the most studied to date are antioxidant, anti-tumour, anti-inflammatory and antimicrobial activities [6–13].

In this review on bLf, both the main characteristics and the major biological functions of this pleiotropic nutraceutical protein will be summarized. In particular, the use of exogenous bLf as a therapeutic agent and the mechanisms responsible for its various actions will be taken into consideration in order to identify new research perspectives.

#### **2. Bioavailability, Metabolism, Absorption and Delivery of Bovine Lactoferrin**

As previously mentioned, bLf, from milk or whey, is used to improve immunity, resistance to infection, control of non-communicable diseases, iron absorption and human health in general. Since many of these functional properties are highly dependent on the structural integrity of the protein, it must be remembered that when bLf is taken orally it can be largely digested in the stomach [14]. In particular, since bLf receptors are found in the intestinal mucosa and in the cells of the lymphatic tissue of the intestine [15,16], it is important that bLf maintain its structural integrity to bind its receptors. However, it has been shown that bLf directly induces the growth and proliferation of enterocytes, depending on its concentration [17], so intestinal absorption of lactoferrin can be different in different periods of life. It is noteworthy that at the beginning of life the intestinal lumen of the baby who is breastfed or fed with infant formula fortified with bLf will have a high concentration of lactoferrin attributable to very limited proteolytic degradation and high cell proliferation [18]. The mucosal development induced by lactoferrin can, thus, increase the mucosal surface and not only improve the absorption of iron but also of other nutrients. Later, as the baby grows, the digestion of proteins will be more efficient and the lactoferrin concentration will be much lower, resulting in increased differentiation. Hence, in adulthood, as previously mentioned, bLf administered orally will be largely digested into small molecules. Since many functions of bLf (such as the ability to bind iron) are highly dependent on the integrity of the protein structure, its gastrointestinal digestion causes a loss of many of these properties. However, protein degradation also has positive aspects as some peptides produced by its digestion, such as lactoferricin, a 25-residue peptide (Lf amino acid residues 17–41) [19], and lactoferrampin, a 20-residue peptide (Lf amino acid residues 265–284) [20], display potent defensive activity. These peptides possess antimicrobial activity due to their hydrophobicity and cationic charge that make them amphipathic molecules. Lactoferricin that in some cases displays a more potent antibacterial and anti-fungal activity than intact bLf [19,21] possess antimicrobial [22–24], anticancer [24–27] and anti-inflammatory properties [28], while lactoferrampin shows a wide antimicrobial action against bacteria, viruses, yeasts and parasites [22,24]. Finally, it has been reported that lactoferricin, incorporated in food supplements, could provide health benefits and reduce the risk of chronic disease [29]. Additional studies are needed to identify all biological activities (together with the molecular mechanisms involved) of these bioactive peptides derived from the digestion of bLf. This is essential in order to optimize their use for human health and well-being. Further insights into the multiple activities of these two peptides can be found in the reviews of Gifford et al. [24], Bruni et al. [30] and Drago-Serrano et al. [31].

As mentioned above, bLf receptors are found in the intestine [15,16], so the orally administered protein must be protected to pass through the stomach and reach the intestine without being degraded. In order to improve its oral bioavailability, the formulation of bLf oral delivery systems has been approached with different approaches. Among the most commonly used methods to protect bLf during the oral and gastric passage phases we find: Iron saturation, microencapsulation, PEGylation and absorption enhancers [14,32]. While it is believed that iron saturation is one of the methods for slowing the enzymatic hydrolysis of bLf, it is not considered an effective method of delivering bLf in its structurally intact form to small intestine by oral administration [33]. Microencapsulation is a commonly used method to protect bLf from protease digestion. This method involves the formation of a protective structure (protein or polysaccharide shell) around the bLf core. This core-shell system effectively protects the bLf from gastric digestion and, by using appropriate shell materials, can also allow for specific and controlled release of the protein. In addition to microencapsulation with proteins or carbohydrates, liposomes have also been shown to prevent gastric degradation of bLf [34]. PEGylation, i.e., the covalent attachment of polyethylene glycol (PEG) to therapeutic proteins, is used to protect bLf from the gastric environment. This technique increases bLf resistance to proteases through steric hindrance and, by increasing molecular mass, inhibits renal clearance [32]. As for absorption stimulators, these are a group of chemicals that increase the permeability or transport of molecules across biological membranes. In the field of bLf, research on absorption stimulators focused on chitosan, a linear polysaccharide composed of randomly distributed beta-(1->4)-linked D-glucosamine (deacetylated unit) and N-acetyl-D-glucosamine (acetylated unit). Chitosan has been reported to increase bLf uptake in the gut by opening the intercellular junctions [35]. However, chitosan tends

to dissolve at acidic gastric pH, so, to overcome this problem, chitosan derivatives which are poorly soluble in acidic conditions, such as chitosan-succinate and chitosan-phthalate, have been used [32]. Therefore, regarding the oral bioavailability of bLf, we can conclude that, at present, microencapsulation and PEGylation appear to be the most efficient methods to deliver bLf to gut absorption sites.

#### **3. Lactoferrin, Iron, Oxidative Stress and Anaemia**

Lactoferrin, as the other transferrins, has a molecular weight of about 80 kDa and its structure includes two lobes, each capable of reversibly chelating two Fe+<sup>3</sup> ions per molecule. Both lobes have the same fold, consistent with their sequence identity of ~40%. In each lobe, two domains, referred to as N1 and N2, or C1 and C2, enclose a deep fissure containing the conserved iron-binding site [36]. It is usually only about 15% saturated with iron, indicating that the two lobes are not entirely occupied by iron. bLf, possessing twice the serum transferrin's affinity for iron, is also able to act on systemic iron homeostasis by modulating the synthesis of the two key proteins hepcidin and ferroportin through the down-regulation of interleukin-6 (IL-6) [37]. Figure 1 shows the two Fe+<sup>3</sup> binding domains of bLf.

**Figure 1.** Cartoon representation of bovine lactoferrin. N-lobe is blue (N1 pale blue and N2 dark blue) and C-lobe is green (C1 dark green and C2 pale green). The hinge helix is represented in a pale cyan. Iron ions are reported as red spheres.

#### *3.1. bLf Protection against Iron Deregulation and Oxidative Stress*

Iron, an essential nutrient for cell growth, can become toxic when too abundant, leading to the generation of free radicals by interconverting between its most common oxidative forms, ferrous (Fe2+) and ferric (Fe3+) forms [38]. Free iron is toxic because it can donate or receive an electron from adjacent molecules, causing damage to cellular components or generating reactive oxygen species (ROS) that are themselves cytotoxic.

bLf controls the physiological balance of ROS production and their elimination rate through iron sequestration. Many researchers have demonstrated that bLf is able to modulate the adaptive immune system, and that it possesses significant regulation activity on cellular redox via upregulation of key antioxidant enzymes [39–42]. Oxidative stress plays a role in numerous chronic degenerative processes, such as those that affect tumour development, inflammation and aging [12]. Notwithstanding factors responsible for the ROS production imbalance having not been fully elucidated, it is known that the rate and extent of ROS development removal is dependent on the efficiency of superoxide dismutase (SOD), glutathione peroxidase (GPx) and catalase (CAT). SOD converts the superoxide radical (•O2 −) into hydrogen peroxide (H2O2); GPx or CAT transform H2O2 into water (H2O) or into H2O and molecular oxygen (O2), respectively. In the presence of free ferric ions (Fe3<sup>+</sup>) the superoxide radical (•O2 −) can be

degraded through two phases: In the first, a superoxide molecule reacts with Fe3<sup>+</sup> to form ferrous ion (Fe2<sup>+</sup>) and O2; in the second (Fenton reaction), Fe<sup>2</sup><sup>+</sup> reacts with H2O2 to form Fe3+, a hydroxyl radical (•OH) and a hydroxide ion (OH−). The reaction of the hydroxyl radical with polyunsaturated fatty acids, causing the removal of a hydrogen atom, starts the lipid peroxidation and the production of new radicals.

bLf by sequestering Fe3<sup>+</sup>, is able to prevent the harmful effects of oxidative stress and many studies have demonstrated that it contributes to general homeostasis by disrupting the production of these dangerous radicals [12]. Figure 2 shows Fenton and Haber–Weiss reactions.

**Figure 2.** Lactoferrin protects against cellular damage induced by oxidative stress. It inhibits free ferric ion reactivity with superoxide molecules, thus limiting the formation of ferrous salt and ground state oxygen. This effect prevents the Fenton reaction in which the ferrous ion is oxidized by hydrogen peroxide to ferric ion, forming a hydroxyl radical and a hydroxide ion.

An example of these studies on bLf protection against iron deregulation and oxidative stress is the one conducted by Okazaki et al. [43] that examined the antioxidant property of bLf oral administration in a rat model of ferric nitrilotriacetate-induced renal tubular oxidative injury. Results of this research showed that bLf pre-treatment suppressed elevation of either serum creatinine or blood urea nitrogen levels and exerted protective effects against renal oxidative tubular damage. These results not only demonstrated the antioxidative effect of bLf but also indicate that lactoferrin consumption is useful in the prevention of iron-mediated renal tubular oxidative damage [43].

#### *3.2. bLf in the Prevention and Treatment of Iron Deficiency Anaemia*

Homeostasis is the maintenance of balance in a biological system and is controlled by several factors including bLf which, for its ability to bind ferric ions, plays a central role. Indeed, iron homeostasis is regulated in part by bLf which plays a safety role by protecting against oxidative stress and reducing the amount of cell damage induced by insult. An increasing amount of data have shown the association between disruption of iron homeostasis and different pathophysiologic conditions such as anaemia and, in particular, Fe-overload related disorders [44].

Anaemia, defined as number of red blood cells or haemoglobin concentration below established cut-off levels [45], is a worldwide disease that should not be underestimated as it has important consequences for human health. Its prevalence in pre-school aged children, pregnant women and women of reproductive age is approximately 50%, 40% and 30%, respectively [46]. WHO has estimated that about 50% of all cases of anaemia can be attributed to iron deficiency [47]. It is well known that iron, an essential component of haemoglobin, is found both in plant and animal foods but it is better absorbed from animal sources [48]. In this view, lactoferrin, being one of the main iron-binding proteins also responsible for its transport and release into cells, represents a key element of the iron absorption process.

On the basis of these considerations, several studies have been carried out to evaluate the efficacy of oral administration of bLf for the treatment of iron deficiency anaemia (IDA). In these researches the effectiveness of bLf has often been compared to that of ferrous iron preparations, sometimes leading to partially conflicting results.

