Pomegranate peel extract alters the microbiome in mice and dysbiosis caused by Citrobacter rodentium infection

Abstract Treatment of mice with a pomegranate peel extract (PPX) decreased the pathogenicity of Citrobacter rodentium (Cr) infections. Here, we investigate the effects of PPX on the microbiome of uninfected or Cr‐infected C3H/HeNCr mice by 16S rRNA gene sequencing. Mice were treated with water or PPX for 14 days, feces were collected, and then, the mice were infected with Cr and feces collected again at day 6 postinfection. DNA was isolated from the fecal samples and subjected to 16S rRNA gene sequencing to determine the microbial composition. Differences in the composition of the microbiome were observed for untreated and PPX‐treated mice with PPX mice having decreased diversity. PPX treatment decreased the Firmicutes/Bacteroidetes ratio by increasing Bacteroidetes and decreasing Firmicutes levels. The decrease in Firmicutes was driven by a large reduction in Lactobacillus. PPX treatment increased the abundance of Proteobacteria and Verrucomicrobiae and decreased Actinobacteria. The relative abundance of Cr reached 22% in water‐treated but only 5% in PPX‐treated infected mice. These results suggest that consumption of pomegranate polyphenols altered the microbiome, making it more resistant to displacement by infection with Cr, indicating that pomegranate polyphenols may mitigate the pathogenic effects of food‐borne bacterial pathogens.

2010). Analysis of data from 16S rRNA gene sequencing showed a significant increase in the family Ruminococcaceae in feces from DSS-treated rats receiving pomegranate extract compared with DSS-treated rats receiving water (Kim, Banerjee, Sirven, & Minamoto, 2017).
Citrobacter rodentium (Cr), an E. coli-like bacterium that naturally infects mice, is a robust model to study bacterial pathogenesis and health benefits of food additives and their impact on the microbiota because it replicates many aspects of enteropathogenic bacterial infections in humans, such as the clinically important gastrointestinal pathogens enteropathogenic E. coli or enterohemorrhagic E. coli (Collins, Keeney, et al., 2014). Infection with Cr produces attaching and effacing lesions in the colon, epithelial cell proliferation, an uneven apical enterocyte surface, crypt dilation and hyperplasia, increased cellularity, and mucosal thickening (Collins, Keeney, et al., 2014;MacDonald, Frankel, Dougan, Goncalves, & Simmons, 2003).
Cr growth is restricted to the colon with little bacterial translocation to systemic compartments, except in highly susceptible mouse strains (Vallance, Deng, Jacobson, & Finlay, 2003). Several lines of evidence suggest that the microbiome plays a significant role in determining the severity of Cr infection. Fecal transplant experiments demonstrated that highly susceptible mice can be rendered less susceptible by transplantation of the microbiome from resistant mice (Ghosh et al., 2011;Willing, Vacharaksa, Croxen, Thanachayanont, & Finlay, 2011). Administration of a Lactobacilli-enriched culture to mice reduced Cr disease severity (Vong, Pinnell, Maattanen, & Yeung, 2015) while dysbiosis induced by feeding a fiber-free diet exacerbated disease activity (Desai et al., 2016

| Pomegranate peel extract (PPX) preparation and treatment of mice
Pomegranate peel extracts were prepared and the composition determined by HPLC (John et al., 2017) as described previously (Smith et al., submitted). In experiment 1, two groups of five mice were given PPX by oral gavage, either 0.2 ml twice a day (morning and afternoon) or 0.3 ml once a day on weekends. To serve as controls, two groups of five mice were given water by oral gavage. Mice were treated for 2 weeks, and then, fecal pellets were collected to determine the microbiome composition using 16S rRNA gene sequencing.
In experiment 2, two groups of mice (n = 4-5) were treated for 14 days with water or PPX and fecal pellets collected to examine microbiome composition, and then, the mice were infected with Cr. For the infection, a nalidixic acid-resistant mutant of Cr strain DBS100 (ATCC 51459) was used. A frozen stock of Cr was streaked out on an LB agar plate and used to inoculate LB broth that was grown overnight at 37°C with shaking. The following morning, the culture was expanded and grown to an OD 600 of approximately 1.5. The culture was harvested by centrifugation and resuspended in LB broth. After fasting for 4-6 hr, mice were infected by oral gavage with 0.2 ml (5.0-10 × 10 9 cfu) of the bacterial suspension. The dose was confirmed by retrospective plating. At day six postinfection, when Cr colonization is at peak levels, fecal pellets were collected to study the microbiome composition. At day 7 postinfection, fecal pellets were collected, homogenized, and serial dilutions plated on LB/nalidixic acid agar plates to determine the level of Cr fecal excretion.

