Genetic Variation of the SusC/SusD Homologs from a Polysaccharide Utilization Locus Underlies Divergent Fructan Specificities and Functional Adaptation in Bacteroides thetaiotaomicron Strains

Dietary polysaccharides play a dominant role in shaping the composition and functionality of our gut microbiota. Dietary interventions using these microbiota-accessible carbohydrates (MACs) serve as a promising tool for manipulating the gut microbial community. However, our current gap in knowledge regarding microbial metabolic pathways that are involved in the degradation of these MACs has made the design of rational interventions difficult. The issue is further complicated by the diversity of pathways observed for the utilization of similar MACs, even in closely related microbial strains. Our current work focuses on divergent fructan utilization pathways in two closely related B. thetaiotaomicron strains and provides an integrated approach to characterize the molecular basis for strain-level functional differences.

A key step toward assessing gut microbiota structure and dynamics is to define the diversity that encompasses this complex community. A gold standard for gut microbiota profiling is the 16S rRNA gene amplicon sequencing method, followed by either taxonomical classification or de novo clustering into operational taxonomic units (OTUs) at 97% identity. This identity cutoff value is based on the premise that phylogenetic relatedness can serve as a proxy for whole genomes and assumes similar functional profiles above this predetermined cutoff value (1). More-recent studies using novel single-nucleotide-resolution approaches for cross-sample comparisons of 16S rRNA gene amplicons such as oligotyping (2) and cluster-free sub-OTU analysis (3) showed that even sequences that share Ͼ99.0% identity at the 16S rRNA locus can occupy distinct ecological niches, thus representing unique strains or ecotypes with distinct functionalities. In recent years, an explosion in whole-genome sequencing of representative bacterial strains from diverse ecosystems has revealed substantial intraspecies gene content variation (4). Although the sizes of the "accessory genomes" differ greatly between species (5), it is safe to assume that the functions encoded by a species that is present in the gut microbiome would be highly strain dependent. Shotgun metagenomics allows higher resolution than the traditional marker gene approaches, thus offering a greater potential for identification of strain-level genomic variation (6). However, this approach too is limited by our current lack of biochemical data linking genetic variants to actual functional differences.
In our current study, we identified the genetic basis for the observed differences in fructan utilization capabilities between two B. thetaiotaomicron strains. The study highlights the need to combine omics-based techniques with traditional culture-based biochemical, structural, and molecular tools to identify and characterize the functional diversity encoded by our gut microbes. B. thetaiotaomicron VPI-5482, a type strain for the species, carries a polysaccharide utilization locus (PUL) responsible for fructan utilization (7). This previously characterized PUL encodes (i) family 32 glycoside hydrolases (GH32s) for glycosidic linkage degradation (8), (ii) outer membrane proteins (SusC/SusD homologs and a surface glycan binding protein [SGBP]) for glycan recognition and import (9), and (iii) a hybrid two-component system (HTCS) sensor regulator for activating expression of the PUL in response to dietary fructose (10). The PUL is specific for degradation and uptake of ␤2-6-linked fructans (levan) (7). In this study, we identified a closely related B. thetaiotaomicron strain (Bt-8736) that utilizes ␤2-1-linked fructans (inulin) instead of levan and that outcompeted B. thetaiotaomicron VPI-5482 in vivo, in the presence of dietary inulin. Whole-genome sequencing, followed by sequence analysis, was used to identify a putative fructan PUL in Bt-8736. The two fructan PULs differ in the genes that are known to encode fructan specificity. Using genetic and biochemical approaches, coupled with in vivo testing, we were able to demonstrate the inulin specificity of the Bt-8736 gene products. Further, the current work reveals that the mechanism of inulin utilization is distinct from levan use, as it occurs in the absence of any cell surface endo-acting enzymes. Instead, Bt-8736 can directly import inulin into its periplasm for intracellular degradation, possibly due to the short average chain length of inulin compared to levan and other polysaccharides. Our current work demonstrates that related genetic loci can encode diversified biochemical pathways in closely related gut commensal strains. The pathways incorporate enzymes and carbohydrate binding and import proteins with distinct substrate specificities, which could not have been predicted previously based on sequence data alone (11).