Fransson et al. in 1983 [49] analysed the efficacy of bLf supplementation in iron-deficient and iron-sufficient young mice demonstrating that this transferrin represents a useful vehicle for iron supplementation. In this research, the efficacy of lactoferrin supplementation was compared with that of iron chloride and no significant differences were observed. Successively, as IDA during pregnancy represents a risk factor for preterm delivery, the effect of bLf supplementation was studied in women at different trimesters of pregnancy and compared with that of ferrous sulphate. Unlike what was previously observed in mice, in this study haemoglobin and total serum iron values increased to a greater extent in women treated with bLf compared to those who received ferrous sulphate [50]. In a subsequent clinical study by the same authors on pregnant women with IDA, it has been demonstrated that the number of red blood cells, haemoglobin and serum iron increased when they received bLf and decreased when they received ferrous sulphate [51]. In particular, during bLf therapy, serum IL-6 concentrations decreased. So, since IL-6 induces hypoferremia and causes anaemia, bLf is likely to improve serum iron and haemoglobin concentrations, rather than providing more absorbable iron [9]. These results of Paesano et al. [51] are partially in disagreement with those obtained by Nappi et al. [52] who, in a prospective, randomized, controlled, double blind trial, compared the effects of bLf with ferrous sulphate on iron nutritional status in 100 pregnant women with IDA. In fact, the results of this trial showed that bLf and ferrous sulphate had the same efficacy in restoring iron deposits. However, it is important to note that bLf had significantly lower gastrointestinal side effects than ferrous sulphate. Recently, a systematic review and meta-analysis performed to evaluate the efficacy of daily oral bLf compared to daily oral ferrous iron preparations for the treatment of IDA in pregnancy confirmed results reported above, suggesting lactoferrin as the iron replacement agent of choice for IDA treatment in pregnancy [53].

Anaemia is also often observed in endurance athletes (sports anaemia). Especially, female long distance runners, who menstruate and accurately control their weight, can easily develop this type of anaemia. Hence, Koikawa et al. [54] conducted a study to verify whether taking bLf could improve or prevent anaemia in these athletes. The results of this study have shown that bLf increases iron absorption among female long distance runners, suggesting that it can be helpful in preventing sports anaemia.

Concerning pre-school aged children, a prospective, multicentre, controlled intervention study on 260 infants (ages 4 to 6 months) evaluated and compared the effect of an iron-fortified formula containing bLf and an iron-fortified formula without bLf on hematologic indexes and iron status in term infants [55]. Results of this study demonstrated that significant increases in total body iron content and iron absorption in the intestine were observed only in infants fed with lactoferrin fortified formula milk. Further information on the role of lactoferrin in the fight against iron deficiency and IDA in newborns and infants are available in the reviews of Ochoa et al. [56] and Cerami [57].

#### **4. Lactoferrin in the Defences of the Babies: Decreased Risk of Sepsis and Necrotizing Enterocolitis in Preterm Infants**

As just mentioned at the end of the previous paragraph, lactoferrin is fundamental in the infant's diet. It is important to note that lactoferrin also plays important functions both in protecting the newborn infants from infections and in promoting the maturation of their innate and adaptive immune system. In fact, term and, in particular, preterm infants are at risk of infections. In preterm neonates, necrotizing enterocolitis (NEC), a destructive inflammatory bowel condition and sepsis are causes

of severe morbidity and represent the most common motives of death in the first weeks of life and breastfeeding is known to reduce the risks of these serious conditions.

Based on the numerous activities of bLf, in particular the antimicrobial, antioxidant and anti-inflammatory ones, some of which will be better described later, and on the observation that bLf is well tolerated, several clinical studies were conducted that examined the usefulness of the administration of lactoferrin (in general commercial bLf added to infant formula) in the prevention of infections in preterm and term neonates as well as in the reduction of mortality or major morbidity [58–64]. Results of these clinical trials are summarized in Table 1.


**Table 1.** Effect of bovine lactoferrin (bLf) in neonates: Clinical trials.

LGG: *Lactobacillus rhamnosus* GG; VLBW: Very low birth weight; CFU: Colony-forming units; Tregs: T-regulatory cells; LOS: Late-onset sepsis; NEC: Necrotizing enterocolitis.

These clinical studies are particularly interesting, not only because they were targeted to the critical VLBW infants, but above all because both mortality and morbidity following sepsis and NEC remain high despite the use of powerful antimicrobial agents [65]. The results of these trials have shown that the administration of bLf in preterm infants, in the absence or in the presence of the probiotic LGG strain, was able to reduce blood infection without adverse effects.

While these results are extremely encouraging, studies are still needed to establish more precisely the dosage, duration of treatment and development of premature babies.

The data obtained so far support the usefulness of further examining the effects of bLf supplementation on the immune response, in particular to infections, in highly vulnerable infants. It is hoped that the results of the numerous on-going studies will definitively demonstrate the benefits of integrating bLf into the preterm baby's diet leading the way the use of bLf in a clinical setting. More insights into the role of lactoferrin in neonatology can be found in the review by Sharma et al. [66].

#### **5. Antimicrobial Activity of bLf**

The antimicrobial effect was the first identified lactoferrin protective activity and has been widely demonstrated both in vitro and in vivo [8,10,67]. The bacteriostatic and bactericidal activity of lactoferrin against a large number of gram-positive and gram-negative bacteria is due to two distinct mechanisms [8,10,67,68]. bLf primary role involves the binding and sequestration of free iron at the infection sites, thus depriving microorganisms of this essential substrate for their growth

and inducing a bacteriostatic effect [36]. Differently, bactericidal activity is independent of iron and involves direct interaction with the infectious agent: Specific interactions have been described both with lipoteichoic acid (LTA) of gram-positive bacteria and with lipopolysaccharide (LPS) of gram-negative bacteria [67]. Iron sequestration by bLf also prevents biofilm formation that represents a crucial step in the development and persistence of infection [69].

Further mechanisms of the antimicrobial action of bLf are: Rupture of the cell membrane of pathogens, proteolysis of microbial virulence factors, inhibition of microbial adhesion to host cells by binding with glycosaminoglycans (GAGs) and improvement of the growth of normal commensal probiotic microflora in the intestine [67,70].

Concerning in vivo preclinical studies, twenty years ago Wada et al. [71] demonstrated in germfree BALB/c mice that the administration of 10 mg bLf for 3–4 weeks significantly reduced the number of *Helicobacter pylori* in the stomach and also inhibited the attachment of bacteria to it. Numerous in vivo studies have been conducted since then, many of which are described in the review of Teraguchi et al. [72]. The satisfactory results obtained in animal models then led to clinical trials. For example in 2005 Okuda et al. [73] confirmed the activity of bLf in inhibiting colonization by *Helicobacter pylori* in humans. In this double-blind placebo-controlled randomized trial, healthy subjects positive for *Helicobacter pylori* received bLf tablets (200 mg/day) or placebo tablets for 12 weeks. After treatments the decrease of the (13) C-urea breath test values in the bLf group was significantly higher than that in the control group suggesting that bLf administration is effective to suppress *Helicobacter pylori* colonization. *Helicobacter pylori* infection, still very frequent, causes chronic active gastritis and can have serious complications such as gastric malignancies. Since antibiotic treatment (mainly clarithromycin and levofloxacin) has led to an increase in antibiotic-resistant strains in recent decades, these results are of particular interest for the development of a new eradication therapy. This represents only one example of the applications of bLf as an antimicrobial agent in humans since many other studies have shown that oral administration of bLf can reduce bacterial and fungal infections mainly in the gastrointestinal tract [74].

Among the many activities carried out by bLf to fight infections, it should be remembered that bLf also acts as a prebiotic by promoting the growth of beneficial bacteria for the host such as probiotics. So, concerning in vivo preclinical and clinical studies, there are a number of experimental observations that oral administration of bLf, alone or in association with probiotic strains, is able to counteract bacterial and fungal vaginal infections [70].

The antiviral activity of bLf has been extensively studied in in vitro systems [67,75,76] and two main mechanisms have been identified by which bLf inhibits viral infection: (i) Competition with the virus for the binding to cell receptors [77,78]; (ii) direct interaction with capsid or viral envelope proteins [67,75,76,79]. An in vivo preclinical study by Shin et al. [80] demonstrated that orally administered bLf reduced pneumonia in mice infected with the Influenza virus by suppressing the infiltration of inflammatory cells in the lung.

Concerning the effects of lactoferrin oral administration against viral infections in humans, its beneficial action has been demonstrated for different viruses such as hepatitis C virus (HCV) [81,82], rotavirus [83], norovirus [84] and common cold infections [85]. Very recently, clinical use of liposomal bLf in seventy-five patients affected by SARS-CoV-2 infection has been reported [86]. The use of liposomes arises from the observation that liposomes loaded with bLf improved the resistance of bLf to digestive enzymes thus enhancing the effect of orally administered bLf [87]. All 75 COVID-19 positive patients were successfully treated with the oral administration of liposomal bLf, which allowed a complete and fast recovery. As aerosol liposomal therapy is widely employed with good results [88,89], in some patients with headache, dry cough and nasal congestions liposomal bLf was also administered by aerosol that was very useful to relieve not only the respiratory symptoms but also the cough, the headache and the smell and taste dysfunction. The results of this study are very encouraging as they indicated that oral treatment with liposomal bLf induces a fast recovery in 100% of patients and that lower dose of the same treatment (half doses) seems to exert a potential preventive effect against

COVID-19 in healthy family members in direct contact with the affected patients [86]. The use of bLf trapped in liposomes will be better discussed in the section on the anticancer activity of bLf. Given the emergence of containing this terrible pandemic, further studies are underway on the use of different forms of Lf to treat COVID-19 patients.

Regarding the antifungal activity of bLf, most of the studies involved *Candida albicans,* known as one of the most dangerous opportunistic pathogens. As for bacteria, bLf can act effectively on a broad spectrum of fungal species due to its strong iron-absorbing property. It has been shown that bLf is capable of killing *Candida albicans* [90]. However, in addition to the iron-depriving effect, bLf is able to directly bind the surface of fungal cells, resulting in increased membrane permeability and inducing their death. The combination of bLf with other antifungal compounds (such as fluconazole) significantly enhanced the inhibitory activity against *Candida albicans* [91] and *Cryptococcus neoformans* [92]. Concerning in vivo studies, it has been reported that, in guinea pigs infected with *Trichophyton mentagrophytes*, orally administered bLf did not prevent development of symptoms during the early phase of infection, but facilitated clinical improvement of skin lesions after the peak of the symptoms [93]. These results indicate the potential utility of bLf as a food component to promote the treatment of dermatophytosis. Other authors developed an experimental model of reproducible oral candidiasis, with immunosuppressed mice, showing local symptoms characteristic of oral thrush in humans and, using this model, demonstrated the efficacy of bLf against experimental *Candida albicans* oral infection [94]. For further information see also the reviews from Superti and De Seta [70] and Fernandes and Carter [95].

