Interbacterial mechanisms of colonization resistance and the strategies pathogens use to overcome them

The communities of bacteria that reside in the intestinal tract are in constant competition within this dynamic and densely colonized environment. At homeostasis, the equilibrium that exists

between these species and strains is shaped by their metabolism and also by pathways of active antagonism, which drive competition with related and unrelated strains. Importantly, these

normal activities contribute to colonization resistance by the healthy microbiota, which includes the ability to prevent the expansion of potential pathogens. Disruption of the microbiota,

resulting from, for example, inflammation or antibiotic use, can reduce colonization resistance. Pathogens that engraft following disruption of the microbiota are often adapted to expand

into newly created niches and compete in an altered gut environment. In this review, we examine both the interbacterial mechanisms of colonization resistance and the strategies of pathogenic

strains to exploit gaps in colonization resistance.

The intestinal microbiota is composed of diverse communities of microorganisms that colonize different regions of the gastrointestinal tract. The composition of these communities varies

along the length of the intestine, in addition to variation from the lumen to the mucous and enterocyte barrier.1 Dramatic transitions in the microbiota occur with age, and a major shift

from the microbes present following initial colonization in the neonatal period to the mature state has been well documented.2,3,4,5 Although the microbiota is highly complex, at higher

taxonomic levels consistent patterns exist between healthy individuals.1 For example, the healthy colon is dominated by strict anaerobes in the Bacteroidetes and Firmicutes phyla.6,7

Furthermore, significant similarity between individuals is observed if diversity is assessed on the basis of the conservation of functional pathways present in the microbiota.6,8

Expansion of some opportunistic species including Enterococcus faecalis, Enterococcus faecium and members of the Enterobacteriaceae family that are typically present in low abundance to very

high levels can have negative consequences for the host. For pathogenic taxa, such as Salmonella enterica serovar Typhimurium (Salmonella) or Clostridium difficile, expansion of these

species in the gut triggers an inflammatory response and diarrhea (reviewed in refs. 9,10). For other species, including Klebsiella pneumoniae, a member of the Enterobacteriaceae, and

Enterococcus faecium, expansion in the gut does not trigger overt intestinal inflammation but is associated with an increased risk for patients to subsequently develop bacteremia.11,12

Increasingly, the strains that dominate the gut in clinical settings are resistant to one or more antibiotics. Indeed, in the United States, the Center for Disease Control has classified

vancomycin-resistant Enterococcus (VRE), carbapenum-resistant Enterobacteriaceae, and extended spectrum β-lactamase producing Enterobacteriaceae (ESBLs) as serious or urgent threats.13

In mouse models and in patients, expansion of these potential pathogens can be triggered by disruption of the microbiota. The ability of the healthy microbiota to prevent expansion of these

potential pathogens is termed colonization resistance, and is a critical function of the healthy microbiota under homeostatic conditions. The mechanisms of microbiota-mediated colonization

resistance are diverse. In general, resistance mechanisms prevent expansion of pathogenic bacteria by either direct, bacteria-to-bacteria pathways, or by activating host immune defenses.

Conversely, opportunistic bacterial species that dominate the gut use diverse strategies to exploit breakdowns in colonization resistance, leading to their expansion into newly created

niches.

In this review, we examine direct interbacterial mechanisms of colonization resistance, as well as the strategies used by opportunistic species to compete and take advantage of the breakdown

in colonization resistance that enables their expansion.

The important function of the microbiota in providing colonization resistance has been recognized for decades.14,15 Indeed, work from Bohnhoff et al. in 1954 demonstrated that

antibiotic-mediated disruption of the microbiota leads to a dramatic increase in susceptibility to Salmonella infection.15 These early pioneering studies and others generated interest in

understanding the mechanisms of colonization resistance. Here we highlight some of the recent advances in our understanding of this critical function.

