**Characterization of** *Sporidiobolus ruineniae* **A45.2 Cultivated in Tannin Substrate for Use as a Potential Multifunctional Probiotic Yeast in Aquaculture**

**Apinun Kanpiengjai 1,2,\*, Chartchai Khanongnuch <sup>3</sup> , Saisamorn Lumyong 2,4,5 , Aksarakorn Kummasook <sup>6</sup> and Suwapat Kittibunchakul <sup>7</sup>**


Received: 4 November 2020; Accepted: 14 December 2020; Published: 18 December 2020 -

**Abstract:** At present, few yeast species have been evaluated for their beneficial capabilities as probiotics. *Sporidiobolus ruineniae* A45.2, a carotenoid-producing yeast, was able to co-produce cell-associated tannase (CAT), gallic acid and viable cells with antioxidant activity when grown in a tannic acid substrate. The aim of this research study was to identify the potential uses of *S. ruineniae* A45.2 obtained from a co-production system as a potential feed additive for aquaculture. *S. ruineniae* A45.2 and its CAT displayed high tolerance in pH 2.0, pepsin, bile salts and pancreatin. Furthermore, its viable cells were characterized by moderate hydrophobicity, high auto-aggregation and moderate co-aggregation with *Staphylococcus aureus*, *Salmonella* ser. Thyphimurium and *Streptococcus agalactiae*. These attributes promoted *S. ruineniae* A45.2 as a multifunctional probiotic yeast. In addition, the intact cells possessed antioxidant activities in a 100–150 µg gallic acid equivalent (GAE)/mL culture. Remarkably, the fermentation broth demonstrated higher antioxidant activity of 9.2 ± 1.8, 9.0 ± 0.9, and 9.8 ± 0.7 mg GAE/mL culture after FRAP, DPPH and ABTS assays, respectively. Furthermore, higher antimicrobial activity was observed against *Bacillus cereus*, *Staphylococcus aureus* and *Strep. agalactiae*. Therefore, cultivation of *S. ruineniae* A45.2 with a tannic acid substrate displayed significant potential as an effective multifunctional feed additive.

**Keywords:** feed additive; probiotic; yeast; *Sporidiobolus ruineniae*; tannase

#### **1. Introduction**

Aquaculture is the most rapidly growing sector of food production throughout the world. Its global demand continues to rise as it is applied to farming practices in ever-increasing proportions [1]. Industrial-scale aquaculture production is recognized as a challenge within the farming sector as it requires sustainable and efficient technologies [2] that address concerns of potential human exposure

to microbial diseases and the possibility of severe economic losses across the industry [3]. The use of antibiotics is a simple approach in the control of diseases, however, excess antibiotics that are discharged into the environment are known to be responsible for the spread of antibiotic-resistant genes of pathogenic and commensal bacteria, all of which can lead to increases in drug resistance among animal and human populations. The addition of additives to antibiotics, vaccines, immunostimulants, prebiotics and probiotics, in particular, is an environmentally friendly alternative and recognized as a sustainable strategy [2]. The Food and Agriculture Organization (FAO)/World Health Organization (WHO) defined probiotics as live microorganisms that, when administered in adequate amounts, confer a health benefit upon the host [4]. In terms of their potential applications in aquaculture, yeasts are considered a second leading source of probiotics after bacteria. However, the use of probiotic yeasts is not as popular as bacteria. Indicative of a greater potential for profit than bacterial probiotics, yeasts are not affected by antibacterial compounds and are known to contain various immunostimulant compounds [3,5]. However, their applicable use among a wide variety of animals is limited due to the fact that the normal body temperature of animals is higher than the temperature for the optimal growth of yeast. To date, two yeast species, namely *Saccharomyces cerevisiae* and *Debaryomyces hansenii*, are widely recognized as potential probiotic yeasts [5]. Additionally, yeasts isolated from fish microbiota exhibit certain probiotic properties. These yeast species include *Candida deformans*, *Rhodotorula mucilaginosa*, *Yarrowia lipolytica*, *Metschnikowia viticola* and *Cryptococcus laurentii* [3].

*Sporidiobolus ruineniae* A45.2, isolated from fermented tea-leaves of northern Thailand, namely Miang, is a pigment-producing and tannin-tolerant type of yeast [6]. It is among the range of yeasts commonly found in the intestines of humans [7]. Based on evidence established by previous studies [6,8], *S. ruineniae* A45.2 is assumed to have a unique cell wall structure that serves its tannin-tolerance and may promote the organism as a potential probiotic yeast. On the other hand, its carotenoid pigment is considered highly valuable in terms of the enhancement of some aquaculture pigmentations [9]. Moreover, *S. ruineniae* A45.2 is capable of producing thermostable and pH-stable cell-associated tannase (CAT) and can degrade tannic acid to gallic acid [10]. Tannase is a feed additive enzyme that plays an important role in the reduction of tannins, an antinutritional factor in animal feed. The enzymatic degradation of tannins releases gallic acid that can be used as both an antimicrobial and an antioxidant agent. In aquaculture feed ingredients, tannins come from plant-derived, alternate fish feed ingredients that are used as protein sources, such as soybean meal, rapeseed meal, pea seed meal and mustard oil cake [11]. This drawback leads to significantly decreased levels of cumulative feed intake and digestibility [12]. However, this can be overcome by the addition of tannase.

In previous studies involving the co-production of gallic acid and CAT derived from tannins, both cells and the culture broth rich in gallic acid may be used as a potential source of tannase and gallic acid in the feed industry, respectively. The aims of this research study were to evaluate *S. ruineniae* A45.2 for its potential to be used as a probiotic in aquaculture. Our objectives were to also investigate the potential for fermented broth cultivated in a tannic acid substrate to be further applied as a multifunctional feed additive. Further carotenoids produced by this yeast were also characterized.

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

#### *2.1. Chemicals*

Bile salts, 40× pancreatin, pepsin, methyl gallate, gallic acid, rhodanine, 2,4,6-Tris (2-pyridyl)-*s*triazine (TPTZ), 2,2-diphenyl-1-picrylhydrazyl (DPPH), 2,2′ -azino-bis (3-ethylbenzothi-azoline-6 ysulfonic acid (ABTS) and potassium persulfate were all of analytic grade and of the highest quality available from Sigma and Sigma-Aldrich (St. Louis, MO. USA). All media used in this research study, including yeast–malt extract broth (YMB), nutrient broth (NB), trypticase soy broth (TSB) and agar, were purchased from HiMedia (Nashik, India).

#### *2.2. Microorganisms and Culture Conditions*

*S. ruineniae* A45.2 was maintained on yeast-malt extract agar (YMA) at 4 ◦C for further use. To prepare the seed inoculum, a single colony of yeast was inoculated in YMB and incubated at 30 ◦C on a 150 rpm rotary shaker for 15–18 h or until the optical density at 600 nm reached 2.0–3.0. *Escherichia coli* TISTR 527, *Salmonella* ser. Thyphimurium TISTR 1472, *Staphylococcus aureus* TISTR 746 and *Bacillus cereus* TISTR 747 were maintained on nutrient agar (NA) and were grown in a nutrient broth (NB) at 37 ◦C on a 100 rpm rotary shaker when necessary. Furthermore, *Listeria monocytogenes* DMST 17303 and *Streptococcus agalactiae* DMST 11366 were maintained on trypticase soy agar (TSA). These pathogenic bacteria were grown in TSB at 37 ◦C on a 100 rpm rotary shaker when necessary.

#### *2.3. Co-Production of Gallic Acid, CAT and Viable Cells of S. ruineniae A45.2*

A total of 10% (*v*/*v*) of the prepared seed inoculum was transferred into a 1 L stirred tank fermenter (B.E. Marubishi Co. Ltd., Tokyo, Japan) with a 60% working volume of the optimized medium [10] that contained 12.3 g/L tannic acid, 6.91 g/L glucose, 10 g/L yeast extract, 2 g/L (NH4)2SO4, 0.5 g/L tween 80 and 1 g/L glutamate (pH 6.0). Culture conditions were administered at 30 ◦C with an aeration rate of 1 vvm and an agitation speed of 250 rpm. After 48 h of cultivation, the cells were harvested by centrifugation at 8000 rpm, 4 ◦C for 10 min and washed twice with phosphate buffer saline (PBS) supplemented with 0.1% (*v*/*v*) triton X-100. The cell pellets were then resuspended in 0.85% (*w*/*v*) NaCl to obtain a concentration of 10<sup>8</sup> cells/mL for further experimentation.

#### *2.4. Tolerance of S. ruineniae A45.2 and Stability of Tannase at Low pH Values*

A total of 0.5 mL of the prepared cell suspension (10<sup>8</sup> cells/mL) was transferred to a 125 mL Erlenmeyer flask containing 49.5 mL of 0.85% (*w*/*v*) NaCl adjusted to pH 2.0 and 3.0 by 0.1 N HCl. The cell suspension in PBS was then used as a control. All mixtures were incubated at 30 ◦C for 4 h. Samples were periodically taken for measurement of viable cells by plate count technique and residual tannase activity. Initial cell concentration and tannase activity without incubation were set to 100%.

#### *2.5. Tolerance of S. ruineniae A45.2 and Stability of Tannase in Simulated Gastric Juice*

A total of 0.5 mL of the prepared cell suspension (10<sup>8</sup> cells/mL) was transferred to a 125 mL Erlenmeyer flask containing 49.5 mL of simulated gastric juice (0.3% (*w*/*v*) pepsin, 0.85% (*w*/*v*) NaCl, pH 2.0). Cell suspension in PBS was used as a control. All mixtures were incubated at 30 ◦C for 4 h. Samples were periodically taken for measurement of viable cells by plate count technique and to determine residual tannase activity. Initial cell concentration and tannase activity without incubation was set to 100%.

#### *2.6. Bile Salt Tolerance of S. ruineniae A45.2 and Stability of Tannase*

A total of 0.5 mL of the prepared cell suspension (10<sup>8</sup> cells/mL) was transferred to a 125 mL Erlenmeyer flask containing 49.5 mL of solution that consisted of 0.85% (*w*/*v*) NaCl and 0.3% (*w*/*v*) bile salts. Cells suspended with PBS were used as a control. All mixtures were incubated at 30 ◦C for 6 h. Samples were periodically taken for measurement of viable cells by the plate count technique to determine residual tannase activity. Initial cell concentration and tannase activity without incubation were set to 100%.

#### *2.7. Tolerance of S. ruineniae A45.2 and Stability of Tannase in Simulated Intestinal Fluid*

A total of 0.5 mL of the prepared cell suspension (10<sup>8</sup> cells/mL) was transferred to a 125 mL Erlenmeyer flask containing 49.5 mL of simulated intestinal fluid (0.3% (*w*/*v*) bile salts, 0.3% (*w*/*v*) pancreatin and 0.85% (*w*/*v*) NaCl). Cells suspended with PBS were used as a control. All mixtures were incubated at 30 ◦C for 8 h. Samples were periodically taken for measurement of viable cells by plate

count technique to determine residual tannase activity. Initial cell concentration and tannase activity without incubation were set to 100%.

#### *2.8. Assay of Tannase*

Tannase activity was assayed using the previously described method [13]. Briefly, 50 µL of enzyme solution was mixed with 50 µL of 12.5 mM methyl gallate in 100 mM citrate–phosphate buffer pH 6.5 and incubated at 30 ◦C. After the incubation procedure, the reaction was stopped by adding 60 µL of 0.667% (*w*/*v*) methanolic rhodanine and the mixture was left at room temperature (25 ◦C) for 5 min. Subsequently, 40 µL of 0.5 M KOH was added to the mixture, which was then left at room temperature for 5 min prior to adding 800 µL of distilled water. Absorbance of the mixture was measured at 520 nm. One unit of tannase was defined as the amount of enzyme that liberated 1 µmol of gallic acid per minute under assay conditions.

#### *2.9. Cell Surface Hydrophobicity*

Yeast adherence was determined by cell surface hydrophobicity. The cell suspension (3 mL) (Ainitial) was transferred to a glass tube (12 × 100 mm) containing 1 mL of chloroform, agitated using a vortex mixer for 2 min and allowed to stand at room temperature for 30 min. The optical density of the aqueous phase (Afinal) was measured at a wavelength of 600 nm. The hydrophobicity index (HPBI) was calculated using the following equation:

$$\text{HPBI} \left( \% \right) = (1 - \frac{\text{A}\_{\text{final}}}{\text{A}\_{\text{initial}}}) \times 100$$

#### *2.10. Auto-Aggregation Assay*

A total of 3 mL of the yeast suspension in PBS (Ainitial) was transferred to a glass tube (12 × 100 mm), vortexed for 10 s and incubated at 30 ◦C for 2 h. Absorbance of the upper part of the mixture (approximately 1 mL) was measured at 600 nm (Afinal). Auto-aggregation was calculated using the following equation:

$$\text{Auto} - \text{aggregation}(\%) = (1 - \frac{\text{A}\_{\text{final}}}{\text{A}\_{\text{initial}}}) \times 100$$

#### *2.11. Co-Aggregation Assay*

Equal volumes (1.5 mL) of the yeast suspension (Ayeast) and pathogenic bacterium (Apathogen) were transferred into a glass tube (12 × 100 mm), vortexed for 30 s and incubated at 30 ◦C for 2 h. Absorbance of the upper part of the mixture (Amix) was measured at 600 nm. Co-aggregation was calculated using the following equation:

$$(\text{Co} - \text{agargaction}(\%) = (1 - \frac{\text{A}\_{\text{mix}}}{(\text{A}\_{\text{year}} + \text{A}\_{\text{pathogen}})/2}) \times 100^{\circ}$$

#### *2.12. Adherence of Bacteria onto Yeast Cells*

Adhesion of bacteria onto yeast cells was accomplished by mixing 1 mL of the yeast suspension (10<sup>8</sup> cells/mL) in PBS and 1 mL of each pathogenic bacteria (10<sup>8</sup> cells/mL). The specimens were then incubated at 30 ◦C. After 2 h of incubation, 10 µL of the mixture was smeared onto a microscopic slide for Gram-staining [14]. The Gram-stained slide was then used to visualize the adherence of the bacteria onto the yeast cells under a phase-contrast light microscope. The pathogenic bacteria used in the adherence test were *B. cereus*, *E. coli*, *Sal.* Thyphimurium, *Staph. aureus*, *L. monocytogenes* and *Strep. agalactiae*.

#### *2.13. Determination of Antimicrobial Activity*

The agar well diffusion method was used to determine antagonistic activity of culture broth obtained from the co-fermentation and fermentation in YMB. Briefly, an overnight culture (approximately 106–10<sup>8</sup> CFU/mL) of the pathogenic bacteria was swabbed onto an NA plate for *B. cereus*, *E. coli*, *Sal.* Thyphimurium and *Staph. aureus* and a TSA plate for *L. monocytogenes* and *Strep. agalactiae*. The wells were prepared by being punched with a 6 mm diameter sterile cork-borer and were filled with 50 µL of sterile culture broth or 50 µg/mL of chloramphenicol as the positive control. The plates were incubated at 30 ◦C for 18 h.

#### *2.14. Analysis of Carotenoids*

The previously obtained cell pellet of *S. ruineniae* A45.2 was lyophilized into a dry cell for carotenoid extraction. The freeze-dried cell (0.25 g) was then placed into a screw-cap glass tube (25 × 150 mm) containing 10 mL of acetone and 30 g of glass beads. Cell disruption was carried out through vigorous mixing for 10 min at room temperature. The mixture was filtered through filter paper to collect the cell extract and then centrifuged at 10,000 rpm for 10 min for the purposes of clarification. To quantify total carotenoids, absorbance of the clear extract was measured at 485 nm. The total carotenoid content of the yeast cells was calculated based on the extinction coefficient (E 1% 1 cm) of 2680 and expressed in terms of total carotenoids (µg)/g dry cell weight. To determine the carotenoid composition, individual carotenoids were separated by Mightysil RP-18 GP prepacked column (150 × 2.0 mm; Kanto Chemical Co., Inc., Tokyo, Japan) equilibrated with a solution of methanol/acetonitrile (90:10 *v*/*v*). The conditions were carried out at 30 ◦C with a flow rate of 1.0 mL/min. The separated carotenoids were detected using a UV detector at 485 nm. Meanwhile, the evaporated cell extract was resuspended in acetone. This was then spotted on a thin layer chromatography (TLC) Silica gel plate (Silica gel 60 F254, Merck Millipore, Germany) and developed in a chamber containing hexane/acetone (70:30 *v*/*v*). The pigments separated by TLC were recovered and dissolved in acetone for measurement of the wavelength of maximum absorbance (λmax) using a UV-visible spectrophotometer.

#### *2.15. Assay of Antioxidants*

A culture of *S. ruineniae* A45.2 obtained from the co-production system was harvested by centrifugation at 8000 rpm for 10 min. The cell pellet was washed twice with PBS solution, suspended in the same solution and the fermented broth was then collected. Both intact cells and cell-free extract were determined for antioxidant activity using three different methods, including ferric-reducing antioxidant power (FRAP), DPPH free-radical-scavenging activity and ABTS free-radical-scavenging activity.

For the FRAP assay, the FRAP reagent consisted of 300 mM acetate buffer pH 3.6, a solution of 10 mM TPTZ in 40 mM HCl and 20 mM FeCl<sup>3</sup> at a ratio of 10:1:1 (*v*/*v*/*v*). The sample solution (0.10 mL) was mixed thoroughly with 3.40 mL of the FRAP reagent for 30 min prior to measuring the absorbance at 593 nm. A standard curve was prepared using different concentrations of gallic acid. The results were expressed in terms of milligram gallic acid equivalent (GAE)/mL culture.

For the DPPH assay, a sample (0.25 mL) was mixed with freshly prepared 40 ppm methanolic DPPH (2.25 mL) and allowed to stand in the dark at room temperature. A decrease in absorbance at 517 nm was determined after 10 min of the incubation process. The concentration of the sample that produced between 20% and 80% inhibition of the blank absorbance was determined and adapted. Radical scavenging activity was expressed as the concentration of the extract required for reduction of the initial concentration of DPPH by 50% (EC50) under specified experimental conditions. DPPH radical scavenging activity was expressed in terms of mg GAE/mL culture.

For the ABTS assay, 0.0384 g of ABTS was prepared in 10 mL of water. Subsequently, 5 mL of the solution was mixed with 88 µL of 140 mM potassium persulfate and adjusted to 25 mL with deionized water in a volumetric flask for further experimentation. A total of 1.75 mL of the ABTS solution was mixed thoroughly with 0.25 mL of the sample and allowed to stand in the dark at room temperature for 10 min. A decrease in absorbance at 734 nm was measured. Radical scavenging activity was expressed as the concentration of the extract required for reduction of the initial concentration of ABTS by 50% (EC50) under specified experimental conditions. ABTS radical scavenging activity was expressed in terms of mg GAE/mL culture.

#### **3. Results**

#### *3.1. Survival of Yeast and CAT Stability under Simulated Gastrointestinal Tract (GIT) Conditions*

Gastric and intestinal conditions were simulated for evaluation of the probiotic properties of *S. ruineniae* A45.2. Under simulated gastric conditions, cell viability of *S. ruineniae* A45.2 and its residual CAT activity were evaluated at pH 2.0 and pH 3.0 (Figure 1a,b) and carried out with and without (Figure 1c,d) the supplementation of pepsin. The results revealed that the values of both pH and pepsin did not significantly affect cell viability of the yeast or its CAT. At a pH value of 2.0, both in the presence and in the absence of pepsin, approximately 90% of the initial viable cells and residual CAT activity were retained. Furthermore, at a pH value of 3.0, the supplementation of pepsin slightly decreased tannase activity but had no effect on cell viability.

**Figure 1.** Effect of pH 2.0 and pH 3.0 on (**a**) survival of *S. ruineniae* A45.2 and (**b**) residual CAT activity and effect of pepsin on (**c**) survival of *S. ruineniae* A45.2 and (**d**) residual CAT activity.

The same results were also obtained when *S. ruineniae* A45.2 and CAT were incubated under simulated intestinal conditions, in which bile salts and pancreatin acted as key factors. It was revealed that *S. ruineniae* A45.2 and CAT were resistant to bile salts (Figure 2a,b) and pancreatin (Figure 2c,d), as they retained 100% residual viable cells and tannase activity after incubation.

**Figure 2.** Effect of bile salts on (**a**) survival of *S. ruineniae* A45.2 and (**b**) residual CAT activity and effect of pancreatin in combination with bile salts on (**c**) survival of *S. ruineniae* A45.2 and (**d**) residual CAT activity.

#### *3.2. Cell Surface Hydrophobicity, Auto-Aggregation and Co-Aggregation*

Cell surface hydrophobicity, auto-aggregation and co-aggregation of *S. ruineniae* A45.2 are presented in Table 1. *S. ruineniae* A45.2 displayed 58.4 ± 2.7% cell surface hydrophobicity. Significantly high values of auto-aggregation were observed at up to 88.2 ± 1.2% along with the ability to be co-aggregated with pathogenic bacteria, including *B. cereus*, *E. coli*, *Staph. aureus*, *Sal.* Thyphimurium, *L. monocytogenes* and *Strep. agalactiae* at different percentages of co-aggregation. *S. ruineniae* A45.2 displayed a stronger co-aggregation ability with *Strep. agalactiae*, *Sal.* Thyphimurium and *Staph. aureus* than other tested pathogenic bacteria. This evidence was in agreement with their adherence ability (Figure 3), wherein pathogenic bacteria obviously adhered to the yeast cells.