University of Maryland Institute for Bioscience and Biotechnology
Research DNA Sequencing laboratory for processing and sequencing on a HiSeq 2500 rapid run using the PE300 protocol (Experiment 2). For in-house library preparation, the quality and size of the libraries were verified by using the Agilent DNA 1000 Kit and run on the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). The NEBNext® Library Quant Kit for Illumina® (New England Biolabs, Ipswich, MA, USA) was used to quantify each library by using the Mx3005P QPCR System (Agilent Technologies, Santa Clara, CA, USA). Libraries were normalized with 10 mM Tris pH 8.5 to 4 nM before pooling equal volumes. The final library concentration was 6 p.m. with PhiX control v3 (15%, v/v) (Illumina, San Diego, CA, USA). Libraries were sequenced using an Illumina MiSeq sequencer and a 600-cycle, MiSeq® Reagent Kit v3.

| 16S reads and data analysis
The demultiplexed sequencing reads from the Illumina MiSeq or HiSeq sequencers were imported into the CLC Genomics Workbench F I G U R E 1 Alpha-and beta-diversity plots for water-or PPX-treated mice. Mice were orally gavaged with water or PPX for 14 days and fecal pellets collected for 16S rRNA gene analysis. Within sample, phylogenetic diversity was quantified using measures of alpha diversity, (a) Shannon entropy, and (b) Chao-1 bias-corrected plots. The mean diversity values are plotted in the inset, and there is a significant reduction in diversity in the PPX samples compared water. The bar graphs show the mean ± SE of values obtained with samples rarefied to 73,684 reads. Beta diversity measured using the (c) weighted UniFrac or (d) Bray-Curtis dissimilarity plot showed distinct separation of water and PPX samples with PCo1 accounting for 59% and 62%, and PCo2 13 and 11% of the diversity, respectively. Blue lines, spheres, and barswater-treated samples; Red lines, spheres, and bars-PPX-treated samples 11.0.1 (CLC Bio, Aarhus, Denmark) and processed accordingly. The quality trimming parameters were set to remove sequences below Phred Score 20, ambiguous nucleotides (maximal 2 nucleotides allowed), and sequences below 100 nucleotides.
The parameters for the open-reference-based operational taxonomic unit (OTU) clustering were set to default. Therefore, unique reads were mapped to the Greengene database (Greengenes v13_5; 97% similarity) with the addition to create de novo OTUs. The obtained OTU table for each experiment was filtered to remove all low abundance OTUs (<10). Before the estimation of alpha and beta diversity, the OTUs were aligned via MUSCLE v3.8.31 (Edgar, 2004) to generate a maximum-likelihood phylogenetic tree (neighbor-joining tree based on the Jukes-Cantor distances). Alpha diversity was measured using phylogenic index diversity, Chao-1 bias-corrected index, and Shannon diversity, whereas Bray-Curtis dissimilarity plot and weighted UniFrac were chosen for the beta diversity. Euclidean heatmaps were generated using CLC Genomics with complete linkage with a fixed number of features set to 200 and minimum count of 10.
Additional CLC Genomics statistical analysis included permutational multivariate analysis of variance (PERMANOVA). The OTU table obtained from CLC Genomics was imported into JMP Genomics.
The count data were TSS normalized and square root transformed (Hellinger transformation), and analyzed for significance using the Basic Expression Workflow with default settings and a FDR of 0.05.
Additional plots and statistics were generated using SigmaPlot 11.0.

