Dietary Xanthan Gum Alters Antibiotic Efficacy against the Murine Gut Microbiota and Attenuates Clostridioides difficile Colonization.

Dietary fiber provides a variety of microbiota-mediated benefits ranging from anti-inflammatory metabolites to pathogen colonization resistance. A healthy gut microbiota protects against Clostridioides difficile colonization. Manipulation of these microbes through diet may increase colonization resistance to improve clinical outcomes. The primary objective of this study was to identify how the dietary fiber xanthan gum affects the microbiota and C. difficile colonization. We added 5% xanthan gum to the diet of C57BL/6 mice and examined its effect on the microbiota through 16S rRNA gene amplicon sequencing and short-chain fatty acid analysis. Following either cefoperazone or an antibiotic cocktail administration, we challenged mice with C. difficile and measured colonization by monitoring the CFU. Xanthan gum administration is associated with increases in fiber-degrading taxa and short-chain fatty acid concentrations. However, by maintaining both the diversity and absolute abundance of the microbiota during antibiotic treatment, the protective effects of xanthan gum administration on the microbiota were more prominent than the enrichment of these fiber-degrading taxa. As a result, mice that were on the xanthan gum diet experienced limited to no C. difficile colonization. Xanthan gum administration alters mouse susceptibility to C. difficile colonization by maintaining the microbiota during antibiotic treatment. While antibiotic-xanthan gum interactions are not well understood, xanthan gum has previously been used to bind drugs and alter their pharmacokinetics. Thus, xanthan gum may alter the activity of the oral antibiotics used to make the microbiota susceptible. Future research should further characterize how this and other common dietary fibers interact with drugs.IMPORTANCE A healthy gut bacterial community benefits the host by breaking down dietary nutrients and protecting against pathogens. Clostridioides difficile capitalizes on the absence of this community to cause diarrhea and inflammation. Thus, a major clinical goal is to find ways to increase resistance to C. difficile colonization by either supplementing with bacteria that promote resistance or a diet to enrich for those already present in the gut. In this study, we describe an interaction between xanthan gum, a human dietary additive, and the microbiota resulting in an altered gut environment that is protective against C. difficile colonization.

the gut to generate the beneficial effects described above. While the community as a whole may remain intact, diet modification can affect subsets of the community that are better suited to utilize the altered nutrient composition (3). This effect is most prominent in hunter-forager societies where seasonal dietary changes modulate the microbiota (4). In Western diets, a great emphasis has been placed on the types and abundance of host indigestible fiber polysaccharides that are only accessible by the microbiota, such as resistant starch, inulin, or the fibers naturally present in fruits, vegetables, and whole grains.
Dietary fiber promotes microbial short-chain fatty acid (SCFA) production. While SCFA profiles are unique from individual to individual, they provide a variety of benefits, including increased colonic barrier integrity and decreased inflammation (5)(6)(7)(8)(9)(10). Depending on the structure of the fiber backbone and side chains, polysaccharides select for unique taxa and, as a result, unique fermentation profiles (11). Several key species may be responsible for degrading the fiber's carbohydrate structure, the by-products of which go on to be metabolized by a number of additional taxa (12). Butyrate, an SCFA and product of fiber degradation, has been linked to increased gut barrier integrity and decreased inflammation (13)(14)(15). Fiber degradation and SCFA production are also associated with clearance of Clostridioides difficile, formerly known as Clostridium difficile, following fecal microbiota transfer (16). Switching mice to a high-fiber diet while colonized with C. difficile increased the SCFA concentrations and also cleared the infection (17). Since C. difficile infection represents a significant health care burden, characterizing how these polysaccharides shape the gut environment and impact C. difficile's ability to colonize will provide insight into how they might be used to improve patient outcomes.
Some polysaccharides included in food are added to alter texture rather than for nutritional benefit. Xanthan gum, synthesized by the bacterium Xanthomonas campestris, is a common food additive used as a thickener, particularly in gluten-free foods, where industrial production is worth approximately $0.4 billion each year. Xanthan gum structure consists of (1¡4)-linked ␤-D-glucose with trisaccharide chains containing two mannose and one glucuronic acid residues linked to every other glucose molecule in the backbone, with possible acetylation on the first branching mannose and 3,6-pyruvylation on the terminal mannose (18). These negatively charged side chains give xanthan gum its viscous, gel-like properties. Although not specifically included in foods for its potential prebiotic activity, bacteria can degrade xanthan gum to increase fecal SCFA concentrations (19,20). However, little is known about what bacterial taxa are involved in these transformations.