**Table 1.** Cell surface hydrophobicity, auto-aggregation and co-aggregation against pathogenic bacteria.

#### *3.3. Antimicrobial Activity*

The antagonistic effects of *S. ruineniae* A45.2 were observed after being exposed to various indicator microorganisms, including *B. cereus*, *E. coli*, *Sal.* Thyphimurium, *Staph. aureus*, *L. monocytogenes* and *Strep. agalactiae*. Culturing periods of 24 and 48 h of the co-production system were tested in comparison with specimens cultured in YMB. Only the supernatant obtained from co-production exhibited antimicrobial activity against some pathogenic bacteria (Figure 4), i.e., *B. cereus*, *Staph. aureus* and *Strep. agalactiae*.

**Figure 3.** Adherence of (**a**) *B. cereus*, (**b**) *E. coli*, (**c**) *Staph. aureus*, (**d**) *Sal.* Thyphimurium, (**e**) *L. monocytogenes* and (**f**) *Strep. agalactiae* on yeast cell walls observed under a phase-contrast light microscope at 100× magnification.

μ **Figure 4.** Antimicrobial activity of cell-free extract of *S. ruineniae* A45.2 cultivated in tannic acid substrate at 30 ◦C for 24 (left) and 48 h (right) of cultivation against (**a**) *B. cereus*, (**b**) *E. coli*, (**c**) *Staph. aureus*, (**d**) *Sal.* Thyphimurium, (**e**) *L. monocytogenes* and (**f**) *Strep. agalactiae* compared to control (top) (50 µg/mL chloramphenicol)

#### *3.4. Carotenoids Produced by S. ruineniae A45.2*

μ *S. ruineniae* A45.2 grown in tannic acid were harvested after 48 h of cultivation, lyophilized and used for carotenoid extraction. Identification and characterization of the carotenoid pigment was performed by HPLC. Three main peaks were separated from the carotenoid extracts (Figure 5a). These peaks corresponded to the three spots that were isolated and visualized by TLC (Figure 5b). The first spot from the bottom was rosy-red in color and migrated with slower mobility than the second spot, which appeared orange to red in color, while the third spot was yellow in color and

migrated with the same degree of mobility as β-carotene. These pigments were identified based on their Visible absorbance maxima (Figure 5c). The least degree of polar pigment was identified as β-carotene due to similar absorbance maxima values recorded at 429 nm and 485 nm, with the maximal degree of absorbance recorded at 456 nm. The most notable polar pigment revealed a spectrum with three absorption maxima at wavelengths of 474, 527 with the absorption optimum at 499 nm, thereby identified as torularhodin. The second most polar pigment was torulene with three absorbance maxima values recorded at 462 nm and 516 nm, with an absorption maximum value recorded at 488 nm. Total carotenoids produced by *S. ruineniae* A45.2 after cultivation at 30 ◦C for 48 h were recorded at 88.0 ± 0.2 µg/g dry cell weight, equivalent to 500 ± 130 µg/L culture.

**Figure 5.** Characterization of pigments produced by *S. ruineniae* A45.2 (**a**) Separation of pigments by HPLC and (**b**) TLC and (**c**) Visible absorption spectra of the major pigments.

#### *3.5. Antioxidant Activity*

Antioxidant activities of intact cells of *S. ruineniae* A45.2 and cell-free extract obtained from the co-production system were measured using FRAP, DPPH and ABTS assays. Overall, there were no significant differences in the antioxidant activities among the different methods. Antioxidant activities ranging from 100–120 µg GAE/mL culture and 110–150 µg GAE/mL culture were detected from the intact cells obtained from 24-h and 48-h periods of cultivation, respectively (Table 2). The cell-free extract revealed significantly higher antioxidant activities than the intact cells. The antioxidant activities of 5.6–6.8 mg GAE/mL culture and 9.0–9.8 mg GAE/mL culture were obtained from the cell-free extract fraction after 24-h and 48-h periods of cultivation, respectively. Activities of both intact cells and cell-free extract increased relative to the incubation time of the co-production system.

**Table 2.** FRAP, DPPH and ABTS antioxidant activity of cell-free extract and intact cells obtained from co-production of gallic acid and viable cells of *S. ruineniae* A45.2.


\* significant difference within a column (*p* < 0.05).

#### **4. Discussion**

In this study, *S. ruineniae* A45.2 and its culture broth obtained from the co-production of cells, gallic acid and tannase were characterized for their potential use in animal feed, specifically in feed prepared for fish and other aquatic organisms. *S. ruineniae* A45.2 was isolated from Miang, which is rich in tannins and considered a microbial inhibitor [6,8]. The cell wall structure and composition of *S. ruineniae* A45.2 are believed to promote its growth along with high concentrations of tannic acid, yet this yeast was found to be a promising probiotic. When used as a functional probiotic yeast, growth temperature is a crucial limitation for the application of probiotics in animals, since the yeast must be able to survive and grow at the animal's normal body temperature in order to enhance the animal's growth performance and promote the health of the animal [15]. Typically, the growth temperatures of yeasts range from 0 to 47 ◦C with an optimal temperature between 25 and 30 ◦C [16], yet probiotic yeasts might actively function when they are used in aquaculture. Probiotic yeasts are less popular than bacteria but can offer some major physiological contributions over bacteria. These include their cell volume and the production of a wide spectrum of simple and more complex compounds that may be beneficial to the health of aquatic organisms. However, only a few varieties of probiotic yeasts have been isolated for aquaculture applications. It was reported that marine and other aquatic environments, along with the gut microbiota of aquatic organisms, are potential sources of probiotic yeasts. In addition to *S. cerevisiae*, *D. hansenii* is a ubiquitous yeast that is frequently associated with fish and marine environments [7]. As of yet, no reports of using *S. ruineniae* as a probiotic yeast have been identified.

To be a good probiotic yeast, it must be able to successfully survive under gastrointestinal tract (GIT) conditions and provide beneficial conditions for the enhancement of the health of the host. In this study, both *S. ruineniae* A45.2 and its CAT were exposed to GIT conditions in order to assess the degree of residual cell viability and CAT. The temperature used in this research study was 30 ◦C, as it was identified as an optimal temperature of *S. ruineniae* A45.2 (data not shown). Considering cell viability, *S. ruineniae* A45.2 resisted low pH values ranging from 2.0 to 3.0, which were within the range found in the stomachs of fish. The degree of acidity in the stomach of a fish can vary depending on the fullness of the stomach and the species of the fish [17,18]. Moreover, *S. ruineniae* A45.2 was not found to be affected by the digestive enzymes we tested, namely pepsin and pancreatin (a mixture of amylase, protease and lipase). These attributes are considered important selection criteria for a good probiotic yeast [19]. On the other hand, the CAT of *S. ruineniae* A45.2 exhibited a good degree of thermostability and pH stability. Surprisingly, positive stability values were observed under simulated GIT conditions by retaining more than 90% of initial activity after treatment. This indicates that the yeast species could be applicable in the aquafeed industry. Plant-based products in fish diets contain valuable proteins used to replace fishmeal. These plant feed ingredients contain considerable amounts of tannins that can have an adverse effect on animals by reducing the nutritional value of the feed [11]. This circumstance can also decrease the palatability of the feed due to an unpleasant taste caused by a high concentration of tannins [12]. The results of this study indicate that both cells of *S. ruineniae* A45.2 might be able to survive in transit through the stomach and small intestines and function effectively in the large intestines. However, its CAT might be stable in stomach environments and could be active in the intestines, as the environments are similar to the known optimal values for pH and temperature.

Cell surface hydrophobicity is defined as a nonspecific interaction in adhesion between probiotic microorganisms onto GIT epithelial cells, where they may provide prophylactic and therapeutic benefits [20]. Colonization in the intestinal epithelial cell wall and mucosal surfaces can prevent pathogenic bacteria adhesion and inflammatory reactions [21]. Yet, hydrophobicity is an important attribute for selecting potential probiotics. *S. ruineniae* A45.2 showed high cell surface hydrophobicity toward chloroform and was comparable with those reported in *Bacillus subtilis* [19,21], various strains of *Lactobacillus* sp. [22] and *Sac. unisporus* [20].

Auto-aggregation is defined as aggregation among yeast cells to form flocs and colonize the intestinal environment of the host when the cells approach harmful conditions [20,23]. Probiotic microorganisms should be associated with higher auto-aggregation than pathogenic microorganisms [22], specifically *Strep. agalactiae*, a representative fish pathogen. Under the same experimental conditions as this study, the percentage auto-aggregation of pathogenic bacteria ranged between 15–35% for *L. monocytogenes*, *Sal.* Thyphimurium and *Staph. aureus* [22]. Within 2 h of the auto-aggregation test, *S. ruineniae* reported 88.2 ± 1.2%, which was higher than previously reported probiotic yeasts, namely *P. kluyveri*, *Issatchenkia orientalis*, *P. kudriavzevii* [24], *Yarrowia lipolytica*, *Wickerhamomyces anomalus* and *Sac. cerevisiae* [23]. Auto-aggregation capacity is strain-specific, while a capacity greater than 50% displayed the potential to prevent the invasion of various other pathogenic microorganisms through film formation.

Co-aggregation is defined as the close interaction between probiotics and different pathogenic bacteria [23]. It was reported that adherence of enteric bacteria onto yeast cells is irreversible, thus transient passage of the bacteria occurs through GIT and subsequent flushing out in the feces [20]. The co-aggregation ability of *S. ruineniae* A45.2 agreed with its adherence ability. This could be explained by the specific fimbriae present on bacteria with mannan on yeast cells and the electrostatic and hydrophobic nonspecific bindings [20].

No antimicrobial activity of *S. ruineniae* A45.2 against the tested pathogenic bacteria was detected when it was cultivated in YMB. It is therefore implied that no antimicrobial metabolite was produced by the organisms typically identified in various yeasts [20,24–26]. Most yeasts scavenge pathogenic infection by indirect mechanisms such as auto-aggregation, co-aggregation and adherence ability [27]. On the contrary, the growing of *S. ruineniae* A45.2 in tannic acid containing medium led to the release of gallic acid, which enhanced the antimicrobial activity of the culture broth against *B. cereus*, *E. coli*, *Staph. aureus* and *Strep. agalactiae*. The results suggest that production of the yeast should be performed in the presence of tannic acid to promote gallic acid production and CAT, thereby gaining antimicrobial activity. Supplementation of gallic acid in animal feed, especially aquatic feed, was scarcely reported. Current research found that the supplementation of gallic acid in broiler diets at levels ranging from 75 to 100 mg/kg improved the performance of broiler chicks in terms of feed utilization, breast muscle yield and oxidative stability, while positively modulating jejunum intestinal morphology [28]. Hence, our results provide supplemental, supportive evidence for the use of gallic acid as an alternative to antibiotics in animal feed or for the determination of synergistic interactions of gallic acid that could enhance the effects of antibiotics.

*S. ruineniae* A45.2 is a basidiomycetous yeast that forms a natural pink-red pigment made up of carotenoids. The pigments extracted from the yeast were separated into three types of carotenoids based on the separation by HPLC and TLC. These pigments displayed distinctively different visible spectra. The most polar pigment showed a rosy-red color and had a similar visible spectrum to torularhodin, while the others displayed a similar spectrum to torulene and β-carotene as the second most polar and the least polar pigments, respectively [29–31]. However, structural elucidation of these compounds must be confirmed. Currently, carotenoid-producing yeasts are mainly represented by the genera *Rhodosporidium*, *Xanthophylomyces Rodotorula* and *Sporobomyces*. The latter genus has a close relationship to the genus *Sporidiobolus* and represents the main source of torulene and torularhodin [32]. The quantity of total carotenoids produced by *S. ruineniae* A45.2 was in the range of those produced by the yeast studied in previously published reports [31,33].

After cell wall components, some probiotic yeasts exhibit multifunctional potential in the production of bioactive compounds with certain antioxidant properties, such as carotenoids, organic acids and glutathione [24]. As *S. ruineniae* A45.2 is a carotenoid-producing yeast, it is likely that it possesses antioxidant capacity. The intact cells and cell-free extract obtained from YMB were evaluated for their antioxidant activity (data not shown). No antioxidant activity was detected in the cell-free extract, while the intact cells possessed approximately 10 times lower the degree of antioxidant activity than that obtained from cultivation in tannic acid. This may have resulted from the presence of β-glucan as a component of yeast cell wall composition [34,35]. Cultivation of *S. ruineniae* A45.2 in tannic acid could potentiate the antioxidant activity of not only intact cells but also cell-free extracts. It was

determined that the fermentation of *S. ruineniae* A45.2 induced the production of CAT which strongly affected the degradation of tannic acid, resulting in gallic acid production. During the degradation of tannic acid, large amounts of gallic acid were released into the fermentation broth and attached to the yeast cell surface, reported in previous studies [10,36]. The antioxidant activity in terms of gallic acid equivalent is likely a consequence of gallic acid content, as reported in previously published studies [10]. This result agrees with previously reported evidence published on the fermentation of plant-based foods [37], including grape seed flour and extracts [38], as well as Miang [39], as sources of *S. ruineniae* A45.2.

Overall, *S. ruineniae* A45.2 may be capable of exhibiting the beneficial characteristics attributed to a probiotic yeast that can be used for aquaculture. Cultivation of the yeast in tannic acid substrate might provide a number of benefits. These benefits include the assertion that yeast cells can be a source of antioxidant agents, tannase and carotenoids for aquatic organisms. Furthermore, it is believed that the resulting culture broth can display strong antioxidant activity as well as the potential to display antimicrobial activity against some pathogenic bacteria, especially fish pathogens. Therefore, this research study described and verified an alternative integrative strategy for the production of feed additives. To our knowledge, this is the first report to suggest that *S. ruineniae* exhibits probiotic properties.

#### **5. Conclusions**

*S. ruineniae* A45.2 was tolerant to simulated GIT conditions, displaying tolerance to pH 2.0, pepsin, bile salts and pancreatin. A high percentage of auto-aggregation was observed, and this species co-aggregated various pathogenic bacteria, specifically *Strep. agalactiae*, and adhered to some specific strains of pathogenic bacteria. These are considered beneficial attributes that support the use of *S. ruineniae* as a probiotic yeast. The fermentation of tannic acid to gallic acid has resulted in the co-production of CAT, gallic acid and viable yeast cells. Moreover, CAT was found to be stable and may be able to function under simulated GIT conditions, while the cells possessed antioxidant activity. Thus, *S. ruineniae* A45.2 as a carotenoid- and CAT-producing yeast could be labeled as a multifunctional probiotic yeast suitable for the feed of animals, particularly aquatic animals. In addition, its cell-free extract derived from the co-production system could be a potential alternative source of natural antioxidants and antimicrobial agents.

**Author Contributions:** Conceptualization, A.K. (Apinun Kanpiengjai); funding acquisition, A.K. (Apinun Kanpiengjai); investigation, A.K. (Apinun Kanpiengjai), C.K., S.L., A.K. (Aksarakorn Kummasook) and S.K.; methodology, A.K. (Apinun Kanpiengjai), A.K. (Aksarakorn Kummasook) and S.K.; project administration, A.K. (Apinun Kanpiengjai); supervision, A.K. (Apinun Kanpiengjai), C.K. and S.L.; writing—original draft, A.K. (Apinun Kanpiengjai); writing—review and editing, A.K. (Apinun Kanpiengjai). All authors read and approved of the final version of this manuscript.

**Funding:** This research study was partially funded by the Thailand Research Fund through the Research Grant for New Scholars (MRG6280057), and Chiang Mai University.

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

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 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/).

**Birgit Keller 1,\*, Henrike Kuder <sup>1</sup> , Christian Visscher <sup>1</sup> , Ute Siesenop <sup>2</sup> and Josef Kamphues <sup>1</sup>**


Received: 30 September 2020; Accepted: 2 December 2020; Published: 4 December 2020

**Abstract:** Liquid feed is susceptible to microbiological growth. Yeasts are said to cause sudden death in swine due to intestinal gas formation. As not all animals given high yeast content feed fall ill, growth and gas formation potential at body temperature were investigated as possible causally required properties. The best identification method for these environmental yeasts should be tested beforehand. Yeasts derived from liquid diets without (LD − S) and liquid diets with maize silage (LD + S) were examined biochemically (ID32C-test) and with MALDI-TOF with direct smear (DS) and an extraction method (EX). Growth temperature and gas-forming potential were measured. With MALDI-EX, most yeast isolates were identified: *Candida krusei* most often in LD − S, and *C. lambica* most often in LD + S, significantly more than in LD − S. Larger colonies, 58.75% of all yeast isolates, were formed at 25 ◦C rather than at 37 ◦C; 17.5% of all isolates did not grow at 37 ◦C at all. Most *C. krusei* isolates formed high gas amounts within 24 h, whereas none of the *C. lambica*, *C. holmii* and most other isolates did. The gas pressure formed by yeast isolates varied more than tenfold. Only a minority of the yeasts were able to produce gas at temperatures common in the pig gut.

**Keywords:** yeasts; liquid swine diets; MALDI-TOF; biochemical identification; growth temperature Ancom Gas Production System; *Candida krusei*; *Candida lambica*

#### **1. Introduction**

Yeasts, about 600 species of which are known [1], are ubiquitous in nature and can also be found on feedstuffs [2]. They pose a risk factor regarding hygiene in liquid diets associated with off-flavor and loss of nutrients [3–5]. Depending on the species or strain, as well as on the growth conditions like temperature, substrate and its aw-value (activity of water), yeasts are able to metabolize numerous sugars, starch, protein, amino acids or even fats, and therefore lead to a loss of nutrients and energy in the feed [3–5]. In pig fattening, these energy losses in the feed are particularly undesirable [4,5]. In addition, the flavor and smell of the feed can be negatively affected [6,7]. High cell counts of yeasts in liquid swine diets due to pronounced metabolic activity are often seen in the presence of easily fermentable, low molecular weight sugars [5]. Choosing maize silage for pig feed was used with the aim of feeding the pigs to increase the feeling of satiety without making them fat [8]. The relatively high initial yeast flora of the feed has to be taken into account [8]. Therefore, feed hygiene related to yeast content was of special concern.

In liquid feeds, mostly microflora develops, which is dominated by lactic acid-producing bacteria [7]. A pH-value lower than 5.0, which significantly reduces several bacteria, is often achieved in a shorter time with the use of starter cultures for fermented liquid feeds [3,9,10]. Yeasts are not only able to stay alive but also continue growing in fermented feeds [11], even if the pH-value is 4.5 [12].

Besides these complications concerning feed composition and quality, animal health may be affected due to the yeast content in the feed [11,13–15]. Hemorrhagic bowel syndrome (HBS), mainly caused by yeasts [13], is supposed to be causally responsible for gastric torsion and gastrointestinal tympani [16], being sometimes associated with liquid feeding [15]. HBS preferentially affects fattened pigs in the second half of the fattening period [16]. Those animals most affected are, as a rule, the better developed pigs in the group [14]. The fact that the affected animals are in excellent health makes this disease of particular economic importance [15].

In feed analyses, yeasts, irrespective of the species, are classified as spoilage indicators in animal feed [17]. A liquid diet with more than 10<sup>6</sup> cfu yeasts/g original substance (OS) is considered as significantly increased, while less than 10<sup>5</sup> cfu/g feed OS in liquid feed is considered as normal [18]. On the other hand, selected yeasts are authorized feed additives in human nutrition and animal feedstuffs as they synthesize vitamin B1, B2, B6, B12, folic acid, niacin, pantothenic acid and biotin, as well as containing some minerals (potassium, sodium, calcium, zinc and iron) [1]. For swine diets, viable *S. cerevisiae* is authorized as a feed additive as intestinal flora stabilizers, digestibility enhancers and microorganisms with a minimum concentration of 1 × 10<sup>9</sup> cfu/kg complete feed (88% DM) [19].

Pathogenicity factors of yeasts have been analyzed to identify high-risk yeasts and their effects on humans and animals. In their study about potential virulence of food-borne yeasts, Rajkowska et al. [20] stated that the ability to grow at 37 ◦C was crucial; hence, they referred to this as preliminary criterion for pathogenicity. Adaptation to pH-value was also suggested to be a key to pathogenicity, especially important for yeasts entering the digestive tract where the pH-value changes from pH 2 to pH 8 [21,22]. The ability to form biofilms also on abiotic surfaces [21] or even to colonize them is a prerequisite for colonizing the liquid feeding system, which allows the yeasts to stay alive even if the hygiene of the liquid feed was improved [23]. Stalljohann et al. [3] distinguished yeasts according to their ability to produce high or low amounts of CO<sup>2</sup> with regard to their pathogenicity for swine, but did not mention which species produced the high gas amounts. Such detailed information on these possible indicators of pathogenicity is provided in the present paper.

The hypothesis of this study was that different yeast species could be found in different feedstuffs. For this reason, a comparison of biochemical differentiation and identification with MALDI-TOF was performed to determine the method with the most reliable identification. Presumably, only distinct species would be able to grow and to produce high amounts of gas at 37 ◦C.

A comparison between gas amounts produced from yeasts measured with the Ancom Gas Production System under defined conditions in a standardized Sabouraud glucose bouillon, had, to the best of the authors' knowledge, never been carried out previously. This permits a comparison of yeast isolates not only to see whether but also how much gas can be produced by yeasts within a certain time period regardless of feedstuffs. Further studies must clarify whether and to what extent these properties have an influence on the development of diseases such as HBS. These new aspects could then allow to make better predictions concerning the ability of high yeast cell counts in liquid diets to cause clinical problems.