| Effect of PPX treatment on the microbiome of uninfected mice
To determine whether treatment with PPX altered the microbiome, Interestingly, the Bray-Curtis dissimilarity plot shows a tight clustering of samples from water-treated mice but two distinct clusters for samples from PPX-treated mice that segregated based on cage of origin. No such difference was noted in the mice prior to the onset of treatment (data not shown), suggesting two slightly different microbiomes evolved in response to PPX treatment and this will be discussed further below.
The relative abundance of OTUs assigned to various bacterial phyla is shown in Figure 2. More than 85% of the OTUs were  The overall percent abundance and microbiota diversity at the class level are shown in Figure 3. Bacilli, Bacteroidia, and Clostridia comprised most of the bacteria from both water-and PPX-treated fecal samples. As was seen at the phylum level with Bacteroidetes, Bacteroidia were elevated by PPX treatment. In contrast, bacilli (essentially all Lactobacillales) were substantially reduced while Clostridia showed similar abundance in water-and PPX-treated fecal samples. Alpha-, Beta-, and Gammaproteobacteria were increased by PPX treatment, going from very low levels (<0.05%) in samples from water-treated mice to the 0.5%-1% range in samples from PPXtreated mice. Members of the class Erysipelotrichia also increased in response to PPX treatment.
The relative abundance at the order, family ( Figure S2), and genus levels is shown as heatmap correlation plots (Figure 4), and differences are summarized in Table S1. As can be seen in the maps, there are distinct patterns of abundance that differentiate samples from the water and PPX groups, indicating that PPX treatment leads to profound changes to the microbiome that included at the order level an increase in abundance of Burkholderiales, Erysipelotrichales, RF-32, Enterobacteriales, and Verrucomicrobiales, and a decrease in abundance of CW040, Lactobacillales, and RF39. At the family level, Alcaligenaceae, Bacteroidaceae, Erysipelotrichaceae, and Verrucomicrobiaceae were increased while Lactobacillacieae and Ruminococcaceae were decreased with these patterns persisting at the genus level ( Figure 4; Table S1).

| Effect of PPX treatment on the microbiome of Cr-infected mice
In the second experiment, mice were treated with PPX or water for 14 days, fecal samples were collected, and then, the mice were infected with Cr. At day six postinfection, when peak colonization has been achieved, additional fecal samples were also collected for analysis. As observed in experiment 1, the overall diversity of samples from uninfected PPX-treated mice was lower than the diversity observed in samples from uninfected water-treated mice. However, no significant differences in diversity were detected in samples from Cr-infected mice treated with water (water-I) or PPX (PPX-I) ( Figure   S3a). In the beta-diversity plots ( Figure S3b), a clear separation of the four groups is shown (p < 0.001). The relative abundance of select phyla is shown in Figure 5. While PPX treatment increased the relative abundance of Bacteroidetes and Verrucomicrobia, levels essentially remained unchanged after infection in both water-I and PPX-I mice while Firmicutes decreased in water-I but not PPX-I mice. Although Proteobacteria were present at low levels in uninfected mice, a significant bloom occurred after infection, increasing in both water-I and PPX-I mice; however, the increase was 3-fold higher in water-I mice reaching 25% relative abundance. In general, samples segregated by treatment and infection in relative abundance heatmaps for orders, families ( Figure S4), and genera ( Figure 6). Table S2 shows the relative percent abundance of various genera. Several minor genera, including RF39, Peptococcaceae, Anaerostripes, Dehalobacterium, and Candidatus Arthromitus, were essentially absent in feces from PPX-I mice but present in water-I mice. While infected and uninfected mice had similar abundance of certain families and genera, there were also families and genera with altered abundance resulting from infection that was more pronounced in water-treated mice. As was observed in uninfected mice, Bacteroides was lower in water-I than PPX-I mice but increased in water-I mice from low levels in uninfected mice while it remained unchanged in PPX-I mice. As previously reported , Lactobacillus levels decreased significantly in response to infection in both groups, but as a percentage was higher in water-I F I G U R E 3 Relative abundance of bacterial classes in water-and PPX-treated mice. Fecal samples were analyzed for the relative abundance of various classes, and the results are shown as a stackplot in panel a. A bar plot showing statistical significance of different classes is shown in panel b. ***indicates p < 0.001 than PPX-I mice. Also, Coprococcus, Desulfovibrio, and Parabacteriodes were lower in water-I than PPX-I mice while Sutterella and Provetella increased in response to infection in water-I but not PPX-I mice.
Of the four classes comprising the phylum Proteobacteria that were found in the fecal samples, the increase in infected samples was mainly due to Gammaproteobacteria (to which Cr belongs). Bacilli levels F I G U R E 4 Heatmap showing the relative abundance of bacterial genera in water-and PPX-treated mice. Euclidean heatmaps were generated and show a unique pattern of abundance for various genera in water-(top blue bar) or PPX-treated (top red bar) mice. Note that for the PPX samples there is a set of genera that elevated in one but not the other cage decreased in both water-I and PPX-I mice but remained higher in water-I than PPX-I mice. Similarly, the relative abundance of the order Enterobacteriales, and family, Enterobacteriaceae (to which Cr belongs) was 25%. Twenty-two and five percent of the sequences mapped to