This study investigated the effect of xanthan gum on the bacterial composition of specific-pathogen-free C57BL/6 mice and its effect during an antibiotic model of C. difficile infection. Our goal was to (i) characterize the effects of xanthan gum on the mouse microbiota and (ii) characterize the effects of xanthan gum on C. difficile colonization. Surprisingly, we found that xanthan gum administration alters mouse susceptibility to C. difficile colonization by maintaining the microbiota during antibiotic treatment.

RESULTS
Xanthan gum maintains an abundance of microbial taxa during cefoperazone treatment. Using C57BL/6 mice, we tested the effects of xanthan gum on the microbiota using mouse models designed to study the effects of antibiotic perturbation. Since our initial goal was to study the effects of xanthan gum on C. difficile infection in mice, these models entailed multiple days of antibiotic treatment necessary to make the microbiota susceptible to C. difficile ( Fig. 1A; see also Fig. S1A in the supplemental material). Some mice were kept on a standard mouse chow diet; the rest were put on an equivalent diet supplemented with 5% xanthan gum.
In the cefoperazone mouse model, 16S rRNA gene analysis of mouse fecal samples revealed a baseline microbiota dominated by Bacteroidetes (ϳ45%) and Firmicutes (ϳ35%), with the remainder of the community composed of Actinobacteria, Proteobac-teria, and Verrucomicrobia ( Fig. 1B and C). After cefoperazone treatment of mice on standard chow, Lactobacillaceae predominated a fluctuating community, as evidenced by increased mean Bray-Curtis distances between time points. Although we observed higher dissimilarity in the xanthan gum chow group following antibiotics, microbial communities were significantly more similar in the xanthan gum chow group compared to the standard chow group, as measured by Bray-Curtis distances. These data also indicated that by day 23 the microbial community in the standard chow group had not returned to the preantibiotic baseline (mean Bray-Curtis distance from day 2 to day 23, 0.86) compared to the xanthan gum chow group (mean Bray-Curtis distance from day 2 to day 23, 0.52). However, the relative abundance of bacterial taxa remained similar after cefoperazone treatment in mice fed 5% xanthan gum (Fig. 1C). These protective effects are reflected in a significantly higher Shannon diversity and an absolute abundance of fecal bacteria in xanthan gum-fed mice following cefoperazone treatment compared to those on standard chow ( Fig. 1D and E). We observed similar antimicrobial activity against an Escherichia coli strain ECOR2 lawn from fecal extracts obtained during antibiotic administration between diet groups and no inhibitory activity in fecal extracts from non-antibiotic-treated control mice (see Fig. S5B in the supplemental material). These data suggest that high concentrations of xanthan gum prevent cefoperazone-mediated alterations of the mouse microbiota. To determine whether xanthan gum had a similar protective effect for other antibiotics, we also used an oral antibiotic cocktail model coupled with intraperitoneal (i.p.) clindamycin, which has also been shown to render mice susceptible to C. difficile colonization (21,22). However, the microbiota differences between chow groups were less pronounced ( Fig. S1B and C). Taken together, our results show that xanthan gum administration maintains both diversity and overall abundance of microbes in the gut during cefoperazone treatment.
Using the linear discriminant analysis effect size method (LEfSe), we identified 35 operational taxonomic units (OTU) that were significantly increased 2 days following the switch from standard to xanthan gum chow (Fig. S3). We also observed a shift in bacterial metabolism marked by significantly higher butyrate and propionate concentrations in mice on xanthan gum chow compared to those on standard chow (Fig. S4). No OTU abundances were identified as being significantly different when comparing the same time points in the standard chow group. After cefoperazone treatment, 4 OTU were increased and 80 OTU were decreased in the xanthan gum group (Fig. S4). In the standard chow group, only 1 OTU (Lactobacillus) significantly increased following cefoperazone treatment (Fig. S6). Unsurprisingly, 48 of the 112 OTU that were negatively correlated with cefoperazone treatment in the standard chow group were also negatively correlated in the xanthan gum group.
