Scientific Report
Strain competition restricts colonization of an
enteric pathogen and prevents colitis
Aaron L Hecht
1,2
, Benjamin W Casterline
1,2
, Zachary M Earley
3
, Young Ah Goo
4
, David R Goodlett
4
&
Juliane Bubeck Wardenburg
1,5,*
Abstract
The microbiota is a major source of protection against intestinal
pathogens; however, the specific bacteria and underlying mecha-
nisms involved are not well understood. As a model of this interac-
tion, we sought to determine whether colonization of the murine
host with symbiotic non-toxigenic Bacteroides fragilis could limit
acquisition of pathogenic enterotoxigenic B. fragilis.Weobserved
strain-specific competition with toxigenic B. fragilis,dependentupon
type VI secretion, identifying an effector–immunity pair that confers
pathogen exclusion. Resistance against host acquisition of a second
non-toxigenic strain was also uncovered, revealing a broader func-
tion of type VI secretion systems in determining microbiota composi-
tion. The competitive exclusion of enterotoxigenic B. fragilis by a
non-toxigenic strain limited toxin exposure and protected the host
against intestinal inflammatory disease. Our studies demonstrate a
novel role of type VI secretion systems in colonization resistance
against a pathogen. This understanding of bacterial competition may
be utilized to define a molecularly targeted probiotic strategy.
Keywords colonization resistance; enterotoxigenic Bacteroides fragilis; in vivo
strain competition; probiotics; type VI secretion
Subject Categories Immunology; Microbiology, Virology & Host Pathogen
Interaction
DOI 10.15252/embr.201642282 | Received 25 February 2016 | Revised 18 June
2016 | Accepted 21 June 2016 | Published online 18 July 2016
EMBO Reports (2016) 17: 1281–1291
See also: M Sassone-Corsi & M Raffatellu (September 2016)
Introduction
Bacterial antagonistic relationships are appreciated as a critical factor
in defining the dense ecosystem of the intestinal microbiota [1,2].
Pathogen exclusion through competition with the microbiota is a long
understood mechanism of host protection, indicating that individual
susceptibility to disease may in part be underpinned by these
relationships [3–9]. Enterotoxigenic Bacteroides fragilis (ETBF) cause
acute diarrhea and are associated with active inflammatory bowel
disease (IBD), late-stage colon cancer, and sepsis through production
of B. fragilis toxin (BFT) [10–16]. Conversely, non-toxigenic B. fragilis
(NTBF) strains are symbiotic, protecting their host against inflamma-
tory disease [17]. Longitudinal human studies show stable predomi-
nancebyeitherETBForNTBFintheB. fragilis population of an
individual microbiota [18], suggesting a competitive interplay between
these strains [7,19]. However, the determinants of this competition
and susceptibility to ETBF colonization remain unknown, representing
a unique model for the study of microbiota–pathogen interactions.
One mechanism of interbacterial competition is type VI secretion
(T6S), whereby an attacking cell injects effector proteins through
the membrane of a target organism [20,21]. Effector-neutralizing
immunity proteins encoded in the genome of the attacking strain
prevent self-intoxication [22–24]. Differential encoding of effector–
immunity pairs has been demonstrated to contribute to in vitro
strain competition in Vibrio cholerae [25]. Broad conservation of
type VI secretion system (T6SS) loci was identified in the
Bacteroidetes phylum, including B. fragilis, allowing for killing of
the closely related B. thetaiotaomicron in vitro [3–9,26,27]. Recent
studies identified B. fragilis strain competition mediated by T6S
both in vitro and in vivo [28,29]. As the Bacteroidetes comprise up
to 50% of the human microbiota, T6S may play a key role in deter-
mining its composition. The complex biogeography of the intestinal
ecosystem necessitates examination of the functional effects of puta-
tive competitive factors on the host [10–15,26,30]. The distinct
biological properties of non-toxigenic and toxigenic B. fragilis
enable an examination of symbiont–pathogen competitive coloniza-
tion in microbiota composition and disease susceptibility.
Results and Discussion
Strain competition reduces ETBF colonization through T6S
To understand competitive dynamics within the B. fragilis species,
we utilized a co-colonization system in specific pathogen-free (SPF)
1 Department of Microbiology, University of Chicago, Chicago, IL, USA
2 Interdisciplinary Scientist Training Program, University of Chicago, Chicago, IL, USA
3 Department of Pathology, University of Chicago, Chicago, IL, USA
4 Department of Pharmaceutical Sciences, School of Pharmacy, University of Maryland, Baltimore, MD, USA
5 Department of Pediatrics, University of Chicago, Chicago, IL, USA
*Corresponding author. Tel: +1 773 834 9763; E-mail: jbubeckw@peds.bsd.uchicago.edu
ª 2016 The Authors EMBO reports Vol 17 |No 9 | 2016
1281
C57BL/6J mice. Following orogastric delivery of B. fragilis,
colonization by non-toxigenic and enterotoxigenic strains of interest
(Appendix Table S1) was monitored over time by fecal colony-
forming unit (CFU) recovery on selective media, utilizing plasmid-
encoded antibiotic resistance markers to distinguish strains [17,31].
