ABSTRACT
The mammalian intestine provides nutrients to hundreds of bacterial species. Closely related species often harbor homologous nutrient utilization genes and cocolonize the gut, raising questions regarding the strategies mediating their stable coexistence. Here we reveal that related Bacteroides species that can utilize the mammalian glycan chondroitin sulfate (CS) have diverged in the manner in which they temporally regulate orthologous CS utilization genes. Whereas certain Bacteroides species display a transient surge in CS utilization transcripts upon exposure to CS, other species exhibit sustained activation of these genes. Remarkably, species-specific expression dynamics are retained even when the key players governing a particular response are replaced by those from a species with a dissimilar response. Bacteroides species exhibiting distinct expression behaviors in the presence of CS can be cocultured on CS. However, they vary in their responses to CS availability and to the composition of the bacterial community when CS is the sole carbon source. Our results indicate that diversity resulting from regulation of polysaccharide utilization genes may enable the coexistence of gut bacterial species using a given nutrient.
IMPORTANCE Genes mediating a specific task are typically conserved in related microbes. For instance, gut Bacteroides species harbor orthologous nutrient breakdown genes and may face competition from one another for these nutrients. How, then, does the gut microbial composition maintain such remarkable stability over long durations? We establish that in the case of genes conferring the ability to utilize the nutrient chondroitin sulfate (CS), microbial species vary in how they temporally regulate these genes and exhibit subtle growth differences on the basis of CS availability and community composition. Similarly to how differential regulation of orthologous genes enables related species to access new environments, gut bacteria may regulate the same genes in distinct fashions to reduce the overlap with coexisting species for utilization of available nutrients.
INTRODUCTION
The mammalian intestine is home to hundreds of gut bacterial species that compete for and cooperate for nutrient sources (1). The gut microbial community benefits the host by degrading otherwise indigestible dietary nutrients (2–4). Alterations in the gut microbial flora have been implicated in numerous disease states, including obesity and diabetes (5, 6). A dominant component of the gut microbiota is the phylum Bacteroidetes (7), which is comprised of a number of species that can break down a wide variety of complex polysaccharides (3). Bacteroides species coexist in the mammalian gut at high densities (109 to 1010 CFU per gram of feces) (8, 9) and share available nutrients that range from intestinal glycans to dietary plant polysaccharides (3). The vast representation of related Bacteroides species colonizing the intestine raises the question: what mechanism(s) drives cohabitation of gut microbial species in an environment in which they share nutrients?
In Bacteroides, the ability to utilize a particular polysaccharide is usually conferred by a cluster of genes organized into a polysaccharide utilization locus (PUL) (10). A given PUL typically encodes proteins that recognize, import, and enzymatically convert the polysaccharide into metabolites that can be broken down via glycolysis (11, 12). A PUL often harbors a gene specifying a regulator that responds to a particular polysaccharide or its breakdown products by altering the expression of PUL genes responsible for the utilization of that polysaccharide (13–16). The PULs mediating utilization of particular mammalian or plant glycans, such as alpha-mannan (17), mucins (18, 19), fructans (13), and xyloglucans (20), are conserved within related Bacteroides species. Some of this conservation may be attributed to the horizontal transfer of PULs between gut-colonizing species (21). While the conservation of PULs suggests that gut species share particular niches, PULs are not identical across species (13, 20). This raises the possibility of shared PULs specifying disparate activities, thereby supporting niche divergence and the coexistence of related Bacteroides species.
Bacteroides thetaiotaomicron is a prominent member of the mammalian gut microbiota due to its ability to utilize a variety of animal and plant polysaccharides as carbon sources (22, 23). We have previously established that the genes mediating utilization of the mammalian glycan chondroitin sulfate (CS) in B. thetaiotaomicron undergo a transcriptional surge (24) that speeds up acquisition of CS (16). This surge in transcription is characterized by a transient peak in CS utilization transcripts followed by a drop to lower steady-state levels, despite the continuous availability of CS (Fig. 1). The regulator of CS utilization genes (BT3334) is activated by a metabolic intermediate in CS breakdown and turned down when the levels and activity of the glucuronyl hydrolase (BT3348) that degrades its activating ligand rise. Even though the glucuronyl hydrolase BT3348 is the rate-limiting enzyme in CS breakdown, the constitutively high level of production of this enzyme actually disrupted the dynamics of PUL gene transcription and delayed growth on CS (16). These findings suggest that the ability of B. thetaiotaomicron to dynamically adjust transcription to the CS catabolic rate boosts CS utilization in environments with fluctuating nutrient levels.
