**A Novel Non-Digestible, Carrot-Derived Polysaccharide (cRG-I) Selectively Modulates the Human Gut Microbiota while Promoting Gut Barrier Integrity: An Integrated In Vitro Approach**

**Pieter Van den Abbeele 1, Lynn Verstrepen 1, Jonas Ghyselinck 1, Ruud Albers 2, Massimo Marzorati 1,3 and Annick Mercenier 2,\***


Received: 30 May 2020; Accepted: 24 June 2020; Published: 29 June 2020

**Abstract:** Modulation of the gut microbiome as a means to improve human health has recently gained increasing interest. In this study, it was investigated whether cRG-I, a carrot-derived pectic polysaccharide, enriched in rhamnogalacturonan-I (RG-I) classifies as a potential prebiotic ingredient using novel in vitro models. First, digestion methods involving α-amylase/brush border enzymes demonstrated the non-digestibility of cRG-I by host-derived enzymes versus digestible (starch/maltose) and non-digestible controls (inulin). Then, a recently developed short-term (48 h) colonic incubation strategy was applied and revealed that cRG-I fermentation increased levels of health-promoting short-chain fatty acids (SCFA; mainly acetate and propionate) and lactate comparable but not identical to the reference prebiotic inulin. Upon upgrading this fermentation model by inclusion of a simulated mucosal environment while applying quantitative 16S-targeted Illumina sequencing, cRG-I was additionally shown to specifically stimulate operational taxonomic units (OTUs) related to health-associated species such as *Bifidobacterium longum*, *Bifidobacterium adolescentis*, *Bacteroides dorei*, *Bacteroides ovatus*, *Roseburia hominis*, *Faecalibacterium prausnitzii*, and *Eubacterium hallii*. Finally, in a novel model to assess host–microbe interactions (Caco-2/peripheral blood mononuclear cells (PBMC) co-culture) fermented cRG-I increased barrier integrity while decreasing markers for inflammation. In conclusion, by using novel in vitro models, cRG-I was identified as a promising prebiotic candidate to proceed to clinical studies.

**Keywords:** prebiotic; microbiome; SCFA; colon; bifidobacteria; pectin; rhamnogalacturonan; transepithelial electrical resistance (TEER)

#### **1. Introduction**

The colon contains a vast number of bacteria that largely impact human health. Next to preventing pathogen colonization through secretion of antimicrobial agents [1–3], the gut microbiota is involved in food processing, synthesis of essential vitamins and production of health-promoting short-chain fatty acids (SCFA), including acetate, propionate and butyrate [4], upon anaerobic fermentation of for instance dietary fibers [5]. While butyrate is an important energy source for colonocytes with anti-inflammatory and intestinal barrier-protecting effects, propionate exerts anti-lipogenic and cholesterol-lowering effects in the liver [6]. In addition, as with butyrate, propionate has been reported

to exert anti-cancer effects in the colon [7,8]. Finally, acetate is used in the liver as a substrate for cholesterol and fatty acid synthesis [9,10]. In terms of composition, the human gut microbiome mainly consists of the *Firmicutes*, *Bacteroidetes*, *Actinobacteria*, *Proteobacteria*, *Fusobacteria*, and *Verrucomicrobia* phyla [11]. Despite having provided key insights, many studies have been limited to (descriptive) analysis of fecal samples as in situ samples from the site of fermentation are difficult to obtain. To allow in-depth research focusing on not only luminal, but also gut wall-associated mucosal microbes, a novel in vitro model (M-SHIME®; Mucosal Simulator of the Human Intestinal Microbial Ecosystem) was recently developed as a complementary in vitro tool [12]. In this model, the mucosal microbiota was enriched with butyrate-producing *Clostridium* cluster XIVa members, correlating with in vivo findings from biopsies [13–18].

Modulation of the human gut microbiome as a route to improve human health has gained a lot of interest over recent years. Prebiotics are defined as non-digestible food ingredients that selectively stimulate health-promoting bacteria [19]. A key feature of prebiotics is their resistance to upper gastro-intestinal digestion so that they reach the colon where they are fermented by the gut microbiome. To assess potential digestibility of polysaccharides, two complementary enzyme sources are to be considered, i.e., amylases [20] and brush border enzymes [21]. α-amylase is present in both saliva and pancreatic juice and can liberate maltose from starch [20]. Furthermore, sucrase-isomaltase and maltase-glycoamylase, collectively known as α-glucosidases, are complexes consisting of 4 enzymes that release glucose from oligosaccharides present at the intestinal brush border [21]. These host-derived enzymes jointly digest carbohydrates and hence their specificity determines whether carbohydrates reach the colon and can exert prebiotic effects. Although critical to the definition of potential prebiotic ingredients, studies confirming their indigestibility are scarce.

Fructans, such as fructooligosaccharides (FOS) and inulin are considered to be "gold standard" prebiotics, with human clinical trials supporting their beneficial effect in acute and chronic diseases such as obesity and type 2 diabetes (T2D), allergy, inflammatory bowel disease (IBD), Traveler's diarrhea and constipation (an overview is given in [19]). As many health-related species belong to the *Bifidobacteriaceae*, prebiotic potential has often been related to an increase of this family. There is however increasing understanding that prebiotics can be fermented by a wider range of gut microbes. Inulin can e.g., also be rapidly fermented by health-promoting *Bacteroidaceae* members, such as *Bacteroides caccae* [22,23]. Currently, there is growing interest to develop novel prebiotics. A specific class of candidates includes pectin-derived polysaccharides enriched in the branched part of pectin, i.e., the rhamnogalacturonan-I (RG-I) domains, which can be extracted from several food crops including carrot. The backbone of RG-I is a repeating unit of the disaccharide [-2)-α-L-rhamnose-(1,4)-α-D-galacturonic acid-(1] and RG-I side-chains consist of galactans (β-1,4-D-galactose (D-Gal) units) and/or arabinans (α-1,5-linked L-arabinofuranose (L-Araf) units with additional L-Araf side-chains), with varying length and composition [24]. Given their structural complexity, RG-I extracts would require fermentation by a consortium of gut microbes with complementary metabolic capabilities [25].

Gut microbial modulation is linked to human health with the concept of a "leaky gut" having gained attention, not only in the context of inflammatory bowel disease, but also in a wider range of psychological and metabolic disorders [26]. Increased intestinal epithelial permeability would allow translocation of bacterial cell wall components, metabolites, or even whole bacteria into the systemic circulation, hence contributing to inflammation and injury, not only in the gut but also in remote organs such as the liver and the brain. Host–microbe interaction studies are increasingly being performed to document this. As an example, microbial fermentation samples can be combined with a human co-culture model [27], including intestinal epithelial cells (Caco-2) and monocytes (THP-1) [28], which demonstrated the gut barrier protective effects together with immuno-modulatory capacity of several prebiotics, including arabinoxylo-oligosaccharides (AXOS) [29], inulin, FOS [30] and a dried yeast fermentate [31]. Despite its usefulness, this model has the limitation that monocytes are only one of the cell lineages involved in the immune response. Therefore, using a co-culture model including Caco-2

cells and peripheral blood mononuclear cells (PBMCs), containing lymphocytes (T-cells, B-cells, and NK-cells), monocytes and dendritic cells [32], could increase the in vivo relevance.

Therefore, the present study investigated whether a carrot-derived RG-I enriched extract (cRG-I) classifies as a potential prebiotic ingredient using a combination of novel in vitro models (Figure 1). First, potential digestion by amylase/brush border enzymes was investigated (Test 1). Then, a recently developed short-term colonic incubation strategy was applied [29] to assess the potential impact of cRG-I on microbial metabolic activity including inulin as a reference (Test 2). Subsequently, after upgrading the fermentation model by inclusion of a simulated mucosal microbiota [12], more in-depth fermentation tests were performed to characterize the prebiotic potential of two carrot RG-I formulations that differed in absence (cRG-I) or presence of low molecular weight carbohydrates (cRG-I+LMWC) (Test 3). To obtain detailed insights in modulation of microbial composition, a novel technique was used where flow cytometry was combined with 16S-targeted Illumina sequencing to obtain quantitative information at high phylogenetic resolution [33]. Finally, fermentation samples were screened for potential beneficial effects on gut barrier integrity and immune modulation in a newly optimized Caco-2/PBMC co-culture model (test 4).

**Figure 1.** Schematic representation of the integrated in vitro approach to investigate the prebiotic potential of cRG-I (carrot-derived rhamnogalacturonan-I). First, potential digestion of cRG-I by amylase/brush border enzymes was investigated (Test 1). In Test 2, short-term colonic batch incubations were used to assess the prebiotic potential on microbial metabolic activity of cRG-I compared to inulin. In Test 3, the prebiotic potential of two formulations that differed in absence (cRG-I) or presence of low molecular weight carbohydrates (cRG-I+LMWC) was assessed in both conventional luminal incubations (L) and incubations including a mucosal compartment (M). Samples were collected to evaluate the effect of the test products on microbial activity, community composition and on intestinal permeability and immunity (Test 4, Caco-2/PBMC co-culture model).

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

#### *2.1. Products*

The two carrot RG-I preparations used in this study (cRG-I [34] and cRG-I+LMWC) were provided by NutriLeads (Wageningen, The Netherlands). Both are pectin-derived polysaccharides and cRG-I+LMWC differs from cRG-I by containing small size sugars, mainly mono- and disaccharides of galactose, glucose, and uronic acids. Pectin is a linear homogalacturonan (HG) interspaced with branched rhamnogalacturonan (RG) regions. HG consists of α-1,4-linked D-galacturonic acid (GalA) monomers while the RG-I backbone is a repeating unit of the disaccharide [-2)-α-L-rhamnose-1,4)- (α-D-galacturonic acid-(1]. The RG-I backbone is decorated with galactans (β-1,4-D-galactose (D-Gal) units) and/or arabinans (α-1,5-linked L-arabinofuranose (L-Araf) units [25]. Inulin (Orafti® HP, 100% inulin, 0% sweetness level, average DP ≥ 23) was generously provided by Beneo GmbH (Mannheim, Germany).

#### *2.2. Digestion by Amylase and Brush Border Enzymes (Test 1)*

Digestion with amylase was performed as described previously [20]. Briefly, cooked starch (positive control), inulin (negative control) and cRG-I were suspended in distilled water at 15 g/L. A stock solution of 1500U α-amylase/mL (10080, Sigma–Aldrich, Bornem, Belgium) was prepared and added to the substrates to simulate the small intestinal phase, while respecting the ratio of units of amylase versus amount of test product according to the Infogest consensus method (1300 units per gram test product [20]). Samples were incubated for 60 at 37 ◦C. Furthermore, digestion with brush border enzymes was performed as previously described in [21]. Briefly, intestinal aceton powder from rat (Sigma–Aldrich, Bornem, Belgium) was dissolved in 0.9% NaCl solution, vortexed, and sonicated. 15 g/L stock solutions of inulin (negative control) and maltose (positive control) were prepared in sodium phosphate buffer (pH 7), while cRG-I was prepared in distilled water. 100 μL enzyme solution and 50 μL substrates were mixed with 100 μL phosphate buffer and incubated for 90 at 37 ◦C. High Performance Anion Exchange Chromatography with Pulsed Amperometric Detection (HPAEC-PAD) was used to measure rhamnose, arabinose, galactose, fructose, glucose, maltose, and galacturonic acid in both digestion experiments. All tests were done in technical triplicate.

#### *2.3. Short-Term Colonic Batch Incubations (Tests 2 and 3)*

Short-term colonic incubations were performed as described in [29]. Briefly, freshly collected fecal material of a healthy human donor (f, 26) was collected and after preparation of an anaerobic fecal slurry inoculated at 10 vol% in a sugar-depleted nutritional medium containing 5.2 g/L K2HPO4, 16.3 g/L KH2PO4, 2.0 g/L NaHCO3 (Chem-lab NV, Zedelgem, Belgium), 2.0 g/L Yeast Extract, 2.0 g/L pepton (Oxoid, Aalst, Belgium), 1.0 g/L mucin (Carl Roth, Karlsruhe, Germany), 0.5 g/L L-cystein and 2.0 mL/L Tween80 (Sigma–Aldrich, Bornem, Belgium). When mucin-coated carriers were added to the reactors during Test 3, 1.0 g/L mucin was omitted from the nutritional medium. Five mucin-coated carriers were added per reactor after being prepared according to Van den Abbeele et al. (2013) [12]. Test products were dosed at 5 g/L and reactors were anaerobically incubated at 37 ◦C for 48 h. All experiments were performed in technical triplicate.

#### *2.4. Microbial Metabolic Activity (Tests 2 and 3)*

Samples were collected upon 0 h, 6 h, 24 h, and 48 h of incubation from each colon reactor. Gas production was measure with a pressure meter (Hand-held pressure indicator CPH6200; Wika, Echt, The Netherlands) and pH measurements were performed with a Senseline pH meter F410 (ProSense, Oosterhout, The Netherlands). Total SCFA were determined as the sum of acetate, propionate, butyrate and branched-chain fatty acids (bCFA; isobutyrate, isovalerate and isocaproate) levels, and were measured as described previously [35]. Lactate production was assessed with a commercially available kit (R-Biopharm, Darmstadt, Germany), according to manufacturer's instructions.

#### *2.5. Microbial Community Composition (Test 3)*

After 48 h of incubation, samples from both lumen and mucus were collected for analysis of the microbial community composition through quantitative polymerase chain reaction (qPCR) and 16S-targeted Illumina sequencing. DNA was isolated as described in [36] from either 1 mL luminal samples or 0.1 g mucus samples. Subsequently, qPCR was performed on a QuantStudio 5 Real-Time PCR system (Applied Biosystems, Foster City, CA, USA). Each sample was run in technical triplicate and outliers with more than 1 CT difference were omitted. The qPCRs were performed as described previously for the following groups: *Lactobacillus* spp. [37], *Bifidobacterium* spp. and *Eubacterium rectale*/*Clostridium coccoides* [38], *Akkermansia muciniphila* [39], *Bacteroidetes* [40], *Enterobacteriaceae* [41], *Faecalibacterium prausnitzii* [42], *Roseburia* and *Eubacterium hallii* [43]. In addition, microbiota profiling was performed using 16S-targeted Illumina sequencing analysis (LGC genomics GmbH, Berlin, Germany) as described in [44] to obtain proportional abundances (%) at different phylogenetic levels (phylum, family, and operational taxonomic unit (OTU) level). Briefly, library preparation and sequencing were performed on an Illumina MiSeq platform with v3 chemistry. The 16S rRNA gene V3-V4 hypervariable regions were amplified using primers 341F (5 -CCT ACG GGN GGC WGC AG-3 ) and 785Rmod (5 -GAC TAC HVG GGT ATC TAA KCC-3 ) [45]. As described in [46,47], the 16S-targeted sequencing analysis was adapted from the MiSeq protocol for read assembly and cleanup using the mothur software (v. 1.39.5) as follows: (1) reads were assembled into contigs, (2) alignment-based quality filtering was performed by alignment to the mothur-reconstructed SILVA SEED alignment (v. 123), (3) chimeras were removed, (4) taxonomy was assigned via a naïve Bayesian classifier [48] and RDP release 14 [49] and (5) contigs were clustered into OTUs at 97% sequence similarity. Sequences classified as Eukaryota, Archaea, Chloroplasts, Mitochondria, and non-classified sequences were also removed. For each OTU, representative sequences were selected as the most abundant sequence within that OTU. Finally, the obtained high-resolution proportional phylogenetic information (i.e., proportional abundances (%)) was combined with an accurate quantification of total bacterial cells via flowcytometry to obtain quantitative data at phylum, family, and OTU level. This was done by multiplying the proportional abundances with absolute cell numbers (cells/mL) obtained via flowcytometry. For flowcytometry analysis, 10-fold serial dilutions were prepared in Dulbecco's Phosphate-buffered Saline (DPBS) (Sigma–Aldrich, Bornem, Belgium) of all samples and stained with 0.01 mM SYTO24 (Life Technologies Europe, Merelbeke, Belgium) for 15 at 37 ◦C in the dark. Samples were analyzed on a BD Facsverse (BDBiosciences, Erembodegem, Belgium) using the high-flow-rate setting and bacteria were separated from medium debris and signal noise by applying a threshold level of 200 on the SYTO channel. Flowcytometry data were analyzed using FlowJo, version 10.5.2.