In summary, numerous in vivo studies have shown that oral administration of bLf is able to counteract various bacterial, viral and fungal infections. With regard to communicable diseases in general, it is important to remember that, due to the frequent use of antimicrobial drugs, numerous pathogens have become prone to drug resistance which represents the main cause of the unsatisfactory results of some conventional antimicrobial treatments. Consequently, research and development of new therapeutic have become urgent. From this point of view, bLf can represent a very promising tool as an alternative or complementary therapeutic approach to conventional therapy.

#### **6. Anti-Inflammatory Activity of bLf**

Inflammation is a complex pathophysiological process involving numerous mediators and various cell types in response to microbial or non-microbial injury [12]. If inflammation is not promptly limited, it can cause damage to the host by establishing systemic and even chronic inflammatory conditions. It is well known that the production of principal immune mediators, such as cytokines and chemokines, depends on the recruitment of inflammatory cells and, in particular, innate immune cells.

Several studies demonstrated that lactoferrin, being a natural immunomodulator, exerts an anti-inflammatory effect [96] supported by the strong increase of its content in body secretions during inflammation [97,98]. There are numerous evidences concerning the capability of lactoferrin to improve injury induced by insult and protect the integrity of organs during the development of inflammation. The anti-inflammatory activity of bLf can be partially ascribed to its positive charge through which it interacts with negatively charged groups (for example proteoglycans) present on the surface of the immune cells. This interaction can activate signalling pathways that induce a physiological anti-inflammatory reaction [41]. bLf is also able to enter cells and translocate to the nucleus [99], so regulating pro-inflammatory gene expression [100]. The anti-inflammatory effect of lactoferrin during bacterial infection is also due to its ability to neutralize negatively charged microbial molecules such as LPS, thus preventing the interaction of the LPS-binding protein with the endotoxin and blocking the binding of LPS with the membrane protein CD14 and the subsequent activation of monocytes and macrophages [101].

It is also likely that lactoferrin controls inflammatory response by preventing iron-mediated free radical injury at inflamed sites [9] so, through the control of oxidative stress, it modulates innate immune responsiveness that alters production of immune regulatory mediators that are important for directing development of adaptive immune function [39,102]. Several mechanisms are involved in the immunomodulating activity of lactoferrin [103,104]. Lactoferrin acts on B cells to allow their successive interaction with T cells, promotes the maturation of T cell precursors into T helper cells and induces the differentiation of immature B cells into antigen presenting cells [103]. It has been also suggested that lactoferrin may play a role in T cell activation through modulation of dendritic cell function [105]. The anti-inflammatory effect is probably due to the inhibition of production of proinflammatory cytokines such as interleukin-1 beta (IL-1 beta), IL-6 and TNF-alpha. This, as mentioned before, can be obtained by the translocation of lactoferrin to the nucleus, where it blocks NF-kB (nuclear factor kappa-light-chain-enhancer of activated B cells) activation. It has long been known that bLf is able to limit irritation both at the level of the skin and within the subcutaneous tissues and internal organs and many studies on the immunomodulatory effects of orally administered bLf have been carried out [106]. For further information see also the reviews from Kruzel et al. [12] and Drago-Serrano et al. [107].

#### *6.1. Lactoferrin and Dermatitis*

Allergic contact dermatitis is an inflammation of the skin resulting from exposure to irritants and allergens present in the environment. The main therapeutic approaches to limit the symptoms of skin allergies include the use of topical corticosteroids and calcineurin inhibitors, which have side effects [108]. Consequently, to overcome the limitations of the currently available treatments, new therapeutic categories including biological ones were considered [109]. In this view, Zimecki et al. [110] carried out a study in BALB/c mice to compare the immunomodulatory actions of bLf on the elicitation phases of the cellular and humoral cutaneous immune responses to oxazolone and toluene diisocyanate, respectively. This study showed that bLf is able to differentially influence the stimulation phases of humoral and cellular immune responses in mouse skin models and that the inhibition of the cellular immune response is probably due to the suppression of Th1 cells.

#### *6.2. Lactoferrin and Inflammatory Bowel Diseases*

Togawa et al. [111] examined the potential ability of bLf to attenuate colitis utilising a 2,4,6-trinitrobenzenesulfonic acid (TNBS)-induced colitis model in rats. This is a well-established model very similar to human inflammatory bowel disease characterized by mucosal infiltration of neutrophils mediated, at least in part, by tumour necrosis factor-alpha (TNF-alpha) and IL-1beta activation [112]. Results obtained showed that bLf administration is able to suppress the activation of proinflammatory cytokines, such as TNF-alpha, IL-1beta and IL-6 in rats with TNBS-induced colitis. Similar results have been obtained by the same research group in a dextran sulphate sodium (DSS) induced-colitis rat model [113]. The ability of bLf to relieve the inflammatory conditions of DSS-induced experimental colitis was later confirmed in BALB/c mice as well [114]. Since, as expected, iron-free bLf (apo-bLf) treatment was better than iron saturated bLf (holo-bLf) treatment, the results of this study suggested therapy with apo-bLf as a helpful tool in clinical management of ulcerative colitis. More recently, it has been demonstrated in in vitro and ex vivo systems that bLf markedly inhibited expression of pro-inflammatory cytokines, such as TNF-alpha, interleukin-8 (IL-8) and IL-6, both in cultured and Crohn-derived intestinal cells [100]. Investigating the dose-dependent effects of bLf, it has been also observed that it is able to modulate neonatal intestinal inflammation [115]. In this study, the effects of bLf at doses comparable to the levels of lactoferrin in bovine and human milk were analysed using intestinal epithelial cells, as the in vitro system, and immature pig intestine, as the in vivo system. Results obtained demonstrated beneficial effects of bLf at low doses (0.1–1 g/L, close to its levels in cow and human milk, respectively) and harmful effects at a high dose (10 g/L, close to hLf levels in colostrum). These researches, demonstrating that moderate doses of bLf increase the proliferation of intestinal cells while high doses trigger inflammation, are fundamental for establishing effective doses of bLf for the integration of formula in preterm infants, in order to support intestinal maturation and prevent inflammation. This study has important biological significance because it shows that bLf

does not always have a beneficial effect but, at high doses and under certain conditions, it can exert a proinflammatory effect.

#### *6.3. Lactoferrin and Pulmonary Inflammation Disorders*

Over the past decades, asthma and allergic lung inflammation diseases have become increasingly common. Asthma is a long-term inflammatory disease of the lungs characterized by airway eosinophilia, mucin secretion, IgE production and airway hyperresponsiveness.

In bronchial asthma, oxidative stress exacerbates airway inflammation by inducing different proinflammatory mediators, enhancing bronchial hyperresponsiveness, stimulating bronchospasm and increasing mucin production. Oxidative stress is a consequence of enhanced ROS production by eosinophils recruited into the lungs during exposure to pro-oxidant environmental molecules or to respiratory viruses [116]. It has been demonstrated that ROS generated by reduced nicotinamide adenine dinucleotide phosphate (NADPH) oxidase from environmental molecules, such as pollen grains or their extracts, provide a signal that enhances antigen-induced allergic airway inflammation in mouse [117]. Successively, it has been shown that bLf, as an iron-binding protein, is able to reduce pollen extract-induced airway inflammation [118]. It is interesting to note that apo-bLf, but not holo-bLf, significantly reduced the accumulation of inflammatory cells and the formation of mucin-producing cells in the inflamed respiratory tract of mice.

Zimecki et al. [119] studied the efficacy of both bLf and human lactoferrin (hLf) to decrease allergen (ovalbumin)-induced pleurisy in BALB/c mice. bLf was given either orally or was administered by gavage intragastrically or by an intraperitoneal injection. The results demonstrated the efficacy of Lfs, bLf more than hLf, in reducing pleurisy in a well-established experimental mouse model of ovalbumin-induced pleurisy. This study is of particular interest as it has increased knowledge of the suppressive efficacy of bLf in allergy, suggesting that oral administration of bLf may be effective in improving allergy symptoms in patients.

bLf has also been used successfully in a cystic fibrosis (CF) mouse model [120]. CF is a multifactorial genetic disease that affects several organs, including the respiratory tract, in which iron imbalance, inflammation as well as bacterial infection, play an important role in the chronicity and gravity of lung disease. Results of this study demonstrated that aerosolized bLf was able to reduce infiltrated leukocytes in CF mice and pulmonary iron overload in both control and CF mice. Above all, a significant reduction was observed in ferroportin (the iron-regulated transporter 1), ferritin (the intracellular protein that stores and releases iron in a regulated manner) and in the luminal iron content.

#### *6.4. Lactoferrin and Hepatitis*

Orally ingested bLf has been shown to provide a wide range of benefits in animal models with inflamed liver [121,122] and clinical use of bLf has also produced several promising outcomes, such as the inhibition of hepatic inflammation in chronic hepatitis C (CHC) patients [81,123].

Concerning in vivo studies, Tsubota et al. [121] utilized Long–Evans Cinnamon rats, which spontaneously develop fulminant-like hepatitis, to evaluate the effect of oral administration of bLf on oxidative liver damage. This study showed that bLf allows the recovery of the reduced base excision repair capacity and reduces the accumulation levels of 8-hydroxy-20-deoxyguanosine (a reliable marker of ROS-induced DNA modifications) and mutations in hepatic mitochondrial DNA, possibly thereby protecting Long–Evans Cinnamon rats from lethal hepatic insufficiency. Based on these observations, it has been suggested that bLf could potentially be useful for the treatment of inflammatory liver diseases induced by oxidative stress.

Successively Kuhara et al. [122] utilized four mouse models of hepatitis induced by D-galactosamine, carbon tetrachloride, D-galactosamine plus lipopolysaccharide and zymosan plus lipopolysaccharide to evaluate the efficacy of oral administration of bLf against hepatitis and to identify its mechanism. Results of this research demonstrated that bLf is able to improve the expression of interleukin 11

(IL-11) and bone morphogenetic protein 2 in the small intestine and to protect mice with hepatitis against inflammation.

Regarding clinical trials, Tanaka et al. [81] carried out a first pilot clinical study demonstrating that lactoferrin could be one potential candidate as an anti-HCV reagent that may be effective for the treatment of CHC patients with low serum concentrations of HCV RNA. Finally, Konishi et al. [123] evaluated the effect of bLf on lipid peroxidation, hepatic inflammation and iron metabolism in patients with CHC. Results of this clinical trial demonstrated that bLf therapy allows improvement in lipid peroxidation and alanine aminotransferase (ALT) levels suggesting its oral administration as a promising therapeutic approach for suppressing oxidative stress and inflammation in patients with CHC non-responders to antiviral therapy.

In conclusion, bLf performs its anti-inflammatory action through different cellular receptors and the activation of various cellular signalling pathways, often via iron-dependent mechanisms. Indeed, its ability to sequester iron and to inhibit ROS formation is a key factor in reducing the damage caused to excessive inflammatory responses. The interaction of bLf with its receptors can trigger several protective effects due to the regulation of enzymatic activities and ROS production, the modification of cell phenotype and cytokine profile, the binding to LPS or the competition with its receptors and the prevention of apoptosis.