Co-colonization of mice with NTBF strain NCTC 9343 (N1) and
ETBF strain ATCC 43858 (E1) resulted in a ~100 fold higher N1 colo-
nization density relative to E1 over a 4-week time period (Fig 1A).
To examine the role of T6S in B. fragilis competition, we generated
an N1 mutant harboring a genomic deletion of the tssC locus (N1
DtssC) that encodes an essential machinery component of the T6SS
[7,19,26]. Co-colonization of N1 DtssC with E1 caused a loss of E1
repression (Figs 1B and EV1A) that was regained by plasmid-based
tssC complementation (N1 DtssC pTssC, Fig 1C). Analysis of bacte-
rial recovery 4 weeks post co-colonization demonstrated an
increased E1 bacterial load achieved with loss of N1 T6SS function
(Fig 1D), concomitant with a decrement in the colonization density
achieved by N1 DtssC (Fig EV1B). Deletion of tssC did not affect N1
mono-colonization (Fig EV1C and D) or bacterial recovery of N1 or
E1 (Fig EV1E–G) 1 day following co-colonization. As E1 mono-
colonization yielded 10
10
CFU/g recovery in the feces (Fig EV1H),
competition with N1 effectively reduced host exposure to
toxin-producing B. fragilis. In vitro plate competition assays
revealed T6S-dependent killing of E1 by N1, confirming a direct
interaction between these strains (Fig EV1I).
A differentially encoded effector–immunity pair mediates
T6S-dependent strain competition
Alignment of the N1 and E1 T6SS loci revealed a non-conserved
region encoding a set of proteins that lack homology to documented
T6SS effector or immunity domains (Fig 2A). We predicted that
these proteins might determine intraspecific competition. Mass spec-
trometry analysis of the N1 secretome revealed decreased secretion
of proteins encoded within the T6SS locus upon mutation of tssC
(Table EV1), including T6S structural components (e.g., Hcp homo-
logs and VgrG). BF9343_1928 demonstrated the greatest fold-change
in the secretome study, leading us to hypothesize that this protein is
a putative effector and that BF9343_1927, encoded immediately
downstream, is its cognate immunity protein (Fig 2A). Congruent
with our studies, BF9343_1928 was demonstrated as a T6S effector,
denoted as Bte2 (B. fragilis T6S effector 2) and BF9343_1927 as a
cognate immunity protein named Bti2a (B. fragilis T6S immunity
2a) [28]. In-frame deletion of bte2 in N1 (N1 Dbte2) phenocopied
the DtssC mutant during co-colonization with E1, as N1 Dbte2 no
longer demonstrated a competitive advantage against E1, permitting
enhanced E1 colonization (Fig 2B, C and E). Heterologous expres-
sion of Bti2a in E1 (E1 pBti2a) conferred full restoration of E1 fecal
CFU 4 weeks post co-colonization with N1 WT (Fig 2D and E).
These results were confirmed in vitro, where N1 Dbte2 exhibits
reduced killing capacity against E1, and E1 pBti2a is protected from
AB
CD
Figure 1. NTBF strain dominance of ETBF through T 6 S.
A–C SPF C57BL/6J mice were co-colonized with E1 and N1 wild type (WT, A, n = 5 mice), N1 T6SS mutant (DtssC,B,n = 4), or N1 complemented (DtssC pTssC, C, n = 5).
Fecal CFU was quantified for E1 (open squares) and N1 (closed squares) weekly.
D Four weeks post-colonization, E1 fecal recovery was compared between the N1 WT, DtssC, and DtssC pTssC groups.
Data information: Results are representative of three independent experiments. Data are presented as mean SEM (A–C) or mean SD (D). n.s., not significant;
**P < 0.01, ***P < 0.001, ****P < 0.0001. Statistical significance was determined by unpaired, parametric, two-tailed Student’s t-test at each time point (applying
Bonferroni correction), comparing the co-colonizing strains (A–C) or one-way ANOVA, Tukey’s multiple comparisons test (D).