#### *2.6. Caco-2*/*PBMC Co-Culture Model (Test 4)*

Caco-2 cells (HTB-37; American Type Culture Collection) were cultured in Dulbecco's Modified Eagle Medium (DMEM) containing glucose and glutamine (Sigma–Aldrich, Bornem, Belgium) and supplemented with HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid) and 20% (v/v) heat-inactivated (HI) fetal bovine serum (FBS) (Gibco, Life Technologies Europe, Merelbeke, Belgium). PBMCs were isolated from buffy coats of healthy donors (Red Cross, Ghent, Belgium) using LymphoprepTM (STEMCELL technologies SARL, Grenoble, France). In brief, blood was collected and diluted (1/5, v/v) in DPBS without Ca/Mg (Sigma–Aldrich, Bornem, Belgium). Then, LymphoprepTM solution was added and samples were centrifuged at 1027× *g* for 20 at room temperature (RT) to separate the mononuclear cells (MNCs) and red blood cells by density-gradient centrifugation. MNCs were collected and washed 3 times with ice-cold DPBS (340× *g*, 7 ). Aliquots of PBMCs were frozen in liquid nitrogen. Baseline IL-8 levels were determined by enzyme-linked immunosorbent assay (ELISA) (Invitrogen, Thermo Fisher Scientific, Merelbeke, Belgium) to eliminate donors with high basal cytokine levels. For experiments, Caco-2 cells were seeded on 24-well semipermeable inserts (0.4 μm pore size) at a density of 1 <sup>×</sup> 105 cells/insert and cultured for 14 days until a functional cell monolayer with a transepithelial electrical resistance (TEER) of more than 300 Ωcm2 was obtained. PBMCs, stimulated

with 2.5 μg/mL pokeweed mitogen (PWM) were added to the basolateral chamber at a concentration of 1 <sup>×</sup> 106 cells/well. PBMCs without PWM stimulation were included as negative control. At the same time, colonic suspensions (filter-sterilized (0.22 μm) and diluted 1/5 (v/v) in culture medium) or 5 mM sodium butyrate (NaB) (Sigma–Aldrich, Bornem, Belgium) were added to the Caco-2 cells at the apical side. Caco-2 cells were also treated with medium as negative control. Cells were incubated for 48 h at 37 ◦C in a humidified atmosphere of air/CO2 (95:5, v/v). TEER was measured at start (0 h timepoint) and after 48 h of incubation using a Millicell ERS2 Voltohmmeter (EMD Millipore, Sigma–Aldrich, Bornem, Belgium). All 48 h values were normalized to their own 0 h value after subtraction of the empty insert value and are presented as percentage of initial value. In addition, basolateral supernatant was collected after 48 h of incubation for cytokine analysis. Human IFN-γ, IL-17A, IL-21, IL-22, IL-4, and IL-9 levels were determined by Luminex® multiplex (Procartaplex, Invitrogen, Thermo Fisher Scientific) and IL-10 levels were measured by ELISA (Invitrogen, Thermo Fisher Scientific), according to the manufacturers' instructions. All experiments were performed in triplicate.

#### *2.7. Statistics*

To evaluate differences in microbial metabolites (tests 2 and 3) and microbial community composition at phylum level between blank and treatment incubations (Test 3), a two-way analysis of variance (ANOVA) with Dunnett's multiple comparisons test was performed. Statistically significant differences between the blank and treatments are presented by (\*, Δ0 h–6 h), (\$, Δ6 h–24 h) or (#, Δ24 h–48 h). 1 sign = *p* < 0.05, 2 signs = *p* < 0.01, 3 signs = *p* < 0.001 and 4 signs = *p* < 0.0001. Statistical analysis was performed with the GraphPad Prism software (version 8.3.0, San Diego, USA). To evaluate differences in microbial community composition at family and OTU level between blank and treatment incubations (Test 3), a Student's t-test was performed (Excel Software). Differences were found significant if *p* < 0.05. To evaluate differences between PWM+ and PWM- or NaB in the Caco-2/PBMC co-culture assay (Test 4), an ordinary one-way ANOVA with Dunnett's multiple comparisons test was performed; while differences between blank and treatment incubations were assessed with an ordinary one-way ANOVA with Tukey's multiple comparisons test. Statistically significant differences are presented by (\*). (\*) = *p* < 0.05, (\*\*) = *p* < 0.01, (\*\*\*) = *p* < 0.001 and (\*\*\*\*) = *p* < 0.0001. Statistical analysis was performed with the GraphPad Prism software (version 8.3.0, San Diego, USA).

#### **3. Results**

#### *3.1. cRG-I is Resistant to Digestion in the Human Upper Gastro-Intestinal Tract (GIT) (Test 1)*

One of the characteristics of a prebiotic is its non-digestibility by host enzymes upon passage along the upper GIT [19]. First, upon exposure to amylase, the positive control cooked starch was readily degraded into maltose, demonstrating its known digestibility (Figure 2A), while the negative control inulin and also cRG-I were not digested to any of the simple sugars measured. Likewise, maltose was digested by brush border enzymes into glucose (Figure 2B), in contrast to the negative control inulin and cRG-I. Therefore, cRG-I can be considered to be a polysaccharide that likely escapes upper GIT digestion in vivo thereby reaching the colon where it could be fermented by the gut microbiota.

#### *3.2. E*ff*ect of cRG-I on Microbial Metabolic Activity in Short-Term Colonic Incubations (Test 2)*

Short-term colonic incubations were performed to investigate the potential prebiotic effect of cRG-I on microbial activity, including inulin as a positive control. A first indication of cRG-I fermentation by the gut microbiota followed from the significant pH decrease and enhanced gas production during the first 6 h of incubation (Δ0–6 h) versus the blank (Figure 3A,B). These changes were even stronger compared to inulin. Between 6 h and 24 h, cRG-I significantly and strongly decreased pH while increasing gas production. pH decreases and gas production were more excessive for inulin. Finally, during the 24 h to 48 h time interval, relatively stable pH values and gas levels indicated substrate depletion.

**Figure 2.** Digestion of cRG-I (carrot-derived rhamnogalacturonan-I) by amylase (**A**) and brush border enzymes (**B**) versus inulin, cooked starch, and maltose. Blank incubations containing all reagents in absence of a substrate were included. Average (± st. dev.) concentrations of different monosaccharides and maltose were measured by High Performance Anion Exchange Chromatography (HPAEC) (*n* = 3).

**Figure 3.** Effect of fermentation of cRG-I on microbial metabolic activity in short-term colonic incubations. Average changes (± st. dev.) in pH (**A**), gas production (**B**), total short-chain fatty acids (SCFA) (**C**), acetate (**D**), propionate (**E**), butyrate (**F**), branched CFA (bCFA) (**G**) and lactate (**H**) levels between 0 and 6 h (light gray), 6 h and 24 h (dark gray) and 24 h and 48 h (stripes) upon dosing cRG-I and inulin to the gut microbiota versus a blank incubation (*n* = 3). Statistically significant differences between blank and treatments for different time intervals are presented by \* for Δ0 h–6 h, \$ for Δ6 h–24 h or # for Δ24 h–48 h. (\*, \$, #) = *p* < 0.05, (\*\*, \$\$, ##) = *p* < 0.01, (\*\*\*, \$\$\$, ###) = *p* < 0.001 and (\*\*\*\*, \$\$\$\$, ####) = *p* < 0.0001.

Upon measuring lactate and SCFA production, acids responsible for aforementioned pH changes were elucidated. During the 0 h to 6 h time interval, cRG-I, but not inulin, significantly increased the production of total SCFA (Figure 3C) which resulted from more strongly elevated acetate levels (Figure 3D) next to significant raises in lactate production for cRG-I (Figure 3H). Furthermore, both treatments strongly augmented total SCFA production during the 6 h to 24 h time interval (merely due to increases in acetate and propionate (Figure 3D,E)), which were more profound for cRG-I compared to inulin. cRG-I, unlike inulin, also stimulated butyrate production during this interval (Figure 3F), which coincided with lactate consumption. Finally, during the 24 h to 48 h time interval, total SCFA production only increased upon treatment with inulin, which was related to increases of acetate, propionate, and butyrate levels, thus indicating slower fermentation of inulin versus cRG-I (Figure 3C–F). The stimulation of butyrate by inulin between 24–48 h correlated with the consumption of lactate during this time interval. Overall, lactate was entirely consumed at the end of the incubations, suggesting optimal conversion to propionate and/or butyrate [50]. Finally, bCFA result from protein fermentation by the gut microbiota [51,52], which is associated with detrimental health effects [51]. bCFA were mainly produced during the 6 h to 48 h time frame (Figure 3G), with both cRG-I and especially inulin significantly decreasing bCFA production.

#### *3.3. E*ff*ect of cRG-I and cRG-I* +*LMWC on Microbial Metabolic Activity in Short-Term Colonic Incubations with or without a Mucosal Compartment (Test 3)*

An in-depth characterization of the effects of cRG-I on microbial metabolic activity (Figure 4A–H) was performed by testing both cRG-I and a modified formulation containing small sugars (cRG-I +LMWC), dosed to colonic incubations including solely a luminal (L) or additionally also a mucosal (M) environment. In consistency with Test 2, cRG-I significantly decreased the pH during the first 6 h of incubation versus the blank due to enhanced acetate and lactate levels. Furthermore, the main fermentation occurred between 6 h and 24 h with strong increases in gas production and further decreases of pH due to stimulation of acetate, propionate, and to a lesser extent butyrate. The latter could again be linked to coinciding lactate consumption. Finally, the absence of marked changes of aforementioned parameters between 24 h and 48 h again indicated substrate depletion, while bCFA were significantly decreased thus illustrating the potential protective effects of cRG-I against toxic by-products of proteolytic fermentation.

cRG-I+LMWC exerted similar effects on microbial activity versus cRG-I with some minor differences. These included a more profound pH decrease for cRG-I+LMWC during the first 6 h of incubation due to enhanced acetate and lactate levels. Furthermore, acetate and propionate production between 6–24 h was less strongly increased with cRG-I+LMWC compared to cRG-I.

Overall, while no major differences were observed between incubations with or without mucosal environment, inclusion of mucus beads led to a tendency to higher gas production, with higher butyrate levels for the blank incubation, suggesting colonization of butyrate-producing bacteria on the mucin-coated carriers.

#### *3.4. E*ff*ect of cRG-I and cRG-I* +*LMWC on Microbial Community Composition in Short-Term Colonic Incubations with or without a Mucosal Compartment (Test 3)*

The data at phylum level are presented both as proportional (Figure 5A) and absolute values (Figure 5B). This demonstrated that for the luminal microbiota, quantitative data revealed greater insight in the true compositional changes since both cRG-I and cRG-I+LMWC largely increased total cell numbers. In contrast, due to the large variation in total cell numbers within identical replicates for the mucosal microbiota resulting in large variations of quantitative numbers, proportional abundances were preferred to draw conclusions on modulation of the mucosal microbiota. Therefore, abundances at the family level are presented as absolute data for the luminal microbiota, whereas they are presented as proportional values for the mucosal compartment (Table 1). Furthermore, proportional abundance of the 25 most abundant OTUs and 7 additional OTUs affected by at least one of the treatments are shown in Supplementary Table S1 to get insights at the highest phylogenetic resolution possible.

**Figure 4.** Effect of fermentation of cRG-I and cRG-I+LMWC on microbial metabolic activity in short-term colonic batch simulations in absence or presence of a mucosal compartment. Average changes (± st. dev.) in pH (**A**), gas production (**B**), total short-chain fatty acids (SCFA) (**C**), acetate (**D**), propionate (**E**), butyrate (**F**), branched CFA (bCFA) (**G**) and lactate (**H**) levels between 0 and 6 h (light gray), 6 h and 24 h (dark gray) and 24 h and 48 h (stripes) upon dosing cRG-I and cRG-I + LMWC to the gut microbiota versus a blank incubation (*n* = 3). Statistically significant differences between the blank and treatments for different time intervals are presented by \* for Δ0 h–6 h, \$ for Δ6 h–24 h or # for Δ24 h–48 h). (\*, \$, #) = *p* < 0.05, (\*\*, \$\$, ##) = *p* < 0.01, (\*\*\*, \$\$\$, ###) = *p* < 0.001 and (\*\*\*\*, \$\$\$\$, ####) = *p* < 0.0001. L = colonic incubations only simulating lumen; M = incubations simulating both lumen and mucus.

**Figure 5.** Effect of fermentation of cRG-I and cRG-I+LMWC on microbial community composition at phylum level in short-term colonic incubations in absence or presence of a mucosal compartment. Average (± st. dev.) proportional (%) (**A**) and absolute (16S gene copies/mL) (**B**) abundance of the different phyla in the original (diluted) inoculum (INO) and after 48 h of incubation upon dosing of cRG-I and cRG-I+LMWC versus a blank control (n = 3). Statistically significant differences between the blank and treatments are presented by \*. (\*) = *p* < 0.05, (\*\*) = *p* < 0.01, (\*\*\*) = *p* < 0.001 and (\*\*\*\*) = *p* < 0.0001. L = colonic incubations only simulating lumen; M = incubations simulating both lumen and mucus.


#### *Nutrients* **2020** , *12*, 1917

**Table 1.** Effect of

fermentation

 of cRG-I and

cRG-I+LMWC

 on microbial composition

 at family level in short-term colonic incubations

 in absence or presence of a

First, upon comparing the luminal and mucosal microbiota, it followed that the Firmicutes phylum was enriched in the mucosal compartment, while *Actinobacteria*, *Proteobacteria* and *Verrucomicrobia* were enriched in the lumen (Figure 5A), in accordance to what has been published for the M-SHIME® model [12]. At family level, the mucosal *Firmicutes* enrichment was due to a marked enrichment in *Clostridiaceae* cluster I and *Lachnospiraceae* (Table 1). At OTU level, this was reflected by an enrichment of OTU14 (related to *Clostridium butyricum*), OTU10 (related to *Clostridium tertium*), OTU8 (related to *Clostridium paraputrificum*), OTU12 (related to *Ruminococcus torques*), OTU3 (related to *Roseburia hominis*) and OTU19 (related to *Ruminococcus lactaris*) (Table S1). Furthermore, the decreased mucosal levels of *Actinobacteria*, *Proteobacteria* and *Verrucomicrobia* were solely related to a decreased mucosal colonization of members of the *Eggerthellaceae*, *Enterobacteriaceae*, and *Akkermansiaceae*. Another overall finding of the in vitro model was that the luminal microbiota of blank incubations in absence and presence of a mucosal environment were highly similar. The introduction of mucin beads only resulted in an enrichment of the luminal *Verrucomicrobia* levels (Figure 5A,B). At family level, this was due to an increased luminal abundance of *Akkermansiaceae* (+0.55 log copies/mL) (Table 1).

With respect to treatment effects in the lumen, cRG-I and cRG-I+LMWC both increased the absolute levels of *Actinobacteria*, *Bacteroidetes* and *Proteobacteria* compared to the blank (Figure 5B). Although the increase of *Actinobacteria* and *Bacteroidetes* was strongest for cRG-I, the increase in *Proteobacteria* was strongest for cRG-I+LMWC. cRG-I additionally increased luminal *Firmicutes* levels. The luminal increase in *Actinobacteria* by cRG-I was due to a significant increase of OTUs related to *Bifidobacterium longum* (OTU7) and *Bifidobacterium adolescentis* (OTU21) (Table S1), thus also strongly increasing *Bifidobacteriaceae* levels upon cRG-I treatment (Table 1). Furthermore, cRG-I also stimulated mucosal *Bifidobacteriaceae*, mostly due to the stimulation of the *Bifidobacterium longum*-related OTU7. The luminal *Bacteroidetes* increase upon treatment with both products was due to the stimulation of the *Bacteroidaceae* and *Prevotellaceae* families, with again the highest levels being reached for cRG-I. At OTU level, a wide spectrum of *Bacteroidaceae* members were stimulated upon dosing both cRG-I and cRG-I+LMWC, including OTUs related to *B. ovatus* (OTU22), *B. plebeius* (OTU6), *B. xylanisolvens* (OTU18) and especially *B. dorei* (OTU2). As a remark, a decrease in abundances of OTUs related to *B. caccae* (OTU13), *B. fragilis* (OTU11) and *B. uniformis* (OTU17) upon cRG-I and cRG-I+LMWC treatment was noted, but these decreases were less profound compared to observed increases in other OTUs. The increased abundance in *Proteobacteria* related to an enrichment of *Desulfovibrionaceae* and *Enterobacteriaceae*, which was most pronounced upon dosing of cRG-I+LMWC. Finally, the luminal increase in *Firmicutes* with cRG-I and cRG-I+LMWC was due to increased levels of *Erysipelotrichaceae*, *Peptostreptococcaceae*, *Streptococcacae*, *Ruminococcaceae*, and *Veillonellaceae*. The enrichment in *Streptococcacae* was most pronounced upon dosing cRG-I+LMWC and linked to an increase in OTU26 (related to *Streptococcus aginosus*). In contrast, the increase in *Veillonellaceae* was more pronounced upon dosing of cRG-I and was attributed to an increase in OTU9 (related to *Dialister succinatiphilus*). Within the *Ruminococcaceae*, two OTUs related to *Faecalibacterium prausnitzii* (i.e., OTU83 and OTU5) increased upon cRG-I, while only OTU5 was stimulated by cRG-I+LMWC and this to a lesser extent.