| D ISCUSS I ON
Polyphenols are micronutrients ubiquitously present in many fruits and vegetables that are thought to have health promoting properties. Pomegranate has been heavily promoted as a "super" food mostly due to its high polyphenol content. Pomegranate juice or peel extracts have been shown to provide protection against chemical-induced colitis (Kim et al., 2016;Larrosa et al., 2010;Rosillo et al., 2012). Pomegranate peel possesses higher concentration of polyphenols than the edible portion, and it reduced Cr-induced colonic damage in PPX-treated mice (Smith et al., manuscript submitted). Here, we demonstrate that oral administration of a PPX can significantly alter the host microbiome of the Cr-susceptible mouse strain, C3H/HeNCr, and reduce overgrowth of Cr.
Of significance, PPX treatment decreased the extent of the Cr bloom after infection from about 22% relative abundance in mice receiving water to 5% in mice treated with PPX. In contrast, the number of viable Cr/g of feces or tissue (data not shown) was the same suggesting similar absolute levels of Cr in the colon. The high relative abundance of Cr reported here is much higher than reported for C57Bl/6 mice (<1% at day 9 postinfection; ).
Infection of mice with Cr, but not Citrobacter jejuni, a bacterium that can colonize the mouse colon but does not induce pathology, caused a marked reduction in total intestinal microbes (Lupp et al., 2007), thus effectively raising the relative abundance of Cr. The fourfold increase in Cr abundance in water-I mice over PPX-I-treated mice suggests that PPX treatment made the microbiome more resistant to displacement by the subsequent infection with Cr, thus reducing the percentage of the total microbiome that Cr occupies rather than reducing the absolute number of Cr and correlates with the increased colon pathology observed in water-treated versus PPX-treated mice.
Alterations and/or reduction in the number of commensals may be deleterious since commensal bacteria have been shown to be important for control of Cr infections (Kamada, Chen, Inohara, & Nunez, 2013;Kamada et al., 2012;Lamas et al., 2018). The microbiota plays a significant role in the infection process, and studies indicate that it is possible to manipulate the microbiome through diet (Desai et al., 2016;Kamada et al., 2012), and probiotics (Collins, Chervaux, et al., 2014;Vong et al., 2015) to minimize the extent/severity of Cr infections. Furthermore, susceptibility to Cr is mouse strain dependent and fecal transplants between resistant and susceptible strains can alter susceptibility (Ghosh et al., 2011;Willing et al., 2011). Notably, protection of mice from lethal colitis was associated with increased levels of Bacteroidetes (Ghosh et al., 2011) and we previously observed decreased mortality and pathology in PPX-treated mice (Smith et al., manuscript submitted).
Here, we found that treatment with PPX significantly increased the presence of Bacteroidetes in both uninfected and infected mice while decreasing Firmicutes thus lowering the Firmicutes/ Bacteroidetes ratio. Increased levels of Bacteroides are generally considered beneficial. Monocolonization of germ-free mice with Bacteroides fragilis protected against DSS-induced acute colitis (Chiu, Ching, Wang, & Liu, 2014). Bacteroides also can provide a pool of short-chain fatty acids that can be utilized by the host and other metabolic substrates that can be used by other commensals (Wexler, 2007). In addition, Bacteroidales were shown to be sufficient to promote intraepithelial cell presence and cytokine secretion in the colon, thus promoting epithelial barrier function in response to Cr infection (Kuhn et al., 2017). Thus, at least part of the protective effects derived from PPX consumption is likely driven by the increase in Bacteroides.
F I G U R E 5 Comparison of the relative abundance of phylum in uninfected and infected water-and PPX-treated mice. Fecal samples were collected prior to and at day six postinfection for analysis. Bacteroidetes levels were affected by treatment but not by infection whereas Firmicutes decreased in water-treated mice in response to infection. There was a large bloom of Proteobacteria in water-treated mice that was significantly reduced by PPX treatment Diet can have significant effects on the microbiome. Feeding mice a high-fat diet led to a decrease in Bacteroidetes (Hildebrandt et al., 2009) and increases the Firmicutes/Bacteroidetes ratio (Turnbaugh, Backhed, Fulton, & Gordon, 2008). In addition, the ratio was increased in obese individuals and decreased with weight loss (Ley, Turnbaugh, Klein, & Gordon, 2006;Turnbaugh et al., 2008).