RESULTS
Bt-8736 encodes a fructan utilization PUL that confers inulin use. Many members of the Bacteroides genus carry a polysaccharide utilization locus (PUL) that encodes the apparatus necessary for uptake and degradation of fructose-based dietary polysaccharides, collectively known as fructans (7). In a previous study, using B. thetaiotaomicron VPI-5482 as a prototypic member of this genus, we defined the genetic and biochemical basis of fructan PUL specificity. That work demonstrated that the fructan PUL is variably conserved among the Bacteroides species, resulting in functional differ-ences in the fructan utilization profiles across the genus. The ecological relevance of these functional differences was demonstrated using gnotobiotic mice harboring defined, two-member gut communities, where dietary fructan dictated the community composition, depending on the fructan degradation capacities of individual members of the microbiota (7).
Here we have assessed whether strains within a given Bacteroides species display similar fructan utilization profiles. We carried out an in vitro growth analysis of B. thetaiotaomicron strains present in our culture collection, in the presence of Bacteroides minimal medium (MM) containing either levan (0.5% [wt/vol]) or inulin (0.5% [wt/vol]) as the sole carbon source. These two polysaccharides were selected because they represent two distinct glycosidic linkages (␤2-6 in levan and ␤2-1 in inulin) that are present in the fructan homopolymers and that are available to the gut microbiota. We identified one strain, Bt-8736, which, unlike all other strains tested (including the type strain, B. thetaiotaomicron VPI-5482), grew in inulin but not levan (Fig. 1A). To identify the genetic basis that defines this ␤2-1 specificity, the Bt-8736 genome was sequenced, Matching gene colors indicate homologs between the two genomes. The yellow box represents genes in the fructan PUL of B. thetaiotaomicron VPI-5482 that either lack orthologs in Bt-8736 or show a low degree of homology with the orthologous genes that are present; amino acid identity is indicated for orthologous genes. (C) Phylogenetic relationship of the 16S rRNA gene sequence (left) and the fructan PUL SusD homolog amino acid sequence (right) for indicated species and strains. Sequences from the two B. thetaiotaomicron strains are highlighted in red. Branch numbers represent percentages of bootstrap values in 1,000 sampling replicates, and the scale depicts branch lengths. and a single fructan PUL orthologous to the one present in the B. thetaiotaomicron VPI-5482 type strain was identified (Fig. 1B). Induction of this putative fructan PUL in the presence of fructose was confirmed by quantitative reverse transcription-PCR (qRT-PCR) (see Fig. S1 in the supplemental material).
A closer look at the PUL encoded by the Bt-8736 genome revealed that it contained an HTCS, a putative fructokinase and inner membrane monosaccharide importer, and GH32 genes that are orthologs of BT1759 and BT1765, supporting the idea of a fructan-associated role for this operon. Notably absent, relative to the B. thetaiotaomicron VPI-5482 strain, was an ortholog of BT1760, a GH32 cell surface endo-levanase, and an ortholog of BT1761, a surface glycan binding protein, both of which exhibit specificity for the ␤2-6 linkage found in levan (7). GH32 enzymes are known to be active only on fructans and to display either endo-activity or exo-activity against either ␤2-1-linked or ␤2-6-linked fructans (12). Our previous work had shown that the presence of the cell surface levanase of B. thetaiotaomicron VPI-5482 was critical for the ability of this strain to use the levan (7). On the basis of this information, our current data indicate that lack of a BT1760 ortholog contributes to the inability of Bt-8736 to utilize levan. Along with GH32s, the susC/susD homolog genes that encode the outer membrane TonB-dependent transporter (TBDT; SusC-like) and associated carbohydrate binding surface lipoprotein (SusD-like) (9, 13) also determine the specificity of the fructans that can be utilized. Accordingly, the SusC/D homologs from the fructan PULs of the three sequenced B. thetaiotaomicron strains (Bt-3731, Bt-7330, and B. thetaiotaomicron VPI-5482) (each with the confirmed ability to utilize levan) exhibit a high level of conservation (all with Ͼ97% amino acid identity in pairwise comparisons) (data not shown). On the other hand, the SusC/D homolog pair from the inulin-utilizing Bt-8736 strain showed a striking lack of conservation compared to B. thetaiotaomicron VPI-5482 (49% and 39% amino acid identity, respectively). Phylogenetic analysis of the divergent SusC/D homologs present in the two PULs revealed that, on the basis of the traditional 16S rRNA gene analysis, their genes indeed do not track with the evolutionary history of the B. thetaiotaomicron strains ( Fig. 1C; see also Fig. S2A). Furthermore, isothermal titration calorimetry (ITC) performed with a recombinant form of Bt-8736 SusD confirmed that the protein bound inulin but not levan (Fig. S2B).