In the mucosal compartment, both treatments slightly increased *Firmicutes*, while decreasing *Bacteroidetes* levels (Figure 5A). Both cRG-I and cRG-I+LMWC strongly enriched *Lachnospiraceae* (Table 1), due to a marked stimulation of OTU3 (related to *Roseburia hominis*) that increased from 9.8% in the blank control incubations to 64.3% and 74.2%, respectively (Table S1). Although both treatments also significantly increased the proportion of *Streptococcacae* in the mucosal environment, only cRG-I increased mucosal *Actinobacteria* abundances.

To confirm several of the above-mentioned observations obtained through 16S-targeted Illumina sequencing, qPCRs on specific bacterial groups of interest were performed (Supplementary Table S2). This confirmed the key aforementioned conclusions obtained through 16S-targeted Illumina sequencing including that both cRG-I and cRG-I+LMWC stimulated levels of (i) luminal/mucosal *Bifidobacteriaceae* (and not *Lactobacillaceae*); (ii) luminal *Bacteroidetes*; (iii) luminal *Faecalibacterium prausnitzii*; (iv) mucosal *Roseburia* (and *Eubacterium rectale*/*Clostridium coccoides* group to which it belongs); while not affecting

*Akkermansia muciniphila*. In addition, both products, but mostly cRG-I+LMWC, increased colonization of *Eubacterium hallii* in the simulated mucus layer.

#### *3.5. E*ff*ect of Fermented cRG-I on Intestinal Epithelial Barrier (Test 4)*

As cRG-I fermentation enriched several health-associated species and increased production of SCFA, it might exert favorable effects at the host level. To address this question, a Caco-2/PBMC co-culture model was developed. Inflammation-induced barrier disruption was obtained upon 48 h co-culturing of Caco-2 cells with PWM-activated PBMCs and measured as a significant decrease in TEER of the Caco-2 monolayers (Figure S1A). In addition, apical treatment with the positive control sodium butyrate (NaB) prevented this TEER decrease. Likewise, apical treatment with fermented cRG-I showed a significant increase in TEER compared to the blank controls (Figure 6A). Furthermore, effects on T-cell dependent cytokine production were assessed. As shown in Figure S1, NaB significantly decreased the secretion of interferon (IFN)γ, interleukin (IL)-17A, IL-21, IL-4, and IL-9; while increasing the secretion of IL-22; a cytokine involved in maintenance of barrier integrity, wound healing and antimicrobial responses [53]. However, in contrast to its positive effects on IL-10 secretion in the Caco-2/THP1 co-culture assay [29], NaB significantly decreased IL-10 secretion possibly due to toxicity of NaB on IL-10 producing cells at the concentration used [54,55]. Compared to the blank control, fermented cRG-I reduced the secretion of IL-17A, IL-4, and IL-9; reaching statistical significance for IL-17A and IL-4 upon luminal incubations and upon both incubations for IL-9 (Figure 6D,F,G). Furthermore, luminal fermentation of cRG-I tended to increase the secretion of IL-22 and IL-10; of which the latter was already increased upon treatment with the blank controls (Figure 6E,H). In addition, all colonic suspensions completely abolished PWM-induced IL-21 secretion (Figure 6C); while no significant differences were observed on IFNγ secretion (Figure 6B). Hence, metabolites generated from colonic fermentation of cRG-I displayed anti-inflammatory and gut barrier protective properties.

**Figure 6.** Effect of fermented cRG-I on transepithelial electrical resistance (TEER) and cytokine production in a Caco-2/PBMC co-culture system. Caco-2 cells, cultured 14 days on transwell inserts, were placed on top of pokeweed mitogen (PWM)-activated PBMCs and incubated for 48 h at the apical side with blank or treatment samples collected during the colonic batch incubations containing only a luminal (L) or also a mucosal (M) environment. Average (±SEM) TEER of the Caco-2 monolayers (**A**) and concentration of secreted interferon (IFN)γ (**B**), interleukin (IL)-17A (**C**), IL-21 (**D**), IL-22 (**E**), IL-4 (**F**), IL-9 (**G**) and IL-10 (**H**) in the basolateral medium are shown (n = 3). Statistically significant differences to PWM are represented by (\*). (\*) = *p* < 0.05, (\*\*) = *p* < 0.01, (\*\*\*) = *p* < 0.001.

Overall, some minor effects of the incorporation of a mucosal compartment were observed as both blank and treatment colonic suspensions slightly increased TEER; while decreased the secretion of all tested cytokines compared to luminal suspensions (Figure 6).

#### **4. Discussion**

In the current study, novel in vitro models and analytical techniques were implemented to investigate whether cRG-I classifies as a potential prebiotic ingredient. First, upon exposure to α-amylase and brush border enzymes (Test 1), unlike the positive controls (cooked starch and maltose, respectively), cRG-I (like the negative control inulin) was not digested to any of the simple sugars measured. This suggests that cRG-I can be considered to be a polysaccharide that likely escapes upper GIT digestion in vivo, thus fulfilling the definition of a prebiotic ingredient that should reach the colon where it could be fermented by the gut microbiota. Secondly, a recently described short-term colonic incubation strategy [29] was upgraded with a simulation of the mucosal microbiota (Test 3). By applying a novel technique to analyze the microbial community composition, i.e., quantitative 16S-targeted Illumina sequencing [33], in-depth quantitative information was obtained at high phylogenetic resolution. Besides elucidating treatment effects of cRG-I, this allowed to validate the implementation of mucin-coated carriers in the short-term incubations by demonstrating a relevant species-specific colonization of the mucosal environment. Indeed, consistent with the well-established long-term M-SHIME® model [12], a wide spectrum of (potential butyrate-producing) *Firmicutes* members specifically colonized the mucosal environment (e.g., OTUs related to *Clostridium butyricum* and *Roseburia hominis*). Furthermore, inclusion of the mucosal environment did not alter the luminal microbiota (except for a minor enrichment in *Akkermansiaceae*), nor did it alter the treatment effects towards luminal microbiota activity and community composition. Overall, including the mucin beads in the current colonic simulations allowed to maintain a higher diversity, thus allowing observation of more complete treatment effects of cRG-I. Finally, application of a model combining gut epithelial and immune cells allowed to point out the gut protective effect of cRG-I fermentation-derived metabolites against an inflammatory stressor (Test 4).

In a first series of short-term colonic incubations (Test 2), cRG-I was found to display prebiotic potential comparable but not identical to that of inulin in modulating microbial activity [19], as followed from increases in health-promoting SCFA (acetate and propionate) and lactate and decreases of bCFA. A side effect of inulin fermentation is the production of high amounts of gas that has been observed in in vitro experiments and clinical studies. Depending on the dose, this can result in mild negative gastro-intestinal symptoms [56–58]. Interestingly, gas production upon dosing of cRG-I was milder compared to inulin, which suggests that in vivo consumption of cRG-I could be less accompanied by adverse side-effects such as bloating and abdominal pain.

Upon fermentation in both short-term colonic incubations (Tests 2/3), cRG-I stimulated acetate, and lactate production which was accompanied with a strong reduction in pH. At community level, this correlated with increases in OTUs related to *Bifidobacterium longum* (OTU7) and *Bifidobacterium adolescentis* (OTU21). *Bifidobacteriaceae* are indeed key acetate and lactate producers [59,60]. Although health effects are to be considered strain-specific, multiple *Bifidobacterium* strains have been associated with beneficial effects on the host health and some strains are widely used as probiotics. A first mechanism by which *Bifidobacteriaceae* contribute to health is by indirectly promoting butyrate production via cross-feeding mechanisms involving for instance *Faecalibacterium prausnitzii* and *Eubacterium hallii* [59,61,62], taxa of which related OTUs were indeed found to be increased upon cRG-I treatment in the current study. In addition, acetate produced by bifidobacteria was also shown to play a key role in their anti-infectious properties against enteropathogens [63]. In terms of host effects, a specific *B. longum* strain has e.g., been shown to exert protective effects in a DSS-induced colitis model in mice by reducing inflammation and enhancing the intestinal epithelial barrier [64]. Moreover, a specific *B. longum* strain reduced chronic mucosal inflammation in ulcerative colitis (UC) patients in a double-blinded, randomized-controlled clinical trial [65]. Likewise, a specific *B. adolescentis* strain protected mice from DSS-induced colitis by increasing IL-10 levels, up-regulation of regulatory T-cells (Treg) and decreasing IL-17A positive T-cells [66]. The strong bifidogenic effect exerted by cRG-I thus supports the prebiotic potential of this novel food ingredient.

Next to acetate and lactate, fermentation of cRG-I increased the production of propionate, which correlated with a stimulation of *Prevotellaceae* and *Bacteroidaceae* due to the stimulation of a wide spectrum of *Bacteroidaceae* OTUs related to e.g., *B. ovatus*, *B. plebeius*, *B. xylanisolvens* and especially *B. dorei*. Indeed, *Bacteriodetes* spp. are known primary fiber degraders that are capable of producing propionate [50]. In the colon, health-related effects of propionate are related to anti-cancer effects [7,8]. Furthermore, propionate was shown to exert anti-lipogenic and cholesterol-lowering effects in the liver [6]. Finally, propionate is a satiety-inducing agent affecting energy intake and feeding behavior [67]. Besides the beneficial effects of propionate, the aforementioned stimulations of specific *Bacteroidaceae* members have also been related to particular health benefits. As an example, in a mouse model of atherosclerosis, *B. dorei* reduced plaque inflammation and decreased intestinal epithelial permeability and systemic endotoxemia [68]. Furthermore, *B. ovatus* reduced mucosal inflammation and stimulated epithelial proliferation and mucin production in a DSS-induced colitis model in mice [69]. Another microbial modulation that could have boosted propionate production upon cRG-I treatment was the increase in *Veillonellaceae* that was solely attributed to an increase in OTU9 (related to *Dialister succinatiphilus*). Interestingly, *D. succinatiphilus* is a succinate-converting, propionate-producing species [70] and as many *Bacteroidetes* spp. are known succinate producers [50], its increase might have contributed to the stronger increase in propionate levels upon dosing of cRG-I. As a final remark, not all *Bacteroidaceae* members increased upon cRG-I treatment, with OTUs relating to *B. caccae* and *B. fragilis* decreasing in abundance upon cRG-I treatment. These species are considered to be opportunistic pathogens carrying virulence factors: enterotoxigenic *B. fragilis* strains secrete the *B. fragilis* toxin (BFT) [71] and *B. caccae* contains the *ompW* gene [72]. Besides its strong bifidogenic effect, cRG-I could further exert health benefits by targeted modulation of specific propionate-producing taxa.

In contrast to propionate, butyrate was not significantly increased upon dosing of cRG-I. However, cRG-I did strongly stimulate *Lachnospiraceae*, containing several butyrate-producing species, in the mucosal compartment. The increase in *Lachnospiraceae* was related to a strong stimulation of an OTU related to *Roseburia hominis*, which has been associated with beneficial effects on barrier function and immune regulation in the gut [73]. Indeed, *R. hominis* showed protective effects in a DSS-model for colitis in mice by reducing pro-inflammatory cytokine expression. Moreover, increased Treg levels were observed in both germ-free and conventional mice fed a supplement containing live *R. hominis*. Also, a significant reduction of *R. hominis* and *F. prausnitzii* was observed in UC and Crohn's disease (CD) patients [74,75]. Interestingly, cRG-I increased the abundance of OTUs related to butyrate-producing species such as *F. prausnitzii* and *Eubacterium hallii*. Like *R. hominis*, *F. prausnitzii* was shown to exert anti-inflammatory potential by increasing IL-10 and promoting Treg differentiation in mice [76]. On the other hand, *E. hallii* improved insulin sensitivity in a mouse model for diabetes [77]. Of note, discrepancy between the stimulation of butyrate producers in the mucosal environment with cRG-I versus the absence of treatment effects on butyrate levels might be explained by the fact that the biofilm which develops on the mucin-coated carriers during short-term colonic incubations (48 h) is still developing during cRG-I treatment. Hence, by the time the biofilm is developed (48 h), all substrate has been consumed (in fact already after 24 h). This may potentially limit the detection of treatment effects resulting from modulation of mucosal microbes on metabolic activity (particularly butyrate production) during short-term incubations. Therefore, testing the impact of cRG-I in a long-term M-SHIME® study could further elucidate the potential impact of cRG-I on butyrate production.

The stimulation of several health-related microbial species and the concomitant increase in health-promoting metabolites suggested that cRG-I may display interesting host beneficial properties in terms of intestinal barrier protection and reduction of inflammation. This hypothesis was confirmed using a Caco-2/PBMC co-culture model in which treatment with fermented cRG-I significantly increased the TEER, indicative for the protective effects of fermentation-derived cRG-I metabolites on inflammation-induced intestinal permeability. To further strengthen this observation, it would be interesting to perform in-depth analysis of the expression and localization of tight junction proteins including occludin, ZO-1, and claudins upon cRG-I treatment in this system. Furthermore, fermented

cRG-I metabolites decreased the secretion of the pro-inflammatory cytokines IL-17A, IL-4, and IL-9, while increasing the secretion of IL-22 and of the anti-inflammatory IL-10. This is suggestive of a possible immunoregulatory effect of cRG-I metabolites on the Treg/TH17 axis; favoring down-regulation of excessive inflammation as IL-10 is necessary for Treg functions [78]. Th17 cells have a dual role in human health as although required for clearance of extracellular pathogens, an elevated frequency of Th17 cells associated with impaired Treg functions has been reported in IBD and other extraintestinal autoimmune disorders [79]. Furthermore, IL-22 plays a role in maintaining the integrity of the mucosal barrier by promoting wound healing and activating antimicrobial responses [53]. Finally, increased levels of IL-4 and IL-9 were associated with UC [80,81]. In addition, IL-9 inhibits wound healing in the intestinal mucosa and impairs intestinal epithelial barrier functions. Also, IL-9 directly regulates tissue recruitment and inflammatory functions of mast cells. Together, these data suggest an interesting immuno-modulatory role of cRG-I metabolites in the gut in terms of increasing barrier tightness and prevention of a "leaky gut".

Finally, a modified formulation of cRG-I was tested which contained small size sugars, i.e., cRG-I+LMWC. Interestingly, the short-term colonic incubation strategy applied in this study was highly sensitive to pick up differences in closely related product compositions. For instance, the presence of these simple sugars resulted in a stronger initial pH decrease upon dosing of cRG-I+LMWC, related to an initial increased production of lactate, which could be linked to a stronger enrichment in a *Streptococcacae* OTU related to *Streptococcus aginosus*. Also, *Enterobacteriaceae* levels were higher upon cRG-I+LMWC treatment. Like *Streptococcaceae*, *Enterobacteriaceae* are expert fermenters of simple sugars [82], specifically present at higher levels in this preparation. Finally, *Coriobacteriaceae* (OTU15 related to *Collinsella aerofaciens*) and *Desulfovibrionaceae* tended to be higher in incubations with cRG-I+LMWC. This suggested a potential co-existence of e.g., *C. aerofaciens* and *Desulfovibrio piger*, based on the fact that the main products of *C. aerofaciens* fermentation (i.e., lactate, H2, and formate) serve as substrates for *D. piger*, which is the main sulfate-reducing bacterial species in the human gut microbiome [83]. This might also indicate that cRG-I+LMWC might contain small quantities of sulfur (sulfate or sulfur-containing amino acids). These data altogether stress the relevance of testing prebiotic candidates in short-term colonic in vitro incubations to understand their potential impact on the human gut microbiome and to support structure-function studies.