F I G U R E 6
Heatmap showing the relative abundance of bacterial genera in infected and uninfected water-and PPX-treated mice. Euclidean heatmaps were generated and show a unique pattern of abundance for various genera in water-(top blue bar) or PPX-treated (top red bar) mice. Infection shifted the pattern of abundance more in PPX-treated mice. Water-infected-cyan bar, PPX-infected-magenta bar F I G U R E 7 The effect of infection on the relative abundance of Citrobacter at the genus level and the level of fecal excretion of Cr in water-and PPX-treated mice. The relative abundance of the genus Citrobacter is significantly higher in water-versus PPX-treated mice (panel a). The level of Cr excretion on day 7 postinfection in feces is shown in panel b A cranberry extract, however, was unable to reverse the high-fat, high-sugar (HF/HS)-induced decline of Bacteroidetes (Anhe et al., 2015). In contrast, a decrease in the Firmicute/Bacteroidetes ratio was observed in mice fed a high-fat diet supplemented with oolong, green, or black tea polyphenols (Cheng et al., 2018; or grape polyphenols (Roopchand, Carmody, Kuhn, & Moskal, 2015). Mice fed a HF/HS diet alone or in conjunction with a commercial extract of pomegranate by-product (POMx, POM Wonderful) had very low levels (<0.01%) of the genus Bacteroides that dropped to undetectable levels upon POMx treatment and high relative abundance of the family S24-7 (58%), a member of the Bacteroidetes that was not altered by POMx treatment (Zhang et al., 2017). We also saw no effect of PPX treatment on the relative abundance of the family S24-7 but saw less of this family (approximately 16%) in both water-and PPX-treated mice and more Bacteroides that was significantly increased by PPX treatment.
Polyphenol consumption can significantly alter the microbiome.
Pomegranate ellagitannins were shown to reduce starch digestibility in vitro (Bellesia, Verzelloni, & Tagliazucchi, 2015) and may alter the amount of starch available for fermentation in the cecum and colon which can alter the microbiome and production of short-chain fatty acids (Wong, Souza, Kendall, Emam, & Jenkins, 2006). Feeding tea polyphenols increased the abundance of Parabacteroides, Bacteroides, and Prevotella, while significantly decreasing several genera such as Roseburia, Lactobacillus, Blautia, Anaerostipes, Shuttleworthia, Bryantella, and Lactococcus in the context of a HF/HS diet in diet-induced obese mice . A. muciniphilia has been associated with improved health outcomes, especially metabolic disorders (Derrien, Belzer, & Vos, 2017;Schneeberger et al., 2015). Dietary polyphenols have been reported to increase the prevalence of A. muciniphilia (Anhe et al., 2015;Roopchand et al., 2015), and in humans, this was associated with the ability to metabolize pomegranate ellagitannins to urolithin A (Henning, Summanen, et al., 2017;. However, POMx treatment of mice did not increase the relative abundance of A. muciniphilia while treatment with inulin did (Zhang et al., 2017). The same study showed that the Turicibacteraceae and Rumminococcaceae were increased by POMx treatment of mice fed a HF/HS diet. We saw an increase in A. muciniphilia in some but not all PPX-treated mice; thus, it is unlikely that it is associated with decreased Cr-induced pathology.
The relative abundance of Lactobacillus was surprisingly high in our experiments but dropped dramatically in response to PPX treatment. Polyphenols have been reported to positively or negatively affect the growth of some strains of Lactobacilli in vitro that was compound and species specific (Duda-Chodak, 2012;Tabasco et al., 2011). Red wine polyphenols increased Lactobacilli levels in obese human subjects (Moreno-Indias et al., 2016). As we observed with our extract, POMx enhanced the growth of Proteobacteria, as well as total bacteria, Bifidobacterium spp., and Lactobacillus spp. from human feces (Bialonska et al., 2010;Li, Summanen, Komoriya, & Henning, 2015) but inhibited the growth of B. fragilis, Clostridia, and Enterobaceriaceae . Crude pomegranate juice and peel extracts inhibited in vitro growth of pathogenic clinical isolates of E. coli and Staphylococcus aureus and other pathogenic bacteria (Howell & D'Souza, 2013;Pagliarulo et al., 2016) and swarming activity of Cr (John et al., 2017). However, fecal excretion and colonic tissue burden of Cr were not affected by PPX treatment (Figure 7b and Smith et al., submitted manuscript), indicating that PPX treatment was not inhibiting growth of Cr. While some Lactobacillus species including L. acidophilus and L. casei are considered probiotics, other species of Lactobacillus (e.g., L. reuteri) have been associated with obesity (Million et al., 2012). In patients with ulcerative colitis, seven Lactobacillus strains, including L. acidophilus, and L. casei, were decreased while L. crispatus, L. delbrueckii, and L. reuteri increased (Cui et al., 2016). Furthermore, mice treated daily with L. crispatus had more severe DSS-induced colitis whereas L. plantarum treatment improved DSS-induced colitis (Cui et al., 2016), suggesting that beneficial effects of Lactobacillus are species dependent and that decreased prevalence of Lactobacilli may not inherently be deleterious.