When food is consumed, host mechanical and chemical processes function to break down large complexes of protein, fat, and carbohydrates, thereby enabling absorption of nutrients. In addition

to host-directed pathways, the microbiota plays an important role in digestion by expressing a complementary suite of catabolic pathways (Fig. 1). Microbiota-dependent digestion supports

the growth of the bacterial community and generates byproducts that are utilized by the host, such as short-chain fatty acids (SCFAs). Competition for nutrients in the gut is intense, and

evolution has shaped the cross-feeding patterns and substrate preferences of the microbiota to maximize utilization of the available substrates.16,17,18 Therefore, under homeostatic

conditions, an exogenous species is unlikely to find an uncontested niche, and will be forced to compete with the established microbiota for nutrients.19,20 Interestingly, the odds of

success for an exogenous species in establishing residence in the gut is increased if the microbiota of the gut harbors related taxa to the new species, suggesting that appropriate nutrient

conditions exist.21

Microbiota metabolism contributes to colonization resistance. a Members of the microbiota, such as C. scindens, convert primary bile salts (blue stars) into secondary bile salts (red stars)

that inhibit the vegetative growth of C. difficile. b The healthy microbiota ferments diet-derived simple sugars (purple/pink circles), complex polysaccharides (blue lines), and

microbiota-liberated metabolites from the mucous layer (brown circles) to produce inhibitory SCFA. c In the absence of competing strains, E. coli and C. rodentium utilize the increased

availability of simple sugars for replication

The abnormal expansion of Citrobacter rodentium (C. rodentium) or Escherichia coli (E. coli) in the gut is fueled by utilization of simple sugars.22,23,24 Kamada et al. found that C.

rodentium competes for simple sugars with the endogenous microbiota.22 Furthermore, shifting the diet to only simple sugars forced competition with strains like Bacteroides thetaiotaomicron

that normally occupy different nutritional niches.22 E. coli is able to utilize a variety of sugars present in the mucous layer simultaneously, including galactose, fucose, mannose, N-acetyl

glucosamine, arabinose, and ribose,23 and commensal strains of E. coli can provide colonization resistance against pathogenic E. coli O157 based on overlapping sugar utilization.25,26

Together, these results highlight the potential importance of carbon and energy source availability. However, additional studies are required to determine the contribution of nutrient

limitation in the context of a normal diet and a complex, unperturbed microbiota in providing colonization resistance. In line with this, studies in both mice and humans have demonstrated

that substrate availability and competition for those substrates along with subsequent microbiota structure is strongly influenced by diet and the presence of appropriate substrate

utilization pathways.27,28,29

In addition to competition for carbon and energy sources, invading bacteria compete with the microbiota for trace metals. For example, the probiotic Nissle 1917 E. coli strain protects mice

from pathogenic Salmonella infection in manner that is dependent on iron uptake.30 Furthermore, zinc uptake pathways in Vibrio cholerae and Campylobacter jejuni are critical to the ability

of these pathogens to compete with the densely colonizing microbiota.31 The importance of this competition for trace metals is highlighted by observations demonstrating that immunization of

mice to generate antibodies against iron-scavenging siderophores is sufficient to reduce the burden of Salmonella infection.32

Bile salts are secreted into the small intestine by the host and function to increase fat solubilization and digestion.33 Bile salts are generated in the liver from enzymatic conversion of

cholesterol, and the majority are conjugated to glycine or taurine through an amide linkage.34 The majority of secreted bile salts are reabsorbed in the small intestine; however, unabsorbed

bile salts continue to the colon where they are metabolized by the microbiota into secondary bile acids. Microbiota-driven metabolism of bile salts begins with deconjugation from glycine or

taurine, in a reaction catalyzed by bile salt hydrolases.35 These hydrolases are expressed by a large number of taxa within the normal microbiota.35 Further metabolism of bile salts is

encoded in bile-acid-inducible (Bai) operons by a small subset of commensal species of the microbiota that includes Clostridium scindens and catalyze 7α/β-dehydroxylation35,36 (Fig. 1).

