species -specific dynamic response s of gut bacteria to a ...38 est ablish that in the case of genes...
TRANSCRIPT
Species-specific dynamic responses of gut bacteria to a mammalian 1
glycan 2
3
4
5
Varsha Raghavana,b and Eduardo A. Groismana,b,c# 6
7
a Department of Microbial Pathogenesis, Yale School of Medicine, New Haven, CT, 8
USA; 9
b Yale Microbial Diversity Institute, West Haven, CT, USA; 10
c Howard Hughes Medical Institute; 11
12
Running title: Response dynamics shapes gut bacterial diversity 13
14
#Correspondence: [email protected] 15
JB Accepted Manuscript Posted Online 17 February 2015J. Bacteriol. doi:10.1128/JB.00010-15Copyright © 2015, American Society for Microbiology. All Rights Reserved.
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
2
Abstract 16
The mammalian intestine provides nutrients to hundreds of bacterial species. Closely 17
related species often harbor homologous nutrient utilization genes and co-colonize the 18
gut, raising questions regarding the strategies mediating their stable coexistence. Here 19
we reveal that related Bacteroides species that can utilize the mammalian glycan 20
chondroitin sulfate (CS) have diverged in the manner in which they temporally regulate 21
orthologous CS utilization genes. Whereas certain Bacteroides species display a transient 22
surge in CS utilization transcripts upon exposure to CS, other species exhibit sustained 23
activation of these genes. Remarkably, species-specific expression dynamics is retained 24
even when the key players governing a particular response are substituted by those 25
from a species with a dissimilar response. Bacteroides species exhibiting distinct 26
expression behaviors in the presence of CS can be co-cultured on CS. However, they 27
vary in their responses to CS availability and to the composition of the bacterial 28
community when CS is the sole carbon source. Our results indicate that diversity 29
resulting from regulation of polysaccharide utilization genes may enable co-existence of 30
gut bacterial species using a given nutrient. 31
32
Importance 33
Genes mediating a specific task are typically conserved in related microbes. For 34
instance, gut Bacteroides species harbor orthologous nutrient breakdown genes, and may 35
face competition from one another for these nutrients. How, then, does the gut 36
microbial composition maintain such remarkable stability over long durations? We 37
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
3
establish that in the case of genes conferring the ability to utilize the nutrient 38
chondroitin sulfate (CS), microbial species vary in how they temporally regulate these 39
genes and exhibit subtle growth differences based on CS availability and community 40
composition. Similar to how differential regulation of orthologous genes enables 41
related species to access new environments, gut bacteria may regulate the same genes in 42
distinct fashions to reduce overlap with coexisting species for available nutrients. 43
44
Introduction 45
The mammalian intestine is home to hundreds of gut bacterial species that compete for 46
and cooperate for nutrient sources (1). The gut microbial community benefits the host 47
by degrading otherwise indigestible dietary nutrients (2-4). Alterations in the gut 48
microbial flora have been implicated in numerous disease states, including obesity and 49
diabetes (5, 6). A dominant component of the gut microbiota is the phylum 50
Bacteroidetes (7), which is comprised of a number of species that can break down a 51
wide variety of complex polysaccharides (3). Bacteroides species co-exist in the 52
mammalian gut at high densities (109-1010 colony forming units per gram of feces) (8, 9) 53
and share available nutrients that range from intestinal glycans to dietary plant 54
polysaccharides (3). The vast representation of related Bacteroides species colonizing the 55
intestine raises the question: what mechanism(s) drives co-habitation of gut microbial 56
species in a shared nutrient environment? 57
In Bacteroides, the ability to utilize a particular polysaccharide is usually conferred by 58
a cluster of genes organized into a polysaccharide utilization locus (PUL) (10). A given 59
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
4
PUL typically encodes proteins that recognize, import, and enzymatically convert the 60
polysaccharide into metabolites that can be broken down via glycolysis (11, 12). A PUL 61
often harbors a gene specifying a regulator that responds to a particular polysaccharide 62
or its breakdown products by altering the expression of PUL genes responsible for the 63
utilization of that polysaccharide (13-16). The PULs mediating utilization of particular 64
mammalian or plant glycans, such as alpha-mannan (17), mucins (18, 19), fructans (13) 65
and xyloglucans (20), are conserved within related Bacteroides species. Some of this 66
conservation may be attributed to horizontal transfer of PULs between gut colonizing 67
species (21). While the conservation of PULs suggests that gut species share particular 68
niches, PULs are not identical across species (13, 20). This raises the possibility of 69
shared PULs specifying disparate activities, thereby supporting niche divergence and 70
coexistence of related Bacteroides species. 71
Bacteroides thetaiotaomicron is a prominent member of the mammalian gut microbiota 72
due to its ability to utilize a variety of animal and plant polysaccharides as carbon 73
sources (22, 23). We have previously established that the genes mediating utilization of 74
the mammalian glycan chondroitin sulfate (CS) in B. thetaiotaomicron undergo a 75
transcriptional surge (24) that speeds up acquisition of CS (16). This surge in 76
transcription is characterized by a transient peak in CS utilization transcripts followed 77
by a drop to lower steady state levels despite continuous availability of CS (Figure 1). 78
The regulator of CS utilization genes (BT3334) is activated by a metabolic intermediate 79
in CS breakdown and turned down when the levels and activity of the glucuronyl 80
hydrolase (BT3348) that degrades its activating ligand rises. Even though the 81
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
5
glucuronyl hydrolase BT3348 is the rate-limiting enzyme in CS breakdown, the 82
constitutive high production of this enzyme actually disrupted the dynamics of PUL 83
gene transcription and delayed growth on CS (16). These findings suggest that B. 84
thetaiotaomicron’s ability to dynamically adjust transcription to CS catabolic rate boosts 85
CS utilization in fluctuating nutrient environments. 86
Species-specific polysaccharide foraging among gut bacteria is often attributed to 87
differences in gene content (25). By contrast, it is presently unclear whether species 88
with orthologous nutrient utilization genes differ in their response to a given 89
polysaccharide. We now report that Bacteroides strains vary in the regulation of genes in 90
the CS PUL. We determine that the temporal response to CS availability is a species-91
specific trait, with certain strains promoting a transient surge in CS utilization 92
transcripts and others displaying sustained expression in response to CS. Surprisingly, 93
the observed regulation is retained when the key genes that mediate the transcriptional 94
surge in a given Bacteroides species are replaced by orthologs from species that exhibit 95
sustained expression. The identified regulatory behaviors are not always conserved 96
among evolutionarily related CS PULs, implicating ecological factors in shaping the 97
dynamics with which PUL genes are transcribed. In addition, we establish that 98
Bacteroides strains display differences in their response to CS availability and to the 99