#### **7. Anticancer Activity of bLf**

The World Health Organization [124] reported that, in 2018, 18.1 million people around the world had cancer, 9.6 million cancer patients died and cancer was the cause of about 30% of all premature deaths from non-communicable diseases (NCDs) among adults aged 30–69. So the incidence of cancer is getting higher and there is still no fully efficacious cure for all different forms of the disease. Therefore, preventing the development of carcinomas and treating them is critical to reduce current cancer mortality.

The anti-tumour activity of hLf and bLf has been extensively studied for both prevention and treatment, and several mechanisms have been suggested such as intra- and extra-cellular effects or immunoregulatory and anti-inflammatory functions.

In vitro studies showed that the intracellular effects are generally associated with the arrest of tumour cell growth, while the extracellular ones are mainly related to the interaction between bLf and cell membranes, and the immunoregulatory action of bLf is obtained through the activation of the cells of the immune system that release tumour cytotoxic effectors [11].

Numerous in vivo studies have provided evidence that oral administration of bLf is effective in reducing the development of chemically induced tumours [125–129]. The chemopreventive anticancer effects are probably due to the multiple functions of bLf and, in particular, to the stimulation of the immune response, to the modulation of the carcinogenic metabolic enzymes [127], to the antioxidant activity [129], the induction of cell death in tumour tissue and to the inhibition of angiogenesis [128,130]. Regulation of the immune system is a key factor in the action of bLf against cancer [11] and both innate and adaptive immunity are involved in immunostimulation induced by bLf [131–134].

It has been demonstrated that orally administered bLf exhibits high bioavailability and selectivity towards tumour cells by inhibiting tumour proliferation, survival, migration, invasion and metastasis [131,135–139]. It is important to underline that bLf is able to promote or inhibit cell proliferation by acting selectively on normal or cancerous cells, respectively [139]. The first study on the suppressive effect of bLf in rat carcinogenesis was carried out by Sekine et al. [125]. These authors demonstrated in male F344 rats treated with azoxymethane that oral administration of bLf (diet containing 2 or 0.2% bLf) induced a significant reduction in the incidence and in the number of adenocarcinomas of the large intestine. Results of this study suggested that bLf might be a promising chemopreventor of colon carcinogenesis. In 1999 Igo et al. [136] examined the effects on tumor growth and metastasis of bLf administered orally to BALB/c mice bearing subcutaneous implants of the highly metastatic colon carcinoma 26. Results of this study showed that bLf demonstrated

significant inhibition of lung metastatic colony formation from subcutaneous implanted tumours without appreciable effects on tumor growth. Subsequently, Kuhara et al. [131] investigated the effects of oral administration of bLf on the lung colonization by the same colon carcinoma 26. In this study bLf was efficacious before and after tumor implantation, demonstrating a significant inhibitory effect on experimental metastasis. bLf oral administration increased CD4+ and CD8+ cells in the spleen and peripheral blood and enhanced their cytotoxic activity against colon carcinoma 26. Morevoer, bLf induced an increase of CD4+ and CD8+ cells and of interleukin-18 production in the small intestinal epithelium. The results of this study indicate that the inhibition of metastases by oral administration of bLF could be due to an increase in cellular immunity, probably mediated by the increase in IL-18 production in the intestinal epithelium. As previously described, in addition to modulating cellular immunity, bLf carries out anti-inflammatory activity by eliminating ROS, pro-oxidant agents capable of contributing to the development of cancer. bLf protects the host from ROS-mediated cell and tissue damage by both binding free iron and regulating key antioxidant enzymes [39–43]. In this regard, a recent study has shown in a mouse model of hepatocarcinogenesis induced by diethylnitrosamine that oral treatment with bLf, by inhibiting in a dose-dependent manner the elevation in serum markers of liver carcinoma and inflammation, induces a significant improvement in hepatic histological structures [138]. This study demonstrated that bLf is effective in inhibiting the oncogenic activity of diethylnitrosamine in a mouse model of hepatocarcinogenesis through its ability to alleviate the hepatic inflammation and apoptosis. As regard the selectivity of bLF towards transformed cells, Chea et al. [137] demonstrated, in oral squamous cell carcinoma cell lines, that bLf is able to reverse programming of epithelial-to-mesenchymal transition (a biological process of invasion and metastasis in cancers) to mesenchymal-to-epithelial transition and observed in vivo both inhibition of tumor cell infiltration and increased E-cadherin expression in xenografts of mice administered orally with bLf.

Since one of the desired properties of an ideal anticancer drug is the ability to selectively target transformed cancer cells, an appropriate delivery system can be extremely useful in releasing bLf into the tumour site. From this point of view, liposomes represent an efficient drug delivery system that can significantly improve the therapeutic potential of the encapsulated compounds. For instance, apo-bLf trapped in positively charged liposomes composed of phosphatidylcholine, dioleoyl phosphatidylethanolamine, cholesterol and stearylamine (ratio 6:1:2:1 M) has been shown to have a greater capacity, compared to protein alone, to inhibit the growth of B16-F10 melanoma cells [140]. In addition, it has been demonstrated, in a brain-targeted chemotherapeutical delivery system, that doxorubicin (DOX)-loaded bLf-modified procationic liposome (PCL), effectively improved both uptake and cytotoxicity of bLf against the glioma C6 cell proliferation, as well as the anti-glioma activity in vivo, compared with DOX solution or DOX-loaded conventional liposomes [141]. In this study, a cholesterol derivative (CHETA, C36H61N3O4S2) was used to prepare negatively charged PCLs and, subsequently, bLf (positively charged at physiological pH) was absorbed onto their surface via electrostatic interaction. This study showed that DOX-Lf-PCLs delivery system was effective and feasible for systemic administration in chemotherapy of glioma. These results confirmed and supported previous researches of the same authors in which this drug carrier for brain delivery, PCLs, was evaluated both in vitro and in vivo. In this study an in vitro model of the blood–brain barrier was developed to assess the ability and mechanisms of PCLs and Lf-PCLs to cross endothelial cells whereas the uptake of PCLs and Lf-PCLs by the mouse brain in vivo was detected by HPLC-fluorescence analysis. Results obtained demonstrated that, compared with the conventional liposomes, PCL and Lf-PCL-8 (CHETA/Lf ratio = 1:8, w/w) showed an improved performance in the uptake efficiency and in the cytotoxicity as well as much improved localization in the brain [142]. Taken together, these results encourage further investigation for the application of Lf-PCLs to treat other brain diseases.

Other authors investigated whether natural bLf or its different iron-saturated forms, as dietary supplements, were able to increase the anti-tumour activity of different recognized anticancer drugs [143]. In this study, bLf was added to the diet of mice that were then challenged with cancer cells and treated with chemotherapy. Results obtained demonstrated that tumours in holo-bLf-fed

mice were totally eradicated with a single injection of known chemotherapy agents whereas apo-bLf (4% iron saturated) or native bLf (about 15% iron saturated) were ineffective. To be fully effective in eradicating tumours, iron-saturated bLf (holo-bLf) had to be administered to mice for more than two weeks before the chemotherapy, indicating that it functions as a competence factor. In particular holo-bLf decreased tumour vascularity and increased anti-tumour cytotoxicity, apoptosis and infiltration of leukocytes in tumours. Holo-bLf bound to intestinal epithelium and enhanced the production of cytokines within the intestine and tumour, as well as nitric oxide that are known to sensitize cancer to chemotherapy. These results that may seem paradoxical are related to the fact that holo-bLf can release iron and trigger an inflammatory reaction. Holo-bLf also restored peripheral blood cell numbers depleted by chemotherapy, thus defending mice from cancer [143].

A subsequent study, based on emerging nanotechnologies, has been carried out to further improving the bioavailability of holo-bLf to tumour sites by developing polymeric-ceramic nanocarriers (NCs) [144]. The authors validated the preclinical efficacies of novel NC oral formulations for the delivery of holo-bLf in colon cancer therapy. Further insights into the therapeutic application of lactoferrin encapsulated in NCs can be found in the review of Sabra and Agwa [145].

In summary, iron-saturated bLf is a powerful natural adjuvant and a fortifying agent capable of improving cancer chemotherapy. As already said, currently the extraction of Lf from cow's milk and its use in various products represents an industrial reality and it is therefore likely that, in the future, the consumption of bLf containing dietary products could be suggested to inhibit or delay the onset of cancer.

#### **8. Other Therapeutic Properties of Lactoferrin**

There are many other potential uses of bLf for improving human health and some of them will be discussed below.

#### *8.1. Lactoferrin and Obesity*

Obesity represents a serious public health problem and is a strong predictor of chronic diseases. It is now recognized that the intestine and its commensal microflora play an important role in the development of chronic inflammation related to obesity. In fact, obesity and a diet rich in fats are associated with an alteration of the gut microbiota and an increase in intestinal permeability that allows the translocation of LPS into the circulation contributing to systemic inflammation. bLf has been used successfully in the prevention and treatment of obesity and inflammation by inducing the reduction of visceral fat, the neutralization of bacteria in the mucous membrane and the reduction of intestinal permeability.

Concerning the control of fat accumulation, it has been demonstrated that bLf oral administration during caloric restriction in mice was able to enhance weight loss and induced a significant reduction in the total fat pad weight and adipocyte size [146]. Moreover, it has been shown in a mice model with unrestricted food intake that bLf administration induced visceral fat reduction and affected mesenteric adipocytes and fatty acid metabolism in the liver, decreasing the size of mesenteric fat without modulating body weight [147].

Based on the observation that the intestinal commensal microflora plays an important role in the obesity control [148], Sun et al. [149] investigated the role of bLf in obesity as a prebiotic compound, demonstrating that oral administration of 100 mg/kg BW bLf for 12 weeks in high-fat diet induced obese mice was able to positively modulate gut microbiota, inhibited inflammation, reduced body weight and fat accumulation, regulated glucose metabolism and relieved liver steatosis. These results are in agreement with previous reports suggesting a healthy role for the supplementation of bLf in the prevention of metabolic complications related to obesity [150].

More recently, Xiong et al. [151] have confirmed the modulatory effects of bLf on lipid metabolism, however the regulatory mechanisms still remain unclear. Moreover, this study has been carried out on high-fat diet-induced obese C57BL/6J mice in which oral administration of bLf for 15 weeks significantly decreased fat tissue weight, visceral adiposity and hepatic lipid accumulation. These effects are probably due to the suppression of lipogenic gene expression and to the improvement of liver and epididymal adipose tissue inflammation.

The effect of bLf administration has also been studied in association with other safe and effective substances. For example, since bLf and metformin both exhibit beneficial effects on body weight management and fat accumulation, their effects, alone or in combination, on lipid accumulation and metabolism in mice fed with high fat diet has been also studied [152]. Results obtained showed that bLf and metformin, both alone or in combination, prevented high fat diet induced obesity and improved lipid metabolism. The actions bLf and metformin that significantly decreased body weight, waist circumference, Lee's index and visceral fat but had no effect on liver weight were partially due to a key kinase regulating cellular energy homeostasis, the AMP-activated protein kinase.