In conclusion, the implementation of novel in vitro models simulating the human colonic environment coupled to cell-based assays mimicking the host gut barrier, allowed to establish the prebiotic potential of cRG-I. This novel fiber is not digested by host enzymes characteristic of the upper gastro-intestinal tract but rapidly fermented by the human colonic microbiota leading to selective stimulation of the growth and activity of intestinal bacterial species associated with human health. cRG-I displays unique properties as it was fermented more rapidly than inulin leading to production of SCFA, especially acetate and propionate, and less gas than inulin. It increases the abundance of several bacterial species reputed for their "anti-inflammatory" profile. In line with this, the metabolites resulting from cRG-I fermentation exhibited a protective effect in an in vitro model of inflamed gut barrier. Overall, the data obtained during this study support future research to further investigate this novel prebiotic candidate in long-term SHIME® models with different fecal sample donors and in clinical trials to confirm the beneficial effect of cRG-I on the microbiota and its impact on human health.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6643/12/7/1917/s1, Figure S1: Effect of Sodium butyrate (NaB) on transepithelial electrical resistance (TEER) and cytokine production in a Caco-2/PBMC co-culture system; Table S1: Effect of fermentation of cRG-I and cRG-I+LMWC on microbial community composition at OTU level in short-term colonic batch simulations in absence or presence of a mucosal compartment; Table S2: Effect of fermentation of cRG-I and cRG-I+LMWC on selected bacterial groups in short-term colonic batch simulations in absence or presence of a mucosal compartment as assessed through qPCR.

**Author Contributions:** Conceptualization, A.M., P.v.d.A., M.M.; Methodology, P.v.d.A.; Formal Analysis, J.G., L.V.; Investigation, J.G., L.V.; Data Curation, P.v.d.A., L.V., J.G.; Writing-Original Draft Preparation, A.M.; P.v.d.A., L.V. Writing-Review and Editing, A.M., R.A., P.v.d.A., L.V.; Supervision, P.v.d.A., M.M.; Project Administration, M.M.; Funding Acquisition, A.M., R.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** The studies described in this manuscript were performed at the request of and were funded by NutriLeads B.V., Wageningen, The Netherlands. This project has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreements No 811592 and Eurostars E! 10574-NIMF.

**Acknowledgments:** We thank C. Rösch and H. Schols for their expert advice in the preparation of the cRG-I formulations and sugar analysis. We acknowledge the support of M. Aparicio-Vergara and M. Tzoumaki in performing preliminary work to the experiments.

**Conflicts of Interest:** A.M. and R.A. are employees of Nutrileads the funder of the study. A.M. participated to the design of the study and the writing of the manuscript. R.A. edited the manuscript.

#### **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 Vitro Evaluation of Prebiotic Properties of a Commercial Artichoke Inflorescence Extract Revealed Bifidogenic E**ff**ects**

**Pieter Van den Abbeele 1, Jonas Ghyselinck 1, Massimo Marzorati 1,2, Agusti Villar 3, Andrea Zangara 3,4,\*, Carsten R. Smidt <sup>5</sup> and Ester Risco 3,6**


Received: 1 May 2020; Accepted: 20 May 2020; Published: 26 May 2020

**Abstract:** Background: Prebiotics used as a dietary supplement, stimulate health-related gut microbiota (e.g., bifidobacteria, lactobacilli, etc.). This study evaluated potential prebiotic effects of an artichoke aqueous dry extract (AADE) using in vitro gut model based on the Simulator of Human Intestinal Microbial Ecosystem (SHIME®). Methods: Short-term colonic fermentations (48 h) of AADE, fructo-oligosaccharides (FOS), and a blank were performed. Microbial metabolites were assessed at 0, 6, 24, and 48 h of colonic incubation via measuring pH, gas pressure, lactate, ammonium, and short-chain fatty acids (SCFAs) levels. Community composition was assessed via targeted qPCRs. Results: After 24 and 48 h of incubation, bifidobacteria levels increased 25-fold with AADE (*p* < 0.05) and >100-fold with FOS (*p* < 0.05) compared to blank. *Lactobacillus* spp. levels only tended to increase with AADE, whereas they increased 10-fold with FOS. At 6 h, pH decreased with AADE and FOS and remained stable until 48 h; however, gas pressure increased significantly till the end of study. Acetate, propionate, and total SCFA production increased significantly with both at all time-points. Lactate levels initially increased but branched SCFA and ammonium levels remained low till 48 h. Conclusion: AADE displayed prebiotic potential by exerting bifidogenic effects that stimulated production of health-related microbial metabolites, which is potentially due to inulin in AADE.

**Keywords:** bifidobacteria; colon; fermentation; microbiota; prebiotic; SHIME®; artichoke

#### **1. Introduction**

Human gut microbiota consist of over 35,000 bacterial strains, encompassing beneficial and pathogenic species; however, the predominance of positively affecting microbes ensure our well-being [1]. Human gut microbiota are dominated by two main phyla, Firmicutes (including *Lactobacillus* spp.) and Bacteroidetes that are susceptible to alterations. Other phyla are Actinobacteria (including *Bifidobacterium* spp.), Proteobacteria, Fusobacteria, and Verrucomicrobia. Spatial and temporal discrepancies in gut microbial distribution contribute toward specific metabolic, immunological, and gut-protective functions throughout an individual's life span [2,3]. Characterization of such discrepancies can help identify gut-related abnormalities and play an important role in ensuring good health [4].

Prebiotics were first defined as, "Nondigestible food ingredients that beneficially affect a host by selectively stimulating growth and/or activity of one or a limited number of bacteria in the colon that are recognized to improve host health" [5]. Dysbiosis of microbial populations has been postulated as one of the reasons for metabolic disorders such as obesity, type 2 diabetes, and nonalcoholic fatty liver diseases. As prebiotics alter microbiota positively, their use as dietary supplements could effectively improve overall host health [6]. Fructo-oligosaccharides (FOS) are prebiotics that are plant-derived, naturally occurring oligosaccharides, indigestible by human enzymes, and can thus reach the colon unaltered [7]. Daily intake of FOS can increase bifidobacteria counts, a member of the indigenous gut microbiota. However, certain individuals are more sensitive to effects of FOS and suffer side effects such as itching in the throat; puffiness in the eyes, face, and mouth; dizziness; light headedness; fainting; gas; bloating; and itching of the skin [8,9].

There is a constant need for new prebiotics that can target specific bacterial species and most approaches have focused on non-digestible oligosaccharides, such as galacto-oligosaccharides, soya-oligosaccharides, isomaltooligosaccharides, gluco-oligosaccharides, xylo-oligosaccharides, lacto-sucrose, and inulin-type fructans. However, they are known to have varied prebiotic potential. Inulin has been demonstrated to positively alter gut microbiota in a dose range of 4 to 40 g/d [10–16]. Whole food sources, such as artichoke (*Cynara scolymus* L.), chicory (*Cichorium intybus*) roots, and garlic (*Allium sativum*) are rich in inulin and dietary fibers. Inulin from artichoke is recognized to have the highest degree of polymerization known in plants. Degree of polymerization directly contributes to prebiotic effects and persistence in the colon [17]. Inulin promotes host health by positively altering the bacterial metabolites mediated via stimulation of different metabolic pathways within the gut microbial community. Acetate, propionate, and butyrate are the most crucial metabolites. By acidifying the colonic environment, short-chain fatty acids (SCFA) promote growth of beneficial bacteria such as bifidobacteria and lactobacilli, which inhibit the growth of pathogenic bacteria [18]. Additionally, bifidobacteria and lactobacilli exert immunomodulatory activity that contributes to the host defense [19]. Prebiotic potential of artichoke has been demonstrated in several clinical studies and the effect was mainly mediated via increase in *Bifidobacterium* spp. in the gut [20,21]. In addition to inulin, artichoke contains polyphenols, such as dicaffeoylquinic acids and flavonoids, which provide additional nutritional values proposing a novel holistic approach to whole digestive health [22,23].

In the present study, we aimed to evaluate the prebiotic effects of artichoke aqueous dry extract (AADE) through an in vitro approach of highly controlled conditions using short-term colonic incubations based on the Simulator of Human Intestinal Microbial Ecosystem (SHIME®) model [18]. In vitro models offer certain advantages, first they allow dynamic monitoring of gut microbiome at the site of fermentation under a controlled environment and second, in vitro models help avoid large variability that arise during in vivo evaluations owing to host-derived factors such as amount of food intake, immune system, enzyme levels, or transit time. Lastly, using molecular detection methods, microbial changes can be evaluated in detail. Artichoke aqueous dry extract is a standardized herbal powder extract prepared from the edible part of artichoke (*Cynara scolymus* L.); cultivated in Spain, and extracted by the Pure-Hydro Process™ using only water (instead of organic solvents), AADE can be safely used in foods and food supplements.

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

#### *2.1. Chemicals and Reagents*

All chemicals were obtained from Sigma-Aldrich (Overijse, Belgium), unless stated otherwise. The test product AADE, also known as Cynamed™, was provided by Euromed S.A. (Mollet del Valles, Barcelona, Spain). It is derived from the edible part of the artichoke plant (*Cynara scolymus* L.) cultivated in Mediterranean regions of Spain. The AADE was prepared in accordance with the European Pharmacopoeia monograph extracts (Extracta) (No. 0765) using a proprietary water-based extraction process [24]. This process starts with the milling of dried Artichoke immature edible

inflorescences that are extracted with ultrapure water at a temperature between 80 ◦C and 90 ◦C. The miscella of extract is filtered until transparency and concentrated under vacuum until a soft paste is obtained that is subsequently dried in a vacuum belt dryer and finally milled to a fine powder. The AADE used in the current study has an exact content of 9.1% caffeoylquinic acids expressed as chlorogenic acid by HPLC and an exact content of 32.2% inulin determined by HPLC. As a nutritional analysis of the AADE, the amount of total carbohydrates is 77% and the amount of protein 8.1% with a negligible content of fat. The FOS preparation used as a positive control in the current study had a purity of 89% FOS with 8% sugar residues. While the degree of polymerization of the ingredient varied between 2 and 10, it was on average 4.

#### *2.2. Short-Term Colonic Fermentation*

Short-term colonic fermentations were performed as described recently [18]. Briefly, colonic background medium containing 5.2 g/L K2HPO4, 16.3 g/L KH2PO4, 2.0 g/L NaHCO3 (Chem-lab NV, Zedelgem, Belgium), 2.0 g/L Yeast Extract, 2.0 g/L pepton (Oxoid, Aalst, Belgium), 1.0 g/L mucin (Carl Roth, Karlsruhe, Germany), 0.5 g/L L-cystein, and 2.0 mL/L Tween80 (Sigma-Aldrich, Bornem, Belgium) was added to incubation reactors (90 vol%), already containing the correct amount of the test products for obtaining a final concentration of 0 g/L (Blank) or 5 g/L (for both AADE and FOS), respectively. The reactors were sealed and anaerobiosis was obtained by flushing with N2. Subsequently, fresh fecal material of a healthy human donor (no history of antibiotic use in the six months preceding the study) was collected (according to the ethical approval of the University Hospital Ghent with reference number B670201836585; 06/08/2018). After preparation of an anaerobic fecal slurry, this was inoculated at 10 vol% in the aforementioned medium. All incubations were performed in biological triplicate for 48 h at 37 ◦C under anaerobic conditions with continuous shaking (90 rpm).

#### *2.3. Microbial Metabolic Activity Analysis*

Microbial metabolic analyses were performed on samples collected at 0, 6, 24, and 48 h of colonic incubation and levels of pH (Senseline F410; ProSense, Oosterhout, The Netherlands), gas pressure (hand-held pressure indicator CPH6200; Wika, Echt, The Netherlands), lactate, ammonium, and short-chain fatty acids (SCFAs) were measured. Acetate, propionate, butyrate, and branched CFAs (BCFAs) (isobutyrate, isovalerate, and isocaproate) were quantified as described by De Weirdt et al. [25] via GC-FID after performing a diethyl ether extraction. Lactate determination was performed using a commercially available enzymatic assay kit (R-Biopharm, Darmstadt, Germany) as per the manufacturer's instructions. Ammonium analysis was performed using a KjelMaster K-375 device (Büchi, Hendrik-Ido-Ambacht, The Netherlands), wherein ammonium in the sample was liberated as ammonia by addition of 32% NaOH. The released ammonia was then distilled from the sample into a 2% boric acid solution and was titrimetrically determined with a 0.02 M HCl solution.

#### *2.4. Microbial Community Analysis*

At the start of colonic incubation and after 24 and 48 h, samples were collected for microbial community analysis. DNA was isolated using the protocol as described by Vilchez-Vargas et al. [26], starting from pelleted cells originating from 1 mL luminal sample. Subsequently, quantitative polymerase chain reaction (qPCR) assays for Bacteroidetes, Firmicutes, *Lactobacillus* spp. (Firmicutes phylum), *Bifidobacterium* spp. (Actinobacteria phylum), and *Akkermansia muciniphila* (Verrucomicrobia phylum) were performed using a StepOnePlus™ real-time PCR system (Applied Biosystems, Foster City, CA, USA). Each sample was analyzed in triplicate. Standard curves for all the different runs had efficiencies between 90–105%. All protocols were initiated for 10 min at 95 ◦C and terminated with a melting curve from 60 ◦C to 95 ◦C. Cycling programs included 40 cycles with a denaturation step of 15 s at 95 ◦C, an annealing step of 30 s at 60 ◦C, and an elongation step of 30 s at 72 ◦C in each cycle. Descriptions of primers used are presented in Table 1.


**Table 1.** Primers used for quantitative polymerase chain reaction (qPCR) quantification of species-specific 16S rDNA.

Besides presenting the absolute levels of the different groups, the ratio between the obtained levels at 24 h and 48 h versus 0 h were calculated for the blank, AADE, and FOS-treated microbiota.

#### *2.5. Statistics*

To evaluate differences in microbial metabolites and microbial community composition between blank and treatment incubations at the different time points, a two-way ANOVA with Tukey multiple comparisons test was performed. Differences were found significant if *p* < 0.05. Statistical analysis was performed with the GraphPad Prism software (version 8.3.0, San Diego, USA).

#### **3. Results**

#### *3.1. Microbial Composition*

While the absolute levels of each of the five targeted microbial groups (*Bifidobacterium* spp., *Lactobacillus* spp., Bacteroidetes, Firmicutes, and *Akkermansia muciniphila*) at each of the three time points (0/6/48 h) are presented in Table 2, the factor increase versus 0 h is presented in Figure 1 for the four microbial groups for which there were significant changes between the treatments (all except *Akkermansia muciniphila*). First, both at 24 h and 48 h, bifidobacteria levels were significantly increased versus the blank for AADE but especially for FOS (Table 2). This was reflected by ~25-fold and ~100-fold increased levels versus 0 h for AADE and FOS, respectively; both after 24 h and 48 h of incubation (Figure 1A). Additionally, *Lactobacillus* spp. were stimulated more profoundly for FOS with ~10-fold increased levels versus 0 h at 24 and 48 h (Figure 1B). AADE exerted more attenuated effects on *Lactobacillus* spp. levels with only statistically significantly increased absolute levels at 48 h. Incubation with FOS increased absolute Firmicutes levels at all time points, while for AADE the increase was only significant at 24 h (Table 2 and Figure 1C). Finally, AADE increased Bacteriodetes levels versus the blank at 48 h (Table 2 and Figure 1D), while FOS decreased *Akkermansia muciniphila* levels versus AADE after 48 h of incubation (Table 2).

**Table 2.** Mean (±standard deviation) levels of microbial groups as measured via quantitative polymerase chain reaction (qPCR) after 0, 24, and 48 h of treatment of a simulated colonic microbiota with 5 g/L AADE (artichoke aqueous dry extract) or FOS (fructo-oligosaccharides). For each microbial group and within each time point (24 h or 48 h), a value indicated with a different letter (a, b, or c) indicates a statistical difference between AADE, FOS, and/or the blank, as tested with a two-way ANOVA with post-hoc Tukey test (*p* < 0.05). In contrast, when at least one letter is shared between two treatments, there was no significant between these groups.


**Figure 1.** Mean (±standard deviation) ratios of (**A**) *Bifidobacterium* spp., (**B**) *Lactobacillus* spp., (**C**) Firmicutes, and (**D**) Bacteroidetes levels after 24 h or 48 h of treatment of a simulated colonic microbiota with 5 g/L AADE (artichoke aqueous dry extract) or FOS (fructo-oligosaccharides) versus the initial levels (24 h/0 h or 48 h/0 h, respectively) as measured via quantitative polymerase chain reaction (qPCR). For each microbial group and within each time point (24 or 48 h), a bar indicated with a different letter (a, b, or c) indicates a statistical difference between AADE, FOS, and/or the blank at a given time point, as tested with a two-way ANOVA with post-hoc Tukey test (*p* < 0.05). In contrast, when at least one letter is shared between two bars, there was no significant between these treatments.