Interestingly, secondary bile salts have been reported to induce the germination of C. difficile spores and inhibit the replication of vegetative cells in a concentration-dependent manner.37

Transfer of C. scindens restores colonization resistance against C. difficile in a bile-salt-dependent manner.38 In agreement, the kinetics of C. difficile germination and the initiation of

pathology correlate with antibiotic-induced changes in bile salt levels and compositions,39 and varying antibiotic regimens leading to different bile salt levels correlated with

susceptibility to C. difficile expansion.40 The relative resistance of various clinical C. difficile isolates to the secondary bile-salt lithocholic acid positively correlates with disease

severity in murine infection models.41,42

The dominant phyla of the large intestine, Bacteroidetes and Firmicutes, are anaerobic bacteria that metabolize diverse substrates including complex polysaccharides, simple sugars, and

glycans released from ingested fiber and also host mucus using fermentation pathways.43,44 This fermentation process leads to the generation of SCFAs (Fig. 1). The most abundant SCFA in the

gut are acetate, propionate, and butyrate and the total concentration of SCFA in the proximal colon can reach concentrations of 70–140 mM.45,46 The majority of the SCFA are produced in the

proximal colon and are absorbed by the host to support enterocyte metabolism.47 In the proximal colon, high levels of SCFA production contribute to acidification of the lumen.45 For example,

using a wireless motility capsule approach in human subjects, Farmer et al. reported average pH values immediately distal to the ileal:caecal junction of 6.16 in healthy individuals.48 In

vitro, numerous studies have demonstrated that SCFAs at an acidic pH are able to inhibit the replication of E. coli and Salmonella,49,50,51 and loss of SCFA production and acidity were

correlated with in vivo susceptibility to Salmonella following streptomycin treatment.14 However, other work has argued against a significant role for direct SCFA-mediated inhibition in

vivo.19,52 During homeostasis, the production of SCFAs by the microbiota also plays a critical role in maintaining the limited availability of oxygen and nitrates in the lumen of the

gut.53,54 Recent work has demonstrated that butyrate production by Clostridia activates PPAR-γ in epithelial cells resulting in the repression of Nos2 expression, and activation of

oxygen-consuming β-oxidation as opposed to anaerobic glycosis.54,55,56 Loss of butyrate production triggers increased oxygen and nitrate release into the lumen of the gut, which can be

exploited by Enterobacteriaceae, as described below. In addition, several studies have demonstrated that the detection of SCFAs triggers a complex modulation of virulence factor expression

in Salmonella.57,58,59,60 Therefore, the relative contribution of direct SCFA-mediated antagonism to colonization resistance compared with other mechanisms of colonization resistance or,

indeed, other indirect effects of SCFAs remains unclear.

Established members of the microbiota are also able to inhibit the expansion of new species through mechanisms of active antagonism. In the context of an exogenously introduced pathogen,

these pathways contribute to colonization resistance. However, in many cases these mechanisms of antagonism can target the same or closely related species, suggesting that, in the absence of

exogenous pathogenic bacteria, these pathways function to allow expressing strains to compete with other commensal strains capable of utilizing similar nutrients.

Microbiota-derived antimicrobials have been identified that have activity against both Gram-positive and Gram-negative pathogens. The original natural variants of clinical antibiotics are

produced through complex and diverse biosynthetic pathways, and include polyketides (erythromycin, tetracycline),61 non-ribosomally produced glycopeptides (vancomycin),61 aminoglycosides

(neomycin),62 and lincosamides (clindamycin).63 Within the gut, the microbiota-produced antimicrobials are most frequently ribosomally synthesized and post-translationally modified peptides

(RiPPs), also termed bacteriocins, which are processed and secreted from the cell64 (Fig. 2).

Mechanisms of active antagonism contribute to colonization resistance: a Constituents of the microbiota can express contact-dependent secretion systems that deliver antagonistic effector

proteins. In Gram-negatives, such as Bacteroides, delivery is driven by the T6SS, while the Esx system has recently been identified as having an analogous role in Gram-positive bacteria. b

In Bacteroides, the GAI-1 and GAI-2 variants of the T6SS are encoded on conjugative plasmids enabling horizontal gene transfer to other strains in the microbiota. c Short peptide

antimicrobials, termed bacteriocins, are secreted by the microbiota and have activity against both related non-immune commensal strains and exogenous species

Some of the best-characterized bacteriocins are those produced by members of Lactobacillales order. Nisin is a prototypical member of the lantibiotic class of bacteriocin that is produced by