B. thetaiotaomicron dynamically adjusts transcription of CS PUL genes to the CS catabolic rate. (Top) The porin BT3332 and its partner, BT3331, transport CS across the outer membrane and into the periplasmic space, where it is converted into unsaturated chondroitin disaccharides by the action of three distinct CS lyases (BT3324, BT3350, and BT4410). Unsaturated disaccharides are broken down by the glucuronyl hydrolase BT3348 into the monosaccharides N-acetylgalactosamine (GalNAc) and 5-keto-4-deoxyuronate (kdu). Particular unsaturated disaccharides that are substrates for the glucuronyl hydrolase serve as activating ligands that bind to the periplasmic domain of the hybrid two-component system regulator BT3334, which governs the transcription of CS utilization genes. (Middle) A cluster of BT3334-regulated genes from the CS PUL is also depicted (the schematic is not drawn to scale). (Bottom) Expression of BT3334-regulated genes, represented by the sulfatase 1 (BT3333) mRNA, initially increases when unsaturated disaccharides are produced and then falls when the levels and activity of the rate-limiting glucuronyl hydrolase rise. The mRNA levels depicted are normalized to a 1,000-fold dilution of 16S rRNA, and the indicated fold change (i.e., 30×) denotes the ratio of mRNA levels at 120 min postinduction to the levels before induction (16).
Species-specific polysaccharide foraging among gut bacteria is often attributed to differences in gene content (25). By contrast, it is presently unclear whether species with orthologous nutrient utilization genes differ in their response to a given polysaccharide. We now report that Bacteroides strains vary in the regulation of genes in the CS PUL. We determine that the temporal response to CS availability is a species-specific trait, with certain strains promoting a transient surge in CS utilization transcripts and others displaying sustained expression in response to CS. Surprisingly, the observed regulation is retained when the key genes that mediate the transcriptional surge in a given Bacteroides species are replaced by orthologs from species that exhibit sustained expression. The identified regulatory behaviors are not always conserved among evolutionarily related CS PULs, implicating ecological factors in shaping the dynamics with which PUL genes are transcribed. In addition, we establish that Bacteroides strains display differences in their response to CS availability and to the presence of other bacterial species when cocultured in the presence of CS as the sole carbon source. Our results suggest that regulatory changes in a PUL diversify the nutrient niches accessible to gut Bacteroides and that this diversification may enable the coexistence of strains utilizing a shared nutrient in the same habitat.
MATERIALS AND METHODS
Bacterial strains, plasmids, and growth conditions.B. thetaiotaomicron strains were derived from strain ATCC 29148 (VPI-5482) (22) and grown under anaerobic conditions at 37°C in tryptone-yeast extract-glucose (TYG) medium containing tetracycline (2 μg/ml), erythromycin (10 μg/ml), or gentamicin (200 μg/ml), when applicable. In this study, we used the following Bacteroides type strains (as recognized by the International Journal of Systematic and Evolutionary Microbiology): B. caccae ATCC 43185, B. ovatus ATCC 8483, B. cellulosilyticus DSM 14383, B. intestinalis DSM 17393, and B. plebeius DSM 17135. All experiments with Bacteroides strains were performed with cells grown anaerobically at 37°C in chemically defined minimal medium (25) supplemented with the appropriate carbon source (0.5% glucose, 0.5% CS, or 0.1% CS) and antibiotics, when required. Escherichia coli strains were derived from strain S17-1 and were grown in LB medium containing 50 μg/ml ampicillin, when applicable. All chemicals were purchased from Sigma (CS from shark cartilage; catalog number C4384). All strains and plasmids used in this study are listed in Table S1 in the supplemental material. All primers used in this study are listed in Table S2 in the supplemental material.
Bioinformatic analyses.BT3334 (regulator) orthologs among sequenced type strains of Bacteroides were initially identified using the NCBI BLASTp program, to locate putative CS PULs. Neighboring gene content at each locus was thereafter assembled manually using BLASTp to identify orthologs of BT3348, a SusC and SusD pair, and the sulfatase (BT3333) and CS lyase (BT3324 and BT3350) genes. Intervening regions between unassembled contigs were manually curated to locate missing or partially annotated genes. For plotting of relatedness trees, protein sequences were retrieved from NCBI using the BLAST program and a maximum likelihood tree was constructed using the MEGA5 program from a multiple-sequence alignment created by MUSCLE software (for the hybrid two-component system [HTCS], the scale bar was 0.2 amino acid substitution per site; for RecA, the scale bar was 0.02 amino acid substitution per site). Bootstrap values are indicated at the branch points. The HTCS tree is drawn to scale, with branch lengths being in the same units as those used for the phylogenetic tree constructed from the conserved RecA sequence. We note that B. fragilis, B. vulgatus, and B. coprocola lack orthologous CS PULs and that nonorthologous hits were included in the HTCS tree for comparison.
Growth assays.For steady-state growth evaluations, Bacteroides strains were grown in minimal medium containing 0.5% glucose, and then the culture was diluted 1:100 into minimal medium containing the desired carbon source (0.5% CS from shark cartilage or 0.5% glucose) and the absorbance at 600 nm was measured, while assays for determination of the numbers of CFU were also performed. B. intestinalis and B. cellulosilyticus displayed distinct and lower ratios of the optical density at 600 nm (OD600) to the numbers of CFU compared to those for the other three Bacteroides species, possibly reflective of differences in cell size. Thus, detailed comparisons of growth curves based on OD600 measurements for the different strains are not presented.