#### *3.2. pH*

A more profound decrease in pH was observed with FOS and to a lesser extent also with AADE compared with blank at 6 h (*p* < 0.05). The pH continued to decrease until 24 h and remained stable thereafter, indicating continued microbial fermentation (Table 3).


**Table 3.** Mean (±standard deviation) pH and gas pressure after 0, 6, 24, and 48 h of treatment of a simulated colonic microbiota with 5 g/L AADE (artichoke aqueous dry extract) or FOS (fructo-oligosaccharides). For each time point (0, 6, 24, or 48 h), a value indicated with a different letter (a, b, or c) indicates a statistical difference between AADE, FOS, and/or the blank as tested with a two-way ANOVA with post-hoc Tukey test (*p* < 0.05).

#### *3.3. Gas Pressure*

As noted in Table 3, as compared with the blank, gas pressure was significant with AADE and FOS on all the time points along the incubation. On all time points, gas production was significantly higher for FOS versus AADE (Table 3).

#### *3.4. Lactate and Carbohydrate (SCFAs, Acetate, Butyrate, and Propionate) and Protein Metabolites (BCFAs and Ammonium)*

Compared with the blank, total SCFAs were significantly increased with AADE and FOS at all time-points of incubation (*p* < 0.05), which reflected enhanced microbial metabolic activity upon AADE/FOS administration. However, the overall increase in total SCFAs was higher with FOS compared with AADE (Table 4). Similar patterns of increase in acetate and propionate levels were observed with AADE and FOS as for the total SCFA (Table 4). In contrast, butyrate concentrations were only significantly increased for FOS, and this after 24 h and 48 h.


*Nutrients* **2020** , *12*, 1552

**Table 4.** Mean (±standard deviation)

carbohydrate-

 and

protein-derived

 metabolites

 after 0, 24, and 48 h of treatment of a simulated colonic microbiota with 5 g/L

As shown in Table 4, no BCFAs were produced during the initial 6 h of incubation in AADE and FOS. After 24 h of incubation, there was a similar production of BCFAs in the blank (1.34 ± 0.24) and upon AADE treatment (1.70 ± 0.37). In contrast, no BCFAs were produced upon FOS administration at the 24 h time point. After 48 h of incubation, BCFAs were produced but were significantly (*p* < 0.05) lower for both AADE (3.42 ± 0.02) and especially FOS (0.34 ± 0.26) when compared with the blank (4.01 ± 0.13). The results for ammonium, another marker for protein fermentation, were similar to those for BCFAs, indicating reduced protein fermentation upon AADE and especially FOS administration. Lactate levels were high (*p* < 0.05) with AADE and even further increased for FOS compared with blank after the initial 6 h of incubation. Thereafter, lactate levels decreased indicating lactate consumption.

#### **4. Discussion**

In the present study, although the effects of AADE on microbial activity and composition were milder as compared to the "gold standard" prebiotic FOS, AADE demonstrated marked prebiotic potential. First, AADE significantly decreased pH and increased gas production, which indicated overall increased microbial activity upon administration of the test product. Saccharolytic metabolites such as acetate and propionate, and thus also total SCFAs, increased, while levels of proteolytic metabolites, BCFAs, and ammonium, significantly decreased upon AADE administration at 48 h. A key finding of this study was the growth-promoting action of AADE, mostly on bifidobacteria which are regarded as health-related members of the intestinal microbiome. Further, AADE also affected Bacteroidetes, Firmicutes, and *Lactobacillus* spp. levels.

Based on results of this study, bifidogenic effects of AADE were milder, yet in the same order of magnitude as those of FOS. These findings were similar to those of a previous in vitro study conducted by Barszcz M et al. [28]. Bifidogenic effects were also reported in healthy volunteers [20,21]. The bifidogenic effect of artichoke has been attributed to its inulin content. Inulin exerts most physiological changes through the bacterial metabolites. SCFAs are some of the important metabolites that acidify the colonic environment promoting growth of beneficial bacteria, such as *Lactobacillus* spp. and bifidobacteria, and prevent growth of pathogenic bacteria [30,31].

Moreover, in our study, lactate was produced during the initial 6 h of incubation. Subsequently, lactate was consumed and coincided with an increase in propionate levels for both AADE and FOS, with most marked stimulations being noted for FOS. Butyrate was not stimulated by AADE, suggesting that the majority of lactate (that can be used as a substrate for both propionate and butyrate), was cross-fed to propionate upon AADE supplementation. Some Negativicutes (family Veillonellaceae, phylum Firmicutes) are shown to form propionate [32] and could potentially explain the increase of Firmicutes that was observed for AADE after 24 h in our study. Bacteroidetes also contain potent propionate producers [33,34] and could have further contributed to propionate production upon AADE supplementation since AADE also stimulated this phylum in our study. These alterations in propionate levels correlate with the inulin content of the artichoke [18]. Propionate metabolites have been shown to reduce cholesterol and fatty acid synthesis in liver, improve glucose metabolism, and regulate immune status in adipose tissue, and thus elicit health-promoting activities [18,35]. Finally, another key propionate producer is the mucin-degrading, acetate and propionate producing, *Akkermansia muciniphila.* This taxon was not increased for either AADE or FOS and even decreased upon FOS administration. This was likely due to the fact that FOS more strongly decreased the pH (to 5.66 within 24 h), which is a pH at which *Akkermansia muciniphila* is unable to grow [36]. In vivo, such lower pH could however boost mucin secretion and result in enhanced mucin degradation by *Akkermansia muciniphila* in the distal colon, as shown for inulin in humanized rats [37].

Similarly, acetate and lactate can be cross-fed to butyrate by members of the Ruminococcaceae, Lachnospiraceae, Clostridiaceae, Eubacteriaceae, all members of the Firmicutes phylum [18,38]. Butyrate is a major energy source for the gut microbiota and may also reduce oxidative stress, improve gut function, and restrict inflammatory response. In this study, butyrate levels were increased majorly with FOS and were usually produced during the later stage of incubation period. As our study duration

was limited to 48 h, additional studies with longer incubation periods are warranted to make accurate conclusions. Moreover, cross-feeding between microbial communities should be taken into account when drawing definite conclusions [39].

Furthermore, propionate and acetate have been shown to stimulate release of peptide hormones leading to short-term signaling of satiation and satiety to appetite centers in the brain, resulting in reduced food intake by the host [35,40,41]. Several metabolic disorders such as obesity, insulin resistance, and metabolic syndrome are associated with impaired carbohydrate and lipid metabolism by the host, and are accompanied by changes in the gut microbiota [32]. Inulin could stimulate different metabolic pathways within the gut microbial community and could potentially elicit varied health-promoting activities [18].

Ammonia and BCFAs are toxic metabolites produced from protein fermentation [39]. In this investigation, the reduction in BCFAs and ammonia production in part explains the increase in carbohydrate metabolism. In vivo studies have also demonstrated that generation and accumulation of ammonia can be reduced by lowering protein supply and by colonic fermentation of suitable non-digestible carbohydrates from food [39,42].

Short-term colonic incubations have often been used to gather information on the prebiotic potential of novel ingredients. Results of the present study using an incubation strategy based on the SHIME® model indicate the prebiotic potential of AADE. These findings could be further validated using different models, such as M-SHIME® (Mucosal Simulator of the Human Intestinal Microbial Ecosystem) which focuses not only on luminal but also mucosal gut-colonizing microbes [43]. Moreover, studies with repeated administration are required in order to simulate gradual changes that occur in vivo with long-term use and to assess any beneficial microbial shift.

#### **5. Conclusions**

The present preliminary evaluation, conducted using the SHIME® model, demonstrated that AADE has promising prebiotic potential. Incubation with AADE resulted in an increase of beneficial microbes, which was correlated with their metabolite profile. The promising results of this study justify future investigations using multiple doses in upgraded models to further validate these findings.

**Author Contributions:** Conceptualization, M.M.; study method, investigation, and data analysis, P.V.d.A. and J.G.; funding acquisition, A.V. and A.Z.; and writing—original draft, reviewing, and editing manuscript, all authors. All authors have read and agreed to the published version of the manuscript.

**Funding:** The study was funded by Euromed S.A.

**Acknowledgments:** Authors acknowledge CBCC Global Research for providing statistical analysis and medical writing assistance in the development of this manuscript, which was funded by Euromed S.A.

**Conflicts of Interest:** A.V., A.Z., and E.R. are employed by Euromed S.A. The study funders (Euromed S.A.) had no role in the design, collection, or analysis of the data.

#### **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* **Possible Protective E**ff**ects of TA on the Cancerous E**ff**ect of Mesotrione**

**Agata Jabło ´nska-Trypu´c 1,\*, Urszula Wydro 1, El ˙zbieta Wołejko 1, Joanna Rodziewicz <sup>2</sup> and Andrzej Butarewicz <sup>1</sup>**


Received: 25 March 2020; Accepted: 6 May 2020; Published: 8 May 2020

**Abstract:** The interaction of different food ingredients is now a very important and often emerging topic of research. Pesticides and their breakdown products, which may be carcinogenic, are one of the frequently occurring food contaminants. Compounds like traumatic acid (TA), which originates from plants, are beneficial, antioxidant, and anticancer food ingredients. Previously obtained results from our research group indicated antioxidative in normal human fibroblasts and prooxidative in cancer cells activity of TA. Since the literature data show an undoubted connection between the presence of pesticides in food and the increased incidence of different types of cancers, we attempted to clarify whether TA can abolish the effect of mesotrione stimulating the growth of cancer cells. In order to study the influence of mesotrione on breast cancer cells, we decided to carry out cytotoxicity studies of environmentally significant herbicide concentrations. We also analyzed the cytotoxicity of TA and mixtures of these two compounds. After selecting the most effective concentrations of both components tested, we conducted analyses of oxidative stress parameters and apoptosis in ZR-75-1 cells. The obtained results allow us to conclude that traumatic acid by stimulating oxidative stress and apoptosis contributes to inhibiting the growth and development of cells of the ZR-75-1 line strengthened by mesotrione. This may mean that TA is a compound with pro-oxidative and proapoptotic effects in cancer cells whose development and proliferation are stimulated by the presence of mesotrione. The presented results may be helpful in answering the question of whether herbicides and their residues in edibles may constitute potential threat for people diagnosed with cancer and whether compounds with proven pro-oxidative effects on cancer cells can have potential cytoprotective functions.

**Keywords:** mesotrione; traumatic acid; breast cancer; herbicide; antioxidant; oxidative stress

#### **1. Introduction**

Different chemical substances from the group of pesticides are used in the food production process to ensure seasonal availability and good quality of products. However, it should be mentioned that the frequent and widespread use of pesticides carries the risk of their penetration into the human body. Pesticides are a group of chemical compounds, both natural and synthetic, that are used in order to destroy plant and animal parasites, reduce the risk of plant diseases, and control weeds. The massive use of pesticides results from the growing number of consumers, and thus from an increased demand for food. Pesticide residues may remain in food after their application to crops. The maximum permissible levels of pesticides residues in food are determined by the regulatory authorities in the European Union, mostly at the level of the European Commission. Exposure of a given population to

pesticides and their residues most often occurs as a result of the consumption of processed food or close contact with pesticide treated areas, such as farms [1,2].

Mesotrione (Mes) is an herbicide which controls most broadleaf weeds and weed grasses in crops cultivation. Due to its frequent use in agriculture, the detectable level of this compound in countries such as North America and Canada fluctuates around 4.1μg/L [3,4]. According to its chemical structure, Mes is 2-[4-methylsulfonyl-2-nitrobenzoyl]-1,3-cyclohexanedione and it belongs to the family of triketone. Its main role is an inhibition of the enzyme 4-hydroxy-phenyl-pyruvate-dioxygenase, which converts tyrosine to plastoquinone (PQ) and alpha-tocopherol [5]. Mes is a quite water soluble compound, which, in combination with its widespread application and soil retention capacity, contributes to the fact that this herbicide easily contaminates surface and groundwater [6]. According to Bonnet et al., Mes decomposition products appear to be more dangerous and harmful than the parent compound [7].

The endocrine-disrupting influence of selected pesticides has caused great concerns due to the hormonal activity of many well-documented risk factors for breast cancer. Literature data suggest that selected pesticides could be related to an increase in breast cancer risk and urge researchers to examine environmental risk factors and possible compounds, preferably of natural origin, that could reduce side effects of pesticide use and thus the incidence of this disease [8–10]. Many pesticides are considered as analogues of human hormones. They may exert influence through estrogen receptors. Therefore, we chose to examine the ZR-75-1 breast cancer cell line, which is commonly used for endocrine-based research, rather than choosing on the other breast cancer cell line or nonhuman cell line. Discordances in scientific data regarding possible herbicides cancerogenic properties may result from the different test models. Therefore, for the experiment, we decided to choose the human estrogen-dependent breast cancer cell line, which is characterized by the presence of the estrogen receptor (ER+). The aim of this paper was to study the mutual interaction mechanisms of two opposite compounds, one of which is highly undesirable food contaminant (Mes), and traumatic acid (TA), which is a promising food ingredient. In already published papers, we indicated that TA is characterized by antioxidative activity in normal human fibroblasts and pro-oxidative properties in malignant cells [11,12]. In our preliminary studies on the toxicological effects of Mes and TA in various breast cancer lines, we showed that TA, depending on the concentration used, has an effect on the selected herbicides including Mes [13]. The literature, which has documented relationship between the consumption of food contaminated with pesticides and the increased incidence of different types of cancers, led us to investigate whether TA can counteract the stimulating influence of mesotrione on the proliferation and growth of malignant cells.

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

#### *2.1. Reagents*

Phosphate buffered saline (PBS), without Ca and Mg, was provided PAN Biotech (Aidenbach, Germany). SDS (Sodium dodecyl sulphate), TCA (trichloroacetic acid), TBA (thiobarbituric acid), Folin-Ciocalteu reagent, andMesotrione were provided by Sigma-Aldrich and DTNB (dithiobis-2-nitrobenzoic acid, Ellman's reagent) by Serva. Dichlorodihydrofluorescein diacetate assay (DCFH-DA) and cell stain double staining kit containing propionium iodide and calcein AM was provided by Sigma-Aldrich, St. Louis, MO, USA. The fluorescein isothiocyanate (FITC) Annexin V Apoptosis Detection Kit I was purchased from BD Pharmingen (San Diego, CA, USA). The Cayman's Catalase Assay Kit, Cayman's Superoxide Dismutase Assay Kit, and Cayman GPx Assay were obtained from Cayman Chemical (Ann Arbor, MI, USA).

#### *2.2. Cell Culture*

The effect of herbicide, TA (Cayman Chemical Company (1180 East Ellsworth Road, Ann Arbor, MI, USA), purity ≥ 98%; formal name: 2E-dodecenedioic acid; CAS number: 6402-36-4; formulation: A crystalline solid) and the mix of herbicide with TA was studied in the ZR-75-1 cell line, which was

obtained from American Type Culture Collection (ATCC, Manassas, VA, USA). The ZR-75-1 cells were cultured in RPMI-1640 Medium containing glucose at 4.5 mg/mL (25 mM) supplemented with 10% fetal bovine serum (FBS) (PAN Biotech), penicillin (100 U/mL) (PAN Biotech), and streptomycin (100 μg/mL) (PAN Biotech) at 37 ◦C in a humidified atmosphere of 5% CO2 in air.

#### *2.3. Cytotoxicity Assay*

Cytotoxicity were studied according to the method of Carmichael using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) (Sigma-Aldrich, St. Louis, MO, USA) [14].

TA cytotoxicity was studied at selected concentrations of 0.5 μM, 0.75 μM, 1 μM, 10 μM, 20 μM, 50 μM, 100 μM, 200 μM, 500 μM, 750 μM, and 1000 μM. Mes cytotoxicity was estimated at concentrations of 0.01 μM, 0.025 μM, 0.05 μM, 0.1 μM, 0.5 μM, 1 μM, 5 μM, 10 μM, 25 μM, and 50 μM. The concentrations of both compounds selected for the analysis of the effect of the tested chemicals mixture on the cells have been chosen on the basis of MTT cytotoxicity tests performed. The concentration of Mes with the highest stimulating effect on the ZR-75-1 cells was selected. The control cells were cultured without test compounds.