Lactococcus lactis that inhibits a broad range of Gram-positive bacteria including Staphylococcus aureus, and Listeria monocytogenes.65 Nisin functions by binding to Lipid II leading to

pore formation and has bactericidal activity at nanomolar levels in Gram-positive bacteria where the peptidoglycan layer is exposed.66 As a result of its potency, Nisin has been used as a

food preservative for decades to prevent the expansion of Gram-positive pathogens.65 Beyond Nisin, bacteriocin production in Lactobacillales is widespread. For example, Lactococcus

salivarius UCC118 production of the bacteriocin Abp118 is able to protect mice from L. monocytogenes infection.67 In comparison to nisin, some bacteriocins target a narrower range of

species.68 Thuricin CD is a bacteriocin produced by commensal Clostridium thuringiensis that has potent antimicrobial activity against C. difficile but does not target other members of the

microbiota including Bifidobacteria or Lactobacillus.68

Although bacteriocins produced by Gram-positive bacteria can have high levels of activity against Gram-positive pathogens, most of these peptides lack significant activity against

Gram-negative bacteria. However, recent work examining the bacteriocin NAI-107 found that this lantibiotic has activity against a small number of fastidious Gram-negative pathogens, and

synergizes with polymyxin to inhibit a broader range of Gram-negative pathogens including Klebsiella pneuomoniae, and E. coli.69 Several groups have also attempted to artificially broaden

the range of targets of bacteriocins.70,71 For example, Nisin that had been complexed to carbon and gold nanoparticles with a net positive charge had increased access to the periplasm of

Gram-negative bacteria and antibacterial activity against E. coli and Pseudomonas aeruginosa.70 Similarly, fusion of Nisin to truncations of Gram-negative targeting antimicrobial peptides

led to increased efficacy against E. coli.71

Using a bioinformatics approach, Donia et al. demonstrated that a large number of biosynthetic gene clusters are widely distributed among the different microbiota communities, and include

pathways for the production of lantibiotics, thiazole/oxazole-modified microcins, and thiopeptides.72 In support of this bioinformatic data, the authors demonstrated that Lactococcus gasseri

produces lactocillin, a thiopeptide with activity against a range of Gram-positive pathogens.72

Some strains of commensal Enterococcus express the bacteriocins AS-48, bac-31, and bac-21 from the pheromone-responsive conjugative plasmids pAD1, pYI17, and pPD1, respectively.73,74,75

Conjugative transfer of these plasmids is triggered by recognition of peptide pheromones secreted by recipient cells,76 which would allow for rapid horizontal propagation of the

bacteriocin-producing phenotype. Interestingly, production of the pheromone cOB1 by commensal strains of Enterococcus can trigger cell death of VRE strains with cOB1 responsive mobile

elements on the pTEF2spc plasmid and an IS-like elements in the main chromosome.77 In vivo, expression of bac-21 increases the long-term colonization of commensal E. faecalis through

increased competitiveness with other related strains.78 Furthermore, the high-density environment of the gut promotes conjugative transfer of bacteriocin-expressing plasmids from positive to

negative strains.78 Finally, by engineering conjugation-defective pPD1 plasmids, Kommineni et al. were able to demonstrate that commensal E. faecalis carrying the engineered plasmids is

able to de-colonize mice of pPD1-negative VRE.78 Potential probiotic Enterococcus strains have been identified that express bacteriocins targeting Gram-positive species including L.

monocytogenes.79

The nasal microbiota provides colonization resistance against S. aureus, and Staphylococcus epidermidis or Streptococcus mitis have been negatively correlated with S. aureus abundance.80,81

To examine the mechanism of protection, Zipperer et al. screened nasal isolates for the ability to inhibit S. aureus and demonstrated that a commensal Staphylococus strain, Staphylococcus

lugdunensis, is able to directly inhibit the growth of S. aureus through the production of a peptide antimicrobial, termed lugdunin.82 In addition to targeting MRSA, lugdunin also inhibits

the growth of Listeria monocytogenes, VRE, and Streptococcus pneuomoniae.82

Members of the Gram-negative Enterobactericeae family can produce microcins, a class of short peptide bacteriocins, that have activity against closely related strains of bacteria.83 A large

number of microcins have been identified in commensal strains of E. coli, including the probiotic Nissle 1917 strain.84 Recent work from Sassone-Corsi et al. demonstrated that probiotic E.