Regarding the administration of bLf in humans, the effect of enteric-coated bLf was also studied in a randomised double-blind placebo-controlled trial. This human clinical trial also reported that eight-week bLf consumption decreased total adiposity and visceral fat accumulation in male and female subjects with abdominal obesity [153].

In conclusion, numerous studies have shown that dietary bLf consumption represents a favourable agent for the control of lipid accumulation however, prospective as well as mechanistic analyses are still necessary to explain the lipolytic role of bLf in the diet and its potential applications for the obesity treatment. Results of these preclinical studies are summarized in Table 2.


**Table 2.** Impact of bLf on the metabolic syndrome components or obesity: Preclinical studies.

Lastly, an important aspect to consider is the relationships between adiposity and serum iron as hypoferremia could be either an actual iron deficiency or a functional iron deficiency mediated by inflammation. In particular, the link among iron levels, bLf and obesity needs further investigations. This is very important, as it is now clear that iron deficiency and obesity are closely linked [154]. It still remains to be clarified the role of bLf in the two aspects of the relationship between iron and obesity: What is the mechanism that leads to an imbalance of iron in the presence of excess adipose tissue and how iron participates in the pathogenesis linked to obesity.

#### *8.2. Lactoferrin and Bone Metabolism*

It is established that the bones and the immune system are closely connected. Both tissues possess common progenitors and produce common cytokines and inflammation and bone loss coexist in several diseases such as arthritis, osteoporosis and periodontitis.

It is also known that both bLf and hLf are anabolic factor for skeletal tissue as they are able to exert strong proliferative and anti-apoptotic actions in osteoblasts, and a reducing or even inhibitory effect on osteoclastogenesis [155–158]. Lactoferrin also stimulated proliferation of primary chondrocytes [156,157].

It has been demonstrated in vitro that, at physiological concentrations, bLf not only stimulates the proliferation of bone forming cells, osteoblasts and cartilage cells, but it is also an effective osteoblast survival factor. In addition, lactoferrin is able to decrease osteoclast in a murine bone marrow culture system as well as in rabbit mixed bone cell culture [155] and, concerning in vivo systems, local injection of bLf in adult mice resulted in increased calvary bone growth [159]. Subsequent research was conducted to evaluate the effects of oral administration of bLf on bone physiology in an osteopenic rat model (ovariectomized rats) [160]. The results of this study showed that bLf oral administration protected osteopenic rats against the ovariectomy-induced reduction of bone volume, trabecular number and thickness. bLf administration also prevented the elevation of trabecular separation and increased bone mineral density in osteopenic rats. Moreover, after bLf treatment, serum TNF-alpha and IL-6 production was suppressed and serum calcitonin increased [160]. Finally, it has been demonstrated that oral administration of bLf in ovariectomized rats strongly stimulated the bone healing following tibial fracture [161].

Lactoferrin, as mentioned above, is also one of the many defence proteins present in saliva where it exerts a defensive activity against both periodontal bacteria and inflammatory processes.

Concerning periodontitis, it has been shown that orally administered liposomal bLf in Wistar rats is an effective preventive and therapeutic agent in decreasing alveolar bone destruction [87] significantly inhibiting LPS-induced alveolar bone reabsorption without interrupting orthodontic tooth movement [162]. Results of this study suggest that liposomal bLf could represent a powerful preventive or therapeutic agent to control periodontal inflammation. In fact, when liposomal bLf was administered orally to periodontitis subjects, a significant improvement in probing depth and a considerable reduction in the production of LPS-induced cytokines from peripheral blood mononuclear cells were obtained [163]. To clarify the mechanism by which bLf is able to prevent LPS-induced osteoclastogenesis without interrupting tooth movement, the same authors investigated its effects on compressive stress (CS)-induced osteoclastogenesis in comparison with those on LPS-induced osteoclastogenesis via osteoblasts in vitro [164]. Results obtained demonstrated that bLf fights bone destruction associated with periodontitis without inhibiting bone remodelling by CS-loading suggesting that its oral administration could be highly beneficial for control of periodontitis in orthodontic patients [164].

All these studies have shown that bLf plays an important physiological role in bone growth and healing and exerts a therapeutic role in various bone diseases. We can therefore conclude that bLf can be considered a useful therapeutic agent both for bone regeneration and for destructive bone diseases.

#### *8.3. Lactoferrin and Dry Eye Disease*

As mentioned above, lactoferrin is secreted into tears by the lacrimal gland. Dry eye, a multifactorial disease causing visual disturbances and instability of the tear film with potential lesion and inflammation of the ocular surface, is a very common condition with a high prevalence among the elderly. Tear lactoferrin level is an indicator of lacrimal secretory function and correlated with the severity of conjunctivocorneal epithelial lesions in patients with primary, secondary and non-Sjögren's syndrome dry eyes [165].

Kawashima et al. [166] studied in a mouse model whether oral administration of bLf was able to influence age-related tear dysfunction. The results obtained showed that lactoferrin, administered orally, can preserve the function of the lacrimal gland in elderly mice by mitigating oxidative damage and suppressing subsequent inflammation of the gland.

Concerning investigations on human subjects, it has been demonstrated that oral administration of bLf (270 mg/day for one month) represents an efficient treatment modality to improve tear stability and preserve ocular surface epithelium in dry eye patients with Sjögren's syndrome. The authors attribute the tear function and ocular surface improvements to the suppression of inflammatory mediators by bLf [167]. More likely these effects of bLf on tear secretion are due to the combination of its direct action on the tear glands and the overall improvement of body metabolism.

Taken together these results suggest that the use of bLf as a dietary supplement may be a new and safe therapeutic alternative for patients with dry eye syndrome. Moreover, as it has been described in the section on bone metabolism, bLf could also prove useful in the prevention and therapy of other age-related diseases such as autoimmune, neurodegenerative and immune hypersensitivity disorders [168].

#### **9. bLf on the Market**

Large-scale preparation of bLf from cheese whey or skim milk has made this protein accessible commercialized health product for human and animals. The first important application of bLf in a commercial product was its supplementation in infant formula and has been subsequently used for the supplementation of various foods (such as probiotic foods to enhance the beneficial intestinal flora or functional foods to increase iron absorption), for skin care (in cosmetics as antioxidant) and oral care products (to provide oral hygiene), and as nutraceutical, to improve the immune system and to inhibit the inflammatory response. bLf is also used for the conservation and safety of food as it delays lipid oxidation [169] and inhibits microbial growth. The use of bLf in the food industry includes: Meat and wine industry, fat process and dairy industry [170]. The increased demand for natural foods has also increased the importance of natural inhibitors such as bLf. Therefore, the applications of bLf in the food sector are growing remarkably. In all these applications, bLf is expected to express its natural antioxidant, anti-inflammatory, immunomodulatory and anticancer properties.

#### **10. Conclusions**

Lactoferrin is an extremely adaptable protein that has been designated by natural selection to be a first-line defence in mammals. This key protein of natural immunity shows many kinds of marvellous biological activities in vitro and in vivo and helps us to defend against external aggressions both of infectious and non-infectious origin. In fact, being positively charged, it can bind numerous surface molecules or metal ions inducing the host's immunomodulatory activation, which in turn affects both adaptive and innate immunity.

The focus of the present review was on several important health-promoting effects of this multi-functional nutraceutical protein although, given the growing array of applications of lactoferrin in the field of human health, coverage is certainly not complete.

Still, the (intentionally diverse) application domains here reviewed should substantiate the main message I meant to convey: This pleiotropic substance accompanies us and defends us throughout our life, from birth to old age, it is safe and is considered by the United States Food and Drug Administration as a GRAS product with no contraindications in patients of all ages. Besides, it represents an ideal nutraceutical product, cheaply produced from bovine milk, and numerous products containing bLf alone or in association with other nutraceuticals, supplements or probiotics are currently being commercialized.

Taken together, the evidence summarized in this review indicate that it would be advisable to embrace a more comprehensive and integrated approach to various diseases, whereby improvement in the patient's quality of life, and even the clinical outcome, may be obtained by combining bLf with conventional therapies, as suggested by the studies examined here.

Such a perspective should motivate the collection of more data in order to improve our understanding of the protective role of bLf, in relation to both non-communicable diseases and infectious diseases.

I hope to have kindled the interest of the reader in the numerous and often interconnected beneficial activities of this surprisingly versatile milk protein.

**Funding:** This research received no external funding.

**Acknowledgments:** Mariangela Agamennone (University "G. d'Annunzio", Chieti, Italy) was kind enough to provide the artwork of Figure 1, for which I thank her warmly.

**Conflicts of Interest:** The author declares no conflict of interest.

#### **References**


© 2020 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Review*

## **Abscisic Acid: A Conserved Hormone in Plants and Humans and a Promising Aid to Combat Prediabetes and the Metabolic Syndrome**

**Mirko Magnone 1,2,\*, Laura Sturla 2, Lucrezia Guida 2, Sonia Spinelli 2, Giulia Begani 2, Santina Bruzzone 2, Chiara Fresia <sup>3</sup> and Elena Zocchi 2,\***


Received: 11 May 2020; Accepted: 3 June 2020; Published: 9 June 2020

**Abstract:** Abscisic acid (ABA) is a hormone with a very long evolutionary history, dating back to the earliest living organisms, of which modern (ABA-producing) cyanobacteria are likely the descendants, well before separation of the plant and animal kingdoms, with a conserved role as a signal regulating cell responses to environmental challenges. In mammals, nanomolar ABA controls the metabolic response to glucose availability by stimulating glucose uptake in skeletal muscle and adipose tissue with an insulin-independent mechanism and increasing energy expenditure in the brown and white adipose tissues. Activation by ABA of AMP-dependent kinase (AMPK), in contrast to the insulin-induced activation of AMPK-inhibiting Akt, is responsible for stimulation of GLUT4-mediated muscle glucose uptake, and for the browning effect on white adipocytes. Intake of micrograms per Kg body weight of ABA improves glucose tolerance in both normal and in borderline subjects and chronic intake of such a dose of ABA improves blood glucose, lipids and morphometric parameters (waist circumference and body mass index) in borderline subjects for prediabetes and the metabolic syndrome. This review summarizes the most recent results obtained in vivo with microgram amounts of ABA, the role of the receptor LANCL2 in the hormone's action and the significance of the endowment by mammals of two different hormones controlling the metabolic response to glucose availability. Finally, open issues in need of further investigation and perspectives for the clinical use of nutraceutical ABA are discussed.