Breast cancer cells were seeded in a 96-well plate at a density of 2 <sup>×</sup> <sup>10</sup><sup>4</sup> cells/well. Cells cultured for 24 h and 48 h were first treated with TA in the concentration range from 0.5 μM to 1000 μM, then Mes in the concentration range from 0.01 μM to 50 μM, and finally TA mixed with Mes: TA in the concentration range from 0.5 μM to 1000 μM mixed with 0.05 μM Mes. The analysis was conducted according to Jabło ´nska-Trypu´c et al. using a microplate reader GloMax®-Multi Microplate Multimode Reader (Promega Corporation, Madison, WI, USA) [13]. The viability of breast cancer cells was presented as a percentage of control cells. All the experiments were done in triplicate.

#### *2.4. Caspase 3*/*7 Activity Assay*

The activity of caspases 3/7 was examined at TA concentrations of 100μM and 200 μM, Mes concentration of 0.05 μM, and the combination of two compounds (TA + Mes, concentrations: 100 μM + 0.05 μM and 200 μM + 0.05 μM, respectively) after 24 h and 48 h of incubation. Breast cancer cells were seeded in 96-well white plate at a density of 2 <sup>×</sup> <sup>10</sup><sup>4</sup> cells/well. Luminescent assay was applied according to manufacturer's instructions (Promega Corporation, Madison, WI, USA) as described previously [15]. A microplate reader GloMax®-Multi Microplate Multimode Reader was used and the experiments were done in triplicate.

#### *2.5. Fluorescent Microscopy Analysis*

For the analysis of apoptotic and necrotic cells nuclear morphology, two fluorescent dyes, propionium iodide and calcein-AM, were applied. Cells were seeded on cell imaging dishes with coverglass bottoms at a density of 2 <sup>×</sup> 105 cells/well with 200 <sup>μ</sup>M TA, 0.05 <sup>μ</sup>M Mes, the mix of two compounds, and without the tested compound for 24 h. Subsequently, cells were washed twice with PBS and then stained with dyes solution in the dark in 37 ◦C for 15 min. The mixture of dyes was removed and the cells were washed with phosphate buffer and analyzed with the use of Olympus IX83 fluorescent microscope with SC180 camera with Cell Sens Dimension 1.17 program (200 × magnification). Calcein-AM stains viable cells and PI pass only through damaged membrane in dead cells. The following criteria were used: Living cells were characterized by regularly distributed green chromatin nucleus and are stained with green color. Dead cells, probably apoptotic cells, were characterized by red nuclei with chromatin condensation or fragmentation. Necrotic cells showed red-stained cell nuclei.

#### *2.6. Analysis of Apoptosis Using Flow Cytometry*

Breast cancer cells were seeded in six-well plates at a density of 2 <sup>×</sup> 105 cells/well. Cells were exposed to 100 μM TA, 200 μM TA, 0.05 μM Mes, and the mix of two compounds (TA + Mes, concentrations: 100 μM + 0.05 μM, 200 μM + 0.05 μM, respectively) and incubated for 24 h and

48 h. Apoptosis was studied by flow cytometry on FACSCalibura II cytometer (Becton-Dickinson). After trypsinization, cells were resuspended in RPMI-1640. Then, cells were suspended in binding buffer for staining with FITC (Annexin V) and propidium iodide (PI) for 15 min at room temperature in the dark following the manufacturer's instructions (FITC Annexin V apoptosis detection Kit I). The signal obtained from cells stained with Annexin V or PI alone was used for fluorescence compensation. Data were analyzed with FACStationTM software (BD PharmingenTM, San Diego, CA, USA).

#### *2.7. Total Protein Content*

ZR-75-1 cells were exposed to 100 μM TA, 200 μM TA, 0.05 μM Mes, and the mix of two compounds (TA + Mes, concentrations: 100 μM + 0.05 μM, 200 μM + 0.05 μM, respectively) and incubated for 24 h and 48 h. ZR-75-1 cells (2.5 <sup>×</sup> 10<sup>5</sup> cells/mL) were cultured with tested compounds. The protein concentration was determined as described previously [15]. All the experiments were done in triplicate.

#### *2.8. Determination of SH Groups*

SH groups content was analyzed using Rice-Evans method (1991) as described previously [15]. ZR-75-1 cells were exposed to 100 μM TA, 200 μM TA, 0.05 μM Mes, and the mix of two compounds (TA + Mes, concentrations: 100 μM + 0.05 μM, 200 μM + 0.05 μM, respectively) and incubated for 24 h and 48 h. ZR-75-1 cells (2.5 <sup>×</sup> <sup>10</sup><sup>5</sup> cells/mL) were cultured with tested compounds. All the experiments were done in triplicate.

#### *2.9. Determination of TBA Reactive Species (TBARS) Level*

The Rice-Evans method (1991) was used for measuring membrane lipid-peroxidation products level (TBARS), as described previously [15]. ZR-75-1 cells were exposed to 100 μM TA, 200 μM TA, 0.05 μM Mes, and the mix of two compounds (TA + Mes, concentrations: 100 μM + 0.05 μM, <sup>200</sup> <sup>μ</sup><sup>M</sup> <sup>+</sup> 0.05 <sup>μ</sup>M, respectively) and incubated for 24 h and 48 h. ZR-75-1 cells (2.5 <sup>×</sup> 10<sup>5</sup> cells/mL) were cultured with test compounds. All the experiments were done in triplicate.

#### *2.10. Determination of GSH*/*GSSG*

GSH/GSSG (GSH–reduced form of glutathione, GSSG–oxidized form of glutathione) ratio was examined at TA concentrations of 100 μM and 200 μM, Mes concentration of 0.05 μM, and the mixture of these two compounds (TA + Mes, concentrations: 100 μM + 0.05 μM and 200 μM + 0.05 μM, respectively) after 24 h and 48 h of incubation. Breast cancer cells were seeded in 96-well white plates at a density of 2 <sup>×</sup> 10<sup>4</sup> cells/well. GSH/GSSG ratio was assayed in triplicate via GSH/GSSG-Glo™ kit (Promega Madison, WI, USA) following manufacturer's instructions as described [15].

#### *2.11. Intracellular ROS Detection*

Intracellular ROS level was examined at TA concentrations of 100 μM and 200 μM, Mes concentration of 0.05 μM, and the mixture of these two compounds (TA + Mes, concentrations: 100 μM + 0.05 μM and 200 μM + 0.05 μM, respectively) after 24 h and 48 h of incubation. Breast cancer cells were seeded in 96-well white plates at a density of 2 <sup>×</sup> 104 cells/well. Dichlorodihydrofluorescein diacetate (DCFH-DA), (Sigma, St. Louis, MO, USA) and GloMax®-Multi Detection System (Promega Corporation, Madison, WI, USA) were used in order to measure the level of intracellular reactive oxygen species (ROS) [16]. The method was described previously [12]. All the experiments were done in triplicate.

#### *2.12. Catalase Activity*

Catalase is an enzyme which is involved in the detoxification processes, mainly the metabolism of hydrogen peroxide, which originates during normal aerobic metabolism and pathogenic reactive oxygen species (ROS) generation. Cells were cultured in six-well plates at 1 <sup>×</sup> 105 cells/well (Sarstedt), treated with 100 μM and 200 μM TA, 0.05 μM Mes, and the mix of two compounds (TA + Mes, concentrations: 100 μM + 0.05 μM, 200 μM + 0.05 μM, respectively) for 24 h and 48 h. For the determination of catalase activity, the Catalase Assay Kit (Cayman Chemical Company Ann Arbor, MI, USA) was used following manufacturer's instructions. The absorbance of final product was read at 540 nm using the GloMax®-Multi Microplate Multimode Reader. All the experiments were done in triplicate.

#### *2.13. Glutathione Peroxidase Activity*

Glutathione peroxidase is involved in cells protection against oxidative stress by catalyzing the reduction of hydroperoxides by reduced GSH. Cells were cultured in six-well plates at <sup>1</sup> <sup>×</sup> <sup>10</sup><sup>5</sup> cells/well (Sarstedt), and treated with 100 μM and 200 μM TA, 0.05 μM Mes, and the mix of two compounds (TA + Mes, concentrations: 100 μM + 0.05 μM, 200 μM + 0.05 μM, respectively) for 24 h and 48 h. Prior to the analysis, cells were washed with phosphate buffer. Cells were collected using rubber policeman and homogenized in a cold buffer, then centrifuged at 10,000× *g* for 15 min at 4 ◦C. Supernatant was used for the assay. For the determination of glutathione peroxidase activity GPx Assay kit (Cayman Chemical Company Ann Arbor, MI, USA) was used following manufacturer's instructions. The absorbance at 340 nm was read using the GloMax®-Multi Microplate Multimode Reader. All the experiments were done in triplicate.

#### *2.14. Superoxide Dismutase Activity*

Superoxide dismutases belong to the group of metalloenzymes. They catalyze the superoxide anion dismutation to molecular oxygen and hydrogen peroxide and therefore play an important role in the cellular antioxidant defense system. Cells were seeded in six-well plates at 1 <sup>×</sup> 105 cells/well (Sarstedt), and treated with 100 μM and 200 μM TA, 0.05 μM Mes, and the mix of two compounds (TA + Mes, concentrations: 100 μM + 0.05 μM, 200 μM + 0.05 μM, respectively) for 24 h and 48 h. Prior to the analysis, cells were washed with phosphate buffer. Cells were collected using rubber policeman and homogenized in a cold buffer, then centrifuged at 10,000× *g* for 15 min at 4 ◦C. Supernatant was used for the assay. For the determination of SOD activity, the Superoxide Dismutase Assay Kit (Cayman Chemical Company Ann Arbor, MI, USA) was applied following manufacturer's instructions. The absorbance (440–460 nm) was read using the GloMax®-Multi Microplate Multimode Reader. All the experiments were done in triplicate.

#### *2.15. Statistical Analysis*

Statistical analysis for the obtained results was performed. The effect of TA, Mes, and the combination of TA and Mes on apoptosis and oxidative stress parameters in ZR-75-1 cells was calculated as a means and compared in analysis of variance using the post-hoc test of ANOVA. The significant differences were estimated by Tukey test at *p* < 0.05. A biplot graph was used in order to present correlation between parameters. Staistica 13 (StatSoft, Kraków, Poland) was used to present data analysis.

#### **3. Results**

#### *3.1. Cytotoxicity*

An MTT assay was applied in order to estimate potential TA cytotoxicity (Figure 1A). The analyzed compound caused significant decreases in relative ZR-75-1 cell viability, which was observed right after 24 h treatment. It was effective even in lower concentrations. At 1 μM TA concentration, a 37% decrease was noticed after 24 h treatment. Concentrations of 100 μM and 200 μM decreased the viability of cells by about 40% and more than 50%, respectively, after 24 h treatment. None of the tested TA concentrations stimulated studied cells viability. On the other hand, Mes significantly increased relative cell viability. The most significant increase in the relatively shorter time of incubation was observed under the influence of 0.05 μM of Mes, which is presented in Figure 1B. Therefore, one of Mes concentrations, 0.05 μM, was selected for further analysis, and was applied together

with all of the studied concentrations of TA. We noticed significant decline in relative cell viability in both incubation times, especially in combination of Mes with TA in 0.05 μM + 100 μM and 0.05 μM + 200 μM, respectively (Figure 1C). Taking into account the results of the above-mentioned experiments, we decided to choose two combinations of analyzed compounds for further analysis of the mechanisms by which they affect breast cancer cells. TA concentrations of 100 μM and 200 μM are close to IC50 value, and more importantly, TA concentrations higher than 400 μM could be potentially cytotoxic for the human organism. For studying oxidative stress parameters and apoptosis, we analyzed the influence of 0. 05 μM Mes + 100 μM TA and 0.05 μM Mes + 200 μM TA.

**Figure 1.** Relative cell viability. The ZR-75-1 cell line was exposed to (**A**) graded concentrations of TA (traumatic acid), (**B**) graded concentrations of Mes (mesotrione), and (**C**) graded concentrations of TA mixed with 0.05 μM of Mes. Mean values from three independent experiments ± standard deviation (SD) are shown.

#### *3.2. Apoptosis*

Before conducting the Caspase-Glo® 3/7 Assay, cells were subjected to 0.05 μM Mes, 100 μM TA, 200 μM TA, and the combination of 0.05 μM of Mes with 100 μM TA and 0.05 μM of Mes with 200 μM of TA (Figure 2A). Mes treatment did not induce apoptosis in studied cell line. TA slightly enhanced the activity of analyzed caspases, especially after 48 h of treatment. However, the most significant changes in caspases 3/7 activity caused by two mixed compounds were observed when TA concentration was about 200 μM.

**Figure 2.** The effect of TA,Mes, and the combination of TA+Mes on apoptosis in ZR-75-1 cells, which were incubated with 100 μM of TA, 200 μM of TA, 0.05 μM of Mes, and a mix of 100 μTA + 0.05 μM Mes, 200 μM TA + 0.05 μM Mes. (**A**) Caspase 3/7 activity in ZR-75-1 cells under the influence of TA, Mes and TA + Mes on, (**B**) Bar graphs presenting the percentage of apoptotic cells. Mean values from three independent experiments ± SD are shown. Different letters indicate statistical differences (*p* ≤ 0.05) between treatments estimated by Tukey's test.

Apoptosis was also estimated using flow cytometry, and the results are presented in Figure 2B. In Figure 2B, the percent of apoptotic cells cultured for 24 h and 48 h with TA, Mes, and the mixture of TA and Mes is depicted. The values obtained for herbicide treatment indicate that tested compound did not enhance apoptosis. However, TA treatment, both alone and in combination with Mes, enhanced apoptosis. The results obtained in the flow cytometry experiment confirmed the studied caspases activity.

A fluorescent microscopy assay was used in order to confirm the occurrence of apoptosis (Figure 3). We evaluated apoptotic and necrotic cells morphology using fluorescent staining. Similar to luminescence and flow cytometry analysis, we observed differences between control, TA, and TA + Mes treatments. We did not observe differences between control and pesticide-treated cells. Calcein—AM stained only viable cells, while propidium iodide stained viable and death cells.

**Figure 3.** The influence of TA (200 μM), Mes (0.05 μM), and the mix of TA + Mes (200 μM + 0.05 μM) on apoptosis and necrosis in the ZR-75-1 cell line estimated using fluorescence microscope assay (200 × magnification). The cells were cultured with TA and Mes for 24 h and stained with Calcein-AM and propidium iodide. Three independent experiments were conducted and representative images are depicted.

#### *3.3. Oxidative Stress*

In the combination of Mes and TA, unsaturated dicarboxylic fatty acid demonstrates anticancer properties against Mes-induced breast cancer development by enhancing the stimulatory effect on oxidative stress parameters.

The influence of TA and Mes on the amount of SH groups is presented in Figure 4A. A statistically significant increase in thiol group content by approximately 68% was noticed under the influence of Mes after 24 h. Exposure to TA after 24 h incubation also caused an increase in analyzed parameter. However, 48 h incubation with tested compounds caused a significant decrease in thiol group content. Notably, the decrease was observed in case of the TA pretreatment of cells. The obtained results indicate that 100 μM of TA delayed the antioxidative effect of Mes on breast cancer cells, because we noticed a decrease in thiol group content caused by 0.05 μM Mes in the culture pretreated with 100 μM TA. The presented data may indicate that TA could be a compound, which intensifies oxidative stress in cancer cells, even in the presence of herbicide.

Lipid peroxidation in cancer cells is a very important process, which consists a source of free radicals inevitable for fast cancer cell proliferation (Figure 4B). Incubation with all of the analyzed compounds after 24 h caused an increase in TBARS content, which was analyzed as an index of lipid peroxidation. However, statistically significant changes in tested parameter were observed after 48 h treatment. Both TA alone and 100 μM TA in combination with Mes caused a very high increase in TBARS level. However, 200 μM TA mixed with Mes caused a decrease of about 75% as compared to the first analyzed mix, which could be explained by the presence of the other lipid peroxidation products, for example HNE. The obtained results suggest that TA may demonstrate protective properties by increasing membrane phospholipid peroxidation to such a high level, which is toxic to cancer cells.

Figure 5B shows the influence of TA and Mes on the production of ROS in the ZR-75-1 breast cancer cells. The intensity of fluorescence of 2 7 -dichlorodihydrofluorescein (DCF) for the ZR-75-1 cells cultured with TA and Mes for 24 h and 48 h is shown as a relative ROS amount. An incubation of cells with tested compounds caused an increase in ROS content in both analyzed times. At a concentration of 200 μM, TA caused an increase, which was very high, at about 95% after 24 h. Treatment with 100 μM and 200 μM of TA significantly reduced the ROS amount in the Mes-treated cell culture as compared to TA-treated cells. The presented data show an enhancing effect of TA on ROS formation.