coli Nissle 1917 could compete with both E. coli and Salmonella in a manner dependent on microcin expression.85 Interestingly, this successful competition occurred only in inflamed

conditions but not in mice treated only with streptomycin. Under inflamed conditions the competition for iron becomes critical for Enterobacteriaceae expansion.30 By linking these

observations, Sassone-Corsi et al. demonstrated that microcin expression is induced by iron deprivation, and that microcin-dependent competition in the gut depends on an iron-limited

environment.85

In contrast to shorter peptides, some species of the microbiota also produce larger proteins with antimicrobial activities. For example, some members of the Bacteroidales, including

Bacteroides fragilis and Bacteroides uniformis, can produce Bacteroidales-secreted antimicrobial proteins that inhibit related species and increase competitiveness in the gut.86,87 In

addition, some B. fragilis strains can secrete a ubiquitin-like protein termed BfUbb that has antimicrobial activity against related strains.88 Similarly, some strains of Enterobacteriaceae

carry plasmids with genes encoding colicins, a class of antimicrobial proteins that target related species lacking the appropriate immunity proteins.89 Unlike the best-characterized peptide

antimicrobials which often target the cell wall, colicins must be transported by specific receptors on target Gram-negative cells into the periplasmic space and can have diverse activities

inside the cell.89 Interestingly, work examining the dynamics of competition between colicin-resistant, colicin-sensitive, and colicin-producing E. coli strains has found that colicin

production can actually promote diversity of related strains in the gut as equilibrium is reached.90

Type VI secretion systems (T6SS) are complex structures that allow for export of effector molecules across the inner and outer membranes of Gram-negative bacteria and into adjacent bacteria

or eukaryotic cells.91 Therefore, unlike antimicrobial compounds or toxic byproducts of metabolism, T6SS activity requires cell-to-cell contact for antagonism (Fig. 2). The

best-characterized T6SS are those expressed by pathogenic Gammaproteobacteria.92 The delivered effectors antagonize neighboring cells through varied mechanisms of action, including

degradation of NAD(P)+ and peptidoglycan hydrolysis.91,93,94,95 Critically, T6SS+ strains must encode an immunity protein that negates the activity of the effector to avoid self-targeting.93

In healthy adult individuals, Bacteroidetes are typically a major constituent of the anaerobic community in the large intestine. Interestingly, orthologs of the well-characterized

Proteobacterial T6SS gene clusters are widespread in Bacteroidetes taxa,96 and can be horizontally transferred between strains97,98 (Fig. 2). T6SS in Bacteroidales are present in three forms

termed GAI-1, GAI-2, and GAI-3.98 GAI-1 and GAI-2 are widely distributed among Bacteriodales while GAI-3 is restricted to B. fragilis.98,99 In vivo, an individual’s microbiota tends to

harbor single variants of GAI-1/GAI-2 immunity/effector genes suggesting that a stable consortium of Bacteroidales in a gut share T6SS by horizontal gene transfer and subsequently resist

colonization from new strains.99 On the basis of initial in silico identification of these orthologs, Russell et al. examined the expression of T6SS in vivo and found robust expression of

T6SS components by both B. fragilis and Bacteroides eggerthii.96 In vitro, B. fragilis was able to inhibit the growth of a T6SS- strain of Bacteroides thetaiotaomicron in a T6SS-dependent

manner.96 More recent reports have shown that B. fragilis T6SS are functional in vivo and mediate competition with related strains including T6SS- B. fragilis and Bacteroides

vulgatus.100,101 Interestingly, both studies found that target strains with susceptibility to in vitro T6SS-antagonism are not necessarily susceptible to in vivo T6SS-antagonism, and species

such as B. thetaiotaomicron can be unaffected by the presence of a T6SS-expressing competitor despite in vitro susceptibility.100,101 Importantly, T6SS-expressing B. fragilis can provide

colonization resistance against pathogenic enterotoxigenic B. fragilis (ETBF).102

Recent work from Whitney et al. has demonstrated the existence of a cell-to-cell contact-dependent mechanism of antagonism in Gram-positive bacteria, including in the genomes of gut resident

species.103 In this pathway, Gram-positive bacteria are able to secrete members of the LxG group of proteins, including toxins, through the Esx secretion system.103,104 For example, TelC is

a cell wall targeting effector of the Esx secretion system,103 and putative NAD(P)+ hydrolyzing toxins with LXG domains have recently been identified.105 This suggests that, similar to

Gram-negative T6SS, contact-dependent secretion systems in Gram-positive bacteria might also secrete a diverse group of effectors. However, it remains unclear if the Esx-type secretion

system is capable of delivering toxins across the thick peptidoglycan layer of target Gram-positive cells in a manner analogous to intracellular delivery toxins by a T6SS.103

Under homeostatic conditions, the microbiota is a highly complex community that as a whole provides colonization resistance to exogenously introduced species. Clinically, fecal microbiota

transplantation has been used successfully to treat recurrent infection with C. difficile.106,107 Therefore, transferring a complex community of bacteria can restore colonization resistance.

As specific pathways of colonization resistance are identified, along with the microbiota constituents responsible for those pathways, it will likely be possible to design minimal consortia

of bacteria that provide colonization resistance or clearance against specific clinically relevant species. In agreement, several papers have already identified small consortia that provide

full or partial colonization resistance against VRE, L. monocytogenes, Salmonella, or C. difficile.38,108,109,110,111,112

As described above, under steady-state conditions, the healthy microbiota effectively excludes newly introduced potential pathogens through an array of mechanisms. In agreement, clinically

relevant expansion of potential pathogens often occurs after the microbiota is perturbed with antibiotics (reviewed in ref. 113). Unfortunately, common nosocomial pathogens, including

members of the Enterobacteriaceae, VRE, and C. difficile are particularly adept at dominating the gut following perturbation of the microbiota.

Perturbation of the healthy gut microbiota creates an opportunity for expansion, and drives intense competition between related strains to expand into those niches. The idea that species

which expand into dysbiotic gut niches exhibit their own forms of colonization resistance and competition is demonstrated by the observation that initial colonization of the gut by a

Salmonella strain prevents subsequent establishment of other isogenic wild-type strains.114 In this section, we examine the strategies utilized by these pathogens to exploit gaps in

colonization resistance, compete for newly created niches, and subsequently maintain dominance in the dysbiotic gut.

Recent work has demonstrated that pathogens that expand into dysbiotic conditions utilize a variety of mechanisms to antagonize both the normal microbiota and other species or strains that

are also seeking to rapidly expand. For example, as described above, the inflamed gut is an iron-limited environment. This environment can signal to initiate production of microcins in

probiotic E. coli;85 however, pathogenic members of the Enterobacteriaceae such as Salmonella also use these environmental signals to upregulate expression of their own antimicrobial

proteins, including Col1b.115 In line with this, pathogenic strains of E. coli causing urinary tract infections carry a greater number of microcin genes, and have higher prevalence of

microcin H47 and colicin E1 compared to nonpathogenic E. coli strains.116 The gut microbiota has a high level of colonization resistance against Listeria expansion.109 Likely in response to

this selective pressure, epidemic Listeria strains have virulence factors that target the microbiota.117,118 In the gut, Listeria expresses Listeriolysin S, which is able to inhibit the

growth of members of the microbiota, including Lactococcus lactis, in order to promote infection.117,118

Many Gram-negative pathogens, including V. cholerae, P. aeruginosa, and S. typhimurium, have T6SS that are expressed during colonization and deliver a wide range of effectors to give a

fitness advantage in a dysbiotic gut119,120,121 (Fig. 3). Strikingly, among Proteobacteria harboring a T6SS and an identifiable effector, 40% of strains had two effectors with different

predicted activities, and 25% had three or more effectors.122 Work from LaCourse et al. found that P. aeruginosa secretes multiple effectors whose activities can be both synergistic and

dependent on environmental conditions.122 In addition to its well-characterized Type III secretion systems SPI-1 and SPI-2, recent work has demonstrated that Salmonella also utilizes a T6SS

to promote its expansion in the gut.121 The Salmonella T6SS is able to target commensal strains in vitro and ΔT6SS strains are deficient in in vivo colonization.121 Both Salmonella and

Pseudomonas tightly regulate the activity of the T6SS and recognize cues present in the gut to activate the T6SS. These cues include both host-derived factors such as bile salts, as well as

cell stress or kin cell lysis.120,121,123

Expansion and engraftment strategies of invading species. a Some pathogens, such as Salmonella and P. aeruginosa, target either the microbiota or other opportunistic species through using

T6SS’s and bacteriocin production. b Pathogens can drive host inflammatory responses (yellow stars) in order to trigger the release of nitrates or oxygen into the lumen of the gut (shaded

gradient), and the formation of an oxidative environment. Together with novel metabolites formed by oxidation, such as tetrathionate, O2 and NO3 are used in anaerobic and aerobic respiration

pathways by pathogens able to use these electron acceptors

Enterobacteriaceae can rapidly expand following initiation of either dysbiosis or inflammation. In 2007, two reports demonstrated that Salmonella and Citrobacter both drive host-driven

inflammation in order to expand to high levels and compete with the microbiota.124,125 Reducing inflammation using avirulent Salmonella or chemically inducing inflammation with DSS is

sufficient to prevent or enable expansion in the gut, respectively.124,125 Subsequent work has found that the ability to utilize nutrients uniquely present in these conditions is essential

for Enterobacteriaceae (Fig. 3). Importantly, and unlike the anaerobic fermenting bacteria of the microbiota, Enterobactericeae are able to perform both aerobic and anaerobic respiration,

utilizing either O2 or other molecules as terminal electron acceptors in a membrane electron transport chain that generates proton motif force to produce ATP.

Two critical components of the inflammatory response to infection are the activation of inducible nitrate oxide synthase, iNOS, encoded by Nos2, and NADPH oxidase encoded in part by

gp91phox. iNOS generates nitric oxide through the oxidation of l-arginine and nitric oxide,126 and NADPH oxidase generates reactive oxygen species.127 The production of these reactive oxygen

and nitrogen species has long been recognized as two of the primary antibacterial effector mechanisms of the myeloid compartment.128 Importantly, the release of these reactive molecules

also creates an oxidative environment, and reactive oxygen and nitrogen species can oxidize molecules present in the gut.

Tetrathionate was the first inflammation-associated terminal electron acceptor identified for Salmonella.129 Although tetrathionate is not present in the healthy gut, the microbiota produces

hydrogen sulfide which is converted to thiosulfate by the host.130 iNOS-generated nitric oxide is capable of oxidizing thiosulfate to tetrathionate in the lumen.129 In this environment,

Salmonella capable of respiring tetrathionate outcompete respiration-deficient mutants.129 The presence of tetrathionate as an anaerobic electron acceptor also enables Salmonella to utilize

ethanolamine as a carbon source both in vivo and in vitro.131,132

Released nitric oxide can also react to form nitrate, and both NO and NO3 increase in the gut lumen during inflammation.133,134 E. coli are able to utilize nitrates, S-oxides, and N-oxides

as terminal electron acceptors under anaerobic conditions through the production of molybdenum-containing reductases.135 As a result, E. coli with nitrate reductase activity have a

competitive advantage over deficient strains in conditions of intestinal inflammation in DSS-treated or Il10fl/fl CD4-cre mice, as well as in mice with low-level inflammation following

streptomycin treatment.54,133,136

The importance of anaerobic electron acceptors for Salmonella replication is highlighted by the observation that Salmonella expresses two chemotactic receptors Aer and Tsr that enable

chemotaxis toward nitrate and tetrathionate in vitro and are critical for Salmonella to take advantage of the available electron acceptors in vivo.137,138 Furthermore, treatment with

tungsten to inhibit the formation of the molybendum cofactor that is necessary for anaerobic reductases counters the competitive advantage gained by Enterobacteriaceae in the inflamed gut

environment.139

The oxidative environment generated by iNOS also leads to the oxidation of carbon sources in the gut and generation of carbon sources that are absent under homeostatic conditions.140 In an