presence of other bacterial species when co-cultured in the presence of CS as the sole 100
carbon source. Our results suggest that regulatory changes in a PUL diversify the 101
nutrient niches accessible to gut Bacteroides; and that this diversification may enable the 102
co-existence of strains utilizing a shared nutrient in the same habitat. 103
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
6
104
Materials and Methods 105
Bacterial strains, plasmids and growth conditions. B. thetaiotaomicron strains were 106
derived from strain ATCC 29148 (VPI-5482) (22) and grown under anaerobic conditions 107
at 37˚C in tryptone-yeast extract-glucose medium containing tetracycline (2 g/ml), 108
erythromycin (10 g/ml), or gentamicin (200 g/ml) when applicable. In this study, we 109
used the following Bacteroides type strains (as recognized by the International Journal of 110
Systematic and Evolutionary Microbiology): B. caccae ATCC 43185, B. ovatus ATCC 8483, 111
B. cellulosilyticus DSM 14383, B. intestinalis DSM 17393 and B. plebeius DSM 17135. All 112
experiments with Bacteroides strains were performed with cells grown anaerobically at 113
37˚C in chemically defined minimal medium (25) supplemented with the appropriate 114
carbon source (0.5% glucose, 0.5% CS, 0.1% CS) and antibiotics when required. 115
Escherichia coli strains were derived from S17-1 and were grown in LB medium 116
containing 50 µg/ml ampicillin when applicable. All chemicals were purchased from 117
Sigma (CS from shark cartilage, C4384). All strains and plasmids used in this study are 118
listed in Table S1. All primers used in this study are listed in Table S2. 119
120
Bioinformatic analyses. BT3334 (regulator) orthologs were initially identified among 121
sequenced type strains of Bacteroides using NCBI BLASTp, to locate putative CS PULs. 122
Neighboring gene content at each “locus” was thereafter assembled manually using 123
BLASTp to identify orthologs of BT3348, a susC/D pair, the sulfatase (BT3333) and CS 124
lyase (BT3324, BT3350) genes. Intervening regions between unassembled contigs were 125
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
7
manually curated to locate missing or partially annotated genes. For plotting 126
relatedness trees, protein sequences were retrieved from NCBI using BLAST and a 127
maximum likelihood tree was constructed using MEGA5 from a multiple sequence 128
alignment created by MUSCLE (HTCS, scale bar, 0.2 amino acid substitutions per site; 129
RecA, scale bar, 0.02 amino acid substitutions per site). Bootstrap values are indicated 130
at branch points. The HTCS tree is drawn to scale with branch lengths in the same units 131
as the phylogenetic tree constructed from the conserved RecA sequence. We note that B. 132
fragilis, B. vulgatus and B. coprocola lack orthologous CS PULs; and that non-orthologous 133
hits were included in the HTCS tree for comparison. 134
135
Growth assays. For steady state growth evaluations, Bacteroides strains were grown in 136
minimal media containing 0.5% glucose and then, the culture was diluted 1:100 into 137
minimal media containing the desired carbon source (0.5% CS from shark cartilage, 138
0.5% glucose) and absorbance at 600 nm was measured while CFU assays were also 139
performed. B. intestinalis and B. cellulosilyticus displayed distinct and lower OD600 to 140
CFU ratio as compared to the other three Bacteroides species, possibly reflective of 141
differences in cell size. Thus, detailed comparisons of growth curves based on OD600 142
measurements for the different strains have not been presented. 143
144
Time course gene expression analysis of Bacteroides induced with CS. Cells were 145
cultured overnight in minimal media with 0.5% glucose. A 1:50 dilution was used for 146
subculture into the same medium for 3.5 to 6 h depending on the species (to an OD600: 147
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
8
0.3 – 0.4). Cells were harvested, resuspended in minimal media lacking carbon source, 148
and induced with 0.5% CS. 2 ml culture was collected before (-5 min time point), and at 149
various times after induction with CS, harvested by centrifugation (for 1 min) and 150
pellets were frozen on dry ice. Frozen pellets were stabilized with RNAprotect bacteria 151
reagent (QIAGEN) and RNA was extracted with RNEasy kit (QIAGEN) for further 152
analysis. 153
154
Quantitative real-time PCR. Quantification of transcripts or bacterial abundance was 155
carried out by real-time PCR using SYBR Green PCR Master Mix (Applied Biosystems) 156
in an ABI 7500 Sequence Detection System (Applied Biosystems). The mRNA 157
abundance represented in the y-axis (throughout the study) is normalized to a 1,000-fold 158
dilution of 16S rRNA abundance to account for cell density. Genomic DNA was used to 159
generate standards for each primer pair used for measuring an mRNA. The BT3333 160
sulfatase gene mRNA was used as a representative transcript because the BT3333 gene 161
was not disrupted in any of the examined mutants. The observed trends were also 162
observed for transcripts corresponding to additional genes (e.g., BT3324 and BT3348) in 163
experiments with complementation of B. thetaiotaomicron mutants. Culture lysate with 164
experimentally determined CFU/mL was used to generate standards for the species-165
specific primer pairs to assign the apparent CFU/mL values denoted in the graphs in 166
co-culture experiments. All experiments depict results obtained with two technical 167
replicates and two biological replicates. 168
169
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
9
Complementation of B. thetaiotaomicron mutants. The region containing the 200 bp 170
sequence upstream of BT3348 was assembled upstream of the BACCELL01621 gene 171
sequence, and was cloned into the vector pNBUtet. The BACCELL01621 gene with a 172
200 bp native upstream sequence was also cloned in a similar fashion and used for 173
complementation. The vectors were introduced into the B. thetaiotaomicron chromosome 174
in a BT3348 mutant background by conjugation as described previosuly (26). The 175
region containing the BACCELL01634/5 gene (the regulator gene annotation is 176
incomplete and divided between two consecutive contigs) including a 200 bp upstream 177
sequence was cloned into vector pNBU tet and used for complementation of the BT3334 178
mutant. 179
180
Co-culture experiments with Bacteroides. Bacteroides species cultured in TYG medium 181
or chemically defined medium for 24 h were frozen and the colony forming units per 182
milliliter (CFU/mL) of the frozen stock was enumerated. Co-culture experiments were 183
performed in 1 mL prereduced minimal media containing the appropriate carbon 184
source (0.5% CS, 0.1% CS, 0.5% glucose) or in TYG medium. Cultures were started with 185
2 x 104 CFU of each species (Figure 8 A-D) or 2 x 105 CFU (Figure 8 E-H) with bacteria 186
being combined in 1:1 ratios and grown for 24 h followed by 1:100 subculture into fresh 187
prereduced media. This process of subculture (1:100) was continued every day for 4 – 5 188
d, and 100 µl volume of 24 h grown cultures were collected and frozen at various days 189
after inoculation. Cells were lysed using the hot-shot lysis method and quantitative 190
real-time PCR was performed with species-specific primers to estimate the number of 191
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
10
bacteria. Initial CFU/mL at time = 0 was determined by plating dilutions for colony 192
counting. Standard curves from cultures with experimentally determined densities 193
were used for estimating CFU/mL of individual species in the co-culture from the 194
measurements obtained via quantitative real-time PCR. All experiments were 195