The influence of Mes, TA, and the mixture of Mes with TA on GSH/GSSG ratio is depicted in Figure 5A. Reduced glutathione belongs to the group of very important antioxidants, which maintain oxidative balance within the cell. At a concentration of 200 μM, TA significantly decreased GSH/GSSG ratio after 24 h of incubation, while 100 μM of TA combined with 0.05 μM Mes significantly increased the tested parameter after 48 h incubation time compared to the control. Treatment with 200 μM of TA caused a reduced ratio of GSH/GSSG after 24 h and 48 h of culturing, even after addition of Mes. Treatment with the Mes and TA mixture for 24 h reduced the level of GSH compared to untreated cells. Based on the results obtained, we conclude that TA had a rather inhibitory effect and Mes had a stimulatory effect on the GSH/GSSG ratio in the ZR-75-1 cell line.

The first line of antioxidant defense that plays a key role in maintaining redox homeostasis in the cell are enzymes such as GPx, catalase and SOD. A significant increase in catalase activity under the influence of TA combined with 0.05 μM Mes was observed after 48 h incubation only (Figure 6A). At concentrations of 100 μM and 200 μM, TA, applied as a pretreatment before adding Mes, increased catalase activity by about 40% and 23%, respectively, as compared to the control untreated cells. At a concentration of 0.05 μM, Mes decreased catalase activity in both treatment times, however statistically insignificantly. The opposite results were observed in case of glutathione peroxidase (Figure 6B). The GPx activity was the highest under the influence of 100 μM TA + 0.05 μM Mes after 24 h incubation. Longer treatment with all of the tested compounds caused declines in GPx activity, but statistically insignificant. Our results showed that SOD activity was enhanced not only by the action of Mes, but also TA. The combination of these two compounds revealed decreases in SOD activity as compared to free Mes or free TA (Figure 6C).

In Figure 7, PCA analysis is depicted. It presents the correlation between the studied variables concerning the oxidative stress and apoptosis resulting from the activity of tested compounds and their mixture in the ZR-75-1 cell line. Figure 7 shows that 24 h treatment with TA was positively correlated with TBARS content, ROS content, and SOD activity, and was also correlated with the first component, which explains the 48.82% variability. However, Mes treatment was correlated with GSH/GSSG ratio, which is represented by the second component, explaining the 36.49% variability. After 48 h treatment, the previously observed correlation between TA and ROS content and TA + M and caspase 3/7 activity was maintained.

**Figure 4.** The influence of TA, Mes, and the mix of TA and Mes on SH group content (**A**) and TBARS content (**B**) in ZR-75-1 cells. The cells were cultured with 100 μM of TA, 200 μM of TA, 0.05 μM of Mes, mix of 100 μM TA + 0.05 μM Mes, and 200 μM TA + 0.05 μM Mes for 24 h and 48 h. Mean values from three independent experiments ± SD are shown. Different letters indicate statistical differences (*p* ≤ 0.05) between each treatment estimated by Tukey's test.

**Figure 5.** The influence of TA, Mes and the mix of TA and Mes on GSH/GSS (GSH–reduced form of glutathione, GSSG–oxidized form of glutathione) ratio (**A**) and reactive oxygen species (ROS) content (**B**) in ZR-75-1 cells. The cells were cultured with 100 μM TA, 200 μM TA, 0.05 μM Mes, and a mix of 100 μTA + 0.05 μM Mes, 200 μM TA + 0.05 μM Mes for 24 h and 48 h. Mean values from three independent experiments ± SD are shown. Different letters indicate statistical differences (*p* ≤ 0.05) between each treatment estimated by Tukey's test.

**Figure 6.** The influence of TA, Mes, and the mix of TA and Mes on catalase activity (**A**) GPx (glutathione peroxidase) activity (**B**) and SOD (superoxide dismutase) activity (**C**) in ZR-75-1 cells. The cells were cultured with 100 μM of TA, 200 μM of TA, 0.05 μM of Mes, and a mix of 100 μM TA + 0.05 μM Mes, 200 μM TA + 0.05 μM Mes for 24 h and 48 h. Mean values from three independent experiments ± SD are shown. Different letters indicate statistical differences (*p* ≤ 0.05) between each treatment estimated by Tukey's test.

**Figure 7.** Biplot showing variables (SH groups (thiol groups), TBARS (thiobarbituric acid reactive species) content, GSH/GSSG ratio, ROS content, SOD, catalase, GPX activity, caspase 3/7 activity, cytotoxicity) and cases (tested compounds: Mes, TA, and TA+Mes—mix of two analyzed compounds) in two dimensions.

#### **4. Discussion**

Due to the increasing popularity of diets based mainly on products of plant origin, we should be aware of the presence in everyday meals of both ingredients having a beneficial effect on the human body, but also compounds that are residues from the method of growing and preserving food of plant origin. The first group of compounds mentioned above includes TA, which is a plant hormone with beneficial antioxidant and probably anticancer effects. While the second group of compounds present in the crop plants are undoubtedly pesticides, an example of which is analyzed Mes. Taking into account its chemical structure TA belongs to the group of unsaturated fatty acids, whose positive effect on the human body has been quite well documented in the scientific literature. Although unsaturated fatty acids are fairly well known and their anti-cancer properties are investigated and described, there is literally scarce of literature data on TA. Our previous papers have shown that TA has a positive effect on healthy human fibroblasts by reducing oxidative stress level and that TA exhibits toxicity towards breast cancer cells by stimulating apoptosis through an increased level of oxidative stress [11,12]. In our preliminary study we also did an experiment regarding the mixtures of TA with selected herbicides frequently used in Poland and in EU on three breast cancer cell lines and one normal healthy cell line obtained from mammary gland [13]. Based on conducted experiments we concluded that TA in a dose–dependent manner may exert some toxicological effects in analyzed cells subjected also to herbicides. Therefore, as a next step, in this study we want to start investigating the mechanisms by which these two compounds may interact with each other and therefore influence growth and development of breast cancer cells.

First, the influence of analyzed compounds on the ZR-75-1 cell line proliferation was investigated. The MTT test was the basic experiment on the basis of which the concentrations were selected for further determinations. Cells under analysis were subjected to wide range of TA and Mes concentrations for 24 h and 48 h. The relative cell viability was monitored, and subsequently, one of Mes concentrations (0.05 μM) was selected for the experiment with the combination of two tested compounds. In the third part of the experiment, cells were pretreated with TA in the wide range of concentrations, and then Mes in the concentration of 0.05 μM was added. The most significant declines in Mes-treated cells viability were noticed as a result of 100 μM and 200 μM of TA treatment. The presented results are in agreement with our previously published data regarding TA influence on MCF-7 cells, where we indicated a decline

in cancer cell proliferation and viability caused by TA [12]. Literature data has indicated that pesticides may stimulate cancer cells proliferation through different mechanisms, e.g., glyphosate stimulates human breast cancer cells growth through estrogen receptors pathways, diuron acts in a tissue-specific manner and ROS play a role in its toxicity, and bifenox and dichlobenil exhibit enhancing effects on oxidative stress, simultaneously stimulating cancer cell proliferation and inhibiting apoptosis [17–19]. There is a large amount of literature data indicating a link between elevated levels of oxidative stress and a simultaneous increase in tumor cell proliferation [20–22]. Therefore, in our work, we focused primarily on the analysis of various parameters of oxidative stress. Cancer cells of different types are characterized by high level of ROS as compared to normal cells. This is primarily due to genetic disorders which appear in cancer cells, resulting in uncontrolled proliferation. In order to analyze the possible inhibiting effect of TA on the proliferation of Mes-treated cells, we conducted an MTT assay. The results indicate that TA exhibits antiproliferative properties.

In our studies, we noticed a decline in the amount of ROS in Mes-treated cells preincubated with TA, which may be associated with the cytotoxic effect of TA in cancer cells. Cancer cells usually use elevated level of oxidative stress caused, among others, by ROS generation in order to reduce the body's antioxidant protection. This allows the initial defeat of the first defense line against metastasis and angiogenesis, which are key stages in the development and progression of cancer [23,24]. This is in agreement with our research results indicating the relationship between an observed increase in ROS content and an increase in proliferation in Mes-treated cells and the decrease in ROS content in TA preincubated cells with a decrease in their proliferation level.

Many compounds of natural origin from the cytokinin group have antioxidant activity in healthy cells and pro-oxidative in cancer cells, which we also showed in our previous papers [12,25]. TA, as an unsaturated fatty acid, shows an increased susceptibility to oxidative processes. According to the literature, breast cancer cells are also definitely predisposed to oxidation of the macromolecules, which build them as compared to healthy cells. Unsaturated fatty acids have been shown to induce increased synthesis in *inter alia* lipid hydroperoxides in lipids that are part of cell membranes [26,27]. According to O'Shea M. et al., conjugated linoleic acid causes an increase in lipid peroxidation in breast cancer cells with a simultaneous decrease in cell proliferation [28]. We observed similar results in our research. We noticed increases in TBARS levels, which are correlated with changes in the content of SH groups. Incubation with TA for 48 h caused a decrease in thiol groups, even in Mes-treated cells. Mes stimulates the growth of cancer cells and causes an increase in the level of SH groups, which can probably also increase their resistance to oxidative stress and possible damage and enhance proliferation. Changes in the level of TBARS content and SH groups were also accompanied by changes in the GSH/GSSG ratio, particularly decreases under the influence of preincubation with 200 μM TA in cells treated with Mes. The correspondingly high level of GSH, which is one of the most important low molecular weight antioxidants in breast cancer cells, is usually correlated with the resistance of these cells to the induction of apoptosis and with their increased proliferation [29]. Our results indicate statistically significant decreases in GSH level under the influence of 200 μM TA in cells treated with Mes, with simultaneous statistically significant increases in the level of effector caspases 3/7 activity. The analysis of the activity of caspases 3/7, confirmed by the results obtained from flow cytometry and fluorescence microscopy, demonstrates that, at a concentration of 0.05 μM, Mes induces a decrease in the percentage of apoptotic cells as compared to control and to TA-treated cells. After both 24 h and after 48 h, we observed that TA at 200 μM significantly induced apoptosis even in combination with Mes, which may mean that TA is capable of overcome activity of pesticide and stimulate apoptosis in breast cancer cells.

Due to the application of TA, especially at a concentration of 200 μM, we noticed a significant increase in the activity of caspase 3/7, a decrease in ROS content, and a decrease in GSH content. However, it should be noted that the cells antioxidant defense system are divided into two parts: Enzymatic and nonenzymatic. Primary endogenous antioxidants are superoxide dismutase (SOD), catalase and glutathione peroxidase [30]. Our results clearly show that TA caused an increase in

oxidative toxicity in Mes-treated ZR-75-1 cells. It was manifested by a decrease in GPx activity and GSH/GSSG ratio and increase in TBARS content. Lipid hydroperoxides and other ROS are a major cause of oxidative damage within the cell membrane lipids, leading to increase in TBARS content. The GSH pool in the cells decreases due to an excessive generation of lipid peroxidation products and other oxygen species [31]. In turn, an excess in TA-induced ROS generation causes a significant decline in GSH synthesis and inhibition of antioxidant enzymes [32]. The reduced form of glutathione is an antioxidant low in molecular mass for appropriate cell integrity and redox balance, although GPx is an antioxidant enzyme, which contains selenium and its main role is scavenging of ROS. Catalase in breast cancer cells is characterized by high activity and expression level [33]. In our research, a decrease in catalase activity was observed under the influence of Mes and both analyzed concentrations of TA. However, under the influence of TA mixed with Mes, significant increases were noticed in both analyzed TA concentrations, especially after 48 h incubation. The presented results are consistent with literature data, indicating that catalase activity is induced in MCF-7 breast cancer cell line exposed to conjugated linoleic acid [34]. However, SOD activity, similar to GPx activity under the influence of Mes in TA-preincubated cells, was significantly lower as compared to the control or to the TA-treated and Mes-treated cells. Mes enhanced the activity of both SOD and GPx, but TA was effective enough to withstand the stimulating effect of the Mes and reduce the activity of the enzymes studied. Ding WQ et al. also observed that cancer cell treatment with docosahexaenoic acid reduced significantly SOD1 expression [35]. Literature data describing in vivo studies have shown divergent results of analyses regarding the effect of fatty acids on antioxidant enzyme activity. Some studies have indicated higher activity of antioxidant enzymes analyzed in animals consuming PUFA-enriched feed, while others have indicated that PUFAs caused a decline in the activity of these enzymes in the tissues of noncancerous rats [36,37]. In the present study, we found that TA inhibits SOD activity in Mes-treated ZR-75-1 cells, which has not been reported previously. Selected unsaturated fatty acids are known to modulate genes expression in malignant cells [38]. After being transported into cell nucleus, fatty acids such as TA are bounded to peroxisome proliferator-activated receptor [39]. This receptor response element is in the SOD1 gene promoter in rats [40]. Therefore, the transcription of the SOD1 gene could be influenced by TA in ZR-75-1 cells. On the other hand, SOD1 mRNA destabilization could be influenced by DHA, which subsequently causes its lower expression. The target reduction of SOD activity was a way to increase intracellular peroxide amount, hence causing an increase in mitochondrial damage and stimulating apoptosis in cancer cells [41].

In our experiments, we observed a 45% decrease of SOD activity in the ZR-75-1 cells treated with 200 μM TA and incubated with Mes. According to our biplot analysis results, this was not correlated the growth rate of cells but was significantly correlated with the effect on TA-induced lipid peroxidation and ROS content after 24 h and with GPx activity, GSH/GSSG ratio, and SH group content. Our results support the idea that compounds, which influence antioxidant enzymes activity and oxidative balance in cancer cells, could be applied in the elimination of tumor cells through the induction of apoptosis.

#### **5. Conclusions**

The presented results allow us to conclude that TA may act as pro-oxidative and pro-apoptotic agent against Mes-stimulated breast cancer growth and development. TA could be considered as a plant, alternative source of unsaturated fatty acids that can eliminate the positive effect of pesticides on the growth and development of breast cancer cells. By stimulating oxidative stress and inhibiting the enzymatic antioxidative defense system in cancer cells, this compound can inhibit the growth and development of breast cancer. It should be also mentioned that TA acts as a pro-oxidative and pro-apoptotic agent in other breast cancer cell lines, simultaneously acting as antioxidant in normal human cells. Due to its unique properties, it could be considered as an important food ingredient. Exposure to herbicides present in food is dangerous for both healthy people and certainly for women diagnosed with breast cancer. This is also evidenced by the results of our research showing the positive and stimulating effect of Mes on the development and growth of cancer cells. However, TA seems to

be a compound with high anti-cancer potential, which may endure the negative impact of herbicides on the human body.

**Author Contributions:** A.J.-T.—corresponding author, wrote the paper, planned experiments; performed experiments; analysed data; U.W.—performed statistical analysis; analysed data; E.W.—analysed data; J.R.—analysed data; A.B.—analyzed the data; funding acquisition. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was financially supported by Ministry of Science and Higher Education, Poland, under the research project number WZ/WBiIS´/3/2019.

**Acknowledgments:** This work was financially supported by Ministry of Science and Higher Education, Poland, under the research project number WZ/WBiIS´/3/2019.

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

**Compliance with Ethical Standards:** The manuscript does not contain clinical studies or patient data.

#### **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*

### **Is Acrylamide as Harmful as We Think? A New Look at the Impact of Acrylamide on the Viability of Beneficial Intestinal Bacteria of the Genus** *Lactobacillus*

### **Katarzyna Petka 1, Tomasz Tarko <sup>2</sup> and Aleksandra Duda-Chodak 2,\***


Received: 20 March 2020; Accepted: 18 April 2020; Published: 21 April 2020

**Abstract:** The impact of acrylamide (AA) on microorganisms is still not clearly understood as AA has not induced mutations in bacteria, but its epoxide analog has been reported to be mutagenic in *Salmonella* strains. The aim of the study was to evaluate whether AA could influence the growth and viability of beneficial intestinal bacteria. The impact of AA at concentrations of 0–100 μg/mL on lactic acid bacteria (LAB) was examined. Bacterial growth was evaluated by the culture method, while the percentage of alive, injured, and dead bacteria was assessed by flow cytometry after 24 h and 48 h of incubation. We demonstrated that acrylamide could influence the viability of the LAB, but its impact depended on both the AA concentration and the bacterial species. The viability of probiotic strain *Lactobacillus acidophilus* LA-5 increased while that of *Lactobacillus plantarum* decreased; *Lactobacillus brevis* was less sensitive. Moreover, AA influenced the morphology of *L. plantarum,* probably by blocking cell separation during division. We concluded that acrylamide present in food could modulate the viability of LAB and, therefore, could influence their activity in food products or, after colonization, in the human intestine.