Somewhat surprising was the decrease in diversity observed in samples from PPX-treated mice compared to those from watertreated mice. Generally, increased diversity is considered to promote a healthy microbiome and dysbiosis is associated with decreased diversity (Levy, Kolodziejczyk, Thaiss, & Elinav, 2017). No difference in diversity between the two treatments was apparent after infection. PPX treatment did, however, result in a significantly altered microbiome as observed in the β-diversity plots and the abundance heatmaps from both uninfected and infected PPX-treated mice. This suggests that the microbiome induced by PPX is more resistant to displacement by Cr thus keeping the relative abundance of Cr lower resulting in less colonic pathology. Also, surprising was the decrease in abundance of Candidatus Arthromitus in PPX-treated mice. This bacterium was originally thought to only be associated with terrestrial arthropods but recent reports indicate that that it is a segmented filamentous bacterium (family Lachnospiraceae) (Bolotin et al., 2014;Thompson, Vier, Mikaelyan, Wienemann, & Brune, 2012) with immunomodulating properties (Gaboriau-Routhiau et al., 2009;Ivanov et al., 2009Ivanov et al., , 2008. Segmented filamentous bacteria are important for the development of Th17 immunity and resistance to Cr (Ivanov et al., 2009) but its reduction by PPX treatment had no apparent effect.
Gastrointestinal infections can induce dysbiosis, and the effect of Cr on the microbiome of C57Bl/6 mice fed standard chow has been investigated (Collins, Chervaux, et al., 2014;Hoffmann et al., 2009). Hoffman et al. (2009 found that the microbiome was more affected late after infection (day 14) than at day 9. In agreement with their findings, in water-I mice we saw reduced abundance of Lactobacillaceae after infection and an outgrowth of some Proteobacteria including Alphaproteobacteria and Gammaproteobacteria while Deltaproteobacteria decreased. In our case, the large increase in Gammaproteobacteria is primarily due to Cr, a member of this class and was much higher than reported by Hoffmann et al. (2009). C3H/HeN mice are more susceptible to Cr infection than C57Bl/6 mice (Vallance et al., 2003), and this may be related to the higher relative abundance of Cr after infection in C3H/HeN mice. The fact that we did not see all the same changes reported by Hoffman et al. (2009) may result from different mouse strains (with different susceptibilities to Cr infection) and sampling times as they showed distinct time-dependent changes. Collins, Chervaux, et al. (2014), Collins, Keeney, et al., 2014 reported that Rumminococcus decreased in response to Cr infection and loss of Rumminococcus is associated with susceptibility to DSS-induced colitis (Zenewicz et al., 2013). PPX, but not water treatment, resulted in a dramatic decline in Rumminococcus levels but this was not associated with increased Cr-induced pathology.
In conclusion, PPX treatment dramatically alters the microbiome of mice. PPX treatment decreases the Firmicutes/ Bacteroidetes ratio primarily by raising Bacteroidetes levels.
Several minor bacterial species also are altered in response to PPX treatment resulting in a modified microbiome. The PPX-induced microbiome is more resistant to overgrowth by Cr infection resulting in significantly lower relative abundance of Cr even though the number of viable Cr/g of feces is the same. Our results indicate that pomegranate peels, an agricultural waste product, may be a source of compounds that are beneficial for treatment of intestinal disorders but further research is required.

ACK N OWLED G EM ENTS
We would like to thank Kamal Chauhan for providing the PPX extract and Kollakondan Muthaiya.MariaJohn for assistance in the PPX compositional analysis.

CO N FLI C T O F I NTE R E S T
The authors have no conflicts of interest to declare.

E TH I C A L R E V I E W
The animal studies described in this manuscript were approved by the USDA/ARS Beltsville Institutional Animal Care and Use Committee.