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

#### *2.1. Sample Origin*

For our project, samples from farms with liquid feeding common to all samples were collected. We obtained these samples either by contacting farms in the area, or from our own studies, which were also carried out on similar farms. In total, 42 liquid feed samples were analyzed. Of these samples, 33 were submitted for diagnostic purposes to the Institute for Animal Nutrition, University of Veterinary Medicine Hannover, Foundation, Germany. These included common liquid feeds to which no silage had been added, referred to as liquid diets without silage (LD − S). The remaining nine samples

obtained from field trials carried out by the Institute for Animal Nutrition, additionally contained whole plant maize-silage (up to 66% DM; liquid diets with silage, LD + S). For collecting the liquid feed samples, a standard laboratory protocol was used for both the submissions and the samples from the studies. The protocol required that the samples were taken fresh, packed directly into a sterile, unbreakable vessel, filled to 2/3 at most, immediately cooled and not sent before the weekend. All samples were processed directly, or in case they arrived late in the afternoon, refrigerated and processed the following morning.

#### *2.2. Detection Techniques*

Yeasts were isolated and morphologically characterized on Sabouraud glucose agar (SAB-Agar, PO 5096A, Thermo Fisher Scientific GmbH, Bremen, Germany) and then incubated at 30 ◦C. Only yeasts that grew at the highest decimal dilution levels of the agar plates were considered.

#### 2.2.1. Biochemical Differentiation

Biochemical differentiation of the yeast isolates was performed by ID 32 C strip (bioMérieux SA, Marcy-l'Ètoile, France). This was performed in accordance with the manufacturer's instructions. The strip consisted of 32 cavities, each containing a dehydrated carbohydrate substrate, testing the assimilation by the yeast. Pure culture yeast material of 44–48 h-grown subcultures was suspended in 3 mL aqua destillatum. Turbidity was set in accordance with a McFarland standard of 2.0 using a Densitometer DEN-1B (BioSan, Riga, Lettland). From this solution, 250 µL were added to the API C medium included in the test kit. After careful vortexing, 135 µL were transferred from this liquid medium to each well of the test strip. The strip was incubated at 30 ◦C for 44–48 h. Yeast growth resulted in turbidity of the liquid medium in the cupules, which was visually evaluated. The obtained results were noted on a result sheet. The values corresponding to the positive reactions were then added up within groups. Three results each were added up for a group. Group values were coded into a numerical profile. This was analyzed by means of identification software (APIWEBTM, bioMérieux). The results of two of these carbohydrates, *N*-acetylglucosamine (NAG) and lactic acid (LAT), are examined in more detail below. Only results that received good, very good or excellent (classified as "Very good identification") ratings were evaluated. Rice agar (Thermo Fisher Scientific GmbH), incubated at 25 ◦C for 44–48 h, was selected for some isolates if the identification software required this deficiency medium, with a cover glass placed over the inoculum for an oxygen-reduced atmosphere.

#### 2.2.2. MALDI-TOF

In the MALDI-TOF analysis, the sample (e.g., bacteria or yeasts) was ionized by a laser beam. These ions were then accelerated differently depending on their mass and charge. The time required to pass through the length of the flight tube was determined [24]. In this way, a characteristic spectrum can be generated for bacteria or fungi, which usually allows a species diagnosis [24,25]. As an advantage, less time is required for this method compared to biochemical methods [24].

MALDI-TOF analyses were performed on a Microflex LT/SH MALDI-MS Biotyper (Bruker Daltonik GmbH, Bremen, Germany) with the direct smear method (MALDI-DS) and with a formic acid–ethanol extraction (MALDI-EX). The latter is used for hardly soluble bacteria or yeasts.

For MALDI-DS, direct on-plate smearing was performed with yeasts incubated 44–48 h on an SAB-plate at 30 ◦C. Small amounts of colony material of every isolate were evenly applied with a toothpick to two circles of the target plate (8280800 MSP 96 Target polished steel BC, Bruker Daltonik). After air drying the sample material at room temperature for five to ten minutes, 1 µL of an α-cyano-4-hydroxycinnamic acid (HCCA, 19182, Sigma Aldrich Inc., St. Louis, MO, USA) matrix solution was applied to each circle of the target plate and dried again at room temperature.

For MALDI-EX, a 1 µL loop of 20–24 h-grown yeast material on an SAB-plate was vortexed in 300 µL deionized water at 30 ◦C; 900 µL of ethanol absolute, HPLC-grade, was added and it was vortexed again. The samples were pelleted by centrifuging for two minutes (13,000 U/min; approximately 3000× *g*); the supernatant was discarded, centrifuged and then the supernatant was discarded again. The pellet in the Eppendorf tube was air dried for three to five minutes and resuspended in 40 µL 70% formic acid. After adding 40 µL acetonitrile (ACN, Acetonitrile HPLC Gradient Grade, 20060.320 VWR International Inc., Radnor, PA, USA), this was followed by the same centrifugation step as above. Eppendorf tubes were taken carefully out of the centrifuge and 1 µL of the supernatant was applied to a circle of the target and air dried for 5 min at room temperature. Immediately afterwards, 1 µL of the same matrix solution that had been previously used for the MALDI-DS method was applied to a circle of the target and air dried for 5 min at room temperature.

For the control of both methods, 1 µL BTS (Bacterial Test-Standard, 8290190 Bruker IVD Bacterial Test-Standard) was placed on every target plate and, additionally, a control strain, *E. coli* DH5 α, was tested on every target plate in two circles.

Each sample was analyzed by a Microflex LT/SH MALDI-TOF MS in the linear mode across a mass-to-charge ratio range between 2000 and 20,000. The obtained data were analyzed automatically by using the MBT Compass Library BDAL and MBT Flex Control software, BTyp2.0-Sec.Library, 1.0. Every strain was tested in two circles, with a decreased concentration in the second circle. The result from the two circles that achieved the higher score value was used for analysis.

The identification cut-off scores were interpreted as per Bruker's recommendation scores as follows: obtaining scoring thresholds between 2.30 and 3.0 suggested highly probable species identification; 2.00–2.29, probable species identification; 1.70–1.99, identification at the genus level was postulated; whereas cut-off scores <1.70 indicated no reliable identification.

#### 2.2.3. Method Comparison

The results "excellent identification" and "very good identification" adopted in the software APIWEBTM were equated to "highly probable species identification" from the MALDI-TOF analysis. "Good identification" was equated to "probable species identification"; "probable genus identification" was equated to "good identification at genus level"; and "doubtful profile" and "no identification" were equated to "unacceptable profile".

#### *2.3. Temperature Comparison*

After cultivating the yeasts from liquid feeds on SAB-agar and subcultivating a single colony, a subculture on two SAB-agar plates was produced for the temperature comparison. One agar plate each was cultivated in an incubator at either 25 ◦C or 37 ◦C for 48 h. The colony growth (diameter) was compared visually.

#### *2.4. PH-Value*

Using a calibrated glass electrode (HI 2211 pH/ORP Meter, Hanna Instruments Inc., Woonsocket, RI, USA), pH-values were measured. Results of 25/33 LD − S and 16/17 LD + S were obtained immediately after dividing the samples for further microbiological testing.

#### *2.5. Gas Pressure Measurement*

To measure yeast activity in gas production, 40 selected yeast isolates were examined for 24 h at 37 ◦C with the ANKOM RF Gas Production System (ANCOM Technology, Macedon, NY, USA), which remotes pressure under controlled pressure measurements and records these on a standard Excel spreadsheet. Gas production curves were generated. In 100 mL glass bottles containing 100 mL Sabouraud glucose broth (SAB-B; CM 0147 B, Thermo Fisher Scientific GmbH), a 10 µL yeast suspension, McFarland standard 0.3 (Densitometer DEN-1B, Biosan, Riga, Lettland), in a physiologic salt solution, was added. SAB-B conforms to the parameters from the harmonized EP/USP/JP Microbial Limit Testing for the microbial enumeration tests and tests for specified microorganisms. The bottles were placed on magnetic stirring panels (MIXdrive magnetic e motion with Mixcontrol 20, 2mag AG, Munich, Germany) for permanent mixing at 210 rpm. Gas pressure was measured over a 24-h period, taking into

account that feed normally does not normally stay longer in animals' gastrointestinal tract. Line charts of the cumulated gas production were generated with the Ancom Gas Pressure Monitor. Each isolate was tested at least twice.

The results were divided into two groups. A very small pressure increase (<100 mbar) at the beginning was also observed if no further gas was produced thereafter. If less than 100 mbar of cumulated pressure within 24 h was observed, the result was determined as negative. If a yeast was able to produce more than 800 mbar within 24 h, the result was determined as gas production. This value was defined based on the results found, because no yeast produced gas amounts between 100 and 800 mbar.

#### *2.6. Statistics*

Data were statistically analyzed using the SAS® Enterprise Guide® (version 7.1, Fa. SAS Institute Inc. Cary, NC, USA). Pearson's chi-square homogeneity test and Fisher's exact test, used to analyze qualitative analytical characteristics, were applied to check if a yeast was found significantly more with one feed; if one of the identification tests found significantly more reliable results; whether yeasts grew better at a certain temperature; or if yeasts built up a distinct pressure at 37 ◦C within 24 h. Fisher's exact test was used especially for low absolute frequencies.

#### **3. Results**

In total, 95 morphologically different yeast colonies (color, size and surface structure) were isolated from a total of 42 feed samples. In each feed sample, one to four different yeast species were found.

#### *3.1. Identification*

In spite of a different morphology, yeast identification led to the same result for 15 yeast isolates. Yeasts diagnosed twice in the same sample were not considered in the evaluation of the number of yeast species or species of yeast-like organisms that were found in the respective feed sample.

The isolates originated from six genera (*Candida* (76.25%; *n* = 61)*, Geotrichum* (7.5%, *n* = 6)*, Trichosporon* (3.75%, *n* = 3)*, Saprochaete* (2.5%, *n* = 2), *Rhodotorula* (1.25%, *n* = 1), *Pichia* (1.25%, *n* = 1) and non-identified yeasts (7.5%, *n* = 6)). A total of 19 different yeast species were identified.

In Table 1, the most often isolated species are listed. Less frequently isolated (one to three times) were *C. pelliculosa* (*n* = 3, 2× LD + S, 1× LD − S), *C. valida* (*n* = 3; 2× LD − S, 1× LD + S), *Sap. suaveolens* (*n* = 2, both from LD − S), *C. rugosa* (*n* = 2, both from LD − S), *C. kefyr* (*n* = 1, LD − S), *C. variabilis* (*n* = 1, LD − S), *C. spherica* (*n* = 1, LD − S), *T. asahii* (*n* = 1, LD + S), *T. coremiiforme* (*n* = 1, LD − S), *T. laibachii* (*n* = 1, LD − S), *P. manshurica* (*n* = 1, LD − S), *Rhodotorula mucilaginosa* (*n* = 1, LD − S) and *Candida* spp. (*n* = 1, LD − S). Six isolates (5× in LD − S; 1× in LD + S) were not reliably identified.



*C.* = *Candida*; *S.* = *Saccharomyces*; *G.* = *Geotrichum*.

Only *C. lambica* was determined to have a significantly higher incidence in LD + S (*p* < 0.0126). The occurrence in LD − S or LD + S did not significantly change (Figure 1).

−

**Figure 1.** Incidence of yeasts according to their feed origin. An asterisk (\*) indicates a significant difference (\* *p* < 0.05). *C*. = *Candida*; *S*. = *Saccharomyces*.

− − If, despite different colony morphology, two identical yeast species were diagnosed in one feed sample, only one yeast species was further examined. In 33 LD − S, 7/70 yeast colonies with different morphologies were selected, but identified as the same yeast species within the same diet. In LD + S, 8/25 morphologically different yeasts were identified as the same yeast within the same diet. Therefore, 63 yeasts from LD − S and 17 yeast from LD + S were further examined. As a result of this, significantly more different yeast morphologies were observed from the LD + S samples (*p* < 0.0211).

− On the other hand, no differences between the two feeding groups were found regarding the actual (not morphologically) different yeasts. In both diet groups, an average of 1.9 different yeasts per feed sample were diagnosed: 63 different yeasts of 33 LD − S diets, and 17 different yeasts of nine LD + S diets.

#### 3.1.1. Method Comparison

The comparison of methods revealed the most reliable results with MALDI-EX (78.75% reliable species identification), closely followed by ID32C with 75.0% reliable results. These test results did not differ statistically significantly (*p* < 0.6762). Among the reliable results, ID32C provided differentiation of ten isolates (8%), seven isolates with species identification and three isolates were at least identified up to genus level, which MALDI-EX was not able to differentiate. Hence, taking the results of both methods together, 71 (88.75%) of all isolates were identified up to species level. A probable identification or rather identification only at genus level was possible in three (3.75%) isolates. No identification with any of the three diagnostic methods was made for 6/80 (7.5%) yeast isolates. Only few results (37.5%) were provided by MALDI-DS (Table 2). When evaluating the reliable results, both MALDI-EX and ID32C differed significantly (both *p* < 0.001) from the MALDI-DS. No identification with any of the three diagnostic methods was made for 6/80 yeast isolates (7.5%). For better comparison, the results of all three methods were presented together in one table. However, the yeast names (according to their teleomorphic or anamorphic growth) given in the results varied sometimes according to the evaluation software.

 Some yeasts were diagnosed more accurately with one or the other method. Five isolates differentiated with MALDI-EX as *C. humilis* were diagnosed as *C. holmii* in ID 32 C. *C. humilis* was not included in the database used to evaluate the ID 32 C. Both of these species had very similar biochemical reactions. Therefore, identification with MALDI-EX was chosen to be more accurate.

 The two yeasts identified as *C. pararugosa* in MALDI-EX were identified as *C. rugosa* in ID 32 C. Results in MALDI-EX were only 1.88 and 1.79, respectively. Therefore, they had to be named according to ID 32 C, where the results for both isolates revealed very good identification scores, namely, 99.8% and 99.5% for *C. rugosa*. On the other hand, *C. pararugosa* was not included in the database used to evaluate the ID 32 C.



a,b Different superscripts differ significantly in a row. \* Reliable identification: Highly probable species identification and probable species identification; \*\* Reliable identification: Good and very good identification together; \*\*\*: No species identification.

Even with the formic acid extraction method, the slimy and red growing *Rhodotorula* (*R*.) *mucilaginosa* could not be detected with MALDI-TOF; there was no reliable identification, although it was registered in the database.

In the ID 32 C, two isolates were diagnosed as *Cryptococcus* (*C. curvatus and C. laurentii*), which showed no mucus capsule in the Indian ink preparation, but showed hyphal growth, arthroand blastosporogenesis on rice agar under the cover glass (Figure 2), and were identified in MALDI-EX as *T. coremiiforme* and *T. laibachii* (Table 3). On the other hand, two *T. asahii* isolates could be recognized well or very well by both methods.

**Figure 2.** *Trichophyton coremiiforme* on rice agar: arthrospores (**a**) and blastospores (**b**). 400× magnification.

**Table 3.** Isolates differentially diagnosed with MALDI-EX and ID 32C.


Yeasts highlighted in bold represent the selected diagnoses.

Six isolates had a score between 1.79 and 1.98 in the MALDI-EX. Of these isolates, *S. cerevisiae* (score 1.98) and *C. holmii* (score 1.97) had the same result in ID32C (see Table 4), with very good identification. Furthermore, two isolates were identified as *C. pararugosa* in the MALDI-EX (with scores 1.88 and 1.63). *C. pararugosa* was not included in the identification software (APIWEBTM, bioMérieux). The remaining two isolates consisted of *Saprochaete* (*Sap.*) *suaveolens* (score 1.88), which was diagnosed as a *Geotrichum* spp. in ID32C, and *Pichia occidentalis* (score 1.83), diagnosed with 99.7% as *C. krusei* in ID32C. Bearing in mind that *Sap. suaveolens* was formerly diagnosed as *Geotrichum fragrans*, the diagnosis made by MALDI-EX was most likely the one with the currently correct name. The name of the yeast in ID32C was probably out of date; the yeast was still correctly identified. Only one isolate was differently identified by the two methods as *P. occidentalis* (MALDI-TOF) and *C. krusei* (ID32C; Table 4).


**Table 4.** Isolates in MALDI-EX rated as probable genus identification (1.70 and 1.99).

Yeasts in bold represent the selected diagnoses *C*. = *Candida*; *Sap*. = *Saprochaete*; *P*. = *Pichia*; *S*. = *Saccharomyces*.

#### 3.1.2. Biochemical Reactions

In total, all the investigated yeasts were able to assimilate glucose and no yeast grew in the cupule where no substrate was present (cupule F). These reactions were considered as the positive growth control or negative control (no contamination). From the large number of biochemical reactions, two of them will be examined in more detail in the following section, since the ability of the yeasts to metabolize them could be an advantage, especially in maize silage.

#### Metabolization of N-Acetylglucosamine (NAG)

In the ID 32 *C*-Test, 27 yeasts from a total of 63 yeasts in the LD − S samples were able to metabolize *N*-acetylglucosamine and 36 yeasts were not. In LD + S, 11 yeasts were able to build *N*-acetylglucosamine and six yeasts were not. Despite the fact that this is insignificant (*p* = 0.0788), the ability to build NAG was more often seen in yeasts from LD + S.

#### Metabolization of Lactic Acid (LAT)

In the ID 32 *C*-Test, 36 yeasts of a total of 63 yeasts in the LD − S samples were able to metabolize lactic acid and 27 yeasts were not. In LD + S, 12 yeasts were able to metabolize lactic acid and five yeasts were not. These results were insignificant (*p* = 0.4075). Nevertheless, the ability to metabolize lactic acid could be found more often with yeasts that had to stay alive or even grow in maize silage than for yeasts in LD − S.

#### *3.2. Temperature Comparison*

Most yeast isolates (*n* = 47; 58.75%) formed larger colonies at 25 ◦C than at 37 ◦C, among them, 14 isolates (17.5%) did not grow at 37 ◦C at all. These included ten isolates from LD − S (3× *C. holmii,* 1× each for *C. humilis*, *C. lambica*, *T. laibachii*, *C. pelliculosa*, *Geotrichum* spp., *S. cerevisiae* and an isolate not identified) and four from LD + S (3× *C. holmii* and 1× isolate not identified). Among all yeasts, which grew better or only at 25 ◦C, many isolates of *C. holmii, C. humilis* and *C. lambica* were found. Only 23.75% of isolates grew better at 3 ◦C compared to 25 ◦C; this was often the case for *C. krusei* (12/19 isolates) and *S. cerevisiae* (4/5 isolates). Additionally, *C. kefyr* (1/1), *C. holmii* (1/11 isolates) and one isolate that could not be identified (1/6) showed better growth at 37 ◦C. All *S. cerevisiae* isolates were harvested from LD − S samples. Larger colonies at 37 ◦C were formed from nine *C. krusei* isolates

gained from LD − S and three from LD + S, while three isolates from LD − S and one from LS + S formed larger colonies at 25 ◦C; one isolate showed equal colony growth at 25 ◦C or 37 ◦C. Hence, for *C. krusei*, no difference was observed, regardless of which feed it was isolated from. When comparing both feeds, it was noticeable that especially yeasts isolated from LD + S grew poorly at 37 ◦C (see Table 5). Better growth at 37 ◦C than at 25 ◦C for yeasts harvested from LD + S was only seen for three isolates, all of which were *C. krusei.* However, there was no statistically significant difference (*p* < 0.3862) between the two feed sources concerning growth performance (colony size) of the yeasts at either of the temperatures.


**Table 5.** Growth performance of the yeasts at different temperatures depending on feed.

#### *3.3. pH-Value in Liquid Swine Diets*

The pH-values of LD − S (n 25/33) ranged from 3.87–5.78, while the pH-values of LD + S (*n* = 8/9) achieved higher pH-values ranging from pH 4.79 to pH 5.61. Related to the feed origin, 44 yeasts isolated from LD − S were harvested from liquid feed, with an average pH-value of 4.59. Yeasts gained from LD + S were harvested from liquid feed, with an average pH-value of 5.51. *C. krusei* was isolated from liquid swine diets with the lowest (pH 3.87) and highest pH-values (pH 5.78) as well. *C. humilis and C. holmii* were found in diets with lower pH-values (pH 3.9 to pH 5.11), whereas *C. lambica* was isolated once from a diet with a pH-value of 4.45. However, other isolates were harvested from diets with higher pH-values (ranging from pH 4.97 to pH 5.61).

### *3.4. Gas Production*

The results of the duplicate testing of each isolate showed small deviations, possibly caused by small differences in cell counts at the beginning as well as differences in replication time and counts of spores formed by each yeast cell during the 24-h incubation period.

Only two groups were formed: yeasts that produced virtually no gas within 24 h at 37 ◦C and yeasts that produced more than 800 mbar. A further subdivision of the yeasts into groups producing little or a lot of gas was omitted, because too little information was available from the literature as to which quantities could be classified as a lot or little.

No yeast produced gas amounts between 100 mbar and 800 mbar. More yeasts harvested from LD − S produced gas than yeasts that were found in LD + S, but the quantity was not significant (*p* < 0.2216).

Gas production with more than 800 mbar was observed for a total of 13 (40.6%) isolates (Table 6): 10/11 *C. krusei*-isolates, 2/3 *S. cerevisiae*-isolates, 1/1 *C. kefyr* and 1/1 *C. humilis*. *C. kefyr* formed the highest gas pressure, with 10,419 mbar, followed by the *C. krusei* isolates (7134.5, 7073, 6659, 6487, 6383, 6164.5, 4839, 4147.5, 3954.5 and 3940.5 mbar), both isolates of *S. cerevisiae* (1160.5 and 1466.5 mbar) and *C. humilis* (888 mbar). Eleven of these isolates grew better at 37 ◦C than at 25 ◦C within 48 h (see Section 3.2). Nonetheless, two isolates, which also grew better at 37 ◦C, were not able to produce more than 100 mbar gas in 24 h. These two yeasts were one *C. holmii* and one *C. krusei* isolate harvested from LD + S. The latter one produced these high gas quantities only after a 40 h incubation time. On the other hand, one isolate of *C. krusei*, harvested from LD − S, which grew better at 25 ◦C than at 37 ◦C, nevertheless produced 3954.5 mbar gas within 24 h. Only one yeast isolate from LD + S could produce significant quantities of gas under the abovenamed circumstances within 24 h (Table 4).