performed twice and in biological replicates. Statistically significant differences in 196
terminal densities i.e. CFU/mL at Day 5 was determined using the paired two-tailed t-197
test: co-culture versus individual culture (Figure 8A – D) and 0.5% CS versus 0.1% CS 198
(Figure 8E – H). 199
200
Results 201
The CS PUL is prevalent and functional among the genus Bacteroides. We used a 202
sequence homology based approach to locate putative CS PULs in Bacteroides genomes, 203
looking for orthologs of the transcriptional regulator BT3334, the SusC/SusD pair 204
BT3332/BT3331 mediating CS import across the outer membrane, and the glucuronyl 205
hydrolase BT3348, specified by genes located in close proximity to one another (10). For 206
this analysis, we focused on type strains of the corresponding Bacteroides species. The 207
CS PULs, which are distributed throughout the genus Bacteroides, display synteny 208
among Bacteroides genomes. Proteins encoded in the CS PUL demonstrate overall 209
sequence divergence reflective of the phylogenetic relationships among Bacteroides 210
species (Figures 2 and S1). 211
Our analysis revealed two key findings. First, Bacteroides species differ in their 212
degree of sequence identity to the proteins specified in the B. thetaiotaomicron PUL: the 213
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
11
CS PULs of B. ovatus and B. caccae are more related to the B. thetaiotaomicron PUL in 214
comparison to CS PULs of B. cellulosilyticus and B. intestinalis (Figure 2 and 3). However, 215
all four species encode functional CS breakdown proteins because they grow when CS 216
is the sole carbon source (25) (Figure 4). A notable exception is B. plebeius where the 217
putative CS PUL is not fully functional because it does not support growth on CS, and 218
also because CS failed to induce certain (but not all) genes in the CS PUL (Figure S2). 219
Homologs of the CS lyase gene BT4410, which is located outside the CS PUL and 220
required for normal CS utilization in B. thetaiotaomicron, are present in CS PUL-221
containing Bacteroides genomes also at a site distal to the CS PUL (Figure 3) (16). 222
And second, Bacteroides strains vary in the presence of additional open reading 223
frames (ORFs) in the CS PUL (Figure 3). For example, the B. thetaiotaomicron CS PUL 224
includes three genes (BT3328, BT3329 and BT3330) encoding proteins of unknown 225
function predicted to localize to the outer membrane. Whereas homologs of these genes 226
are found in certain Bacteroides species, other species that can utilize CS lack these genes, 227
indicating that they are not essential for CS breakdown. These genes display high 228
sequence divergence and variable organization within the CS PUL among Bacteroides 229
genomes, suggesting that they confer species-specific functions. 230
231
CS-utilizing Bacteroides strains display distinct transcriptional responses to CS. The 232
ability of closely related bacteria to explore distinct niches is typically ascribed to 233
differences in gene content (27-30). For example, the lac operon, which enables the 234
mammalian commensal E. coli to use lactose as the sole carbon source, is absent from 235
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
12
the related enteric species Salmonella enterica serovar Typhimurium, which cannot 236
grown on lactose as the sole carbon source (29). However, such ecological properties 237
may result from the distinct regulation of genes that related species have in common 238
(31-33). For instance, differences in quantitative properties, such as the level and timing 239
of expression of polymyxin resistance genes, mediate phenotypic differences in 240
resistance to the antibiotic polymyxin B in enteric bacteria (34). Given that the 241
dynamics with which the CS PUL is expressed are critical for rapid CS utilization in B. 242
thetaiotaomicron (16), we wondered whether the expression dynamics displayed by B. 243
thetaiotaomicron are conserved in B. cellulosilyticus, which contains a CS PUL encoding 244
proteins with considerable sequence differences (i.e., the BT3334 homolog is only 70% 245
identical at the amino acid level, which is much lower than the 91% identity displayed 246
by the BT3334 homolog in B. ovatus). 247
When B. cellulosilyticus was shifted into minimal media containing CS, the mRNA 248
levels of CS utilization genes increased rapidly, and monotonically reached steady state 249
levels within 60 min (Figure 5). The mRNA levels of the B. cellulosilyticus susC-like gene, 250
which represents the most highly induced transcript, continued to increase over the 251
course of 120 min (Figure 5C). This is in contrast to B. thetaiotaomicron, in which 252
exposure to CS results in a rapid increase in the mRNA levels of CS utilization genes, 253
peaking at 30 min, and declining to lower steady state levels by 60 min (Figure 1). Thus, 254
two Bacteroides species with related CS utilization loci differ in the temporal 255
transcriptional response elicited by sudden CS availability: transcriptional surge versus 256
sustained expression. 257
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
13
To ascertain the prevalence of the two temporal transcriptional responses among 258
Bacteroides species, we examined the expression of CS utilization genes in strains 259
sharing varying degrees of identity with the B. thetaiotaomicron or B. cellulosilyticus genes 260
(Figure 2). We determined that when B. ovatus was shifted into minimal media 261
containing CS, the mRNA levels of CS utilization genes increased, reached a peak at 60 262
min, and then declined to lower levels by 120 min (Figure 6A-D). Therefore, the 263
transcriptional surge identified in B. thetaiotaomicron is retained in the closely related B. 264
ovatus. However, the kinetics with which CS utilization mRNAs are produced is 265
delayed in B. ovatus (Figure 6A-D). This delay could be responsible for the diminished 266
growth of B. ovatus on CS in comparison to B. thetaiotaomicron (Figure 4). 267
The CS PUL from B. intestinalis clusters with that from B. cellulosilyticus. We 268
determined that the temporal response of B. intestinalis to CS availability mimics that of 269
B. cellulosilyticus: CS utilization genes increased rapidly and reached a plateau between 270
60 and 120 min (Figure 6E-H). Therefore, the four analyzed Bacteroides strains display 271
two distinct dynamics of CS PUL expression. The lineage containing B. thetaiotaomicron 272
and B. ovatus promotes a transcriptional surge in CS utilization mRNAs, whereas the B. 273
cellulosilyticus lineage does not. This analysis suggested that Bacteroides strains closely 274
related to B. thetaiotaomicron or B. ovatus would promote a transcriptional surge in CS 275
utilization mRNAs. Surprisingly, the dynamics of CS PUL expression in B. caccae, 276
which branches along with B. thetaiotaomicron (their regulators display 89% amino acid 277
sequence identity) and is the most closely related to B. ovatus (the regulators are 92% 278
identical) (Figure 2), revealed an monotonic reach to steady state levels (Figure 6I-L) as 279
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
14
opposed to the anticipated surge behavior. Moreover, the CS utilization genes were 280
induced to lower levels (i.e., 40-90x versus 100-4000x) even though the growth rate and 281
cell densities achieved by B. caccae are comparable to those exhibited by other 282
Bacteroides species (Figure 4). These results demonstrate that, despite sequence 283
conservation of the CS PUL genes, the factors governing the transcriptional dynamics 284
are not evolutionarily conserved, even within closely related Bacteroides strains. 285
286