Time course gene expression analysis of Bacteroides induced with CS.Cells were cultured overnight in minimal medium with 0.5% glucose. A 1:50 dilution was used for subculture into the same medium for 3.5 to 6 h, depending on the species, and cells were grown to an OD600 of 0.3 to 0.4. Cells were harvested, resuspended in minimal medium lacking a carbon source, and induced with 0.5% CS. Two milliliters of culture was collected before (at the −5-min time point) and at various times after induction with CS and harvested by centrifugation (for 1 min), and pellets were frozen on dry ice. Frozen pellets were stabilized with the RNAprotect Bacteria reagent (Qiagen), and RNA was extracted with an RNeasy kit (Qiagen) for further analysis.
Quantitative real-time PCR.Quantification of transcripts or bacterial abundance was carried out by real-time PCR using the SYBR green PCR master mix (Applied Biosystems) in an ABI 7500 sequence detection system (Applied Biosystems). The mRNA abundance represented on the y axes in Fig. 5 to 7 was normalized to a 1,000-fold dilution of 16S rRNA abundance to account for cell density. Genomic DNA was used to generate standards for each primer pair used for measuring the amount of an mRNA. The BT3333 sulfatase gene mRNA was used as a representative transcript because the BT3333 gene was not disrupted in any of the examined mutants. The observed trends were also observed for transcripts corresponding to additional genes (e.g., BT3324 and BT3348) in experiments with complementation of B. thetaiotaomicron mutants. Culture lysate with experimentally determined numbers of CFU/ml was used to generate standards for the species-specific primer pairs to assign the apparent numbers of CFU/ml denoted in the graphs in coculture experiments (see Fig. 8). For all experiments, the results depicted are those from two technical replicates and two biological replicates.
Complementation of B. thetaiotaomicron mutants.The region containing the 200-bp sequence upstream of BT3348 was assembled upstream of the BACCELL01621 gene sequence and was cloned into the vector pNBUtet. The BACCELL01621 gene with a 200-bp native upstream sequence was also cloned in a similar fashion and used for complementation. The vectors were introduced into the B. thetaiotaomicron chromosome in a BT3348 mutant background by conjugation as described previously (10). The region containing the BACCELL01634/5 gene (the regulator gene annotation is incomplete and divided between two consecutive contigs), including a 200-bp upstream sequence, was cloned into vector pNBUtet and used for complementation of the BT3334 mutant.
Coculture experiments with Bacteroides.Bacteroides species cultured in TYG medium or chemically defined medium for 24 h were frozen, and the number of CFU per milliliter of the frozen stock was enumerated. Coculture experiments were performed in 1 ml prereduced minimal medium containing the appropriate carbon source (0.5% CS, 0.1% CS, or 0.5% glucose) or in TYG medium. Cultures were started with 2 × 104 CFU (for the assays whose results are presented in Fig. 8A to D) or 2 × 105 CFU (for the assays whose results are presented in Fig. 8E to H) of each species, with bacteria being combined in 1:1 ratios and grown for 24 h, followed by 1:100 subculture into fresh prereduced medium. This process of subculture (1:100) was continued every day for 4 to 5 days, and a 100-μl volume of cultures grown for 24 h was collected on various days after inoculation and frozen. Cells were lysed using the hot-shot lysis method, and quantitative real-time PCR was performed with species-specific primers to estimate the number of bacteria. The initial number of CFU/ml at time zero was determined by plating dilutions for colony counting. Standard curves from cultures with experimentally determined densities were used for estimating the number of CFU/ml of individual species in the coculture from the measurements obtained via quantitative real-time PCR. All experiments were performed twice and in biological replicates. Statistically significant differences in terminal densities, i.e., the numbers of CFU/ml at day 5, between a coculture and an individual culture (see Fig. 8A to D) and between cultures with 0.5% CS and cultures with 0.1% CS (see Fig. 8E to H) were determined using the paired two-tailed t test.
RESULTS
The CS PUL is prevalent and functional among members of the genus Bacteroides.We used a sequence homology-based approach to locate putative CS PULs in Bacteroides genomes, looking for orthologs of the transcriptional regulator BT3334, the SusC and SusD pair BT3332 and BT3331 mediating CS import across the outer membrane, and the glucuronyl hydrolase BT3348, specified by genes located in close proximity to one another (10). For this analysis, we focused on type strains of the corresponding Bacteroides species. The CS PULs, which are distributed throughout the genus Bacteroides, display synteny among Bacteroides genomes. Proteins encoded in the CS PUL demonstrate overall sequence divergence reflective of the phylogenetic relationships among Bacteroides species (Fig. 2; see also Fig. S1 in the supplemental material).
CS PUL is prevalent among Bacteroides species. The relatedness of CS PULs on the basis of the deduced amino acid sequence of the regulator (the HTCS BT3334 ortholog) is shown. The experimentally examined species are highlighted with colored branches: the blue branch denotes strains that promote a surge in CS utilization mRNAs, and the red branch denotes strains exhibiting sustained activation of CS utilization mRNAs. B. plebeius, whose branch is highlighted in pink, contains an incomplete complement of CS PUL genes. Note that B. fragilis, B. vulgatus, and B. coprocola lack orthologous CS PULs. For a phylogenetic tree based on the deduced amino acid sequence of the conserved recA gene, refer to Fig. S1 in the supplemental material.