**Table 6.** Gas production (mbar) at 37 ◦C within 24 h.

Gas production less than 100 mbar was demonstrated in 5/5 *C. lambica*, 4/4 *C. holmii*, 3/3 *Trichosporon* spp., 3/3 *G. silvicola*, 2/2 *C. pelliculosa*, 2/2 *C. rugosa*, 1/1 *Sap. suaveolens*, 1/1 *Candida* spp., 1/1 *C. valida*, 1/1 *P. manshurica,* 1/1 *C. spherica*, 1/3 *S. cerevisiae* and 1/11 *C. krusei.* Yeasts that showed some signs of growth or grew particularly well at 37 ◦C within 48 h showed different reactions. Some isolates needed more than 24 h to produce high amounts of gas (Figure 3). Some isolates did not produce amounts greater than 100 mbar, even within a given 60-h period. In Figure 3 such yeasts are *C. valida* and *C. lambica*. Therefore, their curves in Figure 3 are so close to the x-axis they are hardly visable, just like the curve of the control (sterile SAB-bouillon without yeast isolate).

− − **Figure 3.** Cumulative pressure graph *C*. = *Candida; C. krusei*, LD − S = isolate of *C. krusei* from LD − S; *C. krusei*, LD-+ = isolate of *C. krusei* from LD + S.

The highest correlation between yeast growth at 25 ◦C and 37 ◦C within 48 h (see Section 3.2) and gas formation was found in yeasts that did not grow at 37 ◦C at all. None of these yeasts (8/8) were able to produce gas during the 24-h incubation period at 37 ◦C.

#### **4. Discussion**

Increased numbers of yeasts in liquid feed for pigs has been the subject of some publications in previous years [10–12,16,26–28]. Some publications compared yeasts in liquid feed for pigs with different feed composition taken from different stables or with and without the addition of starter cultures. However, none have yet compared the yeasts in liquid swine diets with and without maize silage, with the identification results derived from two methods.

#### *4.1. Identification*

− As in most other studies, the genus *Candida* (*C.*) was found most often in our research study. *C. krusei* was found most often in LD − S and in all samples as a whole, whereas *C. lambica* was found most often in LD + S. In addition to these two species, another 17 yeast species from six genera were also diagnosed in this study.

#### 4.1.1. *C. krusei*

Overall, the most often found yeast in this study, *C. krusei*, was isolated from feed samples with the lowest and highest pH-values. From the literature, it is known to grow at low pH-values [29], ferment up to a pH-value of 3.6 [30] and can also form biofilms [31]. These properties are likely to be beneficial for yeast persistence in the liquid feed and the feeding system. In part, *C. krusei* is capable of pseudohyphae formation and mostly of growing at 37 ◦C, both characteristics that could contribute to HBS.

*C. krusei* is responsible for about 2% of yeast infections caused by *Candida* species in humans [32]. *Pichia kudriavzevii, Issatchenkia orientalis* and *Candida glycerinogenes* are proven to be the same yeast with collinear genomes 99.6% identical in DNA sequence. Under these names, the yeast is used for industrial-scale production of glycerol and succinate, and is also used to make some fermented foods [32]. The latter use in fermented foods also explains the frequent occurrence in liquid feed for pigs, which also has a low pH value (see Section 3.3).

#### 4.1.2. *C. lambica*

The significantly higher presence of *C. lambica* in LD + S is possibly due to maize silage in the feed but could also be due to the lower storage temperature of the maize silage outdoors during winter [33]. As LD + S samples were gained from the institute's own research projects, it is known that animals did not develop HBS or any other disease and that they ate a lot more with the ad libitum feeding of LD + S in comparison to the previous feed intake with commercial feed (Jörling, 2017) [8]. This was observed, although the yeast content of the feed was temporarily more than 1 × 10<sup>8</sup> cfu/g feed (Jörling, personal observations, results of which have not yet been published). Olstorpe et al. [26] discovered that *Pichia fermentans* (*C. lambica*) was dominant in all their experiments. They assumed that *C. lambica* was able to improve palatability as it has been described to improve the flavor composition during wine- and cheese-making [26]. Presumably, the yeast species within the diet might be more important than the orientation values, which are the same for all yeasts when the hygiene status of the liquid diets is under debate.

#### 4.1.3. Yeasts from Liquid Diets for Pigs

In the present study, mostly *C. krusei*, *C. holmii*, *C. lambica*, *S. cerevisiae*, *C. humilis* and *Geotrichum* spp. were identified (see Section 3.1, Table 1). Together, they accounted for 66.5% of all yeast isolates. Other species could not be identified (7.5%, Table 1) or were only detected in lower proportions (26%; see Section 3.1).

Middelhoven et al. [33] observed, in whole-crop maize ensiled for two weeks, similar yeast species compared to those found in LD + S. They predominately found *C. holmii, C. lambica, C. milleri* (current name: *C. humilis*), *Hansenula anomala* (current name: *Wickerhamomyces anomalis,* anamorph: *C. pelliculosa*) and *Saccharomyces dairensis* [33], whereby only the latter yeast did not occur in our study. The comparability of the yeast species in both studies could indicate that it is not so much the storage over winter but rather the substrate that influences the yeast occurrence.

The biochemical profiles of both yeasts are very similar and therefore sometimes misidentified [34]. Both yeasts are able to assimilate mostly glucose, lactose, glycerol, inositol and N-acetylglucosamine, while xylose is only metabolized from *C. lambica*. On the other hand, the next frequently identified yeasts, *C. holmii, C. humilis* and *S. cerevisiae*, cannot perform this metabolic function, with the exception of glucose. Instead, they metabolize galactose and raffinose, and, in part, trehalose (*C. holmii and C. humilis*), sucrose (*C. holmii* and partly also *S. cerevisiae*) as well as maltose (*S. cerevisiae*). This could suggest that the assimilative capacities of the yeasts are not essential for their presence or absence in different liquid feeds for pigs.

Likewise, many different yeasts were identified in studies on yeast determination from liquid feed samples for pigs, and different species dominated in the different feed samples [6,26–28].

Similarities to the isolated yeasts in the present study were observed in the studies by Olstorpe et al. [26], who examined liquid feeds based on a cereal grain mix and wet wheat distiller's grain with and without starter cultures. Without starter cultures, they observed *P. fermentans* (*C. lambica*), *C. pararugosa*, *C. rugosa*, *P. galeiformis* (current name: *P. mandshurica*), *T. asahii, Issatchenkia orientalis* (*C. krusei*, *P. kurdriavzevii*), *C. ethanolica* and *C. vini*. These yeasts were found in the present study as well, except the last three mentioned ones.

Olstorpe et al. [26] isolated *C. kefyr* from wheat-based liquid feed as the dominant yeast, which was isolated only once in the present study. This previous publication also found *C. krusei*, *C. pelliculosa* and *Pichia membranaefaciens* (*C. valida*). Plumed-Ferrer and Wright [35] most frequently observed *K. exigua* (*C. humilis*), *Debaromyces hansenii* and *Pichia derserticola* in fresh batches of liquid feed, of which only *K. exigua* was often observed in the present study. On the other hand, in this previous study, other yeasts were also isolated, such as *Pichia kurdriavzevii* (*C. krusei*), *S. exiguous* (*C. holmii*), *Pichia membranaefaciens* (*C. valida*) and *Wickerhamomyces anomalus* (*C. pelliculosa*), which were identified in the present study, too.

Significantly more morphologically different yeasts were observed in LD + S (see Section 3.1), which had lower pH-values than conventional feed and were stored outdoors during winter, which, as a consequence, were exposed to changing temperatures. These different colonial morphologies could be a result of changing environmental conditions, as explained in previous publications [36,37].

The genera *Geotrichum* and *Trichosporon* are classified as yeast or yeast-like organisms, but *Geotrichum* was formerly classified as a mold [38–40]. The colony morphology is very similar to other yeasts and therefore was described in many previous studies concerning yeasts in liquid swine diets [25,27,28], so that comparability with other studies is possible. *Saprochaete suaveolens*, formerly classified as *Geotrichum fragrans*, is also classified as a yeast or yeast-like organism (mycobank.org [41]). Hereafter, for the sake of simplicity, all genera are referred to as yeasts, even if the term yeast or yeast-like organism would be more accurate.

#### 4.1.4. Method Comparison

The present study compared different methods for identifying yeasts to find the best method for the chosen substrate and the yeasts contained in it. From previous studies [6,10,26,28,42], it was known that many tests for identifying yeasts from the environment produce fewer results than those from clinical material [43–45]. Additionally, different databases on which different test procedures are based also influence the obtained results [45,46].

For clinical samples consisting mainly of *Candida* spp., the method of MALDI-TOF outperformed the diagnosis capacities of the phenotypic tests by reducing the delay in results and improving the reliable identification rate at species level [43]. On the other hand, this method requires significantly higher acquisition costs for the equipment. Therefore, this method was compared with the ID 32 C test, which has virtually no purchase costs.

#### ID32C

In our study, 5.2% less reliable results were observed with the ID 32 C test in comparison to MALDI-EX. Nevertheless, in individual cases, correct identification could only be made with this simple biochemical method (see Table 4). *C. rugosa* was twice identified with more than 99.5% accuracy as "very good identification", while MALDI-EX identified these two yeast isolates as *C. pararugosa*. The latter was not included in the ID32C-database. Considering the fact that several authors [44,45,47] proposed a lower identification score for the yeast identification with MALDI (see below), perhaps the MALDI results are the correct ones. In the case of *Geotrichum* spp. and *Saprochaete suaveolens*, the situation is similar. *Saprochaete suaveolens* was not included in the database of the ID32C test. Comparable to the finding in the present study, namely that *Rhodotorula* was better identified with ID32C, Olstorpe et al. [6] reported that, with the applied PCR fingerprinting, two *Rhodotorula glutinis* isolates were incorrectly classified as *Cryptococcus satoi* or *Pichia membranaefaciens*, but correctly identified with ID32C.

The ID32C test can be easily performed in any laboratory and does not require an expensive device. In addition to species identification, the biochemical test has the advantage of showing which enzymes can be produced by the respective yeast isolate. This, in turn, could allow or exclude opportunities for identifying which feed components could be metabolized by the yeast.

#### Selected Biochemical Reactions of the ID32C-Test

**Metabolization of** *N***-Acetylglucosamine (NAG):** More yeasts from LD + S, even if not significant, were able to metabolize the amino sugar NAG. This is the monomeric constituent of chitin, which is one of the most abundant renewable resources found in nature [48]. The uptake of NAG into the yeast cell, its metabolites in the cell and conversion to cell wall formation have already been described for various yeasts [48]. The cell wall reinforced by NAG (chitin) offers protection against low pH-values in the environment [22]. Although the pH-values in the LD − S were not significantly lower than those of LD + S, the prolonged period of survival in silage (see below) may have led to the ability of yeasts to metabolize NAG.

**Metabolization of Lactic Acid (LAT):** More yeasts from LD + S were able to metabolize lactic acid. As this finding is not statistically significant, the ability to metabolize lactic acid obviously is not a prerequisite for yeasts in LD + S. Lactic acid bacteria are the predominant group of bacteria found in maize silages, and are able to multiply in liquid feed, lactic acid being a main product of their metabolism [49]. Maize silage used for LD + S in the present study was kept outdoors during winter and early spring until it was fed to the animals in late spring and early summer. Being able to use a substrate present in the environment is presumed to be an advantage for yeasts [50,51], which have to survive in these conditions for a long time.

**MALDI-DS:** The less time-consuming and less expensive MALDI-DS reduced the identification rate significantly (*p* < 0.001) by more than half compared to MALDI-EX (37.5% vs. 78.75% reliable identification). Thus, the use of this method is clearly limited, at least if different yeasts are to be identified from environmental samples. In contrast to bacteria, yeasts possess a thick and chitinous cell wall, which might lead to the difficulties encountered with the MALDI-DS method [52].

**MALDI-EX:** In our study, 78.75% of the 80 different yeasts could be identified by MALDI-EX and 75.0% by the ID 32 C. While 11 isolates were not identified at all, six isolates achieved only probable results at the genus level (see Table 4). An incorrect diagnosis was observed only once, mistakenly identifying *P. occidentalis* instead of *C. krusei* (see Section 3.1.1).

A comparison of identification of 96 foodborne yeasts with MALDI-TOF and two conventional tests, of which one was ID 32 C, was made by Pavlovic et al. 2014 [53]. In their study, more yeast isolates could be identified with MALDI-TOF than with the ID32C test, too.

The identification rate of the different methods in the present study was comparable to those of others in which yeasts were isolated from the environment rather than from clinical material [6,54–56]. As already shown by these and other authors [1,57], none of the methods were capable of reliably detecting all yeast isolates from the liquid feed. Many authors attribute these differentiation failures to the background of the ID32C, MALDI-TOF and other commercially available tests, as these were developed and established for clinically relevant yeasts in human beings and not for yeasts in animal feed [1,45,57]. None of the available methods can be considered as the golden standard for the differentiation of yeasts from liquid feeds. With respect to the low examination costs, low workload, fast availability of results and available databases, which means the highest rate of correct identification, the different methods exhibit advantages and disadvantages.

Although MALDI-EX was the best method for gaining the most reliable identification results in this study, it has to be considered that results from this method are only as good as the underlying database [46]. Vlek et al. [46] identified 61.5% of their yeasts from human patients using the Bruker Daltonic database (BDAL), but improved their identification rate up to 86.8% by adding their database with the in-house database from the Centraalbureau voor Schimmelcultures (Central Bureau for Fungal Cultures) (BDAL + CBS in-house). This allows the assumption to be made that even more yeasts will be identified with this method in the future, if correspondingly relevant data continue to be added, especially for the non-clinical yeasts found in the surroundings. An improvement in the identification results of 845 environmental yeasts by one third was also described by Augustini et al. [45] after developing a supplementary database.

Besides the databases as reason for missing reliable yeast identification, Augustini et al. [45] stated that identification scores <2.00 are not able to unequivocally affirm that the identification at species level is unreliable. They cited studies that showed identification results under 2.00, but with correct identifications. This observation was underlined in the studies by Tan et al. [44]. Repeating MALDI-TOF attempts in 10.2% of the yeast isolates, which had indicated spectral scores as being unacceptable on the first attempt (scores < 2.00), resulted in acceptable scores (>2.00). Most of these achieved a correct identification on the first attempt [44]. The authors concluded that lowering the identification score from <2.00 to <1.70 could reduce the repetition rate [44]. With a cut-off of <1.70, Lee et al. [50] also improved the identification rate of their 284 pathogenic yeasts from clinical samples compared to the required cut-off value of >2.00 [49]. When comparing the results of two different MALDI-TOF systems (Biotyper from Bruker and ASTA MALDI-TOF MS), Lee at al. [50] found that only 39.5% of the isolates with confirmed identification with molecular sequencing met the cut-off score in both systems. The majority of the isolates (58.6%) ranged between 1.70 and 2.00 when using the Bruker Biotyper and scores > 140 using ASTA MALDI-TOF.

Lee et al. [52] performed a formic acid extraction with a shorter protocol. Most of the yeasts obtained from samples of clinically infected humans were identified correctly, but the method failed to identify the slimy *Cryptococcus* spp. Considering the fact that in our study no *Cryptococcus* spp. were found, possibly this shorter, easier and inexpensive method could have provided as good results as MALDI-EX. On the other hand, different *Cryptococcus* spp. were isolated from liquid swine diets in studies by Olstorpe et al. [6]. Therefore, MALDI-EX seemed to be the best method to reliably identify as many yeasts species as possible.

Extending databases, lowering the identification score for yeasts as well as shorter protocols could improve the ratio of reliable results of environmental yeasts with MALDI-EX in the future, so that the results of this method could be highlighted even more.

Various molecular biological methods described in the literature were not included in this study, although previous authors achieved good results [27]. Gori et al. [27] had difficulties in separating the two most commonly occurring yeasts in their study with 26S rRNA sequencing: *C. humilis* (formerly named *C. milleri*; 58.4%) and *C. holmii* (*Kasachstania exigua*; 17.5%), together accounting for 75.9% of all results (*n* = 766 yeasts). They distinguished the two yeasts biochemically according to their sucrose and raffinose metabolism [27]. In the present study, *C. humilis* and *C. holmii* accounted together for 20% of all results (*n* = 16). In retrospect, it can be assumed that the 26S rRNA method would not have been advantageous in these cases.

#### *4.2. Temperature*

In the present study, clearly more than half of the yeasts (52.9%) grew better at 25 ◦C than at 37 ◦C or did not grow at 37 ◦C at all (23.5%). Those yeasts that did not grow at 37 ◦C at all will presumably not grow in the intestines of pigs, where the internal body temperature normally still exceeds 37 ◦C.

Considering only yeasts isolated from LD + S, there are even more isolates that prefer cooler temperatures (see Table 5). An explanation for these yeasts preferring lower temperatures than yeasts from LD − S could be the chosen time of sampling of LD + S in late spring and early summer in the two projects, when the liquid diets were composed. After harvesting the maize plants and making silage in the fall in the respective projects, this was stored outdoors during winter, where yeasts had to cope with low temperatures. Thus, yeasts may have adapted to these temperatures or died. Storing feed materials or liquid diets at cool temperatures possibly reduces the yeast species, which prefer 37 ◦C, and as a result have little or no impact on gut health.

The present results could also indicate an adaptation of the yeasts to their feed origin and storage temperature. These results were obtained directly after the cultivation of the yeasts from the respective feed (see Section 2.3). Therefore, yeasts had little or no opportunity to adapt to the new temperatures. This is in the broadest sense comparable with the climatic conditions during the long period between the fall and spring. On the other hand, the possibility to adapt would exist at warmer outside temperatures and in case of the pre-fermentation of the liquid feed (24 h, 38 ◦C), as is sometimes practiced, especially with controlled fermentation [58]. Suutari et al. [59] reported morphological changes in some yeasts that had to adapt to cooler or very warm temperatures in a bouillon. The investigations in this previous study on growth performance at different temperatures was made on agar plates. The possible easier adaptation to new temperature conditions in a bouillon could also be an explanation for the observations that some yeasts only produce gas at 37 ◦C after a longer period of time (see Section 3.4, *C. humilis* in Figure 3).

Margesin et al. [60] isolated yeasts and bacteria from cold-adapted habitats and classified 60% of the yeasts but only 8% of bacteria to be true psychrophils, which showed no growth above 20 ◦C, indicating that the remaining microorganisms are able to adapt to warmer temperatures. Yeasts that do not grow or grow very poorly at 37 ◦C are thought to have little or no effect on gut health [20]. As a result, the lack of or partly low clinical symptoms on farms with a high yeast load in the feed are explicable. Correspondingly, yeasts that did not grow at all or worse at 37 ◦C than at 30 ◦C were also found on yeasts obtained from swab samples from milking machines [55]. A large majority of them could not be recovered from the milk collected with these milking machines.

In both groups (LD − S and LD + S), 1.9 different yeasts were identified. On the other hand, significantly more different colonial morphologies of the yeasts were found in LD +S, possibly indicating that temperature could have an influence on morphology, as was also observed by Nadeem et al. [37].

#### *4.3. pH-Value*

In the present study, the LD + S had on average slightly higher pH-values than LD − S and they contained significantly more *C. lambica.* Whether this connection is accidental or related to the higher pH-value can only be suspected due to the small number of farms of origin. Lack of growth at 37 ◦C [34], a good smell/taste [26] but no described ability of biofilm formation, as found by the Olostorpe et al. [26], could mean that this yeast is expected to be less harmful as a feed contaminant and for gut health than other yeasts. On the other hand, some yeasts are known to adapt to pH-values, to temperature and to different media [37], so that the safety of *C. lambica* in liquid swine diets still needs to be tested.

For fungi as well as bacteria, one of the most important environmental conditions is ambient pH. Changes in external pH result in phenotypic, metabolic and physical changes of the microorganisms [22]. The low pH-values in liquid feeds, especially fermented ones or those containing silage compared to normal feed for pigs, in general favor yeasts. This is due to the fact that at pH-values < 5.0, many bacteria are not able to stay alive or to grow as fast as they do at higher pH-values [61]. Molds depend on oxygen, but yeasts are able to grow at low pH-values with and without oxygen [3]. Some yeasts are known to be able to adapt to low pH-values in their surroundings by forming a thicker cell wall with chitin (*N*-acetyl glucuronidase) [22]: the high buffering capacity in the cytosol, high H+-ATP-ase and/or high endogenous energy reserves of *C. krusei* [29]. Therefore, fermented liquid feeds, especially after controlled fermentation, always poses a certain risk of increased yeast content.

#### *4.4. Gas Production*

Quantitatively comparing gas production of different yeasts under standardized conditions with Ancom RF Gas Production System was, to the best of our knowledge, performed for the first time. Investigations in a bouillon, produced in accordance with European and US Pharmacopoeia guidelines, allows for a comparison of gas-producing yeasts irrespective of feed or water. The SAB-bouillon provides ideal conditions for yeasts and contains high amounts of glucose (20 g/L). However, the total

gas quantities measured do not describe quantities that would also be produced in the feed or in the animal, since the competing flora is always different, and feed is not composed like a bouillon or an agar for yeasts.

Different generation times, sizes and numbers of buds make it difficult to precisely calculate the yeast quantity with density determination or even with quantitative cell counting. Hence, the amounts of gas production were not precisely determined but categorized to two major groups, as described above. Additionally, not the exact yeast numbers per milliliter were determined but only the density by means of the McFarland standard. Exemplarily, for some samples with the density of McFarland 0.3, the yeasts were counted, resulting in 1–4 × 10<sup>5</sup> cfu yeasts per mL. Thus, these yeast counts are just about acceptable regarding the requirements in liquid feed according to Kamphues et al. [17].