**Keywords:** abscisic acid; prediabetes; type 2 diabetes mellitus; metabolic syndrome; insulin resistance; adipocyte browning; AMP-activated protein kinase; food supplement

#### **1. Introduction**

#### *1.1. Abscisic Acid (ABA), A Stress Hormone in Plants and Animals*

2-cis, 4-trans-Abscisic acid (ABA) is a 15-carbon weak acid (pKa 4.8) terpenoid hormone (Figure 1) that regulates several pivotal physiological functions in plants, mainly involved in the response to abiotic and biotic stress (water and nutrient availability, UV irradiation; pathogen attack) [1].

**Figure 1.** Structure of ABA. 2-cis, 4-trans-Abscisic acid has an asymmetric carbon atom (arrow) generating two enantiomers S-(+)-ABA and R-(-)-ABA. (+)-ABA is the naturally-occurring form in plants, although (-)-ABA is active in some vegetal functional assays [2].

The function of ABA as a stress signal and its signaling pathway are conserved in all plants, including mosses, and are believed to be the result of the very early adaptation of life to the terrestrial environment. Although the study of ABA in plants has been ongoing for several decades, also in view of its industrial application to improve stress tolerance in crops, it is only in the past decade that evidence has mounted regarding the presence and physiological significance of ABA also in animals. Le Page-Degivry et al. first described the presence of ABA in mammalian tissues, particularly brain [3], but their interesting observation remained isolated. Our group became interested in ABA in our quest for the animal hormonal signal upstream of the second messenger cyclic ADP-ribose (cADPR), a universal Ca2<sup>+</sup> mobilizer from intracellular stores that had been discovered to be involved in insulin release [4]. Intracellular Ca2<sup>+</sup> movements are arguably the most ancient and conserved signaling mechanism throughout the animal and plant kingdoms. Indeed, cADPR had been reported to mediate the effect of ABA in guard cells [5]. Thus, we hypothesized a role for ABA as an animal, as well as a plant hormone, and started our investigation on the simplest and most ancient Metazoa, marine sponges and hydroids; the physiological functions of these animals are limited to water filtration and respiration. However, being sessile, they are particularly exposed to environmental stress (e.g., changes in water temperature or light exposure). We discovered that temperature and light stimulated ABA production in sponges and hydroids, respectively, and that ABA stimulated specific functional responses to these environmental challenges (an increase in water filtration and oxygen consumption in sponges, tissue regeneration in hydroids). These functional responses were mediated by a signaling pathway sequentially involving cAMP, PKA, cADPR and intracellular Ca2<sup>+</sup> movements [6–8]. These discoveries warranted further investigations into the role of ABA in cell-specific functions of higher Metazoa.

#### *1.2. Micromolar ABA Has Pro-Inflammatory and Insulin-Releasing Activity In Vitro and In Vivo*

In several different innate immune cell types exogenous ABA elicits pro-inflammatory functional responses and endogenous ABA is released when cells are challenged with environmental (physical, chemical) stimuli. Briefly, human granulocytes release ABA in response to heat stress or phorbol myristate acetate and exogenous micromolar ABA stimulates cell migration, phagocytosis, ROS and NO production [9]. ABA is released by rat alveolar macrophages stimulated with quartz particles and exogenous micromolar ABA induces the release of pro-inflammatory PGE2 and TNF-α [10]. Micromolar ABA stimulates NO and TNF-α production and cell migration of the mouse microglial cell line N9 and endogenous ABA is released when cells are challenged with bacterial lipopolysaccharide, phorbol myristate acetate, the chemoattractant peptide f-MLP, or β-amyloid [11]. Thus, a plethora of data suggest that ABA in the micromolar concentration range has pro-inflammatory effects on mammalian innate immune cells: a conserved signaling pathway involves the activation of the ADP-ribosyl cyclase CD38 and the production of the Ca2+-mobilizing second messenger cADPR. Starting from the observation that cADPR was involved in the Ca2<sup>+</sup> signaling leading to insulin release, the effect of ABA on β-pancreatic cells was explored. Indeed, micromolar ABA stimulated insulin release in vitro, from rat insulinoma cells and from isolated human islets [12], and also in vivo, in the perfused rat pancreas [13]. In the in vitro study, the role of cADPR as the second messenger of ABA was confirmed, similarly to what observed on innate immune cells.

#### *1.3. ABA Is an Endogenous Mammalian Hormone: Nanomolar ABA Peaks in Human Plasma after Glucose Load*

The studies described above provided ample demonstration that ABA exerts cell-specific functional effects on mammalian cells through the Ca2<sup>+</sup>-mobilizing second messenger cADPR. However, a clear indication that ABA is indeed an endogenous animal hormone was still missing. In their study, Le Page-Degivry et al. observed that ABA was present in the brain of mice fed a synthetic, ABA-free diet at even higher concentrations compared with mice fed a normal (ABA-containing) chow, arguing in favor of the endogenous origin of ABA [3]. A credible indication that ABA is indeed an endogenous animal hormone came from the observation that plasma ABA increases after a glucose load in normal human subjects, reaching concentrations in the low nanomolar range [14]. Moreover, in the same study, the authors showed that nanomolar ABA stimulated glucose uptake, quantitatively similarly to insulin, in rat L6 myoblasts and in murine 3T3-L1 cells differentiated to adipocytes, by increasing GLUT-4 translocation to the plasma membrane. This study provided the first evidence that endogenous ABA could be involved in glycemia homeostasis at hormone-like concentrations.

#### *1.4. ABA Reduces Glycemia without Increasing Insulinemia*

The observations that nanomolar ABA stimulates glucose transport in myocytes and adipocytes in vitro and that endogenous nanomolar ABA increases in the blood of normal subjects after glucose load suggested a role for endogenous ABA in the control of glycemia. Insulin and GLP-1 are the only two peptide hormones known so far to stimulate tissue glucose uptake under conditions of hyperglycemia, and GLP-1 acts by stimulating insulin release. Thus, insulin represents a sort of bottleneck in the response to high blood glucose, apparently a weak spot in glycemia control and a conspicuous example contrary to the principle of redundancy that governs most (if not all) biochemical mechanisms fundamental for cell and organism survival. These considerations suggested to explore the possibility that the effect of ABA on tissue glucose uptake could occur independently of insulin. Several experiments were performed in vivo and ex vivo to address this issue.

In a study performed on rats, the animals were administered an oral glucose load without (control) or with synthetic ABA (at a dose of 1 μg/Kg body weight), or with an aqueous fruit extract yielding the same dose of ABA. The animals receiving the extract and those treated with synthetic ABA showed similar and significantly lower values of the area under the curve (AUC) of the glycemia and insulinemia profiles compared with the control group, suggesting that the active molecule present in the extract and responsible for the observed metabolic effects was indeed ABA [15]. The same fruit extract, providing a dose of approximately 0.5 μg/Kg body weight, was also tested on human volunteers as an adjunct to carbohydrate-rich meals (breakfast and lunch): a significant reduction of the AUC of glycemia and of insulinemia was observed in each subject when the meals were taken with the extract compared with the same meals taken without extract [15]. The plasma ABA concentration measured after intake of the extract in humans increased significantly compared with basal (fasting) values for several hours, but remained in the nanomolar range. The sparing effect on insulinemia of ABA at a dose of 0.5–1 μg/Kg was unexpected, since micromolar ABA had been reported to stimulate insulin release from human β-pancreatic cells and from rat insulinoma cells in vitro [12]. Indeed, when rats were subjected to an OGTT with synthetic ABA at 100 mg/Kg, the AUC of insulinemia was not reduced compared with controls, indicating that at this pharmacological dose ABA did not spare insulin release, although it reduced glycemia [15], in line with previously reported results obtained on mice [16].

The most convincing evidence of the insulin-independent activity of ABA on glycemia control was derived in ex vivo experiments: in mouse skeletal muscle samples taken from fasted animals uptake of 18F-deoxyglucose (FDG) increased two-fold in the presence of nanomolar ABA compared with untreated controls [17]. Stimulation by nanomolar ABA of muscle glucose uptake was confirmed in vivo, on rats undergoing an oral glucose load; as detected by micro-PET, animals receiving a dose of 1 μg/Kg ABA with the glucose load showed a two-fold increase of FDG uptake in skeletal muscle and consequently a significantly reduced glycemia profile after glucose load [17].

A reduction of plasma insulin levels in response to a carbohydrate load would be desirable since insulin stimulates the conversion of glucose into triglycerides in the adipose tissue. Indeed, hyperinsulinemia, either endogenously produced as a result of reduced insulin sensitivity, or exogenously administered as an anti-diabetic drug, is a major factor contributing to an increase in body weight and hepatic steatosis in the pre-diabetic and diabetic patient [18]. Could (endogenous) ABA be a safeguard against excess insulin release?

An indication that ABA can substitute in part for insulin comes from studies on insulin-deficient mice. Administration of 1 μg/Kg ABA together with an oral glucose load reduced the AUC of glycemia in wild-type and also in TRPM2−/<sup>−</sup> mice [17], which have a significantly reduced insulin secretion in response to glucose [19]. The AUC of insulin during the OGTT with ABA was also significantly reduced in the TRPM2−/<sup>−</sup> animals, as in the wild-type mice, compared with the respective OGTT without ABA [17], indicating that the effect of ABA was independent of insulin in both genotypes. Finally, the daily intake of 1 μg/Kg ABA increased muscle glycogen content in TRPM2−/<sup>−</sup> mice fed a high-glucose diet similarly to what observed in wild-type animals, indicating that ABA stimulated muscle glucose uptake also in insulin-deficient animals.

#### *1.5. ABA Improves Lipidemia and Reduces Body Weight and Cardiovascular Risk in Borderline Subjects*

In a recently published human study, the daily intake of a food supplement containing a vegetal extract titrated in ABA in its composition GSECM-50®, sufficient to yield a dose of ABA of approximately 1 μg/Kg body weight, for 75 days significantly improved the metabolic (fasting glycemia, glycated hemoglobin, total, LDL and HDL cholesterol) and morphometric parameters (waist circumference and body mass index) currently employed to evaluate the risk for metabolic syndrome and diabetes. At the end of treatment, fasting blood glucose, glycated hemoglobin, total, LDL and HDL cholesterol, body mass index and waist circumference had significantly improved compared to values at time zero, particularly in those subjects with starting borderline values. The Framingham point score and the 10-year percentage risk calculated for each subject before and at the end of treatment were both significantly reduced in all subjects. In the same study, the daily intake of the same dose of synthetic ABA for four months improved glucose tolerance, and reduced glycated hemoglobin, blood lipids and body weight in mice fed a high-glucose diet [20]. The human and murine parts of the study, taken together, allow to conclude that the improvement of metabolic and bodily parameters observed in the clinical study, performed with the food supplement, can be attributed to ABA present in the vegetal extract of the composition, because similar results were observed on mice fed the synthetic, pure molecule. In addition, mice have certainly not changed their feeding behavior during the study, allowing to rule out this possibility as an explanation for the improvement of the parameters investigated in both studies.