Across species complementation of the temporal transcriptional response to CS. The 287
transcriptional surge displayed by the CS PUL genes in B. thetaiotaomicron requires not 288
only the regulator BT3334 to promote transcription of the CS PUL genes but also the 289
glucuronyl hydrolase BT3348 to cleave the unsaturated chondroitin disaccharides that 290
serve as inducing ligands for BT3334 (16) (Figure 1). We wondered whether the distinct 291
kinetic behaviors displayed by B. thetaiotaomicron and B. cellulosilyticus when they 292
experience CS were due to differences in the amino acid sequences of the regulator 293
and/or glucuronyl hydrolase. To test this possibility, we engineered B. thetaiotaomicron 294
strains by deleting either the regulator or glucuronyl hydrolase genes and introducing a 295
copy of the regulator or glucuronyl hydrolase genes from B. cellulosilyticus 296
(BACCELL_01634/5 and BACCELL_01621, respectively) at a different chromosomal 297
location. This strategy enables the investigation of individual factors in an otherwise 298
isogenic genetic background; and it was previously utilized successfully to examine 299
differential regulation of polymyxin resistance genes in enteric bacteria (34). 300
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
15
B. thetaiotaomicron strains harboring either their own regulator BT3334 or that 301
corresponding to the ortholog from B. cellulosilyticus promoted a surge in the mRNA 302
levels of CS utilization genes upon a shift into minimal media containing CS (Figure 7A). 303
Likewise, the B. thetaiotaomicron strain expressing the B. cellulosilyticus glucuronyl 304
hydrolase exhibited the characteristic surge in the mRNA levels of CS utilization genes 305
(Figure 7B). Thus, the B. cellulosilyticus regulator and glucuronyl hydrolase can 306
substitute for their B. thetaiotaomicron orthologs, indicating that they retain the 307
properties required for promoting a transcriptional surge (16). Even a B. 308
thetaiotaomicron BT3348 variant harboring the B. cellulosilyticus glucuronyl hydrolase 309
gene BACCELL_01621 transcribed from the native B. cellulosilyticus promoter 310
maintained the surge behavior (Figure 7C), demonstrating that the B. cellulosilyticus 311
promoter can undergo a transcriptional surge. Cumulatively, our findings indicate that 312
the kinetic differences among Bacteroides strains cannot be ascribed to the individual 313
players required for the transcriptional surge exhibited by B. thetaiotaomicron in 314
response to CS (16). 315
316
Bacteroides strains compete to different extents when CS is the sole carbon source. 317
Bacteroides strains capable of growth on CS differ in the dynamics with which they 318
respond to CS (Figure 5 and 6). Moreover, the transcriptional surge in B. 319
thetaiotaomicron accelerates CS acquisition (16). To explore the possibility of 320
transcription dynamics providing a growth advantage to individual strains, we co-321
cultured B. thetaiotaomicron, B. cellulosilyticus and B. caccae in various combinations in 322
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
16
the presence of CS as the sole carbon source. To address the question of a how stable 323
the community is, we examined the effects of co-culture (35) over the duration of 5 d by 324
determining the number of bacteria corresponding to each of the species by quantitative 325
real-time PCR using species-specific primers. (Please note that expression differences 326
are manifested 5 min after CS availability.) 327
All three Bacteroides strains achieved similar cell densities when comparing the 328
numbers obtained when cultured alone to those obtained when two species were co-329
cultured (Figure 8A-C). The ability to generate a surge in PUL gene mRNAs does not 330
appear to offer a competitive advantage because B. thetaiotaomicron could be co-cultured 331
with either B. cellulosilyticus or B. caccae without impacting the cell densities attained by 332
the latter two species in the co-cultures (Figure 8A-C). Interestingly, addition of the 333
corresponding third strain to the co-culture harboring B. cellulosilyticus exaggerated the 334
slight disadvantage of B. cellulosilyticus, resulting in it being outcompeted (Figure 8D). 335
However, B. cellulosilyticus was not outcompeted when co-cultured in rich media 336
(Figure S3), arguing against the possibility of a general defect in its ability to compete 337
with other strains. 338
There are two possible sources of CS for intestinal bacteria: host-derived, which is 339
anticipated to be present in relatively constant amounts, and dietary, which is likely to 340
fluctuate based on the nutrients consumed by the host (17, 36, 37). When CS is in excess, 341
multiple species may be able to grow because competition ensues only during limiting 342
nutrient availability. Thus, we re-examined the co-culture experiments by lowering the 343
CS concentration from 0.5% to 0.1% CS. 344
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
17
The cell densities attained by certain Bacteroides strains was reduced in two co-345
cultures when the CS concentration was decreased (Figure 8E-G). Moreover, each of 346
the three investigated species was co-cultured successfully in the presence of one or 347
both species (Figure 8E-G). Strikingly, when the three species were co-cultured, B. 348
cellulosilyticus grew better at the lower CS concentration than at the higher one (Figure 349
8H). These results demonstrate that multiple Bacteroides species are capable of co-350
culture even during limiting CS availability. Our results indicate that Bacteroides 351
growth in mixed strain co-cultures is dependent on both the CS concentration and the 352
composition of the bacterial community. 353
354
Discussion 355
Closely related species often co-exist in a given environment. Our investigations of gut 356
Bacteroides strains have now revealed several unexpected findings. First, we established 357
that the ability to utilize CS (26, 28) is widespread among Bacteroides species (Figure 1). 358
Second, Bacteroides species differ in the content, organization and sequence conservation 359
of the genes responsible for CS uptake and breakdown (Figure 2 and 3). Third, 360
Bacteroides strains with ability to breakdown CS co-exist even when CS is the sole 361
carbon source (Figure 8). Fourth, there is an unanticipated diversity in the regulatory 362
behaviors (dynamics and induction ratios) controlling orthologous nutrient utilization 363
genes in response to the same signal in members of the gut community (Figure 5 and 6). 364
Fifth, the divergence in regulatory dynamics displayed by Bacteroides strains cannot be 365
explained simply by either gene content or sequence conservation of the CS PUL 366
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
18
(Figure 2, 6 and 7). And sixth, competitive fitness of gut bacterial strains can vary 367
depending on CS concentration and community membership (Figure 8). Cumulatively, 368
our findings suggest that changes in the regulation of PUL genes can impact strain-369
specific polysaccharide use and possibly contribute to the diversity and stability of a gut 370
microbial community, properties typically ascribed to differences in gene content (13). 371
372
Expanding the function of orthologous PULs in Bacteroides species. In Bacteroides, 373
genomically encoded traits in related strains often diversify via functional differences 374
encoded within a PUL (13, 17). For instance, despite the apparent PUL conservation, 375
related Bacteroides species may be distinguished by the substrate preferences of PUL-376
encoded proteins (13, 20). Given the natural heterogeneity in the structures of 377