In the present study, slightly increasing gas pressures were also measured for yeasts that did not produce gas at the beginning of the experiment. This could be explained by the rising room temperature during processing to the 37 ◦C in the incubator.

The gas formation capacity of the yeasts differed very clearly between 888 mbar and 10,419 mbar. *C. kefyr* formed over ten times more gas than one of the *S. cerevisiae* isolates. Only one yeast isolate from LD + S was able to produce higher amounts of gas at 37 ◦C within 24 h. This was partly caused by its preference for cooler temperatures, as described above. The reason for the differing amounts of gas production of yeasts may to some extent be seen in the lack of oxygen produced in the Ancom Gas Production System, which is also found in the pig's colon. Some yeasts like *C. sphaerica, C. variabilis, C. kefyr, C. lambica, C. krusei, S. cerevisiae* and *C. pelliculosa* are known to metabolize glucose under anaerobic conditions; variable metabolization is expected from other yeasts like *C. valida, G. candidum* and *G. capitatum*, while *C. rugosa* and *Rhodotorula* spp. are mostly not capable of fermentation [61]. The latter cannot be expected to produce gas amounts under the conditions available in the present study as well as those found in the gastrointestinal tract of pigs. As such, they cannot be expected to cause a disease such as HBS. Comparing the growth of a yeast from liquid feed for pigs at 37 ◦C and 25 ◦C can give a good indication of whether a yeast is likely to cause HBS. However, it is not possible to make an accurate prediction because yeasts are partially capable of adapting to temperatures and some yeasts hardly ferment under anaerobic conditions despite growth at 37 ◦C. On the other hand, no high gas production within 24 h was observed in the present study when a yeast isolate did not grow at 37 ◦C. Presumably, those yeasts are not supposed to cause HBS. A test of growth at 37 ◦C would be easy to perform in every laboratory and could give a hint at whether a yeast would be able to grow in a pig's alimentary tract. Further studies will be needed to clarify which amount of gas production can generally be called high or low. Apart from this, it has to be considered that a yeast, even if it is not able to form a biofilm itself, may colonize the biofilm of the lines of the feeding system. Such yeasts could potentially be capable of adapting to warmer temperatures, especially in the summer months.

#### *4.5. Summary*

In several studies of liquid feed, samples for pigs' yeasts were identified, which were also found in the present study. The most commonly detected yeast in our study was *C. krusei*. This is the first study of liquid feed with and without maize silage. In liquid feed with maize silage (LD + S), significantly more *C. lambica* was found.

MALDI-EX provided the most reliable results (78.75%), but the ID 32 *C*-test, easy to perform in every laboratory, was sufficient for confirming 75.0% of the identified yeasts. Both tests together identified 88.75% of the yeasts because some yeasts were only reliably identified with one or the other test. The quicker MALDI-DS-method provided only 37.5% reliable results, this being significantly less than the other two methods. Thus, a formic acid/acetonitrile extraction (MALDI-EX) before analysis should be preferred.

Clearly more than half of all yeast isolates grew better at 25 ◦C than at 37 ◦C. Fourteen isolates showed no growth at all at 37 ◦C. Gas amounts produced by the different yeast isolates differed more than tenfold within a 24-h incubation period at 37 ◦C in SAB-bouillon measured with the Ancom

Gas Production System. Most of the tested *C. krusei* and *S. cerevisiae* but none of the tested *C. holmii*, *Trichosporon* spp., *G. silvicola* and *C. pelliculosa* were able to produce gas. While only one yeast from LD + S was able to produce gas within 24 h, more yeasts (40.6%) from LD − S were able to do so. None of the yeasts that did not grow on the SAB-agar at 37 ◦C were able to produce high amounts of gas within a 24-h incubation period at 37 ◦C in the bouillon, presuming that those yeasts could only slightly affect the animals' health.

Due to the fact that the majority of *C. krusei* isolates were able to grow at 37 ◦C, produce high amounts of gas, grow in low pH conditions and form biofilms, as is known from the literature, this yeast species seems to be predestinated to grow in liquid diets and to remain in a biofilm in the pipelines serving the liquid diet. Therefore, special interest should be given to this yeast species. The evaluation of yeast levels in liquid feed for pigs has so far only been determined on the basis of the number of yeasts per gram feed. Laboratory values alone could possibly incorrectly estimate the influence of yeasts on the health of the animals as either being too low or too high. Additional investigations are needed to further characterize the effect of each yeast species on pig health. Moreover, investigating the effect of having the storage temperature of the feed significantly below body temperature could be interesting.

**Author Contributions:** Conceptualization, B.K.; methodology, B.K.; software, B.K. and U.S.; validation, B.K., H.K., U.S. and C.V.; formal analysis, B.K. and C.V.; investigation, H.K. and B.K.; resources, J.K. and C.V.; writing—original draft preparation, B.K.; writing—review and editing, B.K., C.V. and U.S.; visualization, B.K. and C.V.; supervision, C.V.; project administration, B.K.; funding acquisition, J.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This publication was supported by the German Research Foundation (Deutsche Forschungsgemeinschaft) and the University of Veterinary Medicine Hannover, Foundation, Hannover, Germany within the funding program Open Access Publishing.

**Acknowledgments:** We would like to thank Frances Sherwood-Brock for proof-reading the manuscript to ensure correct English.

**Conflicts of Interest:** The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 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*

**Alexandros Mavrommatis <sup>1</sup> , Christina Mitsiopoulou <sup>1</sup> , Christos Christodoulou <sup>1</sup> , Dimitris Karabinas <sup>1</sup> , Valentin Nenov <sup>2</sup> , George Zervas <sup>1</sup> and Eleni Tsiplakou 1,\***


Received: 10 November 2020; Accepted: 1 December 2020; Published: 3 December 2020

**Abstract:** This study evaluated the dietary administration of *Saccharomyces cerevisiae* live yeast on milk performance and composition, oxidative status of both blood plasma and milk, and gene expression related to the immune system of lactating ewes during the peripartum period. Chios ewes were fed either a basal diet (BD) (Control, *n* = 51) or the BD supplemented with 2 g of a live yeast product/animal (ActiSaf, *n* = 53) from 6 weeks prepartum to 6 weeks postpartum. Fatty acid profile, oxidative, and immune status were assessed in eight ewes per treatment at 3 and 6 weeks postpartum. The β-hydroxybutyric acid concentration in blood of ActiSaf fed ewes was significantly lower in both pre- and postpartum periods. A numerical increase was found for the milk yield, fat 6% corrected milk (Fat corrected milk (FCM6%)), and energy corrected milk yield (ECM) in ActiSaf fed ewes, while daily milk fat production tended to increase. The proportions of C15:0, C16:1, C18:2n6t, and C18:3n3 fatty acids were increased in milk of ActiSaf fed ewes, while C18:0 was decreased. Glutathione reductase in blood plasma was increased (*p* = 0.004) in ActiSaf fed ewes, while total antioxidant capacity measured by 2,2′ -Azino-bis (3-ethylbenzthiazoline-6-sulfonic acid) (ABTS) method was decreased (*p* < 0.001). Higher ABTS values were found in the milk of the treated group. The relative transcript levels of *CCL5*, *CXCL16,* and *IL8* were suppressed, while that of *IL1B* tended to decrease (*p* = 0.087) in monocytes of ActiSaf fed ewes. In conclusion, the dietary supplementation of ewes with *S. cerevisiae*, improved the energy utilization and tended to enhance milk performance with simultaneous suppression on mRNA levels of pro-inflammatory genes during the peripartum period.

**Keywords:** *Saccharomyces cerevisiae*; livestock; ewes; energy; antioxidant; cytokines

#### **1. Introduction**

Both meat and dairy products consumption are expected to increase in 2050 by 73 and 58%, respectively, compared to their 2010 levels [1,2], due to the rapid population growth rate. Ruminants' milk (67%) and meat (33%) cover 51% of proteins derived from the livestock sector and have a dominant role in food security, which is linked to how efficient animals utilize feed. Ruminants' feed efficiency depends upon the microbes residing within the rumen that ferment and transform feeds into volatile fatty acids (VFAs), proteins, and vitamins which are exploited by the host [3]. This multikingdom ecosystem's efficiency is dependent on various factors, the most prominent being that of diet. The improvement of the rumen microbiome habitat through the advancement

of feed efficiency technologies entails a fundamental stepping stone in the overall improvement of livestock systems sustainability and food security concerns.

In intensive farming systems, high genetic merit animals require higher amounts of concentrate to fulfil their energy and nutrient demands, resulting in metabolic imbalances in rumen function and their microbiome governance. Probiotic yeasts are currently popular and widely used in ruminant feeding systems, especially since some of them have been officially authorized as feed additives in Europe [4]. The main purpose for using such additives in ruminant diets is to prevent rumen flora disorders and disturbances [5]. Dietary supplementation with live yeast (LY), *Saccharomyces cerevisiae*, improves rumen function through several modes of action [6]. This improvement is related to the oxygen scavenging properties of yeast in rumen (anaerobiosis mechanism), which upgrades bacterial viability and therefore the animal production [6]. Amongst the favorable bacteria are cellulolytics, which through the increase in their activity enhance fiber digestion. Moreover, LY can also stabilize the ruminal pH [7], not only after feeding, but also during the peripartum period where animals often find themselves in a negative energy balance and are further sensitive to metabolic diseases. It has been proven that even a low-grade energy deficiency weakens the animals' antioxidant system, which fails to neutralize the formation of Reactive Oxygen Species (ROS) and triggers the pro-inflammatory response [8,9].

By improving ruminants' feed utilization and both energy and nutrient availability during the peripartum period, not only can milk performance and chemical composition be enhanced, but furthermore a downregulation in the immunostimulation response can be achieved through the limitation of lipomobilization metabolites [10,11]. Although LY supplementation in ruminants' diets is a well-established nutritional strategy, previous works have only focused on district parameters instead of a holistic approach. Specifically, except for milk performance [7,10–14], scarce information has been linked to the potential improvement of energy balance and oxidative status and therefore to the immune response under the influence of dietary yeasts inclusion in ruminants.

Taking into account the aforementioned information, the objective of this work was to evaluate the effect of LY *S. cerevisiae* (CNCM I-4407, 10<sup>10</sup> CFU/g, ActiSaf; Phileo Lesaffre Animal Care, France) in dairy sheep during the transition and early lactation period (6 weeks prepartum and 6 weeks postpartum) on milk performance and composition, antioxidant status (determined by Glutathione transferase (GST), Glutathione reductase (GR), Superoxide dismutase (SOD), Glutathione peroxidase (GSH-Px), Catalase (CAT) and Lactoperoxidase (LPO) activities, antioxidant capacity with 2,2′ -Azino-bis (3-ethylbenzthiazoline-6-sulfonic acid) (ABTS) and Ferric reducing ability of plasma (FRAP) methods and oxidative stress indicators such as Malondialdehyde (MDA) and protein carbonyls (PCs)) on both milk and blood plasma and key-gene expression (*CCL5, CXCL16, INFG, IL1B, IL2, IL6, IL8, IL10, TNF, NFKB*) in monocytes and neutrophils which are associated with cytokine production.

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

#### *2.1. Location and Environmental Conditions*

The experiment was conducted from November 2019 to March 2020 on a commercial dairy sheep farm in the region of Chiliomodi in Korithia, Greece. This region has a typical Mediterranean climate with hot dry summers and relatively mild wet winters. During the experimental period, the mean temperatures in November, December, January, February, and March were 12.2, 8, 9.3, 11.1 and 13.1 ◦C, respectively. The selected farm represents the typical intensive dairy sheep production system of Greece.

#### *2.2. Animals and Diets*

Animals' housing, management, handling, and care complied with the latest European Union Directive on the protection of animals used for scientific purposes [15], while taking into account an extended experimental design report, the Bioethical Committee of Faculty of Animal Science (currently known as the Agricultural University of Athens Ethical Committee in Research; FEK 38/A/2-3-2018, eide AUA) approved the experimental protocol. One hundred and twenty (120), 1- to 3-year-old dairy ewes (*Ovis aries*), of pure Chios breed, were physically selected from a flock of six hundred. At approximately 6 weeks before parturition, the ewes were divided into two homogenous groups based on their body weight (BW), number of parturition, and the milk yield from the previous year only for the case of multiparous ewes (2.1 ± 0.68 kg). Both groups had the same number of primi- (*n* = 20) and multiparous (*n* = 40) animals. More specifically, the ewes mean BW in the Control group (*n* = 60) was 61.5 ± 10.70 (SD) kg while in the ActiSaf group (*n* = 60) was 61.5 ± 11.02 (SD) kg. The Control group was fed a basal diet comprising of concentrate mix, alfalfa hay, and oat hay, while the ActiSaf group consumed the same basal diet supplemented with 2 g of *S. cerevisiae* LY/day/ewe (CNCM I-4407, 10<sup>10</sup> CFU/g, ActiSaf; Phileo Lesaffre Animal Care, France) (Table 1). The animals were housed in two pens based on the dietary treatment. Both diets were isonitrogenous and isocaloric and were designed to meet ewes' requirements in the transition period and early lactation according to the flock fat (6%) corrected milk yield [16,17]. The animals were fed on a group basis while forages were offered separately from the concentrate in three equal portions after milking. Diet selectivity did not occur, no refusals of forages and/or concentrate were observed, and all animals had free access to fresh water. The experimental procedure lasted 6 weeks started from each ewes' parturition. After this, each ewe was returned to the commercial farm flock and the experiment ended when the final ewe had completed its 6th week on lactation. Since milk performance was recorded at the same time points, lactation stage had no effect on milk performance. Control ewes (*n* = 60) gave birth to 141 lambs (prolificacy = 2.35; 69 females and 72 males) while those of the ActiSaf (*n* = 60) gave birth to 142 (prolificacy = 2.36; 65 females and 77 males). In addition, since the experimental trial took place on farm-scale conditions, few ewes were unable to be exploited for data curation due to abortions (4), mastitis (10) or dystocia (2), hence the final number of subjects was re-adjusted to 51 and 53 for the Control and ActiSaf groups, respectively.


**Table 1.** Concentrates composition (g/kg), diet intake (g), daily nutrients intake (g/ewe), and feeds chemical composition and fatty acid profile (%).


**Table 1.** *Cont.*

NDF = Neutral detergent fiber; ADF = Acid detergent fiber.

#### *2.3. Feed Samples Analyses*

Samples of the alfalfa hay, oat hay, and concentrate were analyzed for organic matter (OM; Official Method 7.009), dry matter (DM; Official Method 7.007), and crude protein (CP; Official Method 7.016) according to the Association of Official Analytical Chemists (1984) using a Kjeldahl Distillation System (FOSS Kjeltec 8400, Demark). Neutral detergent fiber (NDF) and acid detergent fiber (ADF) expressed exclusive of residual ash according to the method of Van Soest using an ANKOM 2000 Fiber Analyzer (USA) as described by Tsiplakou et al. [18] (Table 1).

#### *2.4. Milk Samples Collection*

The sheep were milked three times per day (at 0700, 1300 and 2000 h) with a milking machine equipped with a digital milk meter and an electronic identification system (Sylco, Greece); thus, milk yield was recorded daily, and software (Sylco, Greece) was set to provide weekly averages. Milk samples were collected from each ewe weekly (at 7, 14, 21, 28, 35 and 42 days from parturition) for a 6-week period with sampling bottles (Sylco, Greece) of 200 mL appropriately for the milking parlor, to receive a representative sample of the milked quantity. Each of the milk samples from the mix of three subsamples was derived from each milking time (at 0700, 1300 and 2000 h) by taking 5% of the milked quantity.

#### *2.5. Milk Chemical Composition*

The milk samples were analyzed for fat, protein, lactose, total solids, and total solids no-fat by IR spectrometry (MilkoScan 120; FOSS, Hillerød, Demark) after proper calibration according to the methods of Gerber [19] and Kjeldahl [20].

#### *2.6. Blood Metabolic Biomarker (B-HBA) Determination*

Four weeks before the expected parturition, blood B-HBA was individually determined (before the morning feeding, 0700 h) once every three days until the lambing to ensure that 15 days before parturition a measurement would be recorded (Table S1). Two weeks postpartum, the sample collection for B-HBA was repeated. Blood ketone concentrations were measured using an electrochemical capillary blood monitoring device (FreeStyle Precision Neo, Abbott Laboratories Hellas S.A) with the corresponding individual foil-wrapped test strips for B-HBA. This method of B-HBA determination possesses 98.4% accuracy for the prediction of both toxemias' pregnancy and ketosis in Chios ewes [21]. After the insertion of a test strip into the device, a drop of blood was applied to the assigned spot,

and the B-HBA concentration was recorded. Data were interpreted using 208 determinations in the two aforementioned sampling time points.

#### *2.7. Antioxidant Status, Immune Response, and Milk Fatty Acid Profile*

Eight (*n* = 8) ewes of each group with comparable weights (Control: 60.2 ± 5.11 kg; ActiSaf: 60.3 ± 4.88 kg), ages (Control: 1.84 ± 0.16 kg; ActiSaf: 1.85 ± 0.18 kg), milk performance (Fat corrected milk 6% (FCM6%) data used up 14th day in milk (DIM), Control: 2.3 ± 0.15 kg; ActiSaf: 2.3 ± 0.21 kg), prolificacy (Control: 2; ActiSaf: 2), and same lactation stage (up to 3 days deviation between animals) were selected for determining the antioxidant status of both milk and blood, for immune system gene response, and for milk fatty acid profile. Milk samples were collected (as mentioned above) in the 3rd and 6th week postpartum and stored at −80 ◦C. Blood samples were also collected, before the morning feeding (0700 h), at the same time points in heparin contained tubes for cell extraction and plasma isolation.

#### *2.8. Enzyme Assays, Oxidative Stress Biomarkers, and Total Antioxidant Capacity*

The enzyme activities, oxidative stress biomarkers, and the total antioxidant capacity were measured spectrophotometrically (Helios alpha, UNICAM, Cambridge, UK) as previously described by Tsiplakou et al. [22]. Briefly, Glutathione transferase (GST) activity in blood plasma was measured according to the method described by Labrou et al. [23] by measuring the conjunction of reduced glutathione to 1-chloro-2,4-dinitrobenzene at 340 nm. Catalase (CAT) activity in blood plasma and milk were assessed using a continuous spectrophotometric rate for the determination of H2O<sup>2</sup> at 520 nm, according to the Sigma-Aldrich Catalase Assay Kit (CAT100). Glutathione peroxidase (GSH-Px) activity in blood plasma was measured according to the method of Paglia and Valentine [24] at 340 nm. Glutathione reductase (GR) activities in both blood plasma and milk were measured according to the method of Mavis and Stellwagen [25] by measuring the reduction in oxidized glutathione at 340 nm. Superoxide dismutase (SOD) activities in both blood plasma and milk were assayed using the method of McCord and Fridovich [26] by measuring the inhibition of cytochrome c oxidation at 550 nm. Lactoperoxidase (LPO) activity in milk was performed according to the methods of Keesey [27] by measuring the oxidation of ABTS present in hydrogen peroxide at 340 nm. Malondialdehyde (MDA) was determined according to the method of Nielsen et al. [28] with some modifications. More specifically, 100 µL blood plasma was added to 700 µL ortho-phosphoric acid (Panreac ITW Companies) and 200 µL aquarius thiobarbituric acid (TBA, Sigma-Aldrich CO USA) and then the samples were heated at 100 ◦C for 60 min. In milk samples, 1 mL of raw milk was added to 7 mL ortho-phosphoric acid (Panreac ITW Companies) and 2 mL of aquarius TBA (thiobarbituric acid, Sigma-Aldrich CO USA) and then incubated at 100 ◦C for 60 min. After that, absorbance was recorded at 532 nm. The protein carbonyl (PC) content was determined according to the method of Patsoukis et al. [29] by measuring the conjunction of 2,4-dinitrophenylhydrazine (DNPH) on protein carbonyls at 375 nm. The 2,2′ -Azino-bis (3-ethylbenzthiazoline-6-sulfonic acid) (ABTS) radical scavenging assay was based on the published methods [30,31]. Ferric reducing ability of plasma (FRAP) assay was used to measure total antioxidant potential according to the method described by Benzie and Strain [32].

#### *2.9. Milk Fatty Acid Profile*

Milk fatty acid profile was determined using Gas Chromatography (Agilent 6890 N GC, Agilent 7683 B autosampler injector), equipped with an HP-88 capillary column (60 m × 0.25 mm i. d. with 0.20 µm film thickness, Agilent Technologies, USA) and a flame ionization detector (FID) as previously described by Mavrommatis and Tsiplakou [33].

#### *2.10. Monocytes and Neutrophils Immune Genes Expression*

Blood monocytes and neutrophils were isolated and then total RNA was extracted as previously described by Tsiplakou et al. [34]. Pure RNA (500 ng) from 64 individual (monocytes (32) and

neutrophils (32)) samples was reverse transcribed with the PrimeScript First Strand cDNA Synthesis Kit (Takara, Japan) according to the manufacturer's instructions using a mix of random hexamers and oligo-dT primers. A pair of primers specific for each target gene was designed using Geneious software (Biomatters, New Zealand) according to the respective *Ovis aries* gene coding sequences (CDSs in GenBank) (Table S7). The specificity of each pair of primers was tested against genomic DNA (positive control) to confirm that a single amplicon would emerge after quantitative real-time PCR. In addition, dissociation curves were generated, and the amplification products were subjected to agarose gel electrophoresis to confirm the production of a single amplicon per reaction. The relative expression levels of the target genes were calculated as (1 + E)−∆Ct, where ∆Ct is the difference between the geometric mean of the two housekeeping genes' Cts and the Ct of the target gene, and the primer efficiency is the mean of each amplicon's efficiency per primer, which was calculated by employing the linear regression method on the log (fluorescence) per cycle number (∆Rn) using the LinRegPCR software [35]. Glyceraldehyde 3-Phosphate Dehydrogenase (*GAPDH*) and Tyrosine 3-monoxygenase/tryptophan 5-monooxygenase activation protein, zeta polypeptide (*YWHAZ*) were used as housekeeping genes to normalize the cDNA template concentrations; the RT-PCR protocols are described in Tsiplakou et al. [18].

#### *2.11. Statistical Analysis*

Experimental data were analyzed using the SPSS.IBM statistical package (version 20.0) and results are presented as mean ± mean standard error (SEM). Dietary treatment effects were determined using a general linear model (GLM) for a repeated measures analysis of variance (ANOVA). With the dietary treatments (D = Control and ActiSaf) used as the fixed factor and the sampling time (S) as the repeated measure, while including their interactions (D\*S) to evaluate differences over time, according to the model:

$$\text{Yijkl} = \mu + \text{Di} + \text{Sj} + \text{Ak} + (\text{D} \times \text{S})\text{ii} + \text{eijkl}$$

where Υijk is the dependent variable, µ the overall mean, Di the effect of dietary treatment (i = 2; Control and ActiSaf), Sj the effect of sampling time (j = 6 for milk performance, 2 for B-HBA concentration, fatty acids profile, antioxidant and immune system), Ak the animal's random effect, (DxS)ij the interaction between dietary treatments and sampling time, and eijk the residual error. Posthoc analysis was performed when appropriate using a Tukey's multiple range test [36]. For all tests, the significance level was set at *p* = 0.05. In order to simplify the visualization of the results, GraphPad Prism 6.0 (2012) was used for interleaved bars while error bars represent the mean standard error (SEM).

Moreover, discriminant analysis was also applied to pooled data to establish those variables capable of distinguishing and classifying samples among the two dietary treatments. Wilk's lambda (λ) criterion was used for selecting discriminant variables [37]. Forty variables were entered to develop a model to discriminate the thirty-two samples of each case. Specifically, five variables were used for grouped fatty acids in milk, ten and 10 for immune system genes' relative expression in monocytes and neutrophils, respectively, and seven and eight in antioxidant indices in milk and blood, respectively.

#### **3. Results**

#### *3.1. Animal Performance*

Dietary supplementation with LY ActiSaf significantly reduced the B-HBA concentrations in ewes' blood by 27% (0.86 ± 0.07 vs. 0.63 ± 0.06 mmol/L, *p* = 0.018) in the prepartum period, and by 17% (0.67 ± 0.04 vs. 0.56 ± 0.03 mmol/L, *p* = 0.028) in the postpartum period. Overall, the B-HBA was reduced by 24% (0.77 vs. 0.59 mmol/L, SEM = 0.04, *p* = 0.003) in the whole experimental period (Figure 1; Tables S1 and S2). Mean BW did not differ between the dietary treatments in the whole experimental period (Tables S3 and S4). However, ewes' BW recovered between lambing and 6th week of lactation tended to increase by 75% (2.00 ± 0.31 vs. 3.44 ± 0.28 kg, *p* = 0.092) in the ActiSaf compared with the Control group (Figure 2; Table S5).

β β β **Figure 1.** Graphical representation of (**A**) β-hydroxybutyric acid (B-HBA) in ewes' blood prepartum (mean ± SE), (**B**) β-hydroxybutyric acid in blood postpartum (mean ± SE), and (**C**) β-hydroxybutyric acid in blood of ewes in Control (black, *n* = 51 ewes) and ActiSaf (grey, *n* = 53 ewes) groups in experimental period of 12 weeks (mean ± mean standard error (SEM)). β β

**Figure 2.** Graphical representation of body weight gain (recovery) between lambing and 6th week of lactation of ewes in Control (black, *n* = 51 ewes) and ActiSaf (grey, *n* = 53 ewes) groups (mean ± SE).

Milk, fat corrected milk 6% (FCM6%), and energy corrected milk yield (ECM) were numerically increased by 7.6 (2.50 vs. 2.69 kg/day, SEM = 0.159, *p* = 0.395), 12 (2.07 vs. 2.32 kg/day, SEM = 0.126, *p* = 0.161), and 10% (1.93 vs. 2.13 kg/day, SEM = 0.116, *p* = 0.231), respectively, in the ActiSaf compared to the Control group (Figure 3; Table S6; Figure S1). Concerning milk chemical composition, fat and protein contents were slightly decreased by 1.2 (*p* = 0.740) and 3% (*p* = 0.381), respectively, in the ActiSaf group, due to higher daily milk yield. However, both daily milk fat (114 vs. 131 g/day, SEM = 7.188, *p* = 0.104) and milk protein production (133 vs. 143 g/day, SEM = 8.288, *p* = 0.434) were increased by 15 and 7.5%, respectively, in the ActiSaf group (Figure 3; Supplementation Table S6). The milk yield in the treated ewes showed a moderate increase after the third week in lactation, and a peak in the fourth week, indicating a more intense milk persistence.

**Figure 3.** Graphical representation of milk yield and chemical composition of ewes in Control (black, *n* = 51 ewes) and ActiSaf (grey, *n* = 53 ewes) groups (mean ± SEM). FCM: Fat corrected milk in 6% according to the equation Y6% = (0.28 + 0.12F) M, where F = fat% and M = milk yield in kg. ECM: Energy corrected milk = milk yield × (0.071 × fat (%) + 0.043 × protein (%) + 0.2224) [38].

#### *3.2. Milk Fatty Acid Profile*

Milk fatty acid profile was not altered among dietary supplementation except for certain minor differences. Specifically, pentadecanoic acid (C15:0), palmitoleic acid (C16:1), trans linoleic acid (C18:2n6t), and linolenic acid (C18:3n3) were increased in ActiSaf milk by 15 (0.82 vs. 0.95%, SEM = 0.045, *p* = 0.042), 13 (0.29 vs. 0.33%, SEM = 0.014, *p* = 0.033), 9 (0.19 vs. 0.22%, SEM = 0.008, *p* = 0.049), and 20% (0.40 vs. 0.48%, SEM = 0.027, *p* = 0.075), respectively, while stearic acid (C18:0) decreased by 5% (8.91 vs. 8.44%, SEM = 0.490, *p* = 0.029) (Table 2).

**Table 2.** The mean individual fatty acids (FAs) (% of total FA), FA groups and Saturated Fatty Acids (SFAs)/Unsaturated Fatty Acids (UFAs) of milk from ewes fed Control (*n* = 8 ewes) and ActiSaf (*n* = 8 ewes) diet throughout the experimental period (21 and 42 experimental days).


\* Effect: The dietary treatment (D), time (T), and the interaction between dietary treatment × time (DxT) effects were analyzed by analysis of variance (ANOVA) using a general linear model (GLM) for repeated measures and posthoc analysis was performed when appropriate using Tukey's multiple range test. †SEM = Standard error of the mean. ‡ *trans-11* C18:1 = these values are not included in the Σ *trans* C18:1 content. §SCFAs: Short-Chain Saturated Fatty Acids = C6:0 + C8:0 + C10:0 + C11:0; ¶MCFAs: Medium-Chain Saturated Fatty Acids = C12:0 + C13:0 + C14:0 + C15:0 + C16:0 + C17:0. ††LCFAs: Long-Chain Saturated Fatty Acids = C18:0 + C20:0. ‡‡MUFAs: Mono-Unsaturated Fatty Acids = C14:1 + C15:1 + C16:1 + C17:1 + C18:1 *cis-9* + *trans-11* C18:1 + *trans* C18:1; §§PUFAs: Poly-Unsaturated Fatty Acids = *cis-9, trans-11* C18:2 (CLA) + C18:2n-6c + C18:2n-6t + C18:3n-3 + C18:3n-6 + C20:3n-3; ¶¶SFAs: Saturated Fatty Acids = SCFA + MCFA + LCFA; †††UFAs: Unsaturated Fatty Acids = PUFA + MUFA; ‡‡‡S/U: Saturated/Unsaturated = (SCFA + MCFA + LCFA)/(PUFA + MUFA), and §§§AI: Atherogenicity index = (C12:0 + 4 \* C14:0 + C16:0)/(PUFA + MUFA).

#### *3.3. Oxidative Status*

Amongst the dietary treatments, we did not report any significant differences in both blood and milk antioxidant enzymes. However, Glutathione Reductase (GR) in blood plasma was significantly increased by 13% (0.067 vs. 0.076 units/mL, SEM = 0.002, *p* = 0.004) in ActiSaf fed ewes. A numerical increase in lactoperoxidase (LPO) and catalase (CAT) activities by 20 and 10%, respectively, in milk of ActiSaf fed ewes was observed. The total antioxidant capacity measured by ABTS assay was significantly higher by 16.7% (37.563 vs. 43.850% inhibition, SEM = 3.564, *p* = 0.001) in the milk of ActiSaf fed ewes (Table 3). A negative correlation between the total antioxidant capacity determined by FRAP assay and the proportions of MUFA and oleic acid (C*18:1 cis-9*) in milk was found. The same trend was reported between Glutathione Peroxidase (GPx) activity in blood plasma and the aforementioned fatty acids of milk. The correlation between blood malondialdehyde (MDA) content and the proportions of milk's PUFA was also negative. On the other hand, the correlations between blood MDA content and the proportions of MUFA and oleic acid, respectively, were positive (Figure 4).


**Table 3.** Enzymes activities (Units/mL), total antioxidant capacity, and oxidative status biomarkers in blood plasma and milk of ewes fed the two diets (Control, *n* = 8 and ActiSaf, *n* = 8) at two sampling times.

† SEM = Standard error of the mean. GST: Glutathione transferase. GR: Glutathione reductase. SOD: Superoxide dismutase. GSH-Px: Glutathione peroxidase. CAT: Catalase. ABTS: 2,2′ -Azino-bis (3-ethylbenzthiazoline-6-sulfonic acid) (inhibition%). FRAP: Ferric Reducing Ability of Plasma (µM ascorbic acid). MDA: Malondialdehyde (µM MDA). PCs: Protein carbonyls (nmol/mL). LPO: Lactoperoxidase. Significance level below 0.05 indicates significant difference. \* Effect: The dietary treatment (D), time (T), and the interaction between dietary treatment × time (DxT) effects were analyzed by ANOVA using a general linear model (GLM) for repeated measures and posthoc analysis was performed when appropriate using Tukey's multiple range test.

γ ′ **Figure 4.** Heat-map represents a Pearson correlation of milk fatty acids, immune system gene expression in monocytes and neutrophils, antioxidant enzymes activities, total antioxidant capacity, and oxidative indices in both blood plasma and milk of ewes. In immune system genes, *M* = *monocytes and N* = *neutrophils,* while in antioxidants B = blood plasma and M = milk. CCL5: C-X-C motif chemokine 5, CXCL16: C-X-C motif chemokine ligand 16, INFG: Interferon γ, IL1B: Interleukin-1 beta, IL2: Interleukin-2, IL6: Interleukin-6, IL8: Interleukin-8, IL10: Interleukin-10, TNF: Tumor Necrosis Factor, NFKB: Nuclear Factor kappa B, GST: Glutathione transferase, GR: Glutathione reductase, SOD: Superoxide dismutase, GPx: Glutathione peroxidase, CAT: Catalase, ABTS: 2,2′ -Azino-bis (3-ethylbenzthiazoline-6-sulfonic acid), FRAP: Ferric Reducing Ability of Plasma, MDA: Malondialdehyde, PCs: Protein carbonyls, LPO: Lactoperoxidase, VA: Vaccenic acid, CLA: Conjugated linoleic acid, SCFAs: Short-Chain saturated fatty acids, MCFAs: Medium-Chain saturated fatty acids, LCFAs: Long-Chain saturated fatty acids, MUFAs: Mono-unsaturated fatty acids, and PUFAs: Poly-unsaturated fatty acids.

#### *3.4. Immune Status*

β The relative transcript levels of both *CCL5* and *CXCL16* in monocytes of ActiSaf fed ewes were significantly suppressed by 30% (0.053 vs. 0.037, SEM = 0.003, *p* = 0.007 and 0.042 vs. 0.029, SEM = 0.008, *p* = 0.008, respectively) (Table 4). Amongst cytokines, Interleukin 8 (*IL8*) relative transcript levels were significantly decreased by 80% (0.0020 vs. 0.0004, SEM = 0.0000, *p* = 0.031), while Interleukin 1β (*IL1B*) showed a tendency to decrease by 43% (0.007 vs. 0.004, SEM = 0.001, *p* = 0.087) in ActiSaf fed ewes (Table 4). A significant downregulation in the relative expression of Interleukin 10 (*IL10*)

by 30% (0.014 vs. 0.010, SEM = 0.002, *p* = 0.047) in the neutrophils of the treated ewes was observed (Table 4). In addition, the relative transcript levels of *IL1B* in monocytes were positively correlated with the proportions of both MUFA and oleic acid in milk. The same trend was found between the relative expression of the Interleukin 2 gene and the proportion of palmitic acid in milk (Figure 4). Monocytes and neutrophil relative transcript levels of *IL1B* were negatively correlated with GR and Glutathione Transferase (GST) activities in blood plasma, respectively (Figure 4).


**Table 4.** Relative transcript levels of several genes in blood monocytes and neutrophils of ewes fed the two diets (Control, *n* = 8 and ActiSaf, *n* = 8) at two sampling times.

Significance level below 0.05 indicates significant difference. † SEM = Standard error of the mean. *CCL5*: C-X-C motif chemokine 5. *CXCL16*: C-X-C motif chemokine ligand 16. *INFG*: Interferon γ. *IL1B*: Interleukin-1 beta. *IL2*: Interleukin-2. *IL6:* Interleukin-6. *IL8*: Interleukin-8. *IL10*: Interleukin-10. *TNF*: Tumor Necrosis Factor. *NFKB*: Nuclear Factor kappa B. \* Effect: The dietary treatment (D), time (T), and the interaction between dietary treatment × time (DxT) effects were analyzed by ANOVA using a general linear model (GLM) for repeated measures and posthoc analysis was performed when appropriate using Tukey's multiple range test.

#### *3.5. Holistic Statistics*

Discriminant analysis was applied to pooled data of two sampling times (3rd and 6th week postpartum) according to fatty acids in milk, immune gene expression in both monocytes and neutrophils, and antioxidant indices in blood plasma and milk (Figure 5) to investigate if the samples can be distinguished according to the type of the diet (Control and ActiSaf). The percentages of the samples that were classified into the correct group, according to the dietary treatment, were 100%. Wilks' lambda was observed at 0.001 for Function 1 (*p* = 0.159), while the relative transcript levels of *CCL5, CXCL16, and IL6* in monocytes', *IL10, IL6,* and *IL8* in neutrophils, and the GR activity in blood plasma were the variables that contributed the most.

**Figure 5.** Discriminant plots separating the Control and ActiSaf fed ewes according to their fatty acid grouped values, immune gene expression in both monocytes and neutrophils, and antioxidant indices in blood and milk.

#### **4. Discussion**

Blood B-HBA concentrations reflect the magnitude of negative energy balance (NEB) and lipid mobilization and are a diagnostic marker for subclinical (SCK) and clinical ketosis (CK) in ruminants. The B-HBA content in the blood of sheep with SCK ranges from 0.5 to 1.6 mmol/L [39–42], while in those with CK from 1.6 to 7 mmol/L [39,40,43]. However, in the case of healthy pregnant sheep these values could be around 0.8–0.9 mmol/L [44,45]. Nonetheless, the values of the Control group, first and foremost during the prepartum period, may indicate a moderate NEB (0.86 mmol/L). On the other hand, results regarding the B-HBA concentration in the blood of ActiSaf fed ewes indicate an improvement in the energetic status of the animals. It should be underlined here that the prolificacy between the two groups was the same (around 2.35), which means that although the number of fetuses affects B-HBA content [46], it had minimum impact in our trial. The same levels of B-HBA content in blood of healthy ewes of the same breed in early lactation have been previously reported [18,38]. An increase in the host energy availability might be due to a better rumen function and microbiome homeostasis. During the peripartum period, the energy and nutrient demands increase exponentially while the dry matter intake decreases. Thus, the optimal rumen function and the balance between VFA for a maximum feed efficiency are momentous in the transition period. The mechanism underling LY contribution in rumen may be down to yeast's oxygen scavenging properties (anaerobiosis). Specifically, the improvement of anaerobiosis in rumen increases the bacteria viability and thus, microbial protein synthesis and fiber digestibility [6]. Energy balance might be improved as a result of the dry matter (DM) and NDF digestion enhancement as have been reported by Plata et al. [47]. Furthermore, Panda et al. [48] also found that crude protein (CP) digestibility was also increased by 4.8% with dietary supplementation of yeast culture on male calves. In agreement with our findings, ActiSaf dietary inclusion (5g/day) in early lactating dairy cows, significantly decreased serum B-HBA and non-esterified fatty acids (NEFA) concentrations [9], while in mid-lactating cows, 4 g/day of LY supplementation did not affect B-HBA concentration since animals were not prone to NEB [49]. In addition, ewes in the ActiSaf group recovered their body weight from lambing until the sixth week postpartum in a more efficient manner, considering the increased available energy, as concluded by B-HBA concentration.

In compliance with our findings (12% FCM6%, *p* = 0.161), Stella et al. [12], reported a significant increase in goats' milk yield by 14% when their diet was supplemented with *S. cerevisiae*. It is worth mentioning that, in the same study, treated goats showed an upward trend in milk yield after the fourth week postpartum which decreased slower compared to the control group, showing a persistence in milk similar to our study. The dietary inclusion of *S. cerevisiae* enhanced cows' milk yield in early [13] mid- [14] and late [7] lactation. However, Dehghan-Banadaky et al. [49], showed that the milk yield was not affected in *S. cerevisiae* supplemented cows after the 145 DIM, possibly due to the absence of NEB. The results from 22 studies with more than 9039 lactating dairy animals showed an increase in their milk production by 7.3% (ranging from 2 to 30%) when their diets were supplemented with Yea-Sacc®1026 yeast [50].

Interestingly, in a meta-analysis study, Dehghan-Banadaky [49] concluded that an enhancement in milk yield was accompanied by an increase in feed intake in supplemented animals with yeast products. Moreover, yeast administration in prepartum cows' diets improved DMI [51]. Additionally, Habeeb et al. [52] reported that an enhancement of animal performance by the inclusion of yeast in their diet was mainly attributed to an increase in feed intake rather than feed digestibility improvement.

The milk fatty acid's profile was not holistically modified; however, certain interesting results, related to the biohydrogenation process (BH), were unveiled. Julien et al. [53] first observed the impact of LY administration on ruminal biohydrogenation processes. Specifically, LY promotes growth and activity of rumen lactate-utilizing bacteria, such as *Megasphera elsdenii* or *Selenomonas ruminantium*, *Actinobacteria*, including *Propionibacterium acnes* as well as fibrolytic bacteria. Consequently, LY could be involved at different stages of BH; firstly, by altering biohydrogenating microorganisms, i.e., improving growth of either t11 or t10 isomer producing bacteria, and secondly by modulating the ruminal biotope, i.e., by stabilizing ruminal pH or favoring stronger reducing conditions. In addition, Julien et al. [53] reported that LY supplementation increased the accumulation of trans C18:1 in vitro and decreased the proportion of C18:0, suggesting an inhibition of the last step of BH of c9c12-C18:2 fatty acids. Thus, in our study it could be hypothesized that the improved rumen conditions by LY administration may favor the isomerisation of c9c12-C18:2 and consequently increased the production of intermediate fatty acids in the rumen, which induced an inhibition or a saturation of the enzyme activity of bacteria involved in the second reduction step [54].

During the peripartum period, animals' augmented requirement for energy and nutrient results in lipid mobilization and blood hyperketonemia which induce oxidative stress [55]. Optimizing nutrition requirements by improving rumen efficiency may suppress the concentration of such trigger metabolites and improve the oxidative status. Glutathione reductase (GR) has a central role in the antioxidant defense system since it catalyzes the conversion of oxidized glutathione disulfide to the reduced form of glutathione, which is a critical molecule in resting oxidative stress [56]. It is known that glutathione is extremely important since it acts as substrate or co-substrate in enzymatic reactions (e.g., the glutathione-S-transferase or glutathione-shuttle enzymes), reacts directly with free radicals and lipid peroxides, and protects cells [57]. The mechanism under which GR increased its activity remains unclear, thus our assumptions are oriented toward a prudent liver function where glutathione is de novo synthesized due to the lower B-HBA concentration. Another possible mechanism that can increase GR activity might be the Flavin Adenine Dinucleotides' (FADs) co-substrate. Specifically, yeasts are sources of B-complex vitamins that act as precursors of the essential co-enzymes NAD and FAD that are responsible for biological oxidation [58]. In addition, high genetic merit dairy animals often burden their metabolism since they require increased levels of energy in order to meet their demands, leading to ROS production and later to the annihilation of the milk oxidative stability [59]. Hence, total antioxidant capacity enhancement in the milk of ActiSaf fed ewes may be important for the dairy industry.

Concerning chemokines, CCL-5 is involved in the activation of T cells, macrophages, eosinophils and basophils, and its enhancement is related to an inflammation response [60]. On the other hand, CXCL-16, a transmembrane protein, is detached from the membrane by metalloproteinase

ADAM10 induced chemotaxis [61]. In this study, the downregulation in the relative transcript levels of both *CCL5* and *CXCL16* in monocytes of ActiSaf treated ewes, indicates a lower inflammatory response during the first 6 weeks of lactation. The *IL1B* which was suppressed in our study, regulates B-cell maturation and proliferation, activates the Natural Killer (NK) cells and is generally related to the acute manifestation of inflammation in immune cells [62]. Interleukin-8, on the other hand, has a chemotaxis-inducer effect mainly in neutrophils. Pro-inflammatory chemokines and cytokine downregulations is directly attributed to B-HBA mitigation. Specifically, blood ketones derived from ketogenesis through acetyl-CoA metabolization have been shown to act as stimulants in chemokines and cytokines in cows' mammary epithelia cells [63].