EMBO reports Vol 17 |No9 | 2016 ª 2016 The Authors
EMBO reports Strain competition restricts a pathogen Aaron L Hecht et al
1282
N1 WT (Fig EV1I). These data demonstrate the importance of a dif-
ferentially encoded effector–immunity locus in modulation of the
colonic microbiota composition, resulting in altered colonization by
toxigenic B. fragilis.
ETBF colonization resistance is strain- and order-dependent
Colonization of gnotobiotic mice with N1 saturates the B. fragilis
intestinal niche and prevents secondary challenge by the same strain
[18,31]. To determine whether colonic establishment of N1 provides
colonization resistance against E1, we examined the N1–E1 interac-
tion in sequential colonization of gnotobiotic mice, monitoring fecal
CFU of each strain through differential encoding of plasmid-borne
antibiotic resistance markers. Primary colonization with N1 restricted
secondary challenge of N1, while E1 challenge produced stable
colonization, suggesting strain-specific colonization resistance within
the B. fragilis species (Fig 3A). Results in SPF mice phenocopied the
gnotobiotic competition (Fig 3B); we therefore utilized the SPF model
to evaluate in vivo strain interactions for all subsequent studies. A
broader pairwise analysis of five B. fragilis strains, three NTBF and
two ETBF (N1, N2, N3, E1, and E2; Appendix Table S1), was
performed to examine B. fragilis competition behavior. While each
strain exhibited similar primary colonization (Fig EV2), we observed
a distinct pattern of intraspecific niche competition between strains
(Figs 3C and EV2), confirmed by PCR-based genomic analysis of the
colonizing strains (Appendix Fig S1). Intraspecific niche competition
is characterized by full restriction of colonization by self-secondary
challenge (Fig 3C, gray box) and strain-specific colonization resis-
tance wherein some strains (N2, N3, and E1) exhibit a dominant
exclusion phenotype (Fig 3C, red boxes). Our data show a strong
A
BC
DE
Figure 2. An effector–immunity pair is required for E1 colonization resistance.
A Nucleotide alignment of the T6SS locus from N1 and E1. Percent identity is indicated as height, green representing high homology with red highlighting non-
conserved regions.
B–E Co-colonization of N1 WT (B and D, n = 4 mice) or N1 Dbte2 (C, n = 4) with E1 WT (B and C) or E1 overexpressing Bti2a(E1 pBti2a, D). Fecal CFU was monitored
over time (B–D) and E1 CFU compared to N1 WT-E1 WT group at 4 weeks post co-colonization (E).
Data information: Results are representative of two independent experiments. Data are presented as mean SEM (B–D) or mean SD (E). **P < 0.01, ***P < 0.001,
****P < 0.0001. Statistical significance was determined by unpaired, parametric, two-tailed Student’s t-test at each time point (applying Bonferroni correction)
comparing the co-colonizing strains (B–D) or one-way ANOVA, Tukey’s multiple comparisons test (E).
ª 2016 The Authors EMBO reports Vol 17 |No9 | 2016
Aaron L Hecht et al Strain competition restricts a pathogen EMBO reports
1283
priority effect of B. fragilis intestinal niche establishment, which can
be overcome when challenged by E2 (Fig 3C, dashed box). This
suggests that modular, genetically encoded factors and environmen-
tally driven gene regulation mediate complex strain–strain interac-
tions. In this context, host protection against ETBF colonization is
dependent upon strain of initial exposure.
T6S provides colonization resistance against challenge strains
T6SSs have been implicated in colonization resistance against
pathogen invasion [20,21,32]; however, in vivo molecular evidence
for this is lacking. The role of T6S observed in our co-colonization
study suggested that this system might govern intraspecific
competition in secondary challenge. To examine this hypothesis, we
generated a tssC deletion mutant in N2 (N2 DtssC), a strain that
demonstrates broad colonization resistance (Fig 3C). Deletion of tssC
relieved N2 colonization resistance against N1, which was restored
upon plasmid-based complementation (Fig 4A and B). Loss of tssC
did not alter N2 primary colonization (Fig 4 and Appendix Fig S2A, C
and E), or self-secondary exclusion of N2 (Fig 4C and D). While N2
DtssC retained colonization resistance against E1, the rate of
elimination was significantly reduced compared to wild-type N2
(Fig 4E and F). Secondary strain recovery one day post-challenge
was not significantly different between groups, emphasizing the
importance of the colonic environment in mediating competition
(Appendix Fig S2B, D and F). These data show that T6S is important
for non-self-colonization resistance in vivo and is a key contributor
to strain stability in the microbiota [22–25].