Our analysis revealed two key findings. First, the sequences of proteins of Bacteroides species differ in their degree of identity to those specified in the B. thetaiotaomicron PUL. That is, the CS PULs of B. ovatus, B. caccae, and B. thetaiotaomicron have diverged from the CS PULs of B. cellulosilyticus and B. intestinalis (Fig. 2 and 3). However, the genomes of all four species encode functional CS breakdown proteins because they grow when CS is the sole carbon source (25) (Fig. 4). A notable exception is B. plebeius, where the putative CS PUL is not fully functional because it does not support growth on CS and also because CS failed to induce certain (but not all) genes in the CS PUL (see Fig. S2 in the supplemental material). Homologs of the CS lyase gene BT4410, which is located outside the CS PUL and required for normal CS utilization in B. thetaiotaomicron, are present in CS PUL-containing Bacteroides genomes also at a site distal to the CS PUL (Fig. 3) (16).
Gene content and organization of CS PULs in Bacteroides strains. Genetic loci correspond to CS PULs based on comparison of their sequences with the B. thetaiotaomicron sequence and on gene expression analysis. The PUL encodes two CS lyases (red), a SusC and SusD pair (black), an HTCS regulator (green), an unsaturated glucuronyl hydrolase, GH88 (blue), and two sulfatases (gray). Another CS lyase (orange) is conserved and located distal to the PUL. Hypothetical proteins predicted to localize to the outer membrane are depicted (purple) and absent from the B. cellulosilyticus and B. intestinalis genomes. The B. plebeius PUL contains the genes encoding sulfatase 1, the HTCS, and GH88 clustered together. The genes for putative orthologs of CS lyase 1 and the SusC and SusD pair are located elsewhere in the genome. The dotted gray symbols above the arrow denoting a B. plebeius gene (03801, 00710, and 02279) indicate genes not induced by CS in B. plebeius that are orthologous to genes induced by CS in other Bacteroides species. Gene numbers refer to identifiers provided by the NCBI Entrez database, and prefixes corresponding to the particular species have been omitted but are as follows: BACOVA (B. ovatus), BACCAC (B. caccae), BACCELL (B. cellulosilyticus), BACINT (B. intestinalis), and BACPLE (B. plebeius). The number below each gene indicates the percent sequence identity of the deduced amino acid sequence of the gene to its corresponding B. thetaiotaomicron ortholog.
Strains with divergent CS PULs are capable of utilizing CS as the sole carbon source. The growth of Bacteroides species subcultured 1:100 into minimal medium with 0.5% glucose for 20 h and 0.5% CS for 20 h and 45 h is indicated. Growth is represented as the OD600. The definitions of the species abbreviations are as follows: Btheta, B. thetaiotaomicron; Baccac, B. caccae; Bacova, B. ovatus; Baccell, B. cellulosilyticus; and Bacint, B. intestinalis.
Second, Bacteroides strains vary in the presence of additional open reading frames (ORFs) in the CS PUL (Fig. 3). For example, the B. thetaiotaomicron CS PUL includes three genes (BT3328, BT3329, and BT3330) encoding proteins of unknown function predicted to localize to the outer membrane. Whereas homologs of these genes are found in certain Bacteroides species, other species that can utilize CS lack these genes, indicating that they are not essential for CS breakdown. These genes display high sequence divergence and variable organizations within the CS PULs among Bacteroides genomes, suggesting that they confer species-specific functions.
CS-utilizing Bacteroides strains display distinct transcriptional responses to CS.The ability of closely related bacteria to explore distinct niches is typically ascribed to differences in gene content (26–29). For example, the lac operon, which enables the mammalian commensal E. coli to use lactose as the sole carbon source, is absent from the related enteric species Salmonella enterica serovar Typhimurium, which cannot grown on lactose as the sole carbon source (28). However, such ecological properties may result from the distinct regulation of genes that related species have in common (30–32). For instance, differences in quantitative properties, such as the level and timing of expression of polymyxin B resistance genes, mediate phenotypic differences in resistance to the antibiotic polymyxin B in enteric bacteria (33). Given that the dynamics with which the CS PUL is expressed are critical for rapid CS utilization in B. thetaiotaomicron (16), we wondered whether the expression dynamics displayed by B. thetaiotaomicron are conserved in B. cellulosilyticus, which contains a CS PUL encoding proteins with considerable sequence differences (i.e., the BT3334 homolog is only 70% identical at the amino acid level, which is much lower than the 91% identity displayed by the BT3334 homolog in B. ovatus).
When B. cellulosilyticus was shifted into minimal medium containing CS, the mRNA levels of CS utilization genes increased rapidly and monotonically reached steady-state levels within 60 min (Fig. 5). The mRNA levels of the B. cellulosilyticus susC-like gene, which represents the most highly induced transcript, continued to increase over the course of 120 min (Fig. 5C). This is in contrast to the findings for B. thetaiotaomicron, in which exposure to CS resulted in a rapid increase in the mRNA levels of CS utilization genes, peaking at 30 min and declining to lower steady-state levels by 60 min (Fig. 1). Thus, two Bacteroides species with related CS utilization loci differ in the temporal transcriptional response elicited by sudden CS availability: transcriptional surge versus sustained expression.
B. cellulosilyticus displays sustained activation of CS utilization genes. The mRNA levels of CS utilization genes normalized to a 1,000-fold dilution of 16S rRNA before and at 20, 40, 60, and 120 min after B. cellulosilyticus was shifted into minimal medium containing 0.5% CS are shown. (A) GH88 gene (BACCELL01621); (B) sulfatase 1 gene (BACCELL01636); (C) susC (CS) gene (BACCELL01637); (D) CS lyase 1 gene (BACCELL01643). Means and SEMs from two biological replicates are shown. The fold change indicated at the upper right of each panel denotes the ratio of the mRNA levels at 120 min postinduction to the levels before induction.
To ascertain the prevalence of the two temporal transcriptional responses among Bacteroides species, we examined the expression of CS utilization genes in strains sharing various degrees of identity with the B. thetaiotaomicron or B. cellulosilyticus genes (Fig. 2). We determined that when B. ovatus was shifted into minimal medium containing CS, the mRNA levels of CS utilization genes increased, reached a peak at 60 min, and then declined to lower levels by 120 min (Fig. 6A to D). Therefore, the transcriptional surge identified in B. thetaiotaomicron is retained in the closely related species B. ovatus. However, the kinetics with which CS utilization mRNAs are produced are delayed in B. ovatus (Fig. 6A to D). This delay could be responsible for the diminished growth of B. ovatus on CS in comparison to the growth of B. thetaiotaomicron (Fig. 4).
Related Bacteroides species differ in the expression dynamics of CS utilization genes. mRNA levels of CS utilization genes normalized to a 1,000-fold dilution of 16S rRNA prepared before and at 20, 40, 60, and 120 min after B. ovatus, B. intestinalis, or B. caccae was shifted into minimal medium containing 0.5% CS are shown. The data shown correspond to those for B. ovatus genes for GH88 (BACOVA01969) (A), sulfatase 1 (BACOVA01997) (B), SusC (CS; BACOVA01998) (C), and CS lyase 1 (BACOVA00969) (D); B. intestinalis genes for GH88 (BACINT01376) (E), sulfatase 1 (BACINT01362) (F), SusC (CS; BACINT01361) (G), and CS lyase 1 (BACINT01356) (H); and B. caccae genes for GH88 (BACCAC01536) (I), sulfatase 1 (BACCAC01519) (J), SusC (CS; BACCAC01512) (K), and CS lyase 1 (BACCAC01509) (L). The fold change indicated at the upper right of each panel denotes the ratio of mRNA levels at 120 min postinduction to the levels before induction.
The CS PUL from B. intestinalis clusters with that from B. cellulosilyticus. We determined that the temporal response of B. intestinalis to CS availability mimics that of B. cellulosilyticus: CS utilization genes increased rapidly and reached a plateau at between 60 and 120 min (Fig. 6E to H). Therefore, the four analyzed Bacteroides strains display two distinct dynamics of CS PUL expression. The lineage containing B. thetaiotaomicron and B. ovatus promotes a transcriptional surge in CS utilization mRNAs, whereas the B. cellulosilyticus lineage does not. This analysis suggested that Bacteroides strains closely related to B. thetaiotaomicron or B. ovatus would promote a transcriptional surge in CS utilization mRNAs. Surprisingly, the dynamics of CS PUL expression in B. caccae, which branches along with B. thetaiotaomicron (their regulators share 89% amino acid sequence identity) and is the most closely related to B. ovatus (the regulators are 92% identical) (Fig. 2), revealed a monotonic reach to steady-state levels (Fig. 6I to L), as opposed to the anticipated surge behavior. Moreover, the CS utilization genes were induced to lower levels (i.e., 40- to 90-fold versus 100- to 4,000-fold), even though the growth rate and cell densities achieved by B. caccae were comparable to those exhibited by other Bacteroides species (Fig. 4). These results demonstrate that, despite the sequence conservation of the CS PUL genes, the factors governing the transcriptional dynamics are not evolutionarily conserved, even within closely related Bacteroides strains.
Across-species complementation of the temporal transcriptional response to CS.The transcriptional surge displayed by the CS PUL genes in B. thetaiotaomicron requires not only the regulator BT3334 to promote transcription of the CS PUL genes but also the glucuronyl hydrolase BT3348 to cleave the unsaturated chondroitin disaccharides that serve as inducing ligands for BT3334 (16) (Fig. 1). We wondered whether the distinct kinetic behaviors displayed by B. thetaiotaomicron and B. cellulosilyticus when they experience CS were due to differences in the amino acid sequences of the regulator and/or glucuronyl hydrolase. To test this possibility, we engineered B. thetaiotaomicron strains by deleting either the regulator or the glucuronyl hydrolase gene and introducing a copy of the regulator or the glucuronyl hydrolase gene from B. cellulosilyticus (BACCELL01634/5 and BACCELL01621, respectively) at a different chromosomal location. This strategy enables the investigation of individual factors in an otherwise isogenic genetic background, and it was previously utilized successfully to examine the differential regulation of polymyxin B resistance genes in enteric bacteria (33).
B. thetaiotaomicron strains harboring either their own regulator (BT3334) or that corresponding to the ortholog from B. cellulosilyticus promoted a surge in the mRNA levels of CS utilization genes upon a shift into minimal medium containing CS (Fig. 7A). Likewise, the B. thetaiotaomicron strain expressing the B. cellulosilyticus glucuronyl hydrolase exhibited the characteristic surge in the mRNA levels of CS utilization genes (Fig. 7B). Thus, the B. cellulosilyticus regulator and glucuronyl hydrolase can substitute for their B. thetaiotaomicron orthologs, indicating that they retain the properties required for promoting a transcriptional surge (16). Even a B. thetaiotaomicron BT3348 variant harboring the B. cellulosilyticus glucuronyl hydrolase gene BACCELL01621 transcribed from the native B. cellulosilyticus promoter maintained the surge behavior (Fig. 7C), demonstrating that the B. cellulosilyticus promoter can undergo a transcriptional surge. Cumulatively, our findings indicate that the kinetic differences among Bacteroides strains cannot be ascribed to the individual players required for the transcriptional surge exhibited by B. thetaiotaomicron in response to CS (16).
B. cellulosilyticus genes rescue a transcriptional surge in B. thetaiotaomicron mutants defective in the BT3334 or BT3348 gene. (A and B) mRNA levels of CS utilization gene BT3333 normalized to a 1,000-fold dilution of 16S rRNA before and at 40 and 120 min after B. thetaiotaomicron was shifted into minimal medium containing 0.5% CS. (A) Wild-type B. thetaiotaomicron carrying the empty vector pNBU was compared to the two ΔBT3334 complemented strains, the ΔBT3334::pBT3334 and ΔBT3334::pBACCELL_01634/5 strains, expressing the B. thetaiotaomicron regulator and its B. cellulosilyticus ortholog, respectively, driven from their native promoters via a vector integrated at a nonnative chromosomal site. (B) Wild-type and ΔBT3348 isogenic strains carrying the empty vector pNBU were compared to the ΔBT3348 pBT-BACCELL01621 strain harboring the B. cellulosilyticus ortholog driven from the B. thetaiotaomicron BT3348 promoter in a vector integrated at a nonnative chromosomal site. (C) mRNA levels of the glucuronyl hydrolase (GH88) gene BACCELL01621 expressed in the B. thetaiotaomicron ΔBT3348 mutant driven from either the B. thetaiotaomicron BT3348 promoter (ΔBT3348 pPBT pPBT-BACCELL01621) or the native B. cellulosilyticus promoter (ΔBT3348 pPBaccell pPBaccell-BACCELL01621) before and at 40 and 120 min after induction of B. thetaiotaomicron with 0.5% CS. The mRNA for the BT3333 sulfatase is used as a representative mRNA for the CS PUL. P values for the differences in the results at 40 min and 120 min were determined by a paired two-tailed t test. *, P < 0.05; **, P < 0.005; ***, P < 0.00005; ns, nonsignificant. Means and SEMs for two biological replicates are shown.
Bacteroides strains compete to different extents when CS is the sole carbon source.Bacteroides strains capable of growth on CS differ in the dynamics with which they respond to CS (Fig. 5 and 6). Moreover, the transcriptional surge in B. thetaiotaomicron accelerates CS acquisition (16). To explore the possibility that transcription dynamics provide a growth advantage to individual strains, we cocultured B. thetaiotaomicron, B. cellulosilyticus, and B. caccae in various combinations in the presence of CS as the sole carbon source. To address the question of how stable the community is, we examined the effects of coculture (34) over a duration of 5 days by determining the number of bacteria corresponding to each of the species by quantitative real-time PCR using species-specific primers. (Note that expression differences are manifested 5 min after CS availability.)
All three Bacteroides strains achieved similar cell densities when comparing the numbers obtained when they were cultured alone to those obtained when two species were cocultured (Fig. 8A to C). The ability to generate a surge in PUL gene mRNAs did not appear to offer a competitive advantage because B. thetaiotaomicron could be cocultured with either B. cellulosilyticus or B. caccae without impacting the cell densities attained by B. cellulosilyticus and B. caccae in the cocultures (Fig. 8A to C). Interestingly, addition of the corresponding third strain to the coculture harboring B. cellulosilyticus exaggerated the slight disadvantage of B. cellulosilyticus, resulting in it being outcompeted (Fig. 8D). However, B. cellulosilyticus was not outcompeted when cocultured in rich medium (see Fig. S3 in the supplemental material), arguing against the possibility of a general defect in its ability to compete with other strains.
Bacteroides strains cocultured on CS differ in their response to CS concentration and community composition. (A to H) Abundance of Bacteroides species represented as the number of CFU per milliliter (y axis). The numbers of CFU/ml were estimated using species-specific real-time PCR and comparison with the numbers of CFU/ml for standards from experimentally enumerated cultures of the respective species: B. thetaiotaomicron (Btheta; black), B. cellulosilyticus (Baccell; red), and B. caccae (Baccac; blue). (A to D) Bacteroides species were cultured in minimal medium with 0.5% CS from frozen cultures with defined numbers of CFU/ml in minimal medium with 0.5% glucose. Bacteria were cultured either alone (dotted lines), in a coculture (solid lines) with one other species (A to C), or in a coculture with three species (D). (E to H) Bacteroides species were cultured in minimal medium with CS from frozen cultures with defined numbers of CFU/ml in TYG medium. Bacteria were cultured either in 0.1% CS (dashed lines) or in 0.5% CS (solid lines) in cocultures of two species (E to G) or in a three-species coculture (H). All cocultures were started with equal ratios of constituent species. Cultures were serially passaged in fresh medium containing the appropriate carbon source at a 1:100 dilution every day for 5 days. The dotted black lines parallel to the x axes denote the lower limit of sensitivity of the assay. Differences in terminal densities (i.e., numbers of CFU/ml) at day 5 were examined using the paired two-tailed t test for the coculture versus individual culture (A to D) and 0.5% CS versus 0.1% CS (E to H). *, P < 0.05; **, P < 0.01; the differences for the other comparisons were nonsignificant.
There are two possible sources of CS for intestinal bacteria: the host, in which case CS is anticipated to be present in relatively constant amounts, and the host's diet, in which case CS is likely to fluctuate on the basis of the nutrients consumed by the host (3, 17, 35). When CS is in excess, multiple species may be able to grow because competition ensues only during limiting nutrient availability. Thus, we reexamined the coculture experiments by lowering the CS concentration from 0.5% to 0.1% CS.
The cell densities attained by certain Bacteroides strains were reduced in two cocultures when the CS concentration was decreased (Fig. 8E to G). Moreover, each of the three investigated species was cocultured successfully in the presence of one or both species (Fig. 8E to G). Strikingly, when the three species were cocultured, B. cellulosilyticus grew better at the lower CS concentration than at the higher one (Fig. 8H). These results demonstrate that multiple Bacteroides species are capable of coculture even during limiting CS availability. Our results indicate that Bacteroides growth in mixed-strain cocultures is dependent on both the CS concentration and the composition of the bacterial community.
DISCUSSION
Closely related species often coexist in a given environment. Our investigations of gut Bacteroides strains have now revealed several unexpected findings. First, we established that the ability to utilize CS (10, 27) is widespread among Bacteroides species (Fig. 1). Second, Bacteroides species differ in the content, organization, and sequence conservation of the genes responsible for CS uptake and breakdown (Fig. 2 and 3). Third, Bacteroides strains with the ability to break down CS coexist even when CS is the sole carbon source (Fig. 8). Fourth, there is an unanticipated diversity in the regulatory behaviors (dynamics and induction ratios) controlling orthologous nutrient utilization genes in response to the same signal in members of the gut community (Fig. 5 and 6). Fifth, the divergence in regulatory dynamics displayed by Bacteroides strains cannot be explained solely by the gene content and/or the degree of conservation of the CS PUL sequence (Fig. 2, 6, and 7). Sixth, the competitive fitness of gut bacterial strains can vary depending on the CS concentration and community membership (Fig. 8). Cumulatively, our findings suggest that changes in the regulation of PUL genes can impact strain-specific polysaccharide use and possibly contribute to the diversity and stability of a gut microbial community, properties typically ascribed to differences in gene content (13).
Expanding the function of orthologous PULs in Bacteroides species.In Bacteroides, genomically encoded traits in related strains often diversify via functional differences encoded within a PUL (13, 17). For instance, despite the apparent PUL conservation, related Bacteroides species may be distinguished by the substrate preferences of PUL-encoded proteins (13, 20). Given the natural heterogeneity in the structures of polysaccharides (i.e., sugar content, linkage, modification, and chain length) (17), it is possible that differences in the chemical specificities of orthologous gene products in PULs lead to the emergence of strain-specific glycan selection (13). This may be manifested in differences in the substrate repertoires among Bacteroides CS PULs (10) and/or the deployment of the same PULs, depending on the presence of additional nutrients (36, 37).
We previously determined that a surge in transcription of the CS PUL promotes rapid acquisition of CS in B. thetaiotaomicron (16). However, not all Bacteroides species exhibit this expression behavior (Fig. 5 and 6). Furthermore, they display CS concentration-dependent differences in growth and competitive behavior when cocultured with other species (Fig. 8E to H). This disparity raises the possibility that regulatory changes promote specialization under conditions of a low or high abundance of a nutrient and/or during the transient or steady availability of a nutrient. Such fluctuations in CS availability likely result from variation in dietary CS. Strain-specific regulatory dynamics, such as expression kinetics and induction ratios, may thereby enable allocation of a given resource (1) among separate niches available in the same habitat.