polysaccharides (i.e., sugar content, linkage, modification and chain length) (17), it is 378
possible that differences in chemical specificities of orthologous gene products in PULs 379
lead to the emergence of strain-specific glycan selection (13). This may be manifested in 380
differences in the substrate repertoires among Bacteroides CS PULs (26) and/or the 381
deployment of the same PULs depending on the presence of additional nutrients (38, 382
39). 383
We previously determined that a surge in transcription of CS PUL promotes rapid 384
acquisition of CS in B. thetaiotaomicron (16). However, not all Bacteroides species exhibit 385
this expression behavior (Figure 5 and 6). Furthermore, they display CS concentration-386
dependent differences in growth and competitive behavior when co-cultured with other 387
species (Figure 8E-H). This disparity raises the possibility that regulatory changes 388
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
19
promote specialization during conditions of low or high abundance of a nutrient 389
and/or during transient or steady availability of a nutrient. Such fluctuations may 390
underlie variation in CS availability spanning from apparently constant host mucosal 391
sources to huge variations in CS depending on dietary sources. Strain-specific 392
regulatory dynamics, such as expression kinetics and induction ratios, may thereby 393
enable allocation of a given resource (1) among separate niches available in the same 394
habitat. 395
396
Ecological factors driving differential regulation of PULs. In principle, the distinct 397
regulation of PULs exhibited by different Bacteroides strains could be the result of 398
genetic drift or of selective pressures that drive variation in polysaccharide utilization 399
among coexisting species (31, 40). The ability to use specific polysaccharides (via 400
commensal colonization factors) is required not only for the occupation of a strain-401
specific intestinal niche by a given Bacteroides species but also for its ability to maintain 402
stable and resilient colonization of the gut (41). This finding supports the notion that 403
niche divergence arises from competitive pressures (42, 43) that influence nutrient 404
utilization in gut bacteria. 405
Mapping the transcription dynamic patterns onto relatedness of CS PULs revealed 406
that, surprisingly, transcription dynamics does not diverge according to the sequence 407
identity of the key players responsible for CS PUL expression or to the phylogenetic 408
relationships of the species containing the CS PUL (Figure 2, 5 and 6). For instance, 409
while B. caccae and B. ovatus are more related to one another than to B. thetaiotaomicron, 410
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
20
B. caccae exhibits an expression behavior distinct from those of the other two species 411
(Figure 2). While the genetic origins of transcription dynamics remain unknown, it is 412
clear that regulatory attributes of single gene products are conserved across strains that 413
exhibit disparate behaviors (Figure 7). This dispersed phylogenetic distribution may be 414
reflective of traits that have been influenced by ecological factors (44) such as 415
distinctions in the gut environment between individual hosts or in the community 416
membership of the microbiota. 417
Strain-level variation of Bacteroides species is greater within isolates from different 418
hosts than in those from an individual host gut (8). Our analysis of the response to CS 419
focused on Bacteroides species found to coexist in a human host (45), but the particular 420
strains used were isolates obtained from different host sources. Representatives of 421
these species also constitute abundant members in artificial communities co-colonizing 422
the mouse gut (25). Thus, the distinct regulatory behaviors we identified may reflect 423
variation across intestinal environments of different hosts (5) and/or species-specific 424
traits. 425
It is also possible that the CS response dynamics of individual strains have been 426
shaped via interactions with members of the gut microbial community. Syntrophic 427
interactions are widespread among bacteria inhabiting a variety of environments such 428
as the gut, where coexisting microbes enable other strains to utilize previously 429
inaccessible polysaccharides (46). Such a network of polysaccharide utilization among 430
gut Bacteroides (35) raises the possibility of interrelationships with community members, 431
whereby each possesses a unique array of nutrient breakdown capabilities (25, 42). 432
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
21
Although there is no cooperative foraging of CS amongst the three Bacteroides species 433
examined (Figure 8), it is possible that they participate in cooperative interactions for 434
another nutrient and/or that they are involved in a CS utilization network with other 435
CS PUL-containing Bacteroides species such as B. plebeius, which possesses a partially-436
functional CS PUL (Figure S2). Ecological interactions promote formation of stable 437
microbial communities such as in the case of marine microbes (Vibrio or cyanobacteria) 438
where certain species rely on neighboring related strains to share functions essential for 439
survival in those environments (44, 47-49). Thus, cooperative utilization of CS among 440
co-existing Bacteroides strains may drive species-specific responses (50). 441
442
Implications of phenotypic diversity in gut Bacteroides. Finally, sequence-based 443
microbial profiling is helping define microbiome composition during health and disease. 444
This profiling has revealed that the majority of bacterial strains in a gut community are 445
maintained stably over long durations (8, 9). Investigations of gut microbial ecology 446
and evolution are beginning to uncover strategies by which related members of 447
Bacteroides species coexist at high cell densities in the microbiota (1). While PUL 448
acquisition imparts nutrient utilization ability (21), Bacteroides species may use different 449
polysaccharides to avoid direct competition (15, 25). Orthologous systems among 450
species (13), as well as paralogous gene families (51), have undergone changes in 451
substrate specificities to target separate nutrients (17). Related species form 452
interdependencies based on cooperative nutrient foraging (35), lending stability to this 453
microbial community (44, 50). The diversity of dynamic behaviors (16) elaborated in 454
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
22
the response of a PUL to the availability of a polysaccharide indicates that regulatory 455
changes are likely to expand foraging abilities of PUL-containing strains (52). 456
Diet-dependent changes in microbiota composition are often predicted on the basis 457
of the genetically encoded ability of individual strains to breakdown polysaccharides 458
consumed by the host (11, 13, 20, 21). Our findings indicate that, in addition to 459
sequence diversity and gene content, an in-depth understanding of the activities of the 460
participating gene products (17), their patterns of gene expression (39) and the 461
community ecology (35) will be required for the effective manipulation of the gut 462
microbial community to improve health. 463
464
Acknowledgments 465
We thank Guy Townsend and Nathan Schwalm for experimental support and 466
discussions; Whitman Schofield, Andrew Goodman, and Jeffrey Gordon for valuable 467
advice, reagents, and strains necessary for culturing and manipulating Bacteroides 468
species. This work was supported by funds provided by the Howard Hughes Medical 469
Institute to EAG, who is an investigator of this institute. V.R. conducted the 470
experiments, V.R. and E.A.G. designed the study and wrote the paper. 471
References 472
1. Ley RE, Peterson DA, Gordon JI. 2006. Ecological and evolutionary forces 473
shaping microbial diversity in the human intestine. Cell. 124:837-848. doi: 474
10.1016/j.cell.2006.02.017. 475
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
23
2. Sonnenburg JL, Xu J, Leip DD, Chen CH, Westover BP, Weatherford J, Buhler 476
JD, Gordon JI. 2005. Glycan foraging in vivo by an intestine-adapted bacterial 477
symbiont. Science. 307:1955-1959. doi: 10.1126/science.1109051. 478
3. Koropatkin NM, Cameron EA, Martens EC. 2012. How glycan metabolism 479
shapes the human gut microbiota. Nat. Rev. Microbiol. 10:323-335. doi: 480
10.1038/nrmicro2746; 10.1038/nrmicro2746. 481
4. Nicholson JK, Holmes E, Kinross J, Burcelin R, Gibson G, Jia W, Pettersson S. 482
2012. Host-gut microbiota metabolic interactions. Science. 336:1262-1267. doi: 483
10.1126/science.1223813 [doi]. 484
5. Ridaura VK, Faith JJ, Rey FE, Cheng J, Duncan AE, Kau AL, Griffin NW, 485
Lombard V, Henrissat B, Bain JR, Muehlbauer MJ, Ilkayeva O, Semenkovich 486
CF, Funai K, Hayashi DK, Lyle BJ, Martini MC, Ursell LK, Clemente JC, Van 487
Treuren W, Walters WA, Knight R, Newgard CB, Heath AC, Gordon JI. 2013. 488
Gut microbiota from twins discordant for obesity modulate metabolism in mice. 489
Science. 341:1241214. doi: 10.1126/science.1241214. 490
6. Vrieze A, Van Nood E, Holleman F, Salojarvi J, Kootte RS, Bartelsman JF, 491
Dallinga-Thie GM, Ackermans MT, Serlie MJ, Oozeer R, Derrien M, Druesne 492
A, Van Hylckama Vlieg JE, Bloks VW, Groen AK, Heilig HG, Zoetendal EG, 493
Stroes ES, de Vos WM, Hoekstra JB, Nieuwdorp M. 2012. Transfer of intestinal 494
microbiota from lean donors increases insulin sensitivity in individuals with 495
metabolic syndrome. Gastroenterology. 143:913-6.e7. doi: 496
10.1053/j.gastro.2012.06.031. 497
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
24
7. Eckburg PB, Bik EM, Bernstein CN, Purdom E, Dethlefsen L, Sargent M, Gill 498
SR, Nelson KE, Relman DA. 2005. Diversity of the human intestinal microbial 499
flora. Science. 308:1635-1638. doi: 10.1126/science.1110591. 500
8. Faith JJ, Guruge JL, Charbonneau M, Subramanian S, Seedorf H, Goodman AL, 501
Clemente JC, Knight R, Heath AC, Leibel RL, Rosenbaum M, Gordon JI. 2013. 502
The long-term stability of the human gut microbiota. Science. 341:1237439. doi: 503
10.1126/science.1237439 [doi]. 504
9. Arumugam M, Raes J, Pelletier E, Le Paslier D, Yamada T, Mende DR, 505
Fernandes GR, Tap J, Bruls T, Batto JM, Bertalan M, Borruel N, Casellas F, 506
Fernandez L, Gautier L, Hansen T, Hattori M, Hayashi T, Kleerebezem M, 507
Kurokawa K, Leclerc M, Levenez F, Manichanh C, Nielsen HB, Nielsen T, 508
Pons N, Poulain J, Qin J, Sicheritz-Ponten T, Tims S, Torrents D, Ugarte E, 509
Zoetendal EG, Wang J, Guarner F, Pedersen O, de Vos WM, Brunak S, Dore J, 510
MetaHIT Consortium, Antolin M, Artiguenave F, Blottiere HM, Almeida M, 511
Brechot C, Cara C, Chervaux C, Cultrone A, Delorme C, Denariaz G, Dervyn R, 512
Foerstner KU, Friss C, van de Guchte M, Guedon E, Haimet F, Huber W, van 513
Hylckama-Vlieg J, Jamet A, Juste C, Kaci G, Knol J, Lakhdari O, Layec S, Le 514
Roux K, Maguin E, Merieux A, Melo Minardi R, M'rini C, Muller J, Oozeer R, 515
Parkhill J, Renault P, Rescigno M, Sanchez N, Sunagawa S, Torrejon A, Turner 516
K, Vandemeulebrouck G, Varela E, Winogradsky Y, Zeller G, Weissenbach J, 517
Ehrlich SD, Bork P. 2011. Enterotypes of the human gut microbiome. Nature. 518
473:174-180. doi: 10.1038/nature09944; 10.1038/nature09944. 519
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
25
10. Martens EC, Chiang HC, Gordon JI. 2008. Mucosal glycan foraging enhances 520
fitness and transmission of a saccharolytic human gut bacterial symbiont. Cell. 521
Host Microbe. 4:447-457. doi: 10.1016/j.chom.2008.09.007. 522
11. Martens EC, Koropatkin NM, Smith TJ, Gordon JI. 2009. Complex glycan 523
catabolism by the human gut microbiota: the Bacteroidetes Sus-like paradigm. J. 524
Biol. Chem. 284:24673-24677. doi: 10.1074/jbc.R109.022848. 525
12. Bolam DN, Sonnenburg JL. 2011. Mechanistic insight into polysaccharide use 526
within the intestinal microbiota. Gut Microbes. 2:86-90. 527
13. Sonnenburg ED, Zheng H, Joglekar P, Higginbottom SK, Firbank SJ, Bolam 528
DN, Sonnenburg JL. 2010. Specificity of polysaccharide use in intestinal 529
Bacteroides species determines diet-induced microbiota alterations. Cell. 141:1241-530
1252. doi: 10.1016/j.cell.2010.05.005. 531
14. Lowe EC, Basle A, Czjzek M, Firbank SJ, Bolam DN. 2012. A scissor blade-like 532
closing mechanism implicated in transmembrane signaling in a Bacteroides 533
hybrid two-component system. Proc. Natl. Acad. Sci. U. S. A. 109:7298-7303. doi: 534
10.1073/pnas.1200479109. 535
15. Martens EC, Lowe EC, Chiang H, Pudlo NA, Wu M, McNulty NP, Abbott DW, 536
Henrissat B, Gilbert HJ, Bolam DN, Gordon JI. 2011. Recognition and 537
degradation of plant cell wall polysaccharides by two human gut symbionts. 538
PLoS Biol. 9:e1001221. doi: 10.1371/journal.pbio.1001221. 539
16. Raghavan V, Lowe EC, Townsend GE,2nd, Bolam DN, Groisman EA. 2014. 540
Tuning transcription of nutrient utilization genes to catabolic rate promotes 541
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
26
growth in a gut bacterium. Mol. Microbiol. 93:1010-1025. doi: 10.1111/mmi.12714 542
[doi]. 543
17. Martens EC, Kelly AG, Tauzin AS, Brumer H. 2014. The Devil Lies in the 544
Details: How Variations in Polysaccharide Fine-Structure Impact the Physiology 545
and Evolution of Gut Microbes. J. Mol. Biol. 426:3851-3865. doi: S0022-546
2836(14)00338-6 [pii]. 547
18. Benjdia A, Martens EC, Gordon JI, Berteau O. 2011. Sulfatases and a radical S-548
adenosyl-L-methionine (AdoMet) enzyme are key for mucosal foraging and 549
fitness of the prominent human gut symbiont, Bacteroides thetaiotaomicron. J. Biol. 550
Chem. 286:25973-25982. doi: 10.1074/jbc.M111.228841. 551
19. Marcobal A, Barboza M, Sonnenburg ED, Pudlo N, Martens EC, Desai P, 552
Lebrilla CB, Weimer BC, Mills DA, German JB, Sonnenburg JL. 2011. 553
Bacteroides in the infant gut consume milk oligosaccharides via mucus-utilization 554
pathways. Cell. Host Microbe. 10:507-514. doi: 10.1016/j.chom.2011.10.007 [doi]. 555
20. Larsbrink J, Rogers TE, Hemsworth GR, McKee LS, Tauzin AS, Spadiut O, 556
Klinter S, Pudlo NA, Urs K, Koropatkin NM, Creagh AL, Haynes CA, Kelly 557
AG, Cederholm SN, Davies GJ, Martens EC, Brumer H. 2014. A discrete genetic 558
locus confers xyloglucan metabolism in select human gut Bacteroidetes. Nature. 559
506:498-502. doi: 10.1038/nature12907. 560
21. Hehemann JH, Correc G, Barbeyron T, Helbert W, Czjzek M, Michel G. 2010. 561
Transfer of carbohydrate-active enzymes from marine bacteria to Japanese gut 562
microbiota. Nature. 464:908-912. doi: 10.1038/nature08937; 10.1038/nature08937. 563
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
27
22. Xu J, Bjursell MK, Himrod J, Deng S, Carmichael LK, Chiang HC, Hooper LV, 564
Gordon JI. 2003. A genomic view of the human-Bacteroides thetaiotaomicron 565
symbiosis. Science. 299:2074-2076. doi: 10.1126/science.1080029. 566
23. Xu J, Chiang HC, Bjursell MK, Gordon JI. 2004. Message from a human gut 567
symbiont: sensitivity is a prerequisite for sharing. Trends Microbiol. 12:21-28. 568
24. Shin D, Lee EJ, Huang H, Groisman EA. 2006. A positive feedback loop 569
promotes transcription surge that jump-starts Salmonella virulence circuit. 570
Science. 314:1607-1609. doi: 10.1126/science.1134930. 571
25. McNulty NP, Wu M, Erickson AR, Pan C, Erickson BK, Martens EC, Pudlo NA, 572
Muegge BD, Henrissat B, Hettich RL, Gordon JI. 2013. Effects of diet on 573
resource utilization by a model human gut microbiota containing Bacteroides 574
cellulosilyticus WH2, a symbiont with an extensive glycobiome. PLoS Biol. 575
11:e1001637. doi: 10.1371/journal.pbio.1001637. 576
26. Martens EC, Chiang HC, Gordon JI. 2008. Mucosal Glycan Foraging Enhances 577
Fitness and Transmission of a Saccharolytic Human Gut Bacterial Symbiont. Cell 578
Host & Microbe. 4:447-457. doi: 10.1016/j.chom.2008.09.007. 579
27. Wiedenbeck J, Cohan FM. 2011. Origins of bacterial diversity through horizontal 580
genetic transfer and adaptation to new ecological niches. FEMS Microbiol. Rev. 581
35:957-976. doi: 10.1111/j.1574-6976.2011.00292.x [doi]. 582
28. Xu J, Mahowald MA, Ley RE, Lozupone CA, Hamady M, Martens EC, 583
Henrissat B, Coutinho PM, Minx P, Latreille P, Cordum H, Van Brunt A, Kim 584
K, Fulton RS, Fulton LA, Clifton SW, Wilson RK, Knight RD, Gordon JI. 2007. 585
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
28
Evolution of symbiotic bacteria in the distal human intestine. PLoS Biol. 5:e156. 586
doi: 10.1371/journal.pbio.0050156. 587
29. Ochman H, Lawrence JG, Groisman EA. 2000. Lateral gene transfer and the 588
nature of bacterial innovation. Nature. 405:299-304. doi: 10.1038/35012500 [doi]. 589
30. Pal C, Papp B, Lercher MJ. 2005. Adaptive evolution of bacterial metabolic 590
networks by horizontal gene transfer. Nat. Genet. 37:1372-1375. doi: 591
10.1038/ng1686. 592
31. Hoekstra HE, Coyne JA. 2007. The locus of evolution: evo devo and the genetics 593
of adaptation. Evolution. 61:995-1016. doi: EVO105 [pii]. 594
32. McAdams HH, Srinivasan B, Arkin AP. 2004. The evolution of genetic 595
regulatory systems in bacteria. Nat. Rev. Genet. 5:169-178. doi: 10.1038/nrg1292 596
[doi]. 597
33. Wagner GP, Lynch VJ. 2008. The gene regulatory logic of transcription factor 598
evolution. Trends Ecol. Evol. 23:377-385. doi: 10.1016/j.tree.2008.03.006 [doi]. 599
34. Winfield MD, Groisman EA. 2004. Phenotypic differences between Salmonella 600
and Escherichia coli resulting from the disparate regulation of homologous genes. 601
Proc. Natl. Acad. Sci. U. S. A. 101:17162-17167. doi: 0406038101 [pii]. 602
35. Rakoff-Nahoum S, Coyne MJ, Comstock LE. 2014. An ecological network of 603
polysaccharide utilization among human intestinal symbionts. Curr. Biol. 24:40-604
49. doi: 10.1016/j.cub.2013.10.077 [doi]. 605
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
29
36. Koropatkin NM, Cameron EA, Martens EC. 2012. How glycan metabolism 606
shapes the human gut microbiota. Nat. Rev. Microbiol. 10:323-335. doi: 607
10.1038/nrmicro2746; 10.1038/nrmicro2746. 608
37. Goodman AL, McNulty NP, Zhao Y, Leip D, Mitra RD, Lozupone CA, Knight 609
R, Gordon JI. 2009. Identifying genetic determinants needed to establish a 610
human gut symbiont in its habitat. Cell. Host Microbe. 6:279-289. doi: 611
10.1016/j.chom.2009.08.003. 612
38. Lynch JB, Sonnenburg JL. 2012. Prioritization of a plant polysaccharide over a 613
mucus carbohydrate is enforced by a Bacteroides hybrid two-component system. 614
Mol. Microbiol. 85:478-491. doi: 10.1111/j.1365-2958.2012.08123.x. 615
39. Rogers TE, Pudlo NA, Koropatkin NM, Bell JS, Moya Balasch M, Jasker K, 616
Martens EC. 2013. Dynamic responses of Bacteroides thetaiotaomicron during 617
growth on glycan mixtures. Mol. Microbiol. 88:876-890. doi: 10.1111/mmi.12228. 618
40. Blount ZD, Barrick JE, Davidson CJ, Lenski RE. 2012. Genomic analysis of a key 619
innovation in an experimental Escherichia coli population. Nature. 489:513-518. 620
doi: 10.1038/nature11514 [doi]. 621
41. Lee SM, Donaldson GP, Mikulski Z, Boyajian S, Ley K, Mazmanian SK. 2013. 622
Bacterial colonization factors control specificity and stability of the gut 623
microbiota. Nature. 501:426-429. doi: 10.1038/nature12447 [doi]. 624
42. Fischbach MA, Sonnenburg JL. 2011. Eating for two: how metabolism 625
establishes interspecies interactions in the gut. Cell. Host Microbe. 10:336-347. 626
doi: 10.1016/j.chom.2011.10.002. 627
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
30
43. Sonnenburg JL, Chen CT, Gordon JI. 2006. Genomic and metabolic studies of 628
the impact of probiotics on a model gut symbiont and host. PLoS Biol. 4:e413. 629
doi: 10.1371/journal.pbio.0040413. 630
44. Morris JJ, Lenski RE, Zinser ER. 2012. The Black Queen Hypothesis: evolution 631
of dependencies through adaptive gene loss. MBio. 3:10.1128/mBio.00036-12. 632
Print 2012. doi: 10.1128/mBio.00036-12 [doi]. 633
45. Coyne MJ, Zitomersky NL, McGuire AM, Earl AM, Comstock LE. 2014. 634
Evidence of extensive DNA transfer between Bacteroidales species within the 635
human gut. MBio. 5:e01305-14. doi: 10.1128/mBio.01305-14 [doi]. 636
46. Rakoff-Nahoum S, Coyne MJ, Comstock LE. 2014. An ecological network of 637
polysaccharide utilization among human intestinal symbionts. Curr. Biol. 24:40-638
49. doi: 10.1016/j.cub.2013.10.077. 639
47. Cordero OX, Polz MF. 2014. Explaining microbial genomic diversity in light of 640
evolutionary ecology. Nat. Rev. Microbiol. 12:263-273. doi: 10.1038/nrmicro3218 641
[doi]. 642
48. Cordero OX, Ventouras LA, DeLong EF, Polz MF. 2012. Public good dynamics 643
drive evolution of iron acquisition strategies in natural bacterioplankton 644
populations. Proc. Natl. Acad. Sci. U. S. A. 109:20059-20064. doi: 645
10.1073/pnas.1213344109 [doi]. 646
49. Cordero OX, Wildschutte H, Kirkup B, Proehl S, Ngo L, Hussain F, Le Roux F, 647
Mincer T, Polz MF. 2012. Ecological populations of bacteria act as socially 648
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
31
cohesive units of antibiotic production and resistance. Science. 337:1228-1231. 649
doi: 10.1126/science.1219385 [doi]. 650
50. Morris JJ, Papoulis SE, Lenski RE. 2014. Coexistence of Evolving Bacteria 651
Stabilized by a Shared Black Queen Function. Evolution. 68:2960-2971. doi: 652
10.1111/evo.12485 [doi]. 653
51. Degnan PH, Barry NA, Mok KC, Taga ME, Goodman AL. 2014. Human gut 654
microbes use multiple transporters to distinguish vitamin B(1)(2) analogs and 655
compete in the gut. Cell. Host Microbe. 15:47-57. doi: 10.1016/j.chom.2013.12.007 656
[doi]. 657
52. Raghavan V, Groisman EA. 2010. Orphan and hybrid two-component system 658
proteins in health and disease. Curr. Opin. Microbiol. 13:226-231. doi: 659
10.1016/j.mib.2009.12.010. 660
Figure Legends 661
Figure 1. B. thetaiotaomicron dynamically adjusts transcription of CS PUL genes to 662
catabolic rate. The porin BT3332 and its partner BT3331 transport CS across the outer 663
membrane and into the periplasmic space where it is converted into unsaturated 664
chondroitin disaccharides by the action of three distinct CS lyases (BT3324, BT3350 and 665
BT4410). Unsaturated disaccharides are broken down by the glucuronyl hydrolase 666
BT3348 into the monosaccharides N-acetyl galactosamine (GalNAc) and 5-keto 4-667
deoxyuronate (kdu). Particular unsaturated disaccharides that are substrates for the 668
glucuronyl hydrolase serve as activating ligands that bind to the periplasmic domain of 669
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
32
the hybrid two-component system regulator BT3334, which governs transcription of CS 670
utilization genes. A cluster of BT3334-regulated genes from the CS PUL is depicted at 671
the bottom of the figure (not drawn to scale). Expression of BT3334-regulated genes, 672
represented by the Sulfatase 1 (BT3333) mRNA, initially increases when unsaturated 673
disaccharides are produced and then falls when the levels and activity of the rate-674
limiting glucuronyl hydrolase rises. mRNA levels depicted are normalized to a 1,000-675
fold dilution of 16S ribosomal RNA (rRNA) and the fold change (i.e., 30x) denotes the 676
ratio of mRNA levels at 120 min post-induction to levels before induction (16). 677