To test whether the divergent SusC/D homologs dictate ␤2-1 linkage specificity in Bt-8736, we transferred the susC/susD homolog genes, along with the upstream intergenic region carrying a putative promoter sequence, into the levan-utilizing strain, B. thetaiotaomicron VPI-5482 (Fig. S3A). Interestingly, despite the observed low homology between the SusC/D homologs of the two PULs, the putative promoter sequence from Bt-8736 displayed 91% identity with the sequence of the corresponding promoter found upstream of the BT1763 gene in B. thetaiotaomicron VPI-5482, indicative of promoter conservation at the species level. The presence of the SusC/D homologs from Bt-8736 enabled inulin utilization by the resultant transgenic strain, Bt(8736-2), without affecting its ability to utilize levan, as tested by its growth in MM containing either inulin (0.5% [wt/vol]) or levan (0.5% [wt/vol]) ( Fig. 2A). Expression of the transferred Bt-8736 genes, along with the native fructan PUL genes of B. thetaiotaomicron VPI-5482 (fructokinase, BT1757; susC homolog, BT1763) was confirmed in MM-fructose (0.5% [wt/vol]), with fructose being a known inducer of the fructan PUL (7) (Fig. S3B). Notably, the growth kinetics of Bt-8736 and Bt(8736-2) in MM-inulin are reproducibly distinct; however, the basis for this difference remains to be determined. Further, transfer of the susC/susD homologs in a previously generated B. thetaiotaomicron VPI-5482 mutant (7), lacking the native hybrid two-component sensor regulator (HTCS) gene (BT1754), prevented growth on inulin (and other fructans; data not shown) (Fig. 2B).
Together, these data confirm that the fructan PUL from Bt-8736 lacks the genes required for effective levan use. Instead, the divergent susC/susD homolog genes confer inulin specificity to Bt-8736, and to Bt(8736-2), when they were transferred along with their native promoter region.
The susC/susD homolog genes from the fructan PUL of Bt-8736 improve accessibility of periplasmic exo-GH32s to inulin. To provide insight into the mechanism of inulin use by Bt-8736 and the transgenic type strain Bt(8736-2), the cells were grown in MM-fructose to the mid-log phase in order to induce the expression of their fructan PULs. Washed cells were then incubated in phosphate-buffered saline (PBS) (1ϫ) containing inulin (0.5% [wt/vol]) under growth-arresting aerobic conditions and were monitored for cell surface inulin degradation activity via the appearance of mono-, di-, tri-, and oligosaccharides using high-performance anion-exchange chromatography with pulsed amperometric detection (HPAEC-PAD). The HPAEC data showed that incubation of the whole cells with inulin resulted in the generation of low levels of free fructose in all the three B. thetaiotaomicron strains tested ( Fig. 3A; see also Fig. S4). Notably, inulin degradation by these strains was not accompanied by the short-chain fructo-oligosaccharides generated by species that express endo-inulinases on the cell surface, such as Bacteroides caccae ( We next tested whether the generation of small amounts of free fructose contributes to the growth of the Bt-8736 and Bt(8736-2) strains on MM-inulin. B. thetaiotaomicron VPI-5482, which is able to utilize fructose as a sole carbon source, was cultured in reconstituted, filter-sterilized spent media, derived from mid-log cultures of Bt-8736 and Bt(8736-2) in MM-inulin (0.5% [wt/vol]). Spent media from B. thetaiotaomicron VPI-5482, an inulin nonutilizer, and B. caccae, which releases short-chain fructooligosaccharides, served as controls. Our data showed that the spent media derived from the B. thetaiotaomicron strains were incapable of supporting the growth of B. thetaiotaomicron VPI-5482 (Fig. 3B). However, addition of fructose at a 0.5% (wt/vol) final concentration to the reconstituted spent media allowed the growth of B. thetaiotaomicron VPI-5482, supporting that the level of extracellular fructose generated by the B. thetaiotaomicron strains during inulin degradation was not sufficient to support growth.