**Keywords:** lactic acid bacteria; probiotic; acrylamide; viability; flow cytometry

#### **1. Introduction**

Acrylamide (AA) is a chemical compound used in many industries. It is produced as a substrate for the synthesis of polymers widely used in the paper, chemical, and cosmetics industries. In 1994, the International Agency for Research on Cancer (IARC) included acrylamide in a group of compounds "probably carcinogenic to humans" after laboratory tests in mice and rats [1].

Acrylamide in foods is formed mainly by the reaction of free asparagine with reducing sugars (especially fructose and glucose) during the Maillard reaction, but it can also be formed by other pathways, e.g., the acrolein pathway [2]. The most important factors for AA formation are time and the temperature of the thermal processing of food products, and it is thought that a prerequisite for AA formation is temperature exceeding 120 ◦C.

Acrylamide has been shown to be a reproductive toxicant in animal models [3,4]. It exerts neurotoxic activity [5–7], and many studies have proved that AA also has genotoxic, cytotoxic, and carcinogenic impacts on the human organism [6,8–10]. However, due to the fact that acrylamide does not exert a mutagenic effect in bacterial cells [3,11], it has been agreed that its carcinogenic activity is related to glycidamide (GA)—an acrylamide metabolite formed in mammalian cells. The mutagenic and genotoxic effects of GA have already been confirmed in various in vitro and in vivo studies, showing that this AA metabolite can induce the formation of DNA adducts, resulting in mutagenesis and the development of cancers [6,8,9,12].

The impact of AA on microorganisms is still unclear. The results of many assays made by various laboratories are consistent in showing that AA is not a mutagen in *Salmonella* Typhimurium tested strains at concentrations up to 5 mg/plate, with or without metabolic activation [3]. However, three epoxide analogs of acrylamide, e.g., glycidamide, have been reported to be mutagenic in *Salmonella* strains ± S9 activation [11,13]. Tsuda et al. [14] reported that AA did not induce any gene mutations in *Salmonella*/microsome test systems (TA98, TA100, TA1535, TA1537) and in *Escherichia coli*/microsome assays (WP2 uvrA−) up to a dose of 50 mg AA/plate, but acrylamide did show a strong positive response in a *Bacillus subtilis* spore-rec assay (induced DNA damage) at 10–50 mg/disc. According to the authors, the results suggested that AA had the potential to induce gross DNA damage rather than point mutations detected by the Ames test. There are also studies demonstrating that after introducing 1%–3% acrylamide into the growth medium, *Escherichia coli* cells undergo various changes, such as blockage of cell division, elongation of cells, inhibition of DNA synthesis, decreased osmotic stability, and ultrastructural alterations of the outer membrane [15].

Taking into account eukaryotic cells, it is worth citing the research of Kwolek-Mirek et al. [16]. They demonstrated that acrylamide caused impairment of growth of *Saccharomyces cerevisiae* yeast deficient in Cu, Zn-superoxide dismutase (Δsod1) in a concentration-dependent manner. This inhibitory effect was not due to cell death but to decreased cell vitality and proliferative capacity. Exposing Δsod1 yeast to acrylamide caused the increased generation of reactive oxygen species and decreased glutathione levels.

It has also been proven that some microorganisms have the ability to use acrylamide as a carbon and nitrogen source for their growth and that amidases are the main factor involved in AA degradation. Amidases are enzymes (EC. 3.5.1.4) that occur ubiquitously in nature and are characterized by a broad spectrum of catalyzed reactions [17]. Classification on the basis of catalytic activity takes into account the substrate specificity profile of the particular amidase and divides known amidases into six classes. During the amidase-catalyzed deamination reaction of acrylamide, acrylic acid and ammonia are formed. Then, acrylic acid can be reduced to propionate or transformed into β-hydroxypropionate, lactate, or CO2, in a pathway involving coenzyme-A [2,5,8].

To date, laboratory tests have shown the ability to degrade AA by many environmental microorganisms, mainly bacteria, such as *Ralstonia eutropha* [18], *Pseudomonas chlororaphis* [19], *Enterobacter aerogenes* [20], *Pseudomonas aeruginosa* [21,22], *Bacillus cereus* [23], *Rhodococcus* sp., *Klebsiella pneumoniae* [24,25], and *Burkholderia* sp. [26]. It is worth highlighting that among the amidase producers are certain species that naturally occur in human organisms or are delivered with food, such as *Escherichia coli* [27], *Bacillus clausii* [28], *Enterococcus faecalis* [29], and *Helicobacter pylori* [30,31]. However, the substrate specificity of their amidases and the potential for reaction with acrylamide have not yet been confirmed. In some cases, it has even been proved that those bacteria produce only cell wall amidases, such as N-acetylmuramoyl-L-alanine amidase [32], with no affinity to acrylamide. Either way, there is a possibility that members of microbiota could degrade acrylamide directly in the human intestine.

Lactic acid bacteria (LAB) constitute very important members of intestinal microbiota and play an important role in proper organism functioning and maintenance of our health [33–36]. Representatives of LAB are also important in the food industry, both as starter culture added during production and as native microbiota of raw materials used for food production [37–39]. The positive role of LAB could also be related to their ability to reduce AA levels in organisms or foodstuffs. To date, the possibility of degrading AA by amidase production has not been confirmed, although synthesis of N-acetylmuramoyl-L-alanine amidase, involved in the degradation of peptidoglycan and hydrolysis of the amide bond between N-acetylmuramic acid and L-amino acids of the bacterial cell wall, has been reported in LAB [40,41]. Other studies [42,43] have shown that *Lactobacillus reuteri* NRRL 14171 and *Lactobacillus casei* Shirota are able to remove acrylamide in aqueous solution by

physically binding the toxin to the bacterial cell wall, probably with a significant role of the teichoic acid structure. Later, Rivas-Jimenez [44] demonstrated that both mentioned bacterial strains were able to remove dietary AA (commercial potato chips with an average AA content of ~34,000 μg/kg) under different simulated gastrointestinal conditions. The percentage of AA removed by each bacterium exposed to different concentrations of the toxin (10–350 μg/mL) had a similar tendency; the lower the concentration of AA, the higher the percentage of toxin removed. The results showed that *L. casei* Shirota showed a higher percentage (68%) of AA removed than *L. reuteri* (53%) when bacteria were exposed to the lowest concentration of toxin (10 μg/mL), but no significant differences (*p* < 0.05) were observed in the percentage of toxin removed by both strains (~2%) when ≥100 mg/mL of AA was used. These findings proved that strains of the genus *Lactobacillus* could be employed to reduce the bioavailability of dietary AA. However, the strong dependence on AA concentration suggests that the mechanism of AA reduction is still the physical binding of AA by bacteria.

To the best of our knowledge, no one has investigated how acrylamide affects the viability of lactic acid bacteria so far, and this is an important issue considering their important role in the human body. First of all, lactic acid bacteria can be exposed to acrylamide just in food products. There are many fermented milk products that contain various "additives" rich in AA, such as biscuits, muesli, roasted almonds, nuts and seeds, dried fruit, breakfast cereals, and bran flake cereals. Also, so-called pro-health foods, such as probiotic bars and cereals, contain live strains of LAB, as well as crispy cereals, roasted nuts, almonds and seeds, almond and peanut butter, dried fruits, flakes, etc. Moreover, intestinal LAB can also be exposed to dietary acrylamide after intake of various fried, grilled, toasted, roasted, or baked foods. Although acrylamide is rapidly absorbed from the intestine, there are studies suggesting that some food matrices (or components) can reduce the intestinal absorption of AA. For example, a high protein concentration in the human diet may reduce acrylamide uptake [45], causing unmetabolized acrylamide to reach the colon. Therefore, the aim of this study was to evaluate whether acrylamide could influence the growth and viability of lactic acid bacteria belonging to the *Lactobacillus* genus.

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

#### *2.1. Bacteria*

Pure cultures of lactic acid bacteria belonging to the *Lactobacillus* genus were used in the study. For the experiments, 4 strains constituting a typical microbiota of fermented milk products and 2 probiotic strains were chosen: *Lactobacillus plantarum* DSMZ 20205, *Lactobacillus brevis* DSMZ 20054, *Lactobacillus lactis* subsp. *lactis* DSMZ 20481, and *Lactobacillus casei* DSMZ 20011. All were purchased from Leibniz Institut DSMZ (Deutsche Sammlung von Mikroorganismen und Zelkulturen GmbH, Braunschweig, Germany). Two probiotic strains—*Lactobacillus acidophilus* LA-5 and *L. casei* LC01—were obtained from Christian Hansen (Hørsolm, Denmark).

Bacteria were delivered as freeze-dried cultures and were handled according to supplier protocol. Briefly, after opening the ampoule, bacteria were rehydrated and then transferred to a tube with sterile liquid De Man, Rogosa, and Sharpe (MRS) agar medium (BioMaxima, Lublin, Poland) and incubated at a temperature optimal for strain. For *L. acidophilus* LA-5 and both *L. casei* strains, the optimal temperature was 37 ◦C, while, for other *Lactobacillus* species, it was 30 ◦C.

#### *2.2. Measurement of Optical Density of Bacterial Suspension: Calibration*

To tubes containing 5 mL of sterile MRS medium, a volume of 0.1 mL of 24-h liquid bacterial culture was added, the contents were mixed, and the tubes were incubated for 24 h at the optimum temperature for the tested strain. After incubation, bacterial cultures were centrifuged at 194× *g* for 15 min (MPW-35JR centrifuge, MPW MED Instruments, Warsaw, Poland), and the supernatant was discarded. The pellets were rinsed by mixing with 5 mL of sterile distilled water followed by centrifugation (using previous parameters). The resulting pellets were resuspended in sterile water so

as to obtain an optical density of the bacterial suspensions equal to McFarland standard 1.0 (using a Den-1B densitometer, Biosan, Latvia). Then, serial 10-fold dilutions were made in sterile water, and 1 mL of subsequent dilution was spread over the surface of the MRS medium (in triplicate). After 72 h of incubation at an optimal temperature, bacterial colonies were counted, mean bacterial cell density in cfu/mL from 3 replicates was calculated for each tested strain, and the relationship between the optical density of McFarland = 1 and bacterial cell density was determined. The relationships obtained for individual strains were as follows (1 McFarland unit equivalent): *L. plantarum*, 1.55 <sup>×</sup> 108 cfu/mL; *L. brevis*, 4.5 <sup>×</sup> 10<sup>7</sup> cfu/mL; *L. lactis* subsp. *lactis*, 1.6 <sup>×</sup> 108 cfu/mL: *L. casei*, 4.9 <sup>×</sup> 107 cfu/mL; *L. acidophilus* LA-5, 4.45 <sup>×</sup> 107 cfu/mL; *L. casei* LC01, 4.8 <sup>×</sup> 10<sup>7</sup> cfu/mL. Before each experiment, a 24 h culture of adequate *Lactobacillus* strain was centrifuged, washed in sterile water, and resuspended (as described above). The optical density of the bacterial suspension was adjusted to a value corresponding to 2 <sup>×</sup> <sup>10</sup><sup>7</sup> cfu/mL.

#### *2.3. Model Medium for Experiments*

All experiments were carried out in carbon- and nitrogen-limiting conditions because model medium composed of 0.45% NaCl (POCh, Gliwice, Poland), and 0.45% bacteriological peptone (BioMaxima, Lublin, Poland) was used. If a solid medium was required, bacteriological agar was added in a final concentration of 2% (BioMaxima, Lublin, Poland). All media were sterilized using a Microjet Microwave Autoclave (process parameters: 135 ◦C, 80 s, 3.6 bar; Enbio Technology Sp. z o.o., Gdynia, Poland).

#### *2.4. Preparation of Acrylamide "Stock" Solution*

Concentrated (20 g/L) aqueous solution of acrylamide (purum, ≥98% (GC) provided by Sigma-Aldrich Sp. z o.o, Poznan, Poland) was sterilized by filtering through a sterile membrane filter (pore ϕ = 0.22 μm; PES Millex-GP, Bionovo, Poland) and diluted (if needed) with sterile distilled water to obtain "stock" solutions of acrylamide (concentrations: 0.5, 1.0, 2.0, 5.0, 10.0, and 20.0 g/L).

#### *2.5. Preliminary Assessment of Acrylamide Impact on Lactobacillus Growth*

The impact of acrylamide on *Lactobacillus* was assessed by evaluating visible bacterial growth on the solid model medium containing acrylamide at various concentrations: 10, 50, 100, 250, 500, and <sup>1000</sup> <sup>μ</sup>g/mL. Serial 10-fold dilutions of the suspension of tested bacteria (2 <sup>×</sup> 107 cfu/mL) were made in sterile water. Then, a volume of 1 mL of acrylamide "stock" solution of adequate concentration was added to 18 mL of sterile, cooled, but still, liquid, model medium and poured into a sterile Petri plate containing 1 mL of the diluted bacterial suspension. Positive controls were Petri plates with 19 mL of the model medium (without acrylamide) mixed with 1 mL of a diluted suspension of tested bacteria. After media solidification, all plates were incubated for 72 h at a proper temperature optimal for the tested strain, and then the bacterial growth was assessed according to the following scale:


First, the growth of bacteria on plates with positive control was evaluated, and the dilution of bacterial suspension with good growth (30–300 colonies/plate) was chosen. For the same dilution, growth in the presence of AA was assessed. The experiment was performed in 5 replicates.

#### *2.6. Determination of Cell Concentration and Viability by Flow Cytometry*

The *Lactobacillus* strains whose growth was influenced by acrylamide in the preliminary analysis were chosen for this stage of the experiment. A volume of 1 mL of bacterial suspension (containing <sup>2</sup> <sup>×</sup> <sup>10</sup><sup>7</sup> cfu/mL) was inoculated into 19 mL of liquid model medium, with the addition of acrylamide to a final concentration of 7.5, 15, 30, or 100 μg/mL, and incubated for 48 h. The final bacterial cell density was 10<sup>6</sup> cells/mL, which corresponded to the average number of LAB cells found in fermented milk drinks (FAO/WHO Food Standards). The proposed AA concentrations were selected based on the literature [42,46], and the 100 μg/mL concentration is higher than the possible level reached in the human gastrointestinal tract or in food products. The positive control was medium with 1 mL of sterile distilled water added instead of an acrylamide "stock" solution (marked as 0 μg/mL). Immediately after adding bacteria to the medium (marked as 0 h, but taking into account staining times and cytometric measurement, the analysis was actually done about 2 h after adding the bacteria), after 24 h and 48 h of incubation at an optimal temperature, the cell concentration (cell/mL) was evaluated by flow cytometry (BD AccuriTM C6 Flow cytometer, BD Biosciences, Bio-Rad, Poland) equipped with fluorescence detectors FL1 533/30, FL2 585/40, FL3 670LP. For this purpose, the commercially available BD™ Cell Viability Kit with BD Liquid Counting Beads (cat. # 349480, Becton, Dickinson and Company, BD Biosciences, San Jose, CA, USA) was used. According to the protocol, cells were stained with provided dyes, and cytometric analysis was conducted using the following parameters: fluidic flow rate 14 μL/min, the threshold set at 10,000 on (Forward Scatter-Height), sample volume set at 10 μL. The bacterial cells and counting beads were gated based on (Side Scatter) parameters and FL2, while the populations of alive, injured, and dead bacteria were discriminated based on an FL1 (thiazole orange) vs. FL3 (propidium iodide) plot. In live cells, the membrane is intact and impermeable to dyes, such as propidium iodide (PI), while when cells are injured or dead, the propidium iodide can leak into the cells because of their compromised membranes. PI is a nucleic acid intercalator, so it stains nucleic acids. On the other side, thiazole orange is a permeant dye that also reacts with nucleic acids but enters all cells—alive, injured, and dead, to varying degrees. Therefore, it will stain all cells containing nucleic acids. Thus a combination of these two dyes provides a rapid and reliable method for discriminating live, injured, and dead bacteria. To determine the concentrations of cell populations (expressed as cell/mL), Equation (1) was used:
