A Citrus Fruit Extract High in Polyphenols Beneficially Modulates the Gut Microbiota of Healthy Human Volunteers in a Validated In Vitro Model of the Colon

The effect of a Citrus Fruit Extract high in the polyphenols hesperidin and naringin (CFE) on modulation of the composition and activity of the gut microbiota was tested in a validated, dynamic in vitro model of the colon (TIM-2). CFE was provided at two doses (250 and 350 mg/day) for 3 days. CFE led to a dose-dependent increase in Roseburia, Eubacterium ramulus, and Bacteroides eggerthii. There was a shift in production of short-chain fatty acids, where acetate production increased on CFE, while butyrate decreased. In overweight and obesity, acetate has been shown to increase fat oxidation when produced in the distal gut, and stimulate secretion of appetite-suppressive neuropeptides. Thus, the data in the in vitro model point towards mechanisms underlying the effects of the polyphenols in CFE with respect to modulation of the gut microbiota, both in composition and activity. These results should be confirmed in a clinical trial.


Introduction
The gastrointestinal tract (GI tract) consists of various organs, such as the stomach, small intestine, and large intestine. Within the GI tract, nutrients pass various chemical (stomach acidity, intestinal bile) and physical (mucus layer, intestinal epithelial cells) barriers and are digested by salivary, gastric, and pancreatic enzymes in order to be absorbed into the blood [1]. These barriers play a crucial role in maintaining intestinal homeostasis by providing protection against pathogens and simultaneously preserving a symbiotic relationship with commensal microorganisms. The collection of commensal microorganisms in the gut is called the gut microbiota and is unique for each individual. It consists of bacteria, fungi and yeasts, viruses and bacteriophages, and protozoa [2]. The four main bacterial components comprise the phyla Firmicutes, Bacteroidetes, Actinobacteria, and Proteobacteria. Differences in gut microbiota composition are observed due to genetics, age, diet, geographic origin, medication use, and disease [3,4]. Over the last few decades, the role of the gut microbiota in numerous diseases and disorders has been shown, not only including disease in the gut, such as inflammatory bowel disease and colon cancer, but also elsewhere in the body, such as allergy of skin and lungs, obesity, and brain-related disorders [5,6]. Overall, high microbiota diversity, stability of gut microbiota composition,

Collection and Preparation of Fecal Samples
Fecal samples from healthy volunteers (n = 7; 3 M, 4 F, average age = 32 years) were collected and homogenized under anaerobic conditions to create a standardized microbiota pool according to Aguirre et al. [27]. The fecal slurry was snap-frozen in liquid nitrogen and stored at −80 • C until further use. Before inoculation, 4 tubes of 30 mL of fecal slurry each were thawed for 1 h at 37 • C and subsequently mixed with prereduced dialysate to a total volume of 250 mL.

Experimental Set Up and the TIM-2 In Vitro Model
Each TIM-2 unit was inoculated with 60 mL of the standardized microbiota pool, as described in Section 2.2, and 60 mL prereduced dialysate. Subsequently, SIEM was administered to each unit (2.5 mL/h) for an adaptation period of 16 h. Thereafter, the units were continuously supplemented for 72 h with a constant flow of SIEM (2.5 mL/h) in three conditions: SIEM (control), SIEM + 250 mg citrus extract/day, SIEM + 350 mg citrus extract/day. Throughout the experiment, lumen samples were removed from the units after 24 and 48 h to simulate passage from the proximal to the distal colon. Lumen and dialysate samples were obtained at 0, 24, 48, and 72 h, and analyzed for metabolite production (see Section 2.4) and microbiota composition and activity.

SCFA, BCFA, and Organic Acid Production
SCFA, branched-chain fatty acids (BCFA), and organic acid production was analyzed through ion exclusion chromatography by Brightlabs (Venlo, The Netherlands) as described previously [26]. Briefly, lumen and dialysate samples were centrifuged at 14,000 rpm for 10 min, filtered, and diluted with 1.5 mM sulfuric acid. Next, 10 µL of this solution was added to a column, and analysis was started on an 883 chromatograph (IC, Methorm, Herisa, Switzerland).

Gut Microbiota Composition
Sequencing of the V3-V4 region of the 16S rRNA gene was performed to determine microbiota composition. In short, from the lumen samples, DNA was isolated, amplified with barcoding, pooled, and subsequently sequenced with the Illumina MiSeq sequencing system according to the manufacturer's instructions (Illumina, Eindhoven, The Netherlands). The Binary Base Call text-based format for storing biological sequence and corresponding quality scores pipeline (BCL2FASTQ, v. 1.8.3, Illumina, San Diego, CA, United States of America) was used to convert the sequences into text-based format for storing biological sequence and corresponding quality score (FASTQ) files after quality checking. Subsequently, Quantitative Insights Into Microbial Ecology 2 (QIIME-2) software was used to analyze the results [28]. Classification of the sequences into amplicon sequence variants (ASVs) was performed using the Silva database (version 132 (available online: https://www.arb-silva.de/documentation/release-132/ acessed on 27 October 2021)) as a reference 16S rRNA database.

Statistical Analysis
The software package R (version 3.6.2, R Foundation for Statistical Computing, Vienna, Austria (R Core Team, 2013; https://cran.r-project.org/bin/windows/base/old/3.6.2/ accessed on 27 October 2021)) was used for statistical analyses. To determine changes in the microbial community composition, various indexes were calculated and the abundances of microbial species in the total microbial community were calculated and shown as relative abundance (RA). Kruskal-Wallis analysis was performed to determine differences in species abundance between the treatment conditions and its dose dependence. The correlation of RA of ASVs and microbial metabolites were calculated with Spearman's correlation. Multiple comparisons were adjusted with the Benjamini-Hochberg false discovery rate, and q-values (FDR-corrected p-values) were considered significant at q < 0.20.

Changes in Microbiota Composition
Specific alterations in microbiota compositions between control and CFE addition were investigated. Since the inoculum was standardized by pooling [27], the different experimental conditions commenced with the same microbiota composition after the adaptation period at both the phylum and genus level ( Figure 1A,E, respectively). Overall, there was a slight increase in Bacteroidetes and a corresponding decrease in Firmicutes over time for the two CFE doses compared to SIEM ( Figure 1B-D).
as relative abundance (RA). Kruskal-Wallis analysis was performed to determine differences in species abundance between the treatment conditions and its dose dependence. The correlation of RA of ASVs and microbial metabolites were calculated with Spearman's correlation. Multiple comparisons were adjusted with the Benjamini-Hochberg false discovery rate, and q-values (FDR-corrected p-values) were considered significant at q < 0.20.

Changes in Microbiota Composition
Specific alterations in microbiota compositions between control and CFE addition were investigated. Since the inoculum was standardized by pooling [27], the different experimental conditions commenced with the same microbiota composition after the adaptation period at both the phylum and genus level ( Figure 1A,E, respectively). Overall, there was a slight increase in Bacteroidetes and a corresponding decrease in Firmicutes over time for the two CFE doses compared to SIEM ( Figure 1B-D).  Table S1 for the top 25 genera, together with the quantitative data for all genera.
At the genus level, when compared to control, supplementation with the CFE doses resulted in a significant increase in the relative abundance of the genera Enterococcus (q = 0.134; Figure 2A) and Roseburia (q = 0.134; Figure 2D) which both belong to the phylum Firmicutes, despite an overall slight reduction in Firmicutes over time ( Figure 1).  Table S1 for the top 25 genera, together with the quantitative data for all genera.
At the genus level, when compared to control, supplementation with the CFE doses resulted in a significant increase in the relative abundance of the genera Enterococcus (q = 0.134; Figure 2A) and Roseburia (q = 0.134; Figure 2D) which both belong to the phylum Firmicutes, despite an overall slight reduction in Firmicutes over time ( Figure 1).
Roseburia (q = 0.134; p = 0.04953) showed a dose-dependent increase upon addition of CFE, while B. eggerthii (q = 0.184, p = 0.2752) and E. ramulus (q = 0.134, p = 0.2752) presented a similar trend ( Figure 2). In contrast, the uncharacterized species of the Bacteroides S24-7 group showed an inverse dose dependency with higher production at the lower dose of CFE, and L. mucosae and Enterococcus seemed to follow this trend. Apart from these increases in taxa on both doses, a decrease in Ruminococcaceae UCG-014 was observed after CFE feeding, and an increase only with the low dose CFE was observed for Lachnospiraceae UCG-004 and Bifidobacterium longum subsp. longum ( Figure S2). Figure 3 shows the changes in relative abundance over time for B. eggerthii, Roseburia, and Enterococcus. Whereas for the two doses of CFE, the RA of these taxa increases over time (or remains relatively the same for the 250 mg dose for Roseburia), the RA of these taxa is reduced in the control (for Enterococcus to below the level of detection). This change over time for the uncharacterized species of the Bacteroidales S24-7 group, E. ramulus, and L. mucosae is shown in Figure S3. Essentially similar phenomena as observed for the taxa in Figure 3 are observed for the latter three taxa, although the graph is more inconsistent, with some peaks either up or down (e.g., Figure S3B,C, respectively). In addition, the relative abundance of the species Eubacterium (E.) ramulus (q = 0.134; Figure 2B), Limosilactobacillus (formerly Lactobacillus) (L.) mucosae (q = 0.198; Figure  2E), from the phylum Firmicutes, and Bacteroides (B.) eggerthii (q = 0.184; Figure 2C) and an uncharacterized species of the Bacteroides S24-7 group (q = 0.198; Figure 2F), from the phylum Bacteroidetes, significantly increased after supplementation with CFE.
Roseburia (q = 0.134; p = 0.04953) showed a dose-dependent increase upon addition of CFE, while B. eggerthii (q = 0.184, p = 0.2752) and E. ramulus (q = 0.134, p = 0.2752) presented a similar trend ( Figure 2). In contrast, the uncharacterized species of the Bacteroides S24-7 group showed an inverse dose dependency with higher production at the lower dose of CFE, and L. mucosae and Enterococcus seemed to follow this trend. Apart from these increases in taxa on both doses, a decrease in Ruminococcaceae UCG-014 was observed after CFE feeding, and an increase only with the low dose CFE was observed for Lachnospiraceae UCG-004 and Bifidobacterium longum subsp. longum ( Figure S2). Figure 3 shows the changes in relative abundance over time for B. eggerthii, Roseburia, and Enterococcus. Whereas for the two doses of CFE, the RA of these taxa increases over time (or remains relatively the same for the 250 mg dose for Roseburia), the RA of these taxa is reduced in the control (for Enterococcus to below the level of detection). This change over time for the uncharacterized species of the Bacteroidales S24-7 group, E. ramulus, and L. mucosae is shown in Figure S3. Essentially similar phenomena as observed for the taxa in Figure 3 are observed for the latter three taxa, although the graph is more inconsistent, with some peaks either up or down (e.g., Figure S3B,C, respectively).

Production of SCFA, BCFA, and Other Organic Acids
After 72h of continuous supplementation with CFE, cumulative production of the most abundant SCFA acetate, propionate, and butyrate was higher compared to the SIEM control condition for both the 250 and 350 mg CFE addition (Table 1; Figure 4). However, when corrected for the number of carbon-atoms present in acetate (2 carbon atoms), propionate (3 C-atoms), and butyrate (4 C-atoms), it is clear that the CFE primarily led to a shift in production of these acids towards more acetate. The fact that the same number of C-atoms is present in these metabolites (which were the major metabolites produced) is logical, as the amount of carbohydrate provided in each of the experiments was equal. Acetate has the lowest acid logarithmic dissociation constant (pKa) of the three major SCFA and a higher amount of acetate is expected to lead to greater inhibition of pathogenic microorganisms.

Production of SCFA, BCFA, and Other Organic Acids
After 72 h of continuous supplementation with CFE, cumulative production of the most abundant SCFA acetate, propionate, and butyrate was higher compared to the SIEM control condition for both the 250 and 350 mg CFE addition (Table 1; Figure 4). However, when corrected for the number of carbon-atoms present in acetate (2 carbon atoms), propionate (3 C-atoms), and butyrate (4 C-atoms), it is clear that the CFE primarily led to a shift in production of these acids towards more acetate. The fact that the same number of C-atoms is present in these metabolites (which were the major metabolites produced) is logical, as the amount of carbohydrate provided in each of the experiments was equal. Acetate has the lowest acid logarithmic dissociation constant (pKa) of the three major SCFA and a higher amount of acetate is expected to lead to greater inhibition of pathogenic microorganisms. C-atoms is present in these metabolites (which were the major metabolites produced) is logical, as the amount of carbohydrate provided in each of the experiments was equal. Acetate has the lowest acid logarithmic dissociation constant (pKa) of the three major SCFA and a higher amount of acetate is expected to lead to greater inhibition of pathogenic microorganisms.  Next to the SCFA acetate, propionate, and butyrate, production of several other organic acids was determined, such as lactate, formate, succinate, valerate, and caproate Next to the SCFA acetate, propionate, and butyrate, production of several other organic acids was determined, such as lactate, formate, succinate, valerate, and caproate ( Figure 5) as well as the branched-chain fatty acids (BCFA) iso-valerate and iso-butyrate ( Figure 6).     Production of lactate, a precursor for propionate, was increased by nearly 100% after 72 h of 350 mg CFE supplementation, compared to control. A similar pattern was observed for valerate production, with a >300% increase after addition of 350 mg/day CFE and an 85% increase after addition of 250 mg/day CFE. Iso-valerate and iso-butyrate production was increased by both concentrations of CFE, while formate, a substrate for acetate production via the Wood-Ljundahl pathway, was decreased. Succinate and caproate production were decreased after 250 mg/day CFE supplementation.

Discussion
In the last few decades, the modulation of the composition and activity of the intestinal microbiota by polyphenols, such as flavonoids, has been a topic that has received increasing attention from the scientific community. In this study, we aimed at investigating the effect of supplementation with CFE on gut microbiota composition and activity using a validated, dynamic, computer-controlled in vitro model: TIM-2.

Changes in Microbiota Composition
Changes over time in microbiota composition were found between control and our study product. After CFE supplementation, genera belonging to the phylum Firmicutes (i.e., Enterococcus and Roseburia) and Bacteroidetes (i.e., Bacteroides) showed an increase in relative abundance compared to the control. Enterococcus has been shown before to be stimulated by polyphenols from grapes in a study with broiler chickens [29]. Roseburia has

Discussion
In the last few decades, the modulation of the composition and activity of the intestinal microbiota by polyphenols, such as flavonoids, has been a topic that has received increasing attention from the scientific community. In this study, we aimed at investigating the effect of supplementation with CFE on gut microbiota composition and activity using a validated, dynamic, computer-controlled in vitro model: TIM-2.

Changes in Microbiota Composition
Changes over time in microbiota composition were found between control and our study product. After CFE supplementation, genera belonging to the phylum Firmicutes (i.e., Enterococcus and Roseburia) and Bacteroidetes (i.e., Bacteroides) showed an increase in relative abundance compared to the control. Enterococcus has been shown before to be stimulated by polyphenols from grapes in a study with broiler chickens [29]. Roseburia has been shown to be increased by polyphenols (amongst others from grape seed [30,31] red apple [32], and tea [33,34]), as well as reduced (e.g., by soy isoflavones [35], decaffeinated green and black tea [36], or pomegranate peels [37].) Moreover, previous studies evaluating microbiota composition in overweight subjects have reported beneficial effects of polyphenol intake through red wine consumption [38] and sorghum bran intake [39] on Roseburia growth. Roseburia is one of the most abundant intestinal butyrate-producing bacteria and has been linked with a reduction in inflammation and anti-obesity effects [40]. In addition, administration of naringenin in a letrozole-induced polycystic ovary syndrome model in rats has shown a positive impact on the relative abundance of the genus Roseburia [41]. Moreover, in a randomized controlled human intervention trial, three weeks of supplementation with virgin olive oil enriched with 500 mg phenolic compounds increased abundance of Roseburia, though it did not reach statistical significance [42]. E. ramulus has the capacity to metabolize numerous polyphenols, through ring-opening by use of the enzyme chalcone isomerase [43], as well as further reductive metabolism by an NADHdependent reductase [44,45]. Through these activities and the metabolites produced as a consequence, E. ramulus is thought to contribute to alleviation of obesity [46]. L. mucosae is Nutrients 2021, 13, 3915 9 of 13 one of the well-known bacteria capable of converting the soy isoflavone daidzein into equol and/or O-desmethylangolensin metabolite [47]. Similarly, it may be able to degrade other polyphenols, in casu the CFE used here. Equol, as well as hesperidin, has been linked with bone loss prevention and serum and a decrease in hepatic lipids [48,49]. Lastly, B. eggerthii is one of the major phenylpropanoid-derived metabolite producers in the gut, and these metabolites are frequently found upon polyphenol metabolism, although B. eggerthii has been shown to produce these from other aromatic substrates, such as phenolic amino acids [50].
The amount of hesperidin in the 250 and 350 mg CFE doses translates into approximately 0.45-0.63 L of orange juice [51]. Not only by the consumption of citrus extract but also by consumption of citrus fruit juice seems to have a positive effect on microbiota composition. In a controlled clinical study, ten healthy women were evaluated after continuous consumption of commercial pasteurized orange juice for two months. The authors showed that orange juice affected the growth of intestinal bacteria (mainly for Lactobacillus spp. and Bifidobacterium spp.). These results suggest a prebiotic effect of daily consumption of orange juice, with a positive effect on the intestinal microbiota and metabolic biomarkers [52]. No data on SCFA were presented in this study.

Production of SCFA, BCFA, and Other Organic Acids
Cumulative SCFA production from both CFE-fed microbiota was higher than the control, mainly in relation to acetate production. Acetate has been shown to be the main energy source for the liver and is also used for lipogenesis in adipose tissue, and oxidized by muscle and brain cells [53,54]. In addition, it has been shown that acetate is capable of reducing appetite via a homeostatic mechanism, through changes in the expression profiles of regulatory neuropeptides that favor appetite suppression [55]. Moreover, in overweight/obese men, acetate has been shown to promote fat oxidation and improve metabolic markers [17]. When cross-feeding mechanisms for conversion of acetate into butyrate do not occur, more acetate is produced and less butyrate. This might explain why butyrate production is observed to be lower after supplementation with CFE, despite observed increases in butyrate-producing taxa. Beneficial properties have also been attributed to propionate (for reviews see [13,56]), although high concentrations have also been linked to autism spectrum disorder [57,58]. Propionate is thought to be mainly metabolized in the liver, where it acts as a precursor for gluconeogenesis, thus influencing metabolic homeostasis [59]. Propionate also interacts with host receptors stimulating the release of satiety signals [60], and may therefore reduce body weight. In addition, immune-modulatory, and in particular, anti-inflammatory effects of propionate have been observed [56]. Of the SCFA, butyrate has been studied the most. In multiple studies, it has been shown to be beneficial, as it is the preferred energy substrate of colonocytes and is thought to reverse colon cancer by induction of differentiation of transformed cells (for a review see [16]). In our study, Spearman correlations have shown that many butyrate-producing taxa correlated with butyrate production. The majority of bacteria with potential to produce butyrate belong to the phylum Firmicutes where the acetyl-coenzyme A (CoA) pathway is the most prevalent [61]. Studies have shown strong co-occurrence between mucolytic bacteria (i.e., Bacteroides spp. and Ruminococcus spp.) and butyrate producers (i.e., Anaerostipes caccae and Eubacterium spp.) [62][63][64] as a possible indication that these different microbial groups shared metabolic networks [65].
Regarding to the organic acids and BCFA production, CFE supplementation has shown an increase in cumulative production for lactate, valerate, iso-valerate, and isobutyrate. The organic acids are usually considered precursors or alternative end-products of acetate, propionate, and butyrate. As a result, variations in production could provide information on cross-feeding mechanisms that take place within the gut. The BCFA are markers for protein fermentation, as they are exclusively derived from the fermentation of the branched-chain amino acids [66]. Valerate and caproate are amongst others involved in cross feeding mechanisms, and are produced by extending propionate and butyrate with acetyl-CoA to produce valerate and caproate, respectively [67,68]. Moreover, production of these metabolites has been linked to protein fermentation [69]. In addition to their effects as precursors, beneficial signaling effects of lactate and valerate have been described recently. For instance, lactate has been shown to play a key role in multiple cellular processes, such as energy regulation, immune tolerance, memory formation, wound healing, ischemic tissue injury, and cancer growth and metastasis [70,71]. Valerate has been shown to protect for eczema, and protects against colitis and necrotic enteritis [72][73][74].

Conclusions
The citrus extract with 88.2% hesperidin and 6.5% naringin modulated the gut microbiota in a validated, dynamic in vitro model of the colon (TIM-2), both with respect to microbiota composition as well as microbiota activity. Roseburia, Eubacterium ramulus and Bacteroides eggerthii were dose-dependently increased. Metabolically, an increase in acetate was observed. Nevertheless, the three increased taxa did not correlate with acetate production, but Ruminococcaceae UCG-010 did. Several butyrate-producing taxa correlated with butyrate production, which overall was slightly lowered by the CFE treatment. Several beneficial traits have been ascribed to acetate, including anti-microbial activity against pathogens, increase in fat oxidation and increase in secretion of regulatory neuropeptides that favor appetite suppression. The latter two traits are beneficial for overweight and obese individuals. Although the validated in vitro model that was used has been shown on many cases to be predictive for the in vivo situation, it remains to be seen whether the CFE has a similar effect in human volunteers. This is currently under investigation.