In another set of experiments, the effect of a single dose of the same ABA-containing food supplement used in the chronic study was tested on the glycemia profile after intake of a standardized carbohydrate-rich breakfast. A significant reduction of the mean glycemia profile and of the mean AUC of glycemia (measured over 120 min) was observed in each subject when breakfast was taken with the food supplement compared with the same breakfast taken without supplement [20]. Intake of the ABA-rich food supplement significantly increased ABAp 5- to 16-fold over fasting levels (5–15 nM), indicating that oral ABA was readily absorbed and contributed to increase endogenous ABAp [20].

A similar significant reduction of the AUC of glycemia as observed with the formulated food supplement was observed when the same standardized carbohydrate-rich breakfast was taken with the ABA-containing vegetal extract alone (ABAMET®), indicating that the active ingredient of the food supplement was indeed the ABA-containing vegetal extract (Figure 2).

**Figure 2.** The active ingredient of the food supplement is the ABA-containing vegetal extract. Six volunteers introduced a standardized carbohydrate-rich breakfast, one without any supplement (control), one with one tablet of the formulated food supplement (Food suppl) and another taking only the amount of vegetal extract (ABAMET®) present in one tablet of the food supplement. The three experiments were performed one week apart. The food supplement or ABAMET® were taken before the meal [20] Mean ± SD values of the incremental AUC of glycemia in the time frames 0–60 and 0–120 min are shown. *p* values by paired, two-tailed *t*-test of the comparison between each bar with the respective control bar are shown.

A daily dose of 0.5–1 μg ABA/Kg body weight, sufficient to improve glucose tolerance in humans in the nanomolar range, is approximately 1 log higher than the amount of hormone that can be ingested daily with a fruit- and vegetable-rich diet [15]. The nutraceutical composition GSECM-50® containing the ABA-rich vegetal extract ABAMET® can provide the optimal daily dose of ABA sufficient to increase endogenous ABAp levels to the extent producing a beneficial effect on human metabolism [20].

#### *1.6. ABA Stimulates White Adipocyte Browning and BAT Activity*

The reduction in body weight in the face of unchanged dietary habits observed in humans and in mice treated with low-dose ABA could be attributed to the sparing effect of ABA on insulin release; however, adipose tissue also seems to be a direct target of ABA, along with skeletal muscle, as suggested by in vitro and in vivo studies. Unlike insulin, ABA does not induce preadipocyte differentiation into triglyceride-rich adipocytes in vitro; instead, treatment with ABA induces adipocyte remodeling in differentiated cells, reducing cell size and increasing mitochondrial content, O2 consumption and expression of the brown adipocyte-specific genes UCP-1, PGC-1α, TMEM26, PRDM16 and CIDE-A. In vivo, a single oral dose of ABA 1 μg/Kg increased BAT glucose uptake 2-fold in rats, as detected by micro-PET, and treatment of mice for 30 days with the same dose significantly increased expression of BAT genes in the WAT and in WAT-derived preadipocytes, isolated from the treated animals [21]. Mitochondrial DNA increased 20-fold in the WAT from ABA-treated mice compared with untreated controls and expression of UCP-1 in the BAT was also significantly higher in ABA-treated as compared with control animals [21]. Thus, oral low-dose ABA stimulates BAT activity and induces browning features in the WAT of chronically treated mice. These actions of ABA on AT could be responsible for the observed reduction of body weight in chronically ABA-treated mice compared with controls: female C57Bl/6 mice fed a high-fat diet and treated with ABA (1 μg/Kg body weight/day) for 12 weeks showed a significantly lower weight gain compared with untreated controls, in the face of a higher food intake in the ABA-treated animals: the body weight relative to time zero was 1.30 ± 0.1 vs. 1.53 ± 0.1 in ABA-treated vs. untreated animals (*n* = 5; *p* < 0.01 by two-tailed, unpaired *t* test); the daily food intake during the period of observation was 2.7 ± 0.2 vs. 2.4 ± 0.2 g/animal/day in ABA-treated vs. untreated animals (*n* = 5; *p* < 0.01 by two-tailed, unpaired *t* test) (Magnone M. unpublished result).

#### *1.7. The Plasma ABA Response to A Glucose Load Is Impaired in T2D and in GDM*

Plasma ABA increases in normal human subjects after an oral glucose load [14], but not in patients with type 2 diabetes mellitus (T2D) nor in pregnant women with gestational diabetes mellitus (GDM). Resolution of GDM one month after childbirth coincides with a restoration of the normal ABAp response to glucose load [22].

Interestingly, a significant increase of ABAp relative to pre-surgical values was observed in obese patients one month after bilio-pancreatic diversion (BPD), a type of bariatric surgery performed to reduce body mass and improve glucose tolerance [22]. The increase of fasting ABAp was observed both in normal glucose tolerant (NGT) and in T2D obese subjects, in parallel with a reduction of fasting blood glucose and a significant decrease of the HOMA-IR and fasting insulinemia in the diabetic subjects [22]. Another difference observed between T2D and NGT subjects regarded the fasting ABAp levels, which were significantly higher in T2D compared with NGT subjects; the respective median values were 1.15 (0.19–4.77) vs. 0.66 (0.13–1.72) (*n* = 21 T2D and 27 NGT; *p* = 0.013 by Mann–Whitney test). Moreover, the distribution of the ABAp values was normal in NGT, but not in T2D subjects [22]. These abnormalities may be caused by a heterogeneity of ABA-related dysfunctions occurring in T2D, such as resistance to the effect of ABA (inducing an increase of ABAp, as occurs with insulin in insulin-resistant subjects), or the inability of ABAp to increase in response to hyperglycemia (causing ABAp levels to be in the normal range despite hyperglycemia).

Collectively, these observations on diabetic patients suggest a role for ABAp dysfunction in the development of glucose intolerance and obesity and a beneficial effect of elevated ABAp on glycemic control; indeed, one can expect that insufficiency of either one of the hormones regulating tissue glucose uptake and its metabolic disposal (insulin and ABA) should negatively affect glycemia control.

#### *1.8. The ABA Signaling Pathway Is Di*ff*erent from That of Insulin*

Conservation of ABA between plant and animal kingdoms suggested to explore whether the ABA receptor could be also conserved. Among the several receptors identified in different plant tissues, a putative G-protein coupled receptor (GCR2) [23] indeed showed a significant amino acid sequence identity with a mammalian family of proteins, the lanthionine synthetase C-like protein (LANCL) family. GCR2 has since been disputed as a G protein-coupled receptor, and its homology with the bacterial lanthionine synthetase protein superfamily has instead been advocated [24], a homology that also pertains to the mammalian LANCL proteins. Although the current general consensus is that the PYR/PYL/RCAR family of intracellular receptors are the principal ABA receptors in land plants [25], the homology between mammalian LANCL proteins and plant GCR2 suggested to explore the possibility that they were implicated in ABA sensing. A role for the LANCL protein family in lanthionine biosynthesis has since been ruled out [26]. The LANCL family comprises three proteins: LANCL1 is the most highly expressed in mammals, particularly in the brain, and is a cytosolic protein, LANCL2 is membrane-anchored through its myristoylated N-terminal [27] and LANCL3 has the lowest expression levels of the LANCL proteins and appears to be cytosolic. The membrane-bound location of LANCL2 first attracted interest in this protein as a putative mammalian ABA receptor; indeed, several in vitro studies indicate that human recombinant LANCL2 binds ABA with a high affinity (Kd 2.6 nM) [28,29] and is required for ABA action in several different mammalian cell types [30]. LANCL2 has an unusual behavior as a hormone receptor as it is coupled to a G protein when membrane bound, but can also detach from the membrane when de-myristoylated and accumulate in the cell nucleus [27]. Indeed, the nuclear translocation of LANCL2 occurs following ABA binding [31]. This behavior appears to combine features typical of the receptors for peptide (G protein coupling) and for steroid hormones (nuclear translocation), perhaps a heritage of the primordial origin of the hormone, or the result of the solubility features of ABA.

ABA is a weak acid (pKa = 4.8). Protonated ABA can diffuse through the lipid bilayer [32]; however, a very low percentage of ABA is protonated at the near-neutral pH of plasma and interstitial liquid. For this reason, the presence of a transport system is essential for ABA trafficking between

extra- and intracellular fluids. Conversely, the strongly acidic pH present in the stomach probably allows the diffusion of protonated ABA through the gastric lipid bilayer, accounting for the rapid absorption of the hormone after oral intake [15].

Binding of ABA to LANCL2 bound to the inner plasma membrane layer requires influx of the hormone through the plasma membrane, which occurs through transporters of the anion exchanger (AE) superfamily [33].

The signaling pathway downstream of LANCL2 has been studied in the target cells of the immune (monocytes, macrophages and T lymphocytes) [34], and of the metabolic actions of the hormone (adipocytes and muscle cells) [17,21]. In innate immune cells, ABA binding to its receptor leads to the activation of adenylate cyclase and the subsequent phosphorylation and activation of the ADP-ribosyl cyclase CD38 by protein kinase A (PKA). The product of the enzyme action of CD38 on its substrate NAD<sup>+</sup>, cADPR, then triggers an intracellular Ca2<sup>+</sup> rise due to both intracellular Ca2<sup>+</sup> release from ryanodine-sensitive endoplasmic Ca2<sup>+</sup> channels and to extracellular Ca2<sup>+</sup> influx due to ADPR-gated plasma membrane Ca2<sup>+</sup> channels [9]. This sequence of events closely parallels the ABA signaling pathway first described in marine sponges [6]. The transcriptional effects of ABA observed on hemopoietic progenitors are also likely mediated by the observed increase of intracellular cAMP and the consequent activation of the cAMP-responsive transcription factor CREB [35]. A role for NF*k*B in the micromolar ABA-induced activation of the transcription of cyclooxigenase-2 has also been observed in quartz-stimulated rat alveolar macrophages [10], again pointing to intracellular Ca2<sup>+</sup> movements as an important feature of the ABA signaling pathway in inflammatory cells. The effect of ABA on cells of the hemopoietic lineage (progenitors and innate immune cells) occurs at low micromolar concentrations, thus it is possible that a different signaling pathway is activated by the low nanomolar concentrations exerting its metabolic actions. The signaling pathway downstream of LANCL2 in macrophages and Treg was studied in silico by Leber et al., suggesting a role for LANCL2 in the induction of regulatory responses in macrophages and T cells during *H. pylori* infection [36].

In adipocytes, stimulation by ABA of glucose uptake via GLUT4 involves the activation of phosphoinositide 3-kinases (PI3K). Interestingly, the N-terminal sequence of LANCL2 has been shown to bind to phosphatidylinositol phosphates (PIPs), particularly PI3P, on the plasma membrane, suggesting a spatial as well as functional correlation between LANCL2-dependent and PI3P-mediated signaling. In muscle cells, AMP-activated protein kinase (AMPK) appears to mediate the nanomolar ABA-induced increase of glucose transport, since the effect of ABA is abrogated by pre-treatment of cultured myocytes and of murine muscle biopsies with the AMPK inhibitor dorsomorphin [17]. Activation of AMPK in the ABA-signaling pathway is in sharp contrast with the signal transduction activated by insulin, which results in the inactivation of AMPK by Akt/PKB-mediated phosphorylation.