Figure 2. CS PUL is prevalent among Bacteroides. Relatedness of CS PULs based on 678
deduced amino acid sequence of the regulator (HTCS BT3334 ortholog). The 679
experimentally examined species are highlighted with colored branches: blue branch 680
denotes strains that promote a surge in CS utilization mRNAs and red branch denotes 681
strains exhibiting sustained activation of CS utilization mRNAs. B. plebeius highlighted 682
in pink contains an incomplete complement of CS PUL genes. Note: B. fragilis, B. 683
vulgatus and B. coprocola lack orthologous CS PULs. For a phylogenetic tree based on 684
the deduced amino acid sequence of the conserved recA gene refer to Figure S1. 685
Figure 3. Gene content and organization of CS PULs in Bacteroides strains. Genetic 686
loci corresponding to CS PULs based on sequence comparison with B. thetaiotaomicron, 687
and on gene expression analysis. The PUL encodes two CS lyases (red), a susC and 688
susD pair (black), a hybrid two-component system (HTCS) regulator (green), an 689
unsaturated glucuronyl hydrolase GH88 (blue) and two sulfatases (grey). Another CS 690
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
33
lyase (orange) is conserved and located distal to the PUL. Hypothetical proteins 691
predicted to localize to the outer membrane are depicted (purple) and absent from the B. 692
cellulosilyticus and B. intestinalis genomes. The B. plebeius PUL contains the genes 693
encoding the sulfatase 1, the HTCS and the GH88 clustered together. Putative orthologs 694
of the CS lyase 1 and the susC and susD pair are located elsewhere in the genome. The 695
dotted grey symbols above the arrow denoting a B. plebeius gene (03801, 00710, 02279) 696
indicate genes not induced by CS in B. plebeius that are orthologous to genes induced by 697
CS in other Bacteroides species. Gene numbers refer to identifiers provided by the NCBI 698
Entrez database, prefixes corresponding to the particular species have been omitted but 699
are as follows: BACOVA (B. ovatus), BACCAC (B. caccae), BACCELL (B. cellulosilyticus), 700
BACINT (B. intestinalis) and BACPLE (B. plebeius). The number below each gene 701
indicates the percentage sequence identity of the deduced amino acid sequence of the 702
gene to its corresponding B. thetaiotaomicron ortholog. 703
Figure 4. Strains with divergent CS PULs are capable of utilizing CS as the sole 704
carbon source. Growth of Bacteroides species subcultured 1:100 into minimal media 705
with 0.5% glucose for 20 h, and 0.5% CS for 20 h and 45 h. Growth is represented as 706
OD600. Species abbreviations are as follows: Btheta (B. thetaiotaomicron), Baccac (B. 707
caccae), Bacova (B. ovatus), Baccell (B. cellulosilyticus) and Bacint (B. intestinalis). 708
Figure 5. B. cellulosilyticus displays sustained activation of CS utilization genes. 709
mRNA levels of CS utilization genes normalized to 1,000-fold dilution of 16S rRNA; 710
before and at 20, 40, 60 and 120 min after B. cellulosilyticus was shifted into minimal 711
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
34
media containing 0.5% CS; (A) GH88 gene, BACCELL01621 (B) Sulfatase 1 gene, 712
BACCELL01636 (C) susC (CS) gene, BACCELL01637 (D) CS lyase 1 gene, 713
BACCELL01643. Mean and SEM from two biological replicates are shown. The fold 714
change “x” denotes ratio of mRNA levels at 120 min post-induction to the levels before 715
induction. 716
Figure 6. Related Bacteroides species differ in expression dynamics of CS utilization 717
genes. mRNA levels of CS utilization genes normalized to 1,000-fold dilution of 16S 718
rRNA prepared before and at 20, 40, 60 and 120 min after B. ovatus, B. intestinalis or B. 719
caccae were shifted into minimal media containing 0.5% CS. Figure shows data 720
corresponding to the B. ovatus genes (A) GH88, BACOVA01969; (B) Sulfatase 1, 721
BACOVA01997; (C) susC (CS), BACOVA01998; and (D) CS lyase 1, BACOVA00969; to 722
B. intestinalis genes (E) GH88, BACINT01376; (F) Sulfatase 1, BACINT01362; (G) susC 723
(CS), BACINT01361; and (H) CS lyase 1, BACINT01356; and to B. caccae genes (I) GH88, 724
BACCAC01536; (J) Sulfatase 1, BACCAC01519; (K) susC (CS), BACCAC01512; and (L) 725
CS lyase 1, BACCAC01509. The fold change “x” denotes ratio of mRNA levels at 120 726
min post-induction to the levels before induction. 727
Figure 7. B. cellulosilyticus genes rescue transcriptional surge in B. thetaiotaomicron 728
mutants defective in the BT3334 or BT3348 genes. (A-B) mRNA levels of CS 729
utilization gene BT3333 normalized to 1,000-fold dilution of 16S rRNA before, and at 40 730
and 120 min after B. thetaiotaomicron was shifted into minimal media containing 0.5% CS. 731
(A) Wild-type B. thetaiotaomicron carrying the empty vector pNBU was compared to the 732
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
35
two ∆BT3334 complemented strains, ∆BT3334: pBT3334 and ∆BT3334 733
pBACCELL01634/5, expressing the B. thetaiotaomicron regulator or its B. cellulosilyticus 734
ortholog, respectively, driven from their native promoters via a vector integrated at a 735
non-native chromosomal site. (B) Wild-type and ∆BT3348 isogenic strains carrying the 736
empty vector pNBU were compared to the strain ∆BT3348 pBT-BACCELL01621 737
harboring the B. cellulosilyticus ortholog driven from the B. thetaiotaomicron BT3348 738
promoter in a vector integrated at a non-native chromosomal site. (C) mRNA levels of 739
the glucuronyl hydrolase (GH88) gene BACCELL01621 expressed in the B. 740
thetaiotaomicron ∆BT3348 mutant driven from the either B. thetaiotaomicron BT3348 741
promoter (pPBT – BACCELL01621) or the native B. cellulosilyticus promoter (pPBaccell – 742
BACCELL01621); before, and at 40 and 120 min after induction of B. thetaiotaomicron 743
with 0.5% CS. BT3333 sulfatase is used as a representative mRNA for the CS PUL. ns, 744
non-significant. Paired two-tailed t-test (40 min and 120 min). * P<0.05, **P<0.005, 745
***P<0.00005. Mean and SEM from two biological replicates are shown. 746
Figure 8. Bacteroides strains co-cultured on CS differ in their response to CS 747
concentration and community composition. (A-H) Abundance of Bacteroides species 748
represented as colony forming units per milliliter (CFU/mL) values (y-axis). CFU/mL 749
are estimated using species-specific real-time PCR by comparison with standards from 750
experimentally enumerated cultures of the respective species: B. thetaiotaomicron (Btheta, 751
black), B. cellulosilyticus (Baccell, red) and B. caccae (Baccac, blue). (A-D) Bacteroides 752
species were cultured in minimal media with 0.5% CS from frozen cultures of defined 753
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
36
CFU/mL in minimal media with 0.5% glucose. Bacteria were either cultured alone 754
(dotted line) or in co-culture (solid line) with one other species (A-C) or in three species 755
co-culture (D). (E-H) Bacteroides species were cultured in minimal media with CS from 756
frozen cultures of defined CFU/mL in TYG. Bacteria were cultured either in 0.1% CS 757
(dashed line) or in 0.5% CS (solid line) in co-cultures of two species (E-G) or in a three 758
species co-culture (H). All co-cultures were started at equal ratios of constituent species. 759
Cultures were serially passaged in fresh media containing the appropriate carbon 760
source at 1:100 dilution everyday for 5 d. Dotted black line parallel to the x-axis denotes 761
lower limit of sensitivity of the assay. Differences in terminal densities (i.e., CFU/ml) at 762
Day 5 examined using the paired two-tailed t-test: co-culture versus individual culture 763
(Figure 8A – D) and 0.5% CS versus 0.1% CS (Figure 8E – H). * P<0.05, **P<0.01, others 764
are non-significant. 765
on April 14, 2020 by guest
http://jb.asm.org/
Dow
nloaded from