Ecological factors driving differential regulation of PULs.In principle, the distinct regulation of PULs exhibited by different Bacteroides strains could be the result of genetic drift or of selective pressures that drive variation in polysaccharide utilization among coexisting species (30, 38). The ability to use specific polysaccharides (via commensal colonization factors) is required not only for the occupation of a strain-specific intestinal niche by a given Bacteroides species but also for its ability to maintain stable and resilient colonization of the gut (39). This finding supports the notion that niche divergence arises from competitive pressures (40, 41) that influence nutrient utilization in gut bacteria.
Mapping the patterns of transcription dynamics onto the relatedness of CS PULs revealed that, surprisingly, transcription dynamics do not diverge according to the sequence identity of the key players responsible for CS PUL expression or according to the phylogenetic relationships of the species containing the CS PUL (Fig. 2, 5, and 6). For instance, while B. caccae and B. ovatus are more related to one another than to B. thetaiotaomicron, B. caccae exhibits an expression behavior distinct from the expression behaviors of the other two species (Fig. 2). While the genetic origins of transcription dynamics remain unknown, it is clear that regulatory attributes of single gene products are conserved across strains that exhibit disparate behaviors (Fig. 7). This dispersed phylogenetic distribution may be reflective of traits that have been influenced by ecological factors (42), such as distinctions in the gut environment between individual hosts or in the community membership of the microbiota.
Strain-level variation of Bacteroides species is greater among isolates from different hosts than among those from an individual host gut (8). Our analysis of the response to CS focused on Bacteroides species found to coexist in a human host (43), but the particular strains used were isolates obtained from different host sources. Representatives of these species also constitute members abundant in artificial communities cocolonizing the mouse gut (25). Thus, the distinct regulatory behaviors that we identified may reflect variation across the intestinal environments of different hosts (5) and/or species-specific traits.
It is also possible that the CS response dynamics of individual strains have been shaped via interactions with members of the gut microbial community. Syntrophic interactions are widespread among bacteria inhabiting a variety of environments, such as the gut, where coexisting microbes enable other strains to utilize previously inaccessible polysaccharides (34). Such a network of polysaccharide utilization among gut Bacteroides species (34) raises the possibility of interrelationships with community members, whereby each possesses a unique array of nutrient breakdown capabilities (25, 40). Although there is no cooperative foraging of CS among the three Bacteroides species examined (Fig. 8), it is possible that they participate in cooperative interactions for another nutrient and/or that they are involved in a CS utilization network with other CS PUL-containing Bacteroides species, such as B. plebeius, which possesses a partially functional CS PUL (see Fig. S2 in the supplemental material). Ecological interactions promote the formation of stable microbial communities, such as in the case of marine microbes (Vibrio or cyanobacteria), where certain species rely on neighboring related strains to share functions essential for survival in those environments (42, 44–46). Thus, cooperative utilization of CS among coexisting Bacteroides strains may drive species-specific responses (47).
Implications of phenotypic diversity in gut Bacteroides.Finally, sequence-based microbial profiling is helping to define the microbiome composition during health and disease. This profiling has revealed that the majority of bacterial strains in a gut community are maintained stably over long durations (8, 9). Investigations of gut microbial ecology and evolution are beginning to uncover strategies by which related members of Bacteroides species coexist at high cell densities in the microbiota (1). While PUL acquisition imparts the nutrient utilization ability (21), Bacteroides species may use different polysaccharides to avoid direct competition (15, 25). Orthologous systems among species (13), as well as paralogous gene families (48), have undergone changes in substrate specificities to target separate nutrients (17). Related species form interdependencies based on cooperative nutrient foraging (34), lending stability to this microbial community (42, 47). The diversity of dynamic behaviors (16) elaborated in the response of a PUL to the availability of a polysaccharide indicates that regulatory changes are likely to expand the foraging abilities of PUL-containing strains (49).
Diet-dependent changes in the microbiota composition are often predicted on the basis of the genetically encoded ability of individual strains to break down polysaccharides consumed by the host (11, 13, 20, 21). Our findings indicate that, in addition to sequence diversity and gene content, an in-depth understanding of the activities of the participating gene products (17), their patterns of gene expression (37), and the community ecology (34) will be required for the effective manipulation of the gut microbial community to improve health.
ACKNOWLEDGMENTS
We thank Guy Townsend and Nathan Schwalm for experimental support and discussions and Whitman Schofield, Andrew Goodman, and Jeffrey Gordon for the valuable advice, reagents, and strains necessary for culturing and manipulating Bacteroides species.
This work was supported by funds provided by the Howard Hughes Medical Institute to E.A.G., who is an investigator of this institute.
V.R. conducted the experiments, and V.R. and E.A.G. designed the study and wrote the paper.
FOOTNOTES
- Received 6 January 2015.
- Accepted 6 February 2015.
- Accepted manuscript posted online 17 February 2015.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/JB.00010-15.
- Copyright © 2015, American Society for Microbiology. All Rights Reserved.