#### **5. Conclusions**

In conclusion, supplementing dairy sheep diets with 2 g of the ActiSaf live yeast/day/ewe during the transition and early lactation periods have a beneficial impact on animals' performance whilst simultaneously portraying an improvement on pro-inflammatory responses attributed to a lower lipomobilization. This overall stress suppression during this turning point for the ruminants' period may unveil the potential of live yeasts as health modulators towards the collective effort of reducing antibiotic dependance at the farm scale. However, further research is needed to deeply understand the mechanism under the enhancement of energy supply in small ruminants.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2309-608X/6/4/334/s1, Table S1. Analysis of variance in blood β-hydroxybutyric acid in prepartum and postpartum period; Table S2. Repeated measure analysis of variance in blood β-hydroxybutyric acid in overall experimental period using prepartum and postpartum sampling time as repeated factor; Table S3. Analysis of variance in body weight at the start of the experiment, at lambing, and at the end; Table S4. Repeated measure analysis of variance in body weight in overall experimental period using the weighing at the start of the experiment, at lambing and at the end, as repeated factor.; Table S5. Analysis of variance in body weight recovery from lambing to 6 weeks postpartum. Table S6. Milk yield and milk chemical composition of ewes fed the Control and ActiSaf diets in the six sampling times; Table S7. Sequences of primers for target genes used in real-time qPCR; Figure S1. Graphical representation of milk chemical composition of the Control and ActiSaf groups.

**Author Contributions:** Conceptualization, G.Z. and V.N.; methodology, A.M., C.M. and C.C.; software, A.M.; validation, A.M., C.M., C.C., V.N. and E.T.; formal analysis, A.M.; investigation, G.Z. and E.T.; resources, V.N.; data curation, A.M. and D.K.; writing—original draft preparation, A.M.; writing—review and editing, E.T., C.C. and G.Z.; visualization, A.M.; supervision, E.T.; project administration, G.Z.; funding acquisition, G.Z. All authors have read and agreed to the published version of the manuscript.

**Funding:** The study was partially funded by Phileo Lesaffre Animal Health, Lille, France.

**Acknowledgments:** Authors would like to thank Flessas Dairy Farm for their collaboration.

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

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 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* **Assimilation of Cholesterol by** *Monascus purpureus*

#### **Theresa P. T. Nguyen \* , Margaret A. Garrahan** † **, Sabrina A. Nance** † **, Catherine E. Seeger** † **and Christian Wong** †

Department of Chemistry & Biochemistry, Loyola University Maryland, Baltimore, MD 21210, USA; magarrahan@loyola.edu (M.A.G.); sanance@loyola.edu (S.A.N.); ceseeger@loyola.edu (C.E.S.); ckwong@loyola.edu (C.W.)

**\*** Correspondence: tptnguyen@loyola.edu; Tel.: +1-410-617-2862

† These authors contributed equally to this work.

Received: 31 October 2020; Accepted: 8 December 2020; Published: 9 December 2020

**Abstract:** *Monascus purpureus*, a filamentous fungus known for its fermentation of red yeast rice, produces the metabolite monacolin K used in statin drugs to inhibit cholesterol biosynthesis. In this study, we show that active cultures of *M. purpureus* CBS 109.07, independent of secondary metabolites, use the mechanism of cholesterol assimilation to lower cholesterol in vitro. We describe collection, extraction, and gas chromatography-flame ionized detection (GC-FID) methods to quantify the levels of cholesterol remaining after incubation of *M. purpureus* CBS 109.07 with exogenous cholesterol. Our findings demonstrate that active growing *M. purpureus* CBS 109.07 can assimilate cholesterol, removing 36.38% of cholesterol after 48 h of incubation at 37 ◦C. The removal of cholesterol by resting or dead *M. purpureus* CBS 109.07 was not significant, with cholesterol reduction ranging from 2.75–9.27% throughout a 72 h incubation. Cholesterol was also not shown to be catabolized as a carbon source. Resting cultures transferred from buffer to growth media were able to reactivate, and increases in cholesterol assimilation and growth were observed. In growing and resting phases at 24 and 72 h, the production of the mycotoxin citrinin was quantified via high-performance liquid chromatography-ultraviolet (HPLC-UV) and found to be below the limit of detection. The results indicate that *M. purpureus* CBS 109.07 can reduce cholesterol content in vitro and may have a potential application in probiotics.

**Keywords:** *M. purpureus*; red yeast rice; filamentous fungi; cholesterol reduction; probiotic potential

### **1. Introduction**

*Monascus purpureus* is a filamentous fungus that produces a variety of secondary metabolites, including pigments, lipids, and monacolins. *M. purpureus* is most widely known for the fermentation of white rice to produce a deep red rice known as angkak or beni koji [1–3]. More commonly, the fermented product is called "red yeast rice", though *Monascus* species are more accurately molds. *M. purpureus* fermented rice is used in food preparation for flavoring, coloring, and preservation, and is also consumed in traditional Chinese medicine to improve ailments of circulation and heart health [2,4–7]. Modern research explored the health claims, and found a plausible cause: *M. purpureus* can synthesize monacolins, naturally occurring compounds capable of decreasing cholesterol levels by inhibiting HMG-CoA reductase, the rate-limiting step in cholesterol biosynthesis [7,8]. The most potent of the *Monascus* monacolins, monacolin K, was isolated and patented as lovastatin and is widely prescribed to treat hypercholesterolemia [8]. Statins are effective treatments for high cholesterol; however, side effects, low tolerance, and the cost of the drug have led patients to pursue alternative options to lower their cholesterol levels [9]. As lyophilized red yeast rice (RYR) supplements emerged as a naturopathic alternative in the U.S., the Food and Drug Administration (FDA) restricted the

market, determining that without standardization and quality control for the amount of monacolin K, RYR was not a dietary supplement, but rather an unauthorized new drug [5,10]. As a result, current U.S. supplements marketed as RYR may contain, at most, only trace amounts of monacolin K [5,10–12]. Nevertheless, several clinical studies report that RYR supplements may be an effective treatment option for hypercholesterolemia as significant reductions in low-density lipoprotein (LDL) cholesterol and total cholesterol levels were observed in patients taking RYR [4,10–14].

Two-thirds of cholesterol required for cell membranes and biosynthesis of steroid hormones and bile acids is endogenously synthesized; however, an excess of cholesterol is a major risk factor for cardiovascular disease [15–17]. The World Health Organization (WHO) projects that by 2030, nearly 23.6 million people will die from cardiovascular disease, which includes heart disease and stroke as the leading causes of death [18]. Research into the efficacy of therapies such as RYR supplements in the treatment of elevated cholesterol levels is ongoing and critical for treating cardiovascular disease.

*M. purpureus* strains administered via RYR supplements in clinical trials often have unspecified viability [19,20]. This led us to ask if living *M. purpureus*, and not simply its secondary metabolites, have cholesterol-lowering properties. Several studies have demonstrated that active probiotic microorganisms introduced into intestinal microbiota can improve the overall lipid content in human blood serum [15,21–25]. Probiotics are defined as live microorganisms that, when administered in adequate amounts, can confer a health benefit on the host [26]. The fermented product of *M. purpureus* can be labeled as "contains live and active cultures," but clinical studies on the safety, viability, and dosage of live *M. purpureus* strains are still necessary before characterizing strains of *Monascus* as probiotic [26–28]. Additionally, factors like mycotoxins should considered; *Monascus* species, like *Aspergillus* and *Penicillium*, naturally synthesize the cytotoxic citrinin at low levels [29,30]. In the European Union (EU) and US, *Monascus* pigments are prohibited from use in food industries [3,31,32]. Still, many *Monascus* strains are generally regarded as safe (GRAS) in many Asian countries and the commercial interest in fermentation has led to easy access of RYR outside of Asia [6,11,33].

While the cholesterol-lowering benefits of probiotics have been highlighted in in vivo studies, the mechanisms of probiotics remain not fully understood [34–36]. Potential mechanisms have been proposed in vitro, such as the removal of cholesterol from media by the assimilation or uptake of cholesterol by probiotic strains of *Lactobacilli* and *Bifidobacteria* [35,37–41]. The ability of cholesterol removal appears to be growth- and strain-specific, with nearly all research focused on bacterial strains [35–41]. We anticipate that as beneficial fungi are characterized, greater attention will be drawn to the roles of fungi in health, nutrition, and the mycobiome. Currently, *Saccharomyces boulardii* is the only fungus with a strain commercially labeled in the U.S. as a probiotic, as it can survive the passage through the gastrointestinal tract, present good growth at 37 ◦C, complement treatment of gastrointestinal diseases, and also assimilate cholesterol [37,42–46].

In this study, we evaluated the ability of the fungus *M. purpureus* CBS 109.07 to assimilate cholesterol in vitro. We developed new sample collection methods and used gas chromatography to quantify the levels of cholesterol remaining after incubation with *M. purpureus* CBS 109.07. Our data indicate that this strain of *M. purpureus* is capable of cholesterol assimilation, a cholesterol-lowering mechanism separate from its ability to produce monacolins and which has not been previously reported. The present study does not establish the safety and efficacy of CBS 109.07 as a therapeutic agent, however, these results lay the groundwork for the possibility that active *M. purpureus* CBS 109.07 may have probiotic potential based on its ability to assimilate cholesterol.

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

#### *2.1. Strains and Media Conditions*

*M. purpureus* Went teleomorphic type strain CBS 109.07 (ATCC 16365) was obtained from the American Type Culture Collection (ATCC) strain bank and chosen for its ability to grow at 30 ◦C. *M. purpureus* CBS 109.07 has been used in food-grade studies for human and animal food applications [6,47,48]. *M. purpureus* was grown in a malt extract media (MEA) containing: 2% soluble Bactomalt extract (BD Bioscience), 2% glucose, 1% peptone, and pH adjusted to pH 7. Plate media contained 2% agar. Phosphate-buffered saline (PBS) solution contained 8.0 g NaCl, 0.2 g KCl, 1.44 g Na2HPO4, 0.24 g KH2PO<sup>4</sup> for every 1.0 L solution and was pH adjusted to pH 7. Where indicated, PBS was supplemented with 6.72 g/L of yeast nitrogen base without amino acids (BD Difco) or 5 g/L ammonium sulfate, and pH was adjusted to 7 before sterilization. Bile salt supplemented media contained 0.3% (*w*/*v*) oxgall (BD Difco).

#### *2.2. Submerged Culture Preparation*

A sterilized 5 mm cork-borer was used to remove an agar plug of *M. purpureus* CBS 109.07 grown on MEA plate media. The agar plug was subcultured into 4 mL of liquid MEA media and incubated at 30 ◦C. After four day incubation at 30 ◦C at 150 rpm, a colorless, spherical fungal pellet was formed. The pellet was transferred into new media with 0.3% oxgall at 37 ◦C and 60 rpm for growth curve, cholesterol assimilation, and citrinin production experiments.

#### *2.3. Cholesterol Assimilation*

#### 2.3.1. Cholesterol Reagents

A stock solution of cholesterol (Lipids Cholesterol Rich from adult bovine serum; Sigma-Aldrich, St. Louis, MO, USA) at 10 mg/mL was used to prepare cholesterol assimilation assays and to prepare a 6-point calibration curve as described in Section 2.3.6 [37]. A stock solution of 5-α-cholestane (Sigma-Aldrich, St. Louis, MO, USA) at 2.5 mg/mL was used as an internal standard in the lipid extractions.

#### 2.3.2. Culture Preparation for Growing, Resting, Dead, and Control Conditions

Growing, resting, and *M. purpureus* control conditions contained *M. purpureus* CBS 109.07 pellets that were homogenized using a sterilized glass douncer in a sterile 50 mL conical tube and divided into replicates. Dead culture conditions contained *M. purpureus* CBS 109.07 pellets that were autoclaved at 121 ◦C for 20 min under 15 psi pressure and transferred to fresh media. Growing and dead cultures contained 10 mL MEA media in sterile 50 mL borosilicate tubes; resting cultures contained 10 mL of PBS in sterile 50 mL borosilicate tubes; resting cultures supplemented with nitrogen sources contained 10 mL of PBS with ammonium sulfate or 10 mL of PBS with yeast nitrogen base. Cholesterol assimilation and dry weight experiments were supplemented with 0.3% oxgall and incubated at 37 ◦C and 60 rpm. With the exception of the *M. purpureus* control, all conditions were incubated with 120 µg/mL cholesterol. Media control without *M. purpureus* contained 120 µg/mL cholesterol and 0.3% (*w*/*v*) oxgall in 10 mL MEA.

#### 2.3.3. Cholesterol Assimilation and Dry Weight Growth Curve

Cholesterol assimilation and dry weight experiments for each growth condition were prepared in triplicate, where three independent sets of culture were homogenized and divided into six sterile 50 mL borosilicate tubes to account for six timepoints (0, 24, 36, 48, 60, and 72 h). At designated time point, a 1.0 mL aliquot of culture supernatant was collected, centrifuged at 2000× *g* for 15 min, and stored in a 50 mL borosilicate glass tube with a PTFE-lined cap at −20 ◦C. The remaining 9.0 mL of the culture was then harvested and filtered via vacuum flask and a pre-weighed Whatman filter #1. The contents were allowed to air dry for five days and dry weight was measured on an analytical balance.

#### 2.3.4. *M. Purpureus* Dormancy Experiment

Cholesterol assimilation and dry weight experiments were prepared in triplicate, where three independent sets of cultures were homogenized and divided into five sterile 50 mL borosilicate tubes to account for five timepoints (0, 24, 48, 72, and 96 h). *M. purpureus* was incubated in PBS buffer, pH 7 with 0.3% oxgall and 120 µg/mL cholesterol. At designated time points, a 1.0 mL aliquot of culture supernatant was collected, centrifuged at 2000× *g* for 15 min, and stored in a 50 mL borosilicate glass tube with a PTFE-lined cap at −20 ◦C. *M. purpureus* samples were then washed twice under sterile conditions with 10 mL of PBS + 0.3% oxgall, and transferred to 10 mL MEA media with 0.3% oxgall and 120 µg/mL cholesterol. A 1.0 mL aliquot of culture supernatant was collected at the starting point (t = 0 h), and also at day 4 and day 7 of incubation in MEA media. At day 7, the culture was harvested and filtered via vacuum flask and a pre-weighed Whatman filter #1. The contents were allowed to air dry for five days and dry weight was measured on an analytical balance.

#### 2.3.5. Resting *M. purpureus* Supplemented with Nitrogen Sources Experiment

Cholesterol assimilation and dry weight experiments were prepared in duplicate, where two independent sets of cultures were homogenized and divided into three sterile 50 mL borosilicate tubes to account for three timepoints (0, 24, and 72 h). *M. purpureus* was incubated in PBS buffer supplemented with either ammonium sulfate or yeast nitrogen base without amino acids, pH 7 with 0.3% oxgall and 120 µg/mL cholesterol. At designated time points, a 1.0 mL aliquot of culture supernatant was collected, centrifuged at 2000× *g* for 15 min, and stored in a 50 mL borosilicate glass tube with PTFE lined cap at −20 ◦C. *M. purpureus* samples were then washed twice under sterile conditions with 10 mL of PBS + 0.3% oxgall, and transferred to 10 mL MEA media with 0.3% oxgall. At day 4, the culture was harvested and filtered via vacuum flask and a pre-weighed Whatman filter #1. The contents were allowed to air dry for five days and dry weight was measured on an analytical balance.

#### 2.3.6. Cholesterol Extraction

Cholesterol assimilation samples were thawed and a stock solution of cholesterol (10 mg/mL) was used to prepare cholesterol standards in a range from 10–150 µg/mL in a borosilicate glass tube. To each sample and standard, 20 µL of 2.5 mg/mL internal standard 5-α-cholestane was added. Direct saponification was carried out on all samples and standards based on the method described by Fletouris et al. [49]. Four millimeters of methanolic 0.5 M KOH solution was added to each tube which was then capped and vortexed for 15 s. The samples and standards were heated for a total of 15 min in an 80 ◦C water bath, and removed every 5 min to vortex for 10 s. After cooling to room temperature, 4 mL of hexane was added for lipid extraction and vortexed for 1 min. After incubating at room temp for 10 min to permit phase separation, the entire hexane layer of each sample was transferred to a clean test tube. The hexane layer was evaporated using speed vacuum at −109 ◦C. Dried samples and standards were resuspended in 0.6 mL of hexane and transferred to autosampler vials for gas chromatography (GC) analysis.

#### 2.3.7. Gas Chromatography Methods

Cholesterol was determined using gas chromatography (Shimadzu GC-2014, Kyoto, Japan) with a flame ionized detector (FID) and an autosampler [49]. The separation was completed using an SPB-1 column (15 m × 0.32 mm i.d.; film thickness 1.0 mm) (Supelco Inc., Bellefonte, PA, USA) using helium as a carrier gas at a flow rate of 2 mL/min. The oven temperature was set at 285 ◦C, injection port temperature at 300 ◦C, and flame ionization detector temperature at 300 ◦C. The injection volume was 1 µL with a split ratio of 20:1. Matrix effects were addressed by the addition of an internal standard of 5-α-cholestane to all samples and standards. In addition, extracting the standards for each set of experiments using the same process as for the samples helped account for any errors in the preparation process. This allowed for the determination of the limits of detection (LOD) and quantitation (LOQ) for the experimental conditions of 8.31 µg/mL and 27.71 µg/mL, respectively.

#### 2.3.8. Calculations for Cholesterol Assimilation

The integrated peak areas for cholesterol and the internal standard 5-α-cholestane were used to determine a 6-point calibration curve for cholesterol and used to extrapolate cholesterol recovered. The experimental calibration curve to determine LOD and LOQ were created by combining the calibration curves from five experiments to generate a linear calibration curve with R<sup>2</sup> = 0.9832. Cholesterol assimilated and % cholesterol assimilated were calculated as follows, where Cholesterol*<sup>i</sup>* represents cholesterol content recovered at t = 0 and Cholesterol*<sup>f</sup>* represents cholesterol content recovered at given time point.

$$\text{Cholesterol assimilated} = \text{Cholesterol}\_l - \text{Cholesterol}\_f \tag{1}$$

$$\% \text{ Cholesterol assimilated} = \frac{\text{Cholesterol assimilated}}{\text{Cholesterol}\_{\%}} \times 100\% \tag{2}$$

#### *2.4. Citrinin Production*

#### 2.4.1. Citrinin Reagents

Citrinin, high-performance liquid chromatography (HPLC)-grade, was purchased from Sigma-Aldrich. A stock solution of citrinin at 100 ug/mL was prepared in HPLC-grade methanol (J.T. Baker) and used to construct a 7-point calibration curve as described in Section 2.4.3. Acetonitrile and water used in chromatography were HPLC grade (J.T. Baker), and trifluoroacetic acid (TFA) (Sigma-Aldrich, St. Louis, MO, USA) was analytical grade.

#### 2.4.2. Culture Preparation and Extraction for Citrinin Production

A 5 mm agar plug of *M. purpureus* was pre-cultured at 150 rpm and 30 ◦C for 4 days, and transferred to 10 mL of MEA + 0.3% or PBS + 0.3% oxgall and grown at 60 rpm and 37 ◦C, as described in culture preparations for cholesterol assimilation assays Section 2.3.2. At 24 h, 72 h, and 14 days, cultures were extracted for citrinin as described in Liu and Xu, with some modifications [50]. Briefly, 10 mL cultures were dounced and extracted with 10 mL of ethanol (1:1). Samples were then vortexed for 5 min and sonicated for 20 min. Samples were spun down at 4200× *g* for 10 min. Supernatant was collected, dried down, and resuspended in 1 mL HPLC-grade methanol. Extraction method was validated with recovery controls, where *M. purpureus* grown in MEA + 0.3% oxgall at 24 h and 72 h were spiked with citrinin at 10 µg/mL and extracted as described previously.

#### 2.4.3. High-Performance Liquid Chromatography Methods

Citrinin was determined using high-performance liquid chromatography, HPLC (Agilent 1100 liquid chromatograph) with a diode array detector (DAD) and an autosampler. The separation was completed using a Discovery C18 column (5 µm, 150 × 4.6 mm column) (Supelco Inc., Bellefonte, PA, USA) and an isocratic elution. The mobile phase consisted of acetonitrile:water containing 0.05% TFA and the volume ratio was 35:65 [50,51]. All samples, standards, and solvents were filtered through 0.22 µm membrane filters prior to HPLC analysis. The flow rate was 1 mL/min, and 20 µL sample was injected. The UV-DAD detection was monitored at 254 nm and 334 nm. The integrated peak areas at 334 nm for standard citrinin were used to determine a 7-point calibration curve and used to extrapolate citrinin recovered. The LOD and LOQ for the experimental conditions were 1.11 µg/mL and 3.70 µg/mL, respectively. Recovery of citrinin was determined by dividing citrinin concentration recovered by known citrinin concentration injected.

#### *2.5. Statistical Analysis*

For cholesterol assimilation and growth curves, growing, resting, and dead *M. purpureus* cultures and controls were conducted in triplicate for each time point. For citrinin assays, growing and resting *M. purpureus* cultures were conducted in duplicate for each time point. All GC-FID and high-performance liquid chromatography-ultraviolet (HPLC-UV) samples were measured in duplicate. Two-way ANOVA was carried out to examine the effect of *M. purpureus* × incubation time interaction

on growing, resting, or dead conditions. Tukey's test was used to compare means. Significance was defined at *p* < 0.05 or *p* < 0.01. Standard deviation is calculated as either absolute error, or percent error through propagation of uncertainty from Equation (2) of Section 2.3.8. All statistical analyses were carried out using GraphPad Prism 8.0.

#### **3. Results**

#### *3.1. Cholesterol Assimilation*

The in vitro removal of cholesterol by the filamentous fungi *M. purpureus* CBS 109.07 (hereon referred to in the Results section as *M. purpureus*) was analyzed at 37 ◦C in media containing 0.3% (*w*/*v*) oxgall and 120 µg/mL cholesterol. Three growth phases were assessed: growing, where culture is active in MEA media; resting, where culture is dormant in PBS buffer; and dead, where culture has been heat-killed and incubated in MEA media. At indicated time points, an aliquot of spent media was collected from three independent replicates.

After 36 h, *M. purpureus* removed 18.69 µg/mL or 18.78% of the cholesterol in spent media, which is a significant decrease from the initial concentration (Table 1, Figure 1, *p* < 0.01). The rate of cholesterol removal was most dramatic from 36 to 60 h, and cholesterol removed increased from 18.78% to 50.27% (*p* < 0.05). At 72 h, 69.65% cholesterol was removed.

**Table 1.** Cholesterol assimilated in *M. purpureus* CBS 109.07 at different growth phases. All cultures were incubated at 37 ◦C with 120 µg/mL cholesterol and 0.3% (*w*/*v*) oxgall bile salts. Cholesterol assimilated was calculated from initial cholesterol, and determined from three independent trials conducted for each growth phase at each time point, and measured in duplicate via gas chromatography-flame ionized detection (GC-FID). Standard deviation calculated is absolute error (µg/mL) and percent error (%).


a, b, c Means within a row are significantly different (*p* < 0.01). <sup>+</sup> Means significantly different from the initial value at t = 0 (*p* < 0.01).

In contrast, *M. purpureus* cultures dormant in PBS for resting phase or heat-killed in dead phase removed a negligible amount of cholesterol after 72 h (Table 1) (*p* > 0.05). Non-growing *M. purpureus* conditions removed less than 10% of cholesterol, with a high percent error accounting for the propagation of the standard deviation. The media control containing MEA or PBS incubated without *M. purpureus* similarly showed no significant change in cholesterol content or cholesterol assimilation over 72 h (Table S1). We note that cholesterol content of resting and dead *M. purpureus* and media controls did not change significantly from the initial concentration of cholesterol within each trial (*p* < 0.01).

**Figure 1.** Cholesterol assimilated (%) by *M. purpureus* CBS 109.07 at different growth phases. Growing, resting, and dead *M. purpureus* cultures were incubated with 120 µg/mL cholesterol and 0.3% (*w*/*v*) oxgall bile salts at 37 ◦C. Malt extract media (MEA) control contained 120 µg/mL cholesterol and 0.3% (*w*/*v*) oxgall bile salts incubated without *M. purpureus*. Three independent trials were conducted for each condition at each time point, and cholesterol in samples were measured in duplicate via GC-FID. Standard deviation in cholesterol assimilated is percent error.

#### *3.2. Growth of M. purpureus*

The morphology of *M. purpureus* in submerged media is a compact, smooth, and spherical pellet consisting of intertwined hyphae [52–54]. Consistent with studies on other filamentous fungi, we found that dry weight measurement is the most reproducible method of quantifying *M. purpureus* growth [55–61].

After an aliquot of spent media was collected from three independent replicates, the remaining culture was harvested to measure dry weight (Table 2, Figure 2). Growth phases growing, resting, and dead were assessed and conditions included 0.3% oxgall and 120 µg/mL cholesterol. *M. purpureus* control was grown in MEA media without cholesterol. The dry weight of growing *M. purpureus* was significantly different from that of resting or dead *M. purpureus* (*p* < 0.05). The presence of cholesterol did not significantly enhance or inhibit the growth of *M. purpureus* [62] (*p* > 0.05). In both resting and dead conditions, *M. purpureus* had no significant growth (*p* > 0.05).

**Table 2.** Dry weight of *M. purpureus* CBS 109.07. Growing, resting, or dead *M. purpureus* cultures were incubated in 120 µg/mL cholesterol and 0.3% (*w*/*v*) oxgall bile salts at 37 ◦C. *M. purpureus* incubated without cholesterol served as a control. Three independent trials, corresponding to samples used for cholesterol assimilation, were conducted for each growth phase at each time point.


a, b, c Means within a row are significantly different (*p* < 0.05). <sup>+</sup> Means significantly different from the initial value at t = 0 (*p* < 0.05).

**Figure 2.** Dry weight of *M. purpureus* CBS 109.07. Growing, resting, or dead *M. purpureus* cultures were incubated in 120 µg/mL cholesterol and 0.3% (*w*/*v*) oxgall bile salts at 37 ◦C. *M. purpureus* incubated without cholesterol served as a control. Three independent trials were conducted for each growth phase at each time point. Standard deviation is represented by error bars.

#### *3.3. Reactivating Dormant M. purpureus*

*M. purpureus* incubated in PBS buffer, pH 7 with 0.3% oxgall and 120 µg/mL cholesterol does not assimilate cholesterol during a 96 h incubation (Table 3). To examine if resting conditions correspond to a dormant *M. purpureus*, cultures incubated in PBS were washed and transferred to MEA media with 0.3% oxgall and 120 µg/mL cholesterol. After four days of incubation in MEA, cholesterol assimilation was measured, but was not significantly different than MEA media control (*p* > 0.05). After seven days of incubation, cholesterol assimilation was initiated and cholesterol content is comparable to growing phase *M. purpureus* (Table S1). Dry weight of rescued *M. purpureus* was increased from resting *M. purpureus* (Table 2). The length of time incubated in PBS before the transfer to MEA did not have a significant effect on the ability to assimilate cholesterol (*p* < 0.05).

**Table 3.** Cholesterol content (µg/mL) and dry weight (mg) of *M. purpureus* CBS 109.07 incubated in phosphate-buffered saline (PBS) and rescued in MEA. All cultures were incubated at 37 ◦C with 120 µg/mL cholesterol and 0.3% (*w*/*v*) oxgall bile salts. Resting *M. purpureus* was washed in PBS + 0.3% oxgall and then transferred to MEA with 0.3% oxgall and 120 µg/mL cholesterol, and sample collected after 4 and 7 days. Data were determined from three independent trials conducted at each time point. Cholesterol content was measured in duplicate via GC-FID. Standard deviation in cholesterol content is absolute error.


<sup>1</sup> Media contained 0.3% oxgall and 120 µg/mL cholesterol. <sup>2</sup> Time corresponds to duration incubated in PBS prior to rescue in MEA media. a, b, c Cholesterol content means within a row are significantly different (*p* < 0.05). <sup>+</sup> Means significantly different from the MEA media control Table S1 (*p* < 0.05). ++ Means significantly different from the resting *M. purpureus* dry weight Table 2 (*p* < 0.05).

To determine if cholesterol is catabolized by *M. purpureus* as a carbon source, resting phase cultures in PBS were incubated with a nitrogen source, either ammonium sulfate or yeast nitrogen base without amino acids, at concentrations found in minimal media. We observed that resting *M. purpureus* in PBS supplemented with nitrogen sources had no significant cholesterol assimilation throughout a 24 and 72 h incubation (Table 4, *p* > 0.05).

**Table 4.** Cholesterol content (µg/mL) of *M. purpureus* CBS 109.07 incubated in PBS with nitrogen sources. All cultures were incubated at 37 ◦C with 120 µg/mL cholesterol and 0.3% (*w*/*v*) oxgall bile salts. Cholesterol content was determined from two independent trials conducted for each growth phase at each time point, and measured in duplicate via GC-FID. All means were not significantly different from the initial value at t = 0 (*p* < 0.05). Standard deviation in cholesterol content is absolute error.


Fungi samples were then washed twice and transferred to MEA + 0.3% oxgall. The dry weight at day 4 in MEA of *M. purpureus* previously incubated in yeast nitrogen base was a significant increase from the initial weight before rescue (Table 5, *p* < 0.05), supporting the observation that dormant *M. purpureus* can be reactivated.

**Table 5.** Dry weight (mg) of *M. purpureus* CBS 109.07 after incubation in PBS with 120 µg/mL cholesterol and 0.3% (*w*/*v*) oxgall bile salts and rescued in MEA with 0.3% (*w*/*v*) oxgall bile salts. *M. purpureus* is washed in PBS + 0.3% oxgall and then transferred to MEA with 0.3% oxgall, and sample collected after 4 days. Two independent trials, corresponding to samples used in PBS cholesterol assimilation, were conducted for each growth phase at each time point. Standard deviation in cholesterol content is absolute error.


<sup>1</sup> Time corresponds to duration incubated in PBS prior to rescue in MEA media. \* Initial weight at t = 0. <sup>+</sup> Means significantly different from the initial weight at t = 0 (*p* < 0.05).

#### *3.4. Citrinin Production in M. purpureus*

The production of the mycotoxin citrinin was measured in *M. purpureus* grown under conditions used in cholesterol assimilation assays, where a 5 mm agar plug of *M. purpureus* was precultured in MEA at 150 rpm and 30 ◦C and then transferred to fresh media with oxgall and incubated at 60 rpm and 37 ◦C (Table 6). Samples collected at 24 h and 72 h did not have citrinin production above the limit of detection. The culture broth in these samples were also colorless. After 14 days, *M. purpureus* grown in MEA + 0.3% oxgall began to produce red pigment and culture broth turned reddish. *M. purpureus* in resting phase did not become pigmented. Though 14 days is outside the incubation period for this study's cholesterol assimilation experiments, we extracted the red cultures, and measured 6.77 ± 1.02 µg citrinin per mL of culture broth. To validate extraction methods and show

effectiveness, a set of *M. purpureus* cultures grown in MEA + 0.3% oxgall at 24 h and 72 h were spiked with citrinin at 10 µg/mL and extracted. Recovery of citrinin was 72.9% ± 3.8.

**Table 6.** Citrinin production under experimental conditions. Citrinin concentration (µg/mL) of *M. purpureus* CBS 109.07 under experimental conditions of growing and resting phases was measured at 24 and 72 h via HPLC-UV. Two independent trials were conducted for each time point. Standard deviation in cholesterol content is absolute error.


<sup>1</sup> ND is not detected with a peak below the limit of detection (LOD) of 1.11 µg/mL.

#### **4. Discussion**

As a natural source for monacolins, fermentation products of *Monascus purpureus* are widely used as alternative treatments for hypercholesterolemia. To the best of our knowledge, our data is the first to show that a strain of *M. purpureus* is capable of a cholesterol-lowering mechanism separate from its ability to produce monacolins and other secondary metabolites. We observed that active growing *M. purpureus* CBS 109.07 can assimilate cholesterol in vitro, and after 48 h incubation at 37 ◦C and high bile salt conditions, 36.38% of cholesterol content was removed. The removal of cholesterol by resting or dead *M. purpureus* CBS 109.07 was not statistically significant, and cholesterol was not catabolized as a carbon source. When resting cultures were washed and transferred to MEA media, *M. purpureus* CBS 109.07 became active and cholesterol assimilation and growth were observed. Citrinin production of *M. purpureus* CBS 109.07 incubated in growing or resting phase conditions at 24 h and 72 h was lower than the limit of detection, and we note that CBS 109.07 produced citrinin under our experimental conditions at day 14 when red pigment production was observed.

The ability of microorganisms to remove cholesterol in vitro from growth media is an indicator of probiotic potential, and the range of reduction percentage is wide and dependent on strain. We note that as with any therapeutic dosage, the concentration of microorganisms present will play a major role in cholesterol assimilation percentage. Miremadi et al. tested strains of *Lactobacilli* and *Bifidobacteria* and found 14 strains capable of removing cholesterol with a range of 34–65% assimilation after 24 h [41]. Eukaryotes capable of lowering cholesterol include strains of *S. boulardii*, *S. cerevisiae*, and *I. orientalis*, which after 48 h of incubation was observed to assimilate 90.6%, 96.8%, and 88.1% of cholesterol, respectively [37]. Strains of *P. kudriazevii*, *Galactomyces* sp., and *Y. lipolytica* were observed to assimilate 45.7%, 36.3%, and 30.9% of cholesterol, respectively, after 48 h of incubation in Chen et al. [45]. In the same study, the commercially available yeast probiotic *S. boulardii* lowered cholesterol by 36.5% at 48 h, and 41.5% cholesterol at 72 h. In this study, active growing *M. purpureus* CBS 109.07 was comparable to *S. boulardii* and was able to lower cholesterol from the media by 36.38% at 48 h, and 69.65% cholesterol at 72 h (Table 1). We note that at higher aeration and agitation, *M. purpureus* CBS 109.07 was able to assimilate a higher percentage of cholesterol (Table S2). When *M. purpureus* CBS 109.07 is resting or dead, cholesterol removal is not significant (Figure 1) and ranged from 2.75–9.29% removal after 72 h of incubation (Table 1). Other studies observed similar trends where resting and heat-killed cultures did not significantly reduce cholesterol [41,63–65], suggesting a mechanism where actively growing strains are more efficient at removing cholesterol.

To eliminate the possibility that cholesterol assimilation by *M. purpureus* was an artifact of starvation and the uptake of available carbon sources, we allowed cultures to incubate undisturbed until the entire culture was collected at the designated time point. This procedure differed from other cholesterol assimilation studies, where one-tenth of the culture volume was removed at each time point and, thus, could significantly impact nutrient availability [38,39,41,63,65–67]. We note that in our methods, the presence of cholesterol at 120 µg/mL did not enhance or inhibit the growth of *M. purpureus* CBS 109.07 as measured by dry weight (Figure 2).

We investigated the ability of *M. purpureus* CBS 109.07 to transition out of microbial dormancy after one to four days of incubation in PBS. Resting phase cultures incubated in PBS with 120 µg/mL cholesterol did not show significant cholesterol assimilation (Table 3). When washed and transferred to MEA media with 120 µg/mL cholesterol, previously resting phase *M. purpureus* CBS 109.07 cultures were able to restore cholesterol assimilation and growth at day 7 of rescue (Table 3) at levels comparable to growing cultures (Table S1, Table 2). There was no significant difference in reactivation of cholesterol assimilation between cultures that were incubated for one day or four days in PBS. We also observed that *M. purpureus* CBS 109.07 incubated in PBS with cholesterol and supplemented with nitrogen sources showed insignificant cholesterol assimilation between 24 to 72 h (Table 4), and were able to grow after transfer into MEA media for four days (Table 5). These results on resting phase reveal that *M. purpureus* CBS 109.07 incubated in PBS is indeed dormant and that cholesterol is not significantly taken up as a carbon source during dormancy. Additionally, the absence of cholesterol assimilation in PBS with nitrogen sources supports the assertion that *M. purpureus* CBS 109.07 does not metabolize cholesterol (Table 4). We posit that such an absence of cholesterol removal may reflect a mechanism where growing *M. purpureus* CBS 109.07 is more efficient at assimilating cholesterol.

Microorganisms can utilize the cholesterol-lowering mechanisms of active assimilation and passive adhesion to decrease host absorption of intestinal cholesterol [21,36,38]. Our results suggest that *M. purpureus* CBS 109.07 is capable of an in vitro active assimilation mechanism by growing cells. The dense pellet morphology of *M. purpureus* CBS 109.07 has made it difficult to measure the cholesterol content of the cell membrane, as similarly noted in biosorbent studies on other filamentous fungi such as *Aspergillus niger* and *Penicillium* sp. L1 strains [55–57]. Follow-up experiments will be conducted to lyse the *Monascus* membrane and examine membrane cholesterol content. Other cholesterol-lowering mechanisms by probiotic microorganisms include modulation of lipid metabolism and deconjugation of bile salts [36]. *M. purpureus* is capable of directly modulating lipid metabolism, as it synthesizes monacolins that directly inhibit HMG-CoA reductase, the committed step of cholesterol biosynthesis in the liver [7,8]. In future studies, we will assay *M. purpureus* CBS 109.07 for bile salt hydrolase (BSH) activity, the enzyme responsible for the deconjugation of bile salts found in many probiotic strains [68,69].

Like many strains within *Monascus*, *Aspergillus*, and *Penicillium* genera, *M. purpureus* CBS 109.07 can biosynthesize citrinin, with levels highly dependent on the growth conditions and amount of microorganisms used. In this study, the citrinin production in *M. purpureus* CBS 109.07 under growing and resting phase conditions replicated from our cholesterol assimilation experiments was below our limit of detection of 1.11 µg/mL (Table 6). Our results at 24 h and 72 h are consistent with published studies on other *M. purpureus* strains which measured citrinin production in different growth conditions over time. These studies observed delays in citrinin production, with detection of citrinin beginning as early as day 5 or late as day 10 [70–72]. Notably, the commencement of citrinin production corresponded to the commencement of red pigment production, and increases in agitation and aeration increased citrinin production [72,73]. *M. purpureus* CBS 109.07 studies in particular did not measure citrinin at early time points. However using thin-layer chromatography (TLC), they reported citrinin level to be 5 µg/mL after 14 day incubation in glucose media and unspecified agitation [30], and 65 µg/mL after 7 day incubation in ethanol media and 220 rpm agitation [74,75]. We used HPLC-UV to quantify citrinin production after 14 day incubation in MEA + 0.3% oxgall and 60 rpm. At day 14 under our conditions, the culture broth began to turn reddish, and we measured a citrinin concentration of 6.77 µg/mL [30,74–76]. The differences in citrinin production between CBS 109.07 studies highlight how critical growth conditions are to the control of citrinin levels in *Monascus* strains [77,78]. In future studies, we can target the citrinin issue as many studies have successfully eliminated or reduced the levels of citrinin by disrupting the citrinin biosynthetic genes *pksCT* or *ctnA* in *M. purpureus* [70,79–81].

To be beneficial for human health, probiotic microorganisms must be capable of surviving transit through the human gastrointestinal tract. Absorption of dietary cholesterol into the bloodstream occurs predominantly in the duodenum of the small intestine, where the pH varies from pH 6 to 7 and bile salts excreted from the bile duct assist in solubilizing cholesterol [82,83]. *M. purpureus* CBS 109.07 was cultured in media at physiological temperature and pH, and with a high bile salt concentration and low aeration and agitation. However we recognize the limitations of an in vitro study in reproducing gastrointestinal conditions. Additionally, the clinical safety of *M. purpureus* CBS 109.07 needs to be established—a potentially complicated issue if the restrictive regulations on *Monascus* pigments and red yeast rice supplements by the FDA and European Food Safety Authority are any indication [4,5,31,84]. In the current study, we are only beginning to raise the possibility of an application for *M. purpureus* CBS 109.07 in probiotics; we recognize that additional safety and gastrointestinal survival experiments are required, and that such advancement in understanding *Monascus* biochemistry may improve the restrictions on their usage in the US and EU [85]. We also recognize that other candidate strains of *M. purpureus* or other *Monascus* species may be found [86], and that CBS 109.07 may not be unique or exemplary. However, we note that the human consumption of *M. purpureus* CBS 109.07 has precedents, as several food-grade studies have considered CBS 109.07 an edible filamentous fungus and used it as the representative *Monascus* strain in human and animal food applications of mycoprotein [6,47,48].

#### **5. Conclusions**

Our findings demonstrate that *M. purpureus* CBS 109.07, which can biosynthesize statin-like monacolins, can also reduce cholesterol content in vitro via a mechanism of cholesterol assimilation at 37 ◦C with a high concentration of bile salts. The most effective removal of cholesterol occurred in growing *M. purpureus* CBS 109.07 cultures, while non-growing *M. purpureus* CBS 109.07 minimally adhered to cholesterol and did not metabolize cholesterol. Dormant cultures, once transferred from buffer to nutrient rich media, were able to be resume cholesterol assimilation at levels observed in active cultures. Citrinin production under our experimental conditions was not detected. Our results show that it is valuable to continue examining the cholesterol-lowering potential of active *M. purpureus* CBS 109.07 cultures, as further research may provide a possible insight in the treatment of hypercholesterolemia and will draw attention to the significance of filamentous fungi in human health and nutrition.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2309-608X/6/4/352/s1, Table S1: Cholesterol content of *M. purpureus* CBS 109.07, Table S2. Cholesterol content, cholesterol assimilated, and dry weight of growing *M. purpureus* CBS 109.07.

**Author Contributions:** Conceptualization, T.P.T.N.; methodology, T.P.T.N.; validation, T.P.T.N.; M.A.G.; S.A.N.; C.E.S.; C.W.; formal analysis, T.P.T.N.; investigation, T.P.T.N.; M.A.G.; S.A.N.; C.E.S.; C.W.; resources, T.P.T.N.; data curation, T.P.T.N.; M.A.G.; S.A.N.; C.E.S.; C.W.; writing—original draft preparation, T.P.T.N.; writing—review and editing, T.P.T.N.; M.A.G.; S.A.N.; C.E.S.; C.W.; visualization, T.P.T.N.; supervision, T.P.T.N.; project administration, T.P.T.N.; funding acquisition, T.P.T.N. All authors have read and agreed to the published version of the manuscript.

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

**Acknowledgments:** We thank the members of the Chemistry & Biochemistry Department and the Natural & Applied Sciences College at Loyola University Maryland for support to carry out this project. We are especially grateful to Elizabeth E. Dahl for assistance with GC-FID and quantitative and statistical analyses, Courtney J. Hastings for assistance with HPLC-UV, Heather R. Schmidt for assistance with materials, and Timothy J. McNeese for helpful discussions.

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

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 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*