We then went on to test if the transgenic strain, Bt(8736-2), was reliant upon the native GH32s encoded by the B. thetaiotaomicron VPI-5482 genome for inulin degradation. We previously have shown that B. thetaiotaomicron VPI-5482 encodes three exo-acting fructosidases that release fructose from both ␤2-1-linked and ␤2-6-linked fructans (7). We tested the requirement of these endogenous GHs by transferring the susC/susD homologs from Bt-8736 into a B. thetaiotaomicron VPI-5482 mutant strain that lacked all three native GH32 genes (BT1759, BT1765, and BT3082). The resulting mutant, Bt(8736-2)⌬1759⌬1765⌬3082, was unable to grow on MM-inulin (Fig. 4A). Therefore, at least one of the GH32s from B. thetaiotaomicron VPI-5482 is required for inulin degradation, and the SusC/D homologs from Bt-8736 allow access of these enzymes to extracellular inulin. To test this idea, we generated GH32 double-knockout strains of B. thetaiotaomicron VPI-5482 such that each mutant expresses only one exo-acting GH32. Subcellular fractionation of these mutants, followed by enzymatic assays performed with inulin (0.5 [wt/vol] and 1% [wt/vol]; data combined), confirmed that inulin was being degraded in the periplasm and that BT1759 and BT3082 together were mainly responsible for the inulin degradation in these cells (Fig. 4B). Notably, Bt-8736 encodes close homologs of all three B. thetaiotaomicron VPI-5482 fructosidases but lacks any putative endo-acting inulinases. Collectively, these data support a model of inulin use in which the SusC/D homologs from Bt-8736 enable transport of inulin across the outer membrane, thus making the fructan accessible to the exo-acting GH32 enzymes localized in its periplasm.
The fructan PUL susC/D genes from Bt-8736 are sufficient to drive a dietdependent functional advantage in vivo. Differences in the ability to consume host dietary carbohydrates can dictate competitiveness within members of the gut microbiota, favoring those endowed with the capacity to utilize existing dietary microbiotaaccessible carbohydrates (MACs) (7). We therefore set out to determine whether the difference in inulin utilization capacity between Bt-8736 and B. thetaiotaomicron VPI-5482 resulted in a difference in bacterial fitness in vivo. Germfree mice (n ϭ 12) were cocolonized with equivalent quantities of Bt-8736 and B. thetaiotaomicron VPI-5482 and were fed a standard diet containing diverse MACs for 7 days. Four mice were then switched to a diet in which inulin was the sole carbohydrate source for 14 days (Fig. 5A). Mice were individually housed throughout the experiment to prevent cross inoculation, and the relative abundances of the strains were assessed using strain-specific primers by quantitative PCR. Seven days after cocolonization, Bt-8736 and B. thetaiotaomicron VPI-5482 were present at equivalent densities (47.5 Ϯ 3.5% and 52.5 Ϯ 3.5%, respectively). However, 1 week after switching to the inulin diet, Bt-8736 accounted for 98.6 Ϯ 0.1% of the total population. This increase in abundance was maintained through day 21 (Bt-8736 at 97.2 Ϯ 1.7% and B. thetaiotaomicron VPI-5482 at 2.8 Ϯ 1.7% [P ϭ 4.9 ϫ 10 Ϫ5 , day 21 versus day 7; Student's t test]) (Fig. 5B). The relative abundances of Bt-8736 and B. thetaiotaomicron VPI-5482 in cocolonized mice fed a standard diet were not statistically different even after 21 days (P ϭ 0.7, day 21 versus day 7; Student's t test, Fig. 5B). The susC and susD homologs of the fructan PUL from Bt-8736 were upregulated on day 21 (14 days after the introduction of dietary inulin) relative to day 7 (9.7 Ϯ 0.17-fold and 22.8 Ϯ 0.19-fold, respectively), indicating that Bt-8736 was actively sensing inulin and upregulating the inulin utilization machinery in the presence of dietary inulin. These data are consistent with a model in which the presence of a specific MAC may allow a bacterial strain to gain an advantage over a closely related strain within the gut due to small differences in gene content that confer the ability to utilize that MAC (7).
On the basis of our in vitro data, we then set out to test whether the presence of the ␤2-1-linkage-specific SusC/D homologs alone was sufficient to confer an in vivo advantage to a strain in the presence of dietary inulin. Germfree mice (n ϭ 8) were cocolonized with wild-type B. thetaiotaomicron VPI-5482 and the modified mutant strain harboring the Bt-8736 susC/D [Bt(8736-2)]. The experimental setup was similar to that outlined in Fig. 5A. B. thetaiotaomicron VPI-5482 and Bt(8736-2) started out at similar densities. Upon a shift to the inulin diet, Bt(8736-2) reached relative densities of 98.1 Ϯ 1.0% on day 14 and 99.3 Ϯ 0.4% on day 21 (P ϭ 4.6 ϫ 10 Ϫ4 , day 21 versus day 7; Student's t test) (Fig. 5C). Notably, Bt(8736-2) also exhibited a significant, though less striking, competitive advantage over the wild-type B. thetaiotaomicron VPI-5482 strain with the MAC-rich standard diet over 21 days [49.6 Ϯ 1.5% to 74.8 Ϯ 6.0% relative abundance of Bt(8736-2) on day 7 to day 21; P ϭ 7.6 ϫ 10 Ϫ3 , Student's t test], possibly reflecting the presence of ␤2-1-linked fructans in the standard mouse diet.
Our in vivo data confirmed that two wild-type strains of B. thetaiotaomicron, Bt-8736 and B. thetaiotaomicron VPI-5482, have divergent fructan degradation capabilities attributable to the different specificities in the SusC/D homologs encoded by their respective fructan utilization PULs and that this functional difference provides a competitive advantage to Bt-8736 in vivo in the presence of dietary inulin.

DISCUSSION
Our current work identified genetic variation in a polysaccharide utilization locus in two B. thetaiotaomicron strains, B. thetaiotaomicron VPI-5482 and Bt-8736, which results in differences in the type of fructans that are utilized by these two closely related organisms. These functional differences dictated interstrain competition in a simplified gnotobiotic gut community in the presence of dietary inulin. Using a multidisciplinary approach, we were able to demonstrate that a phylogenetically distinct susC/susD homolog gene pair that is present in Bt-8736 is the genetic determinant of fructan specificity. Our work highlights how structural differences present in dietary polysaccharides such as fructans can result in distinct molecular mechanisms for utilization of these polymers. Levan utilization by B. thetaiotaomicron VPI-5482 has an absolute requirement for a cell surface endo-levanase, BT1760, which degrades the polymer into shorter chains that are channeled through the SusC/D homolog transport apparatus (7). On the other hand, inulin utilization in Bt-8736 could proceed in the absence of an endo-inulinase, most likely due to the average chain length of inulin being shorter than that of levan, allowing importation without prior depolymerization (14). Such a mechanism would enable utilization of dietary inulin by minimizing its loss to the external environment (15). It was previously reported that the fructan PUL in Bacteroides ovatus encodes cell surface endo-inulinases that enable cross-feeding between the species that were mostly likely to cooperate in the human gut. Interestingly, even in this PUL, the endo-inulinases were dispensable for inulin use by B. ovatus, suggesting a similar mechanism for inulin uptake (16).
Our current work highlights the limitations of functional predications based on 16S rRNA gene relatedness alone. Integration of top-down genomics approaches with bottom-up culture-based microbiology and molecular genetics is proving to be a powerful approach to determine the vast diversity of PULs present in gut Bacteroidetes (15-17), many of which are involved in utilization of MACs. Given the significant role played by these MACs in determining gut microbiota structure and dynamics (18), dietary intervention using MACs has a great potential as a noninvasive method for modulating our gut microbes. A key method to enable accurate predictions is that of identifying and characterizing the diverse molecular mechanisms that underlie these MAC utilization pathways.
Carbohydrates used in the study. Glucose, fructose, sucrose, maltotriose, chicory inulin (average degree of polymerization [dp], Ն25) (for ITC), and levan (L8647) were purchased from Sigma. Orafti HP inulin (dp, Ն23), which was used for the rest of the studies, was obtained from the Beneo-Orafti group.
Genome sequencing and de novo assembly. Bt-8736 was sequenced by whole-genome shotgun sequencing using an Illumina system, which achieved 148-fold coverage. There were 25,128,539 reads that were trimmed to remove reads of poor quality, resulting in 24,896,432 reads with an average length of 35.6 nucleotides. A de novo assembly strategy using CLC Genomics Workbench in lieu of a reference assembly with respect to the previously sequenced strain (B. thetaiotaomicron VPI-5482) was used to maximize the likelihood of assembling regions that differed between the two strains. This strategy matched 24,660,279 reads into 376 contigs that ranged in length from 200 to 239,349 nucleotides and that had an average length of 15,744 nucleotides and an N50 of 53,056 nucleotides. The total genome size was 5.9 Mb.
Phylogenetic analysis. Amino acid sequences from Bacteroides SusC and SusD homologs and 16S rRNA were aligned and neighbor-joining phylogenetic trees were generated using MEGA 5.0 software. The accession numbers of all the amino acid sequences used for the analysis are provided in Table S2 in the supplemental material.
Genetic modification of B. thetaiotaomicron VPI-5482. The fructan PUL susC/susD homolog gene pair from Bt-8736, along with the intergenic 5= region containing putative promoters (Table S1), was PCR amplified using high-fidelity Platinum Pfx DNA polymerase (Invitrogen). The resulting PCR product was ligated into either a pNBU2-tetQb or a pNBU2-ErmGb vector, transformed by electroporation into Escherichia coli S17.1 -pir, and conjugated into B. thetaiotaomicron VPI-5482 (19). The resulting clones were screened by PCR and sequenced to confirm stable site-specific integration in either of the two sites located at the 3= end of a serine tRNA gene in B. thetaiotaomicron VPI-5482 (21). In-frame, nonpolar gene deletions for glycoside hydrolase 32 knockouts were generated using a counterselectable allele exchange protocol and specific primers (Table S1) (19).
Cloning and expression of recombinant SusD. Bt-8736 fructan PUL SusD, lacking its native signal sequences, was amplified from genomic DNA using the primers listed in Table S1. The construct was cloned into pET28b (Novagen), expressed in E. coli BL21, and purified by immobilized metal affinity chromatography as described previously (22).
Quantitative RT-PCR. Gene-specific primers were used to measure expression of the fructan PUL and 16S rRNA genes using the threshold cycle (ΔΔCT) method as described previously (Table S1) (19). In vivo expression of the Bt-8736s susC/susD homolog gene pair was quantified by normalizing the expression of each gene to the relative density of Bt-8736 in these samples. The assay was conducted using Brilliant III Ultra-Fast SYBR green (Agilent) in an MX3000P thermocycler (Stratagene).
Isothermal titration calorimetry. Isothermal titration calorimetry (ITC) was carried out at 25°C in 20 mM HEPES buffer (pH 8.0) essentially as described previously (22). Protein concentrations were between 50 and 100 M in the cell, and glycans (inulin or levan; Sigma) in the syringe were at 1% (wt/vol). Binding curves were fitted to a single-site model in Microcal Origin v7.0 with the "ligand in cell" option selected, as the molar concentration of binding sites on the polymeric glycans was not known.
Enzyme localization studies. Cultures (5 ml) grown on fructose (0.5% [wt/vol]) were harvested by centrifugation (OD 600 of~1.0). Cells were washed with PBS and then resuspended in 1 ml 20 mM Tris-HCl (pH 8.0) for whole-cell assays. Washed cells were also used to make the periplasmic fraction, which was prepared as follows. Washed cells from 5-ml cultures were resuspended in 5 ml of ice-cold 20% sucrose-20 mM Tris-HCl (pH 7.5)-1 mM EDTA and incubated on ice for 30 min before being pelleted by centrifugation. Cells were then resuspended in 1 ml ice-cold 1 mM MgCl 2 -PBS and centrifuged, and the supernatant was retained as the periplasmic fraction. This periplasmic fraction was then extensively dialyzed in PBS to remove contaminating sucrose. For enzyme assays, 100 l of washed whole cells or the periplasmic fraction was added to cuvettes containing PBS-washed cells and 0.5% inulin (final concentration)-20 mM Tris-HCl (pH 8.0) at 37°C and fructose release was monitored continuously using a modified fructose detection kit (Megazyme International). Activities of the periplasmic marker alkaline phosphatase and cytoplasmic marker glucose-6-phosphate dehydrogenase were compared to the results seen with lysed cells to ensure that no cell lysis/leakage had occurred.
Gnotobiotic mouse experiments. Germfree Swiss-Webster mice were maintained in gnotobiotic isolators and fed an autoclaved standard diet (Purina LabDiet 5K67) or a custom diet (Bio-Serv) (7), in accordance with the Administrative Panel on Laboratory Animal Care (A-PLAC), the Stanford University School of Medicine IACUC. Mice were bicolonized using a single oral gavage dose of 10 8 to 10 9 CFU of each strain from stationary-phase cultures grown in TYG. Fecal samples were collected on days 4, 7, 14, and 21, and total fecal DNA was extracted using the phenol-chloroform-isoamyl alcohol method, followed by a cleanup performed using a Qiagen DNeasy blood and tissue kit. Bacterial densities were determined by real-time quantitative PCR using strain-specific primers (Table S1).
Data availability. This Whole Genome Shotgun project has been deposited at DDBJ/ENA/GenBank under accession number QAJH00000000. The version described in this paper is version QAJH01000000. Note that NCBI genome classification has assigned Bt-8736 as a Bacteroides faecis based on wholegenome similarity. The 16S rRNA gene sequence of Bt-8736 and B. faecis are both Ͼ97% identical to the B. thetaiotaomicron type strain (24).