Xanthan gum-mediated microbiota protection limits C. difficile colonization. Two days after the mice were removed from cefoperazone, they were challenged with C. difficile strain 630g spores administered by oral gavage. By monitoring feces for CFU (both vegetative cells and spores), we observed approximately 1 ϫ 10 6 CFU/g feces C. difficile in mice on standard chow 1 day postgavage (day 15), which rose to 1 ϫ 10 7 for the duration of the experiment (Fig. 2). However, when on xanthan gum, only a small number of CFU was observed 1 day postgavage, but by day 4 (day 19) all mice had cleared C. difficile (Fig. 2). In the antibiotic cocktail model, C. difficile colonized mice on both standard and xanthan gum chow similarly until 7 days postgavage (day 15) when C. difficile colonization levels were significantly lower in the mice on xanthan gum chow (Fig. S6).

DISCUSSION
The use of dietary polysaccharides for their beneficial health effects, either directly on the host or indirectly through the microbiota, has been widely demonstrated (15,19). In the context of C. difficile, diet may play a role in pathogen evolution, such as with trehalose, or influence colonization resistance, such as with dietary fiber and zinc (17,(23)(24)(25). Dietary alteration may shape the intestinal environment by altering the nutri-ents available or by modulating the concentrations of compounds toxic to C. difficile, such as secondary bile salts. As a common food additive, xanthan gum's physicochemical properties are well known (18). However, its effects on the gut microbiota are poorly understood. Although we were not able to test whether xanthan gum enriches for fiber-degrading bacteria to increase colonization resistance, we did observe that xanthan gum interferes with the activity of orally administered antibiotics to protect mice from C. difficile colonization. These protective effects vary by type of antibiotic. While xanthan gum may have enriched for taxa capable of degrading it, these changes were minor compared to the much larger differences observed between diet groups during antibiotic treatment.
As a third-generation cephalosporin, cefoperazone has broad-spectrum efficacy (26,27). As a result, it is not surprising that, in the standard chow group, it had a significant impact on microbiota community structure. These results agree with previously published work on cefoperazone's ability to disrupt the murine gut microbiota and cause lasting alterations even 6 weeks after the cessation of treatment (28,29). As demonstrated here, diet can affect antibiotic efficacy in unexpected ways. Although both bacterial diversity and abundance were maintained in mice on xanthan gum, the similarities in OTU levels identified by LEfSe between the two groups indicates that cefoperazone affected the microbiota in both groups but was attenuated in the xanthan gum chow group. Since we observed similar antimicrobial activity in feces from each diet group, our data suggest that cefoperazone is still active in the feces from these mice. We also demonstrated that xanthan gum chow itself did not have any inhibitory effect directly on C. difficile (data not shown). These data indicate that cefoperazone retained antibiotic activity in the presence of xanthan gum, but its effect on the microbiota in vivo was somehow interfered with. This decreased antibiotic activity in the gut of xanthan gum-fed mice allowed the bacterial community to recover faster than in animals on standard chow.
By at least partially protecting the microbiota from the effects of cefoperazone, xanthan gum administration preserved colonization resistance to C. difficile. Coloniza- tion resistance comprises a variety of mechanisms, including the metabolism of bile salts and competition for nutrients (30). Microbially modified secondary bile salts inhibit C. difficile outgrowth much more than their primary precursors (31). Microbial metabolism mediates a variety of modifications to primary bile salts, including deconjugation by Lactobacillus and Bifidobacterium species, as well as 7␣-dehydroxylation by Clostridium species (32)(33)(34)(35). The lack of secondary metabolites produced by these taxa has been correlated with a lack of colonization resistance (31,(36)(37)(38). The indigenous microbiota also prevents C. difficile from establishing itself within the colonic environment by limiting the nutrients available for growth (39,40). A number of taxa, including the Lachnospiraceae, have been shown to provide resistance to C. difficile colonization, which may occur through niche competition (41,42). Despite increased SCFA concentrations immediately following xanthan gum administration, direct alterations of the microbiota by xanthan gum did not appear to affect colonization resistance on the day of C. difficile gavage since SCFA concentrations had returned to baseline levels. By protecting the microbiota during antibiotic treatment, xanthan gum likely maintained these metabolic mechanisms to exclude C. difficile from the gut. This suggested that while the community was altered, enough bacterial taxa remained to exclude C. difficile.