It is clear from these findings that T6S alone cannot explain the
complex strain–strain interactions observed (Fig 3C). To date, one
other antibacterial factor has been identified in B. fragilis. Bacteroi-
dales secreted antimicrobial protein 1 (BSAP-1) is a membrane
attack complex/perforin (MACPF)-containing protein, produced by
N2 that displays N1 killing properties in vitro [1,25]. Mutation of
bsap-1 in N2 (N2 Dbsap-1) had no effect on N2 colonization resis-
tance against N1 in vivo (Appendix Fig S3), demonstrating the
importance of the in vivo niche in defining factors that mediate
intraspecific colonization resistance.
The commensal colonization factor (ccf) locus of B. fragilis
enables niche occupancy within the colonic crypt and is required by
N1 for self-colonization resistance [31]. Diverged ccf loci of
B. thetaiotaomicron and B. vulgatus, two species closely related to
B. fragilis, are suggested to define separate niches for those organ-
isms, supported by the observation that N1 does not exhibit
AB
C
Figure 3. B. fragilis provides strain-specific colonization resistance.
A, B Initial colonization of gnotobiotic (A) or SPF (B) mice (n = 4 mice per group) with N1 followed by secondary challenge with N1 (closed squares) or E1 (open
squares). Fecal CFU was determined for the primary and secondary colonization strains through 4 weeks post-secondary challenges.
C All primary colonization and secondary challenge pairs were tested with 3 NTBF and 2 ETBF strains. Stable colonization of the secondary challenge strain
significantly above the limit of detection is denoted as a “+” while non-significance is denoted as a “” (n = 4 mice per group). The diagonal gray bar indicates
self-secondary challenge, the horizontal red bars show strains that provide broad colonization resistance against non-self strains, and the vertical dashed box
indicates a strain that has an enhanced secondary colonization phenotype.
Data information: Results illustrate a single experiment (A) or are representative of at least two independent experiments (B and C). Data are presented as mean SEM.
Arrows denote day of primary colonization and secondary challenge. A dashed line denotes limit of detection. **P < 0.01, ****P < 0.0001. Statistical significance was
determined by unpaired, parametric, two-tailed Student’s t-test at each time point (applying Bonferroni correction) comparing the secondary challenge strains.
EMBO reports Vol 17 |No9 | 2016 ª 2016 The Authors
EMBO reports Strain competition restricts a pathogen Aaron L Hecht et al
1284
colonization resistance against either species [31]. It is unknown
whether these niches are spatially distinct from the B. fragilis niche
or whether these Bacteroides species interact in the colon. N1 kills
B. thetaiotaomicron through T6S in vitro [26], which we confirmed
for N2 (Fig EV3A). N2, however, did not confer resistance against
secondary challenge by B. thetaiotaomicron in vivo (Fig EV3B), indi-
cating the species specificity of colonization resistance. Mutation of
the T6SS of N2 had no effect on fecal recovery of B. thetaiotaomi-
cron after secondary challenge (Fig EV3B–D). Similarly, despite T6S-
dependent killing of B. vulgatus in vitro (Fig EV3E), N2 did not
restrict B. vulgatus in secondary challenge (Fig EV3F–H). As T6S is
contact-dependent, these data suggest a physical niche separation
between these Bacteroides species and implies that intraspecies
competition is a primary function of Bacteroides T6S in vivo.
AB
CD
EF
Figure 4.T6S is required for strain-specific colonization resistance.
A–F Primary colonization of SPF mice with N 2 WT, T6SS mutant (DtssC), and complemented (DtssC pTssC) followed by secondary challenge with N1 WT (A and B, n = 5
mice), N2 WT (C and D, n = 5), or E1 WT (E and F, n = 4) was performed. Fecal CFU for primary and secondary strains was determined for 4 weeks post-seconda ry
challenge (A, C, E). Selected time points were tested for statistical difference of secondary challenge between groups. This includes 4 weeks post-secondary
challenge (B) and 3 days post-challenge (D and F).
Data information: Results are representative of three independent experiments. Data are presented as mean SEM (A, C, and E) or mean SD (B, D, and F). Arrows
denote day of primary colonization and secondary challenge. A dashed line denotes limit of detection. n.s., not significant; ****P < 0.0001. Statistical significance was
determined by unpaired, parametric, two-tailed Student’s t-test at each time point (applying Bonferroni correction, D and F) or one-way ANOVA, Tukey’s multiple
comparisons test (B).
ª 2016 The Authors EMBO reports Vol 17 |No9 | 2016
Aaron L Hecht et al Strain competition restricts a pathogen EMBO reports
1285