Indeed, the effect of ABA on energy metabolism appears to be different from that of insulin, pointing to non-overlapping physiological functions of these hormones.

Glucose intake induces an increase of ABAp and one could hypothesize that this hormone provides the first response to nutrient availability, stimulating muscle glucose uptake and thermogenic energy expenditure. Persistence of hyperglycemia, despite the action of ABA and in excess of muscle energy requirement, then results in insulin release and in the activation of the metabolic responses to nutrient abundance (glycogen and fatty acid synthesis, adipocyte differentiation and accumulation of triglycerides) (Figure 3). Interestingly, in 3T3-L1-derived murine adipocytes the siRNA-mediated downregulation of LANCL2 expression reduces both the ABA- and insulin-induced glucose uptake and downregulates Akt phosphorylation after insulin treatment [21], suggesting that levels of LANCL2 expression in adipocytes could affect insulin sensitivity.

**Figure 3.** Non-overlapping roles of ABA and insulin in the regulation of energy metabolism. Insulin, via the kinase Akt, stimulates glucose uptake in skeletal muscle and white adipose tissue (WAT), triglyceride synthesis in adipocytes, preadipocyte differentiation into WAT and increased hepatic triglyceride synthesis and lipoprotein export into the blood. Nanomolar ABA, via the AMP-dependent protein kinase (AMPK), stimulates muscle glucose uptake similarly to insulin, but the effect on adipose tissue is different. ABA does not induce preadipocytes differentiation; instead, it stimulates the expression of browning genes in the WAT and increases glucose uptake and mitochondrial uncoupling in brown adipose tissue (BAT). The effect of insulin and of ABA on body weight (BW) is opposite, with insulin inducing an increase and ABA instead favoring a decrease. The relative plasma concentrations of these hormones is likely to affect BW homeostasis and energy metabolism. Via Akt, insulin inhibits AMPK and the metabolic responses to low cell energy levels.

Activation by ABA of the transcription and phosphorylation of AMPK opens new perspectives on the signaling pathways activated by the hormone. AMPK phosphorylates and inhibits the transcriptional activity of PPAR-γ, the master regulator of adipogenesis, thereby preventing the differentiation of preadipocytes [37] and triglyceride accumulation [38]. AMPK is also an upstream positive regulator of p38 MAPK [39], which promotes PPAR-γ phosphorylation on Ser122, thus preventing PPAR-γ mediated inhibition of GLUT4 expression [40]. The partial suppression of the transcriptional activity of PPAR-γ in heterozygous PPAR-γ-deficient mice results in an improved insulin sensitivity and in a reduced tendency to obesity [41,42] and mice chimeric for wild-type and PPAR-γ null cells exhibit little or no contribution to adipose tissue formation by null cells [43]. The observations that low-dose ABA significantly reduces body weight in mice fed a high-glucose diet and in humans [20] and improves muscle glucose uptake are in agreement with the activation of AMPK in adipocytes and muscle cells.

#### *1.9. LANCL2 Is Not the Only Mammalian ABA Receptor*

The role of LANCL2 in mediating the stimulatory effect of ABA on innate immune cell function and on energy metabolism appears to be somewhat different.

Unlike inflammatory cells, where LANCL2 silencing abrogates the response to ABA [30] in adipocytes and muscle cells, silencing of LANCL2 reduces, but does not eliminate, the effect of ABA [14,17], suggesting a role for other receptors in the metabolic action of the hormone. A more direct indication that other receptors could contribute to mediate the metabolic actions of ABA comes from studies on LANCL2 knock-out mice. Indeed, in C57Bl/6 mice, the genetic ablation of LANCL2 did not modify fasting glycemia values but resulted in the reduction of glucose tolerance compared with wild-type siblings, as indicated by a significantly increased AUC of glycemia after an oral glucose load; however, LANCL2−/<sup>−</sup> mice were still responsive to exogenous ABA, which significantly reduced the AUC of glycemia after glucose load, to values similar to those of wild-type animals (Figure 4) (Magnone M., unpublished results).

This result clearly indicates that the genetic ablation of LANCL2 negatively affects glucose tolerance. The fact that exogenous low-dose ABA (1 μg/Kg body weight) improved glucose tolerance in LANCL2−/<sup>−</sup> mice suggests that another receptor can substitute for LANCL2 in the stimulation of muscle and AT glucose transport, although at higher ABA concentrations than those reached by the endogenous hormone (Magnone M., unpublished results). Indeed, intake of ABA at a dose of 1 μg/Kg body weight increases ABAp between 10 and 50 times compared to basal, endogenous levels in humans [15]. These high plasma concentrations, obtained by pharmacologic intervention, could activate a low-affinity receptor not normally participating in endogenous ABA signaling.

**Figure 4.** LANCL2−/<sup>−</sup> mice have a reduced glucose tolerance, but respond to ABA. Eight-hour fasted, male LANCL2−/<sup>−</sup> (KO) and wild-type (WT) mice (6/group) were subjected to OGTT (2 g glucose/Kg body weight) without or with ABA (1 μg/Kg body weight). Glycemia was measured before gavage (time zero) and at 30, 60 and 120 min thereafter. The incremental AUC of glycemia was calculated in the time frames 0–60 and 0–120 min by the trapezoidal rule on values relative to time zero. *p* values by two-tailed unpaired *t*-test (KO vs. WT) or by two-tailed paired *t*-test (KO + ABA vs. KO). The genetic ablation of LANCL2 in KO mice was confirmed by genotyping. Absence of LANCL2 protein expression in skeletal muscle, liver, WAT and BAT of KO mice was confirmed by Western blot (not shown), (Magnone M., unpublished results).

The identity of this receptor remains to be determined; however, the high sequence identity (54%) between LANCL2 and LANCL1 is highly suggestive of a role for LANCL1 as a second ABA receptor, in addition to LANCL2. These results suggest a redundancy in ABA-sensing molecules in mammals too, as occurs in plants, which could be expected given the strategic role of the hormone in the response to nutrients. Nonetheless, absence of LANCL2 negatively affects glucose tolerance in mice, indicating that the other ABA receptor(s) do not wholly substitute for this protein. Further studies are definitely needed to deepen our understanding of the physiology of mammalian ABA receptors, their tissue distribution and affinity for the hormone.

He et al. published a study on the triple knock out of LANCL1-2-3 in mice, demonstrating that these proteins are not involved in the synthesis of lanthionine, which conversely could have been hypothesized based on their homology with bacterial LanC enzymes. It would be informative to compare glucose tolerance and insulin sensitivity on the LANCL2 KO and on the double (LANCL1 + 2) KO [26].

#### **2. Conclusions and Future Perspectives**

In this review, we focused our attention on the metabolic function of the ABA/LANCL2 system on glycemic and lipidemic control, neglecting other important aspects of ABA physiology in mammals, such as its regulatory role in inflammation, for which we redirect the reader to the excellent review by Lievens et al. [44].

The results summarized in this review allow to draw some conclusions, but also provide the starting point for future investigations, both clinical and preclinical.

The significant beneficial effect of micrograms of oral ABA observed in volunteers with metabolic and morphometric parameters borderline with the metabolic syndrome allows to forecast similar positive results in trials of the ABA-containing nutraceutical on a higher number of prediabetic subjects and of subjects with the metabolic syndrome. The primary outcomes of this study would be the improvement of glucose and lipid tolerance and the reduction of body weight.

In addition, studies on murine model(s) of insulin-dependent diabetes could provide preclinical evidence supporting the use of oral low-dose ABA to improve glycemia control in combination with insulin. This result in turn would warrant clinical studies aimed at confirming whether ABA-containing nutraceuticals could represent an adjunctive therapy in insulin-dependent diabetic patients (both T1D and T2D), improving the daily glycemia profile, reducing blood lipids and contributing to the control of body weight by reducing the amount of insulin required for glycemic control.

The physiology and possible dysfunction of LANCL2 and of other ABA receptors still to be identified is another important area of research. Do LANCL2 mutations or constitutively low expression levels, particularly in skeletal muscle, predispose to "ABA resistance" and to diabetes? In this case, intake of the ABA-containing nutraceutical would provide the amount of ABA sufficient to increase ABAp 5- to 10-fold over endogenous levels, overtaking ABA resistance.

Another issue requiring further study is the identity of the ABA-producing cells in mammals. Results obtained on T1D patients seem to indicate that β-pancreatic cells are the main ABA producers, since ABAp is reduced by approximately 90% compared to values measured in healthy controls. In addition, rat insulinoma cells and human pancreatic islets release ABA after stimulation with GLP-1 [14]. In mice, a very high ABA concentration has been measured in the BAT, but not in the WAT, suggesting that BAT could be another source of endogenous ABA in mammals. The identification of the main ABA-releasing cells in humans is relevant to understand which physiological stimuli induce an ABA response in humans. In particular, if β-pancreatic cells were indeed the main ABA producers, the demise of these cells in T1D would compromise secretion of both hormones regulating glycemia, insulin and ABA, of which only one is currently replaced by therapy.

Finally, a field of exploration on mammalian ABA physiology which lies at the crossroad between metabolism and inflammation is the role of ABA in the central nervous system. The fact that brain has the highest ABA content among the various tissues, as first reported by Le Page-Degivry et al. [3], raises the possibility that ABA is produced and acts locally in the brain. ABA administration improves neuroinflammation and cognitive impairment and anxiety in rodents [45,46]. Interestingly, phaseic acid, the principal ABA metabolite in plants, is apparently endogenously produced in the brain and protects from ischemic injury by acting as a non-competitive inhibitor of glutamate receptors [47].

**Author Contributions:** M.M. and E.Z. wrote the review and performed the experiments on the LANCL2 KO mice; C.F. generated the LANCL2 KO mouse strain; L.S., G.B., S.S., S.B. and L.G. contributed to performing the experiments reported herein. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was supported by Nutravis S.r.l., which provided the ABA-containing food supplement and ABAMET® tested in the clinical studies and by the University of Genova (FRA).

**Acknowledgments:** This study was supported in part by the University of Genova (FRA) and by the Italian Ministry of Research (FIRB RBFR1299KO\_002).

**Conflicts of Interest:** M.M. is also the C.E.O. of Nutravis S.r.l., a University spin-off whose mission is the development of nutraceuticals; L.S., L.G., S.S., G.B., C.F. and E.Z. declare no conflicts of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Nutrients* Editorial Office E-mail: nutrients@mdpi.com www.mdpi.com/journal/nutrients

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18