While we did not demonstrate a mechanism for xanthan gum's effect, its gel-like nature may interrupt the activity of antibiotics by altering their pharmacokinetics. Several large polysaccharides with negatively charged or polar side chains, such as hydroxypropylmethyl cellulose, mannan oligosaccharides, and guar gum, increase the excretion of cholesterol and bile salts in feces by limiting their absorption (43)(44)(45)(46)(47)(48)(49).
While not previously reported, xanthan gum may also interact with these compounds. Similarities between the chemical structures of these sterol ring-containing compounds and of cefoperazone may result in interactions between xanthan gum and the antibiotic. The greater efficacy of the antibiotic cocktail plus clindamycin model against the microbiota is likely due to varied interactions with the five antibiotics in addition to the effect of the i.p. injection of clindamycin. Although potential alterations to the bile salt pool by xanthan gum may have limited C. difficile germination, we observed more fecal CFU 1 day postgavage (day 15) than what we used to inoculate the mice on day 14, suggesting that any disruptions to enterohepatic circulation did not prevent germination since there was some vegetative cell outgrowth. Furthermore, we have previously observed that few spores (i.e., Ͻ100) are sufficient to infect antibiotic-treated mice, suggesting that even if only a limited number of spores germinated, the mice would still become infected (unpublished data). If xanthan gum did alter bile salt concentrations, the resulting changes would be more likely involved in inhibiting vegetative cells instead of preventing germination, since some spores would likely germinate in spite of the bile salt changes.
Polysaccharide-drug interactions are frequently explored as means to delay drug release in vivo. When mice consume xanthan gum in their chow, orally administered antibiotics may become trapped inside the gel formed by hydrated xanthan gum. Previous research has shown that xanthan gum would provide time-dependent release that occurs at a lower rate than other large, polar polysaccharides. For example, hydroxypropylmethyl cellulose requires three times the concentration to achieve similar drug binding levels as xanthan gum (50,51). The binding affinity of xanthan gum is pH-dependent, where a higher pH limits drug release due to the increased integrity of the polymer structure (51). Furthermore, environments with a higher ionic strength, as well as the presence of other polysaccharides, increases xanthan gum's drug retaining efficiency (52,53). Thus, the colonic environment would be conducive for high xanthan gum affinity for binding compounds such as cefoperazone due to its relatively higher pH.
In our study, dietary xanthan gum administration protected the microbiota during antibiotic treatment, leading to the exclusion of C. difficile from the gut. Although our study suggests that a common dietary polysaccharide interacts with the effects of antibiotics, there are several limitations that merit future research. Since few individuals will consume xanthan gum at the concentrations we used, titering in lower doses of xanthan gum to get closer to physiological levels would elucidate the effects of xanthan gum in a normal human diet. Future research should also characterize how polar polysaccharides such as xanthan gum interact with compounds in the gut. This would be important for understanding drug pharmacokinetics, as well as the impact of xanthan gum on bile salts and enterohepatic circulation. Further work characterizing this common food additive would provide a greater understanding not only of how it is degraded in the gut it but also the potential positive effects of its fermentative by-products.

MATERIALS AND METHODS
Ethics statement. The University Committee on Use and Care of Animals of the University of Michigan, Ann Arbor, approved all animal protocols used in the present study (PRO00008114). These guidelines comply with those set by the Public Health Service policy on Humane Care and Use of Laboratory Animals.
Animals and housing. We obtained 5-to 8-week-old male and female mice from an established breeding colony at the University of Michigan. These mice were originally sourced from Jackson Laboratory. We housed mice in specific-pathogen-free and biohazard AALAC-accredited facilities maintained with 12-h light/dark cycles at an ambient temperature of 22°C Ϯ 2°C. All bedding and water were autoclaved. Mice received gamma-irradiated food (LabSupply 5L0D PicoLab Rodent Diet, a gammairradiated version of LabSupply 5001 Rodent LabDiet) or an equivalent diet with 5% xanthan gum added (95% LabSupply 5001 Rodent LabDiet, 5% xanthan gum [Sigma]; gamma-irradiated by manufacturer). We housed mice in groups of two to five animals per cage, with multiple cages per treatment group.
All cage changes, infection procedures, and sample collections were conducted in a biological safety cabinet (BSC) using appropriate sterile personal protective equipment between cage contacts. The BSC was sterilized with Perisept (Triple S, Billerica, MA) between treatment groups. Gloves were thoroughly sprayed with Perisept between each cage and completely changed between groups. A description of the metadata for the mouse experiments, including the cage numbers and treatment groups, can be found in Table S1 in the supplemental material.
Xanthan gum-cefoperazone mouse model. To investigate the effect of xanthan gum on cefoperazone-treated mice, we switched mice to a diet containing 5% xanthan gum on day zero. Two days later, we gave mice 0.5 mg/ml cefoperazone (MP Biomedicals, Solon, OH) in the drinking water for 10 days, as previously described, to render the mice susceptible to C. difficile colonization (54,55). We changed the antibiotic-water preparation every 2 days. After 10 days of cefoperazone treatment, we switched mice to Gibco distilled water. We orally gavaged mice with between 10 2 and 10 4 C. difficile 630g spores or vehicle control (sterile water) 2 days after removing the mice from antibiotics. Spores were prepared as previously described and then suspended in 200 l of Gibco distilled water and heat shocked (55). Viable spores were quantified immediately after gavage using taurocholate cycloserine cefoxitin fructose agar (TCCFA) as previously described (55). To monitor infection severity, mice were weighed over the course of the model.
Xanthan gum-antibiotic cocktail mouse model. To investigate the effect of xanthan gum on an alternative antibiotic model (antibiotic cocktail with clindamycin), we switched mice to a 5% xanthan gum diet on day 0 and then put on an antibiotic cocktail (0.4 mg/ml kanamycin, 0.035 mg/ml gentamicin, 850 U/ml colistin, 0.215 mg/ml metronidazole, and 0.045 mg/ml vancomycin; Sigma-Aldrich) for 3 days in their drinking water, as previously described (21,22). On day 5, we removed mice from oral antibiotic administration and returned them to regular drinking water. On day 7, mice were given an i.p. injection of clindamycin hydrochloride (10 mg/kg). One day after the i.p. injection, we orally gavaged mice with between 10 2 and 10 4 C. difficile 630g spores and weighed the mice as described above.
Quantitative culture. We suspended fresh fecal pellets in sterile, prereduced Gibco phosphatebuffered saline (PBS; Thermo Fisher) using a ratio of 1 part feces to 9 parts Gibco PBS (wt/vol; Thermo Fisher, Waltham, MA). We serially diluted these suspensions, plated them on TCCFA, and then incubated the plates anaerobically at 37°C for 18 to 24 h before counting the colonies.
Fecal cefoperazone activity assay. We used fecal supernatant obtained from mice 6 days after the beginning of cefoperazone treatment (day 8). The fecal content was diluted by a factor of 10 in PBS to test its activity on a lawn of Escherichia coli strain ECOR2, which is susceptible to cefoperazone. Next, 10 l of supernatant was added to a 0.7-cm-diameter autoclaved Whatman filter paper (Sigma-Aldrich) disk and laid in duplicate onto a Luria-Bertani (LB) agar plate (BD Difco, Miller) streaked for lawn growth of E. coli. After incubation of the plates anaerobically at 37°C for 24 h, we measured the zones of inhibition (ZOI) and then confirmed these measurements after another 24 h of anaerobic growth. The ZOIs from samples were compared to those of fecal supernatant from mice not on antibiotics and to PBS controls.
Lipocalin-2 ELISA. Fecal supernatants were diluted 100-fold in PBS plus 0.1% Tween 20 (USB Corp., Cleveland, OH) and then tested using the standard protocol for the DuoSet enzyme-linked immunosorbent assay (ELISA) kit for Mouse Lipocalin-2/NGAL (R&D Systems, Minneapolis, MN). Sample concentrations were normalized to g of feces and analyzed in duplicate.
E. coli growth curve with cefoperazone. We grew E. coli strain ECOR2 overnight in LB broth (Difco LB Broth, Lennox; BD), pelleted the culture, and then resuspended it in fresh LB medium. We back-diluted this bacterial suspension into LB medium or LB medium containing 0.25% xanthan gum. Finally, we added cefoperazone to the growth medium before placing the cultures in a Sunrise microplate reader (Tecan, Switzerland) and monitoring growth for 48 h. Measurements of the optical density at 600 nm were automatically taken every 15 min, with 60 s of shaking immediately prior to measurement.
16S rRNA gene qPCR. We suspended fecal pellets in PBS as described above and centrifuged them at 6,000 rpm for 1 min. Then, 100 to 400 l of supernatant was removed for metabolite analysis. Using the sedimented fecal content, we performed DNA extractions using a DNeasy PowerSoil kit (Qiagen, Germantown, MD), according to the manufacturer's protocol. We immediately stored extracted DNA at Ϫ20°C until further use. We then performed qPCR on a LightCycler 96 thermocycler (Roche, Basel, Switzerland) using PrimeTime gene expression master mix (IDT, Coralville, IA) and a set of broad-range 16S rRNA gene primers (56). All fecal DNA was amplified in triplicate with E. coli genomic DNA standards in duplicate and negative controls in triplicate. The LightCycler reaction conditions were as follows: 95°C for 3 min, followed by 45 cycles of two-step amplification at 60°C for 60 s and 95°C for 15 s. The quantification cycle (C q ) values for each reaction were determined by using the LightCycler 96 software, and fecal DNA concentrations were determined by comparing C q values to the standards in each plate and normalizing them to each individual sample's fecal mass. We used Welch's two-sample t test to determine significance.
Short-chain fatty acid analysis. Portions (100 l) of fecal supernatants were filtered at 4°C using 0.22-m 96-well filter plates and stored at -80°C until analysis. We transferred the filtrate to 1.5-ml screw cap vials with 100-l inserts for high-performance liquid chromatography analysis (HPLC) and then randomized them. We quantified acetate, propionate, and butyrate concentrations using a refractive index detector as part of a Shimadzu HPLC system (Shimadzu Scientific Instruments, Columbia, MD) as previously described (8). Briefly, we used a 0.01 N H 2 SO 4 mobile phase through an Aminex HPX87H column (Bio-Rad Laboratories, Hercules, CA). The sample areas under the curve were compared to volatile fatty acid standards with concentrations of 40, 20, 10, 5, 2.5, 1, 0.5, 0.25, and 0.1 mM. Through blinded curation, we assessed baseline and peak quality and excluded poor-quality data where necessary.
DNA extraction and Illumina MiSeq sequencing. The detailed protocol for DNA extraction and Illumina MiSeq sequencing was followed as described in previous publications with modifications (55). Briefly, 200 to 300 l of 10-fold-diluted fecal pellets were submitted for DNA isolation using a MagAttract PowerMicrobiome DNA isolation kit (Qiagen, Germantown, MD). Samples were randomized into each extraction plate. To amplify the DNA, we used barcoded dual-index primers specific to the V4 region of the 16S rRNA-gene (57). Negative and positive controls were run in each sequencing plate. Libraries were prepared and sequenced using the a 500-cycle MiSeq V2 reagent kit (Illumina, San Diego, CA). Raw FASTQ files, including the appropriate controls, were deposited in the Sequence Read Archive (SRA) database (accession numbers SRX6897486 to SRX6897789).
Data processing and microbiota analysis. 16S rRNA gene sequencing was performed as previously described using the V4 variable region and analyzed using mothur. Detailed methods, processed read data, and data analysis code are described on GitHub (https://github.com/mschnizlein/xg_microbiota). Briefly, after assembly and quality control, such as filtering and trimming, we aligned contigs to the Silva v.128 16S rRNA database. We removed chimeras using UCHIME and excluded samples with fewer than 5,000 sequences. We binned contigs by 97% percent similarity (OTU) using Opticlust and then used the Silva rRNA sequence database to classify those sequences. Alpha-and beta-diversity metrics were calculated from unfiltered OTU samples. We used LEfSe to identify OTU that significantly associated with changes across diets and antibiotic treatments (58). We performed all statistical analyses in R (v3.5.2).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.

ACKNOWLEDGMENTS
We thank the Host Microbiome Initiative and Microbial Systems Molecular Biology Laboratory at the University of Michigan for support with the 16S rRNA sequencing. We also thank Thomas Schmidt and Kwi Kim for their help with the HPLC analyses and Matt Ostrowski for guidance with the project's direction.
This research was funded by the National Institutes of Health (5-U01-AI124255).