in vitro assay, reactive nitrogen oxidizes galactose to galactarate and glucose to glucarate. In vivo, galactarate and glucarate increase in the lumen following streptomycin treatment, and

this increase is Nos2 dependent.140 Both E. coli and S. Typhimurium are able to use glucarate and galactarate as carbon and energy sources, and mutants that are deficient in this pathway are

outcompeted in the gut following antibiotic treatment.140

As described above, the normal microbiota plays an essential role in maintaining an anaerobic environment in the gut and depletion of Clostridia taxa leads to increased oxygen availability

in the lumen of the gut.53,54 Under microaerophilic conditions, Enterobacteriaceae express cytochrome oxidase variants with increased oxygen affinity that enable them to utilize oxygen as a

terminal electron acceptor at low oxygen tensions.141 Salmonella and E. coli with the ability to utilize oxygen in these conditions have a competitive advantage compared to deficient

strains.53,54 Similarly, Citrobacter actively promotes oxygenation of the lumen by anchoring to enterocytes and promoting crypt hyperplasia using T3SS-delivered effectors.142 This functions

to create an inflammatory environment where wild-type strains able to utilize oxygen outcompete deficient strains.142 Together, the presence of both oxygen and other anaerobic electron

acceptors promotes Enterobacteriaceae expansion. Indeed, the loss of both pathways triggers a synergistic colonization defect in E. coli,54 and both pathways enable the usage of alternate

carbon sources including ethanaloamine or 1,2-propanediol.131,143

Disruption of the microbiota can trigger transient increases in metabolites that are normally present but typically at low levels owing to utilization by the microbiota.144,145,146,147 For

example, both Salmonella and C. difficile can utilize sialic acid that is cleaved from host mucus by microbiota-encoded sialidase enzymes during their enteric expansion.144,148

Alternatively, C. difficile, C. rodentium, and Salmonella can utilize microbiota-derived succinate during expansion.145,147,149 Work from Maier et al. found that Salmonella is able to

utilize microbiota-generated H2 as an electron donor in the anaerobic gut to drive anaerobic respiration.150 In addition, Salmonella can also utilize fucose that is liberated in a

microbiota-dependent manner.144 Fucose is a deoxyhexose sugar that is produced by the mammalian host, and, in the adult gut, glycans containing α(1-2)-linked fucose are exposed to the

intestinal lumen.151 Members of the microbiota, such as Bacteroides acidifaciens or Bacteroides thetaiotaomicron, can cleave fucose from the conjugated proteins, thus enabling utilization by

the microbiome.152,153 Following LPS stimulation of the immune system, fucosylation in the gut is increased in an fut2-dependent manner.152 Unexpectedly, increased availability of fucose

increases tolerance of C. rodentium, without decreasing pathogen burden and correlating with reduced virulence gene expression.152 As described above, Citrobacter actively drives

inflammatory responses partly in order to support its oxidative metabolism. Therefore, the release of fucose by fucosidase activity of the microbiota could provide substrates for

Citrobacter, counteracting the importance of pathogen-driven intestinal inflammation. Overall, pathogens that expand in the gut are well adapted to take advantage of transient increases in

microbiota-derived metabolites to fuel their expansion.

Under homeostatic conditions, the normal constituents of the gut microbiota exist in a constant state of competition for nutrients and niches. The intense competition that results has likely

driven evolution of the bacteria present in this environment to both cooperate and compete. In most cases, the mechanisms of colonization resistance against pathogenic species described in

this review are also active against commensal members of the microbiota and likely help to shape the equilibrium that exists between these species. Therefore, a newly introduced potential

pathogen encounters a pre-existing web of interacting bacteria occupying all available niches and successfully competing with related strains. As a result, in most cases, the microbiota

provides effective colonization resistance, and newly introduced pathogens are unable to gain a foothold and expand. If they are able to gain a foothold in the gut, often due to perturbation

of the microbiota, pathogens that can expand in that environment use a similar collection of antagonistic pathways as the microbiota in addition to metabolic flexibility to successfully

engraft.

In this review, we have focused on interbacterial mechanisms of colonization resistance; however, fungi, protozoa, helminths, and bacteriophage can also be present in the gut. Recent studies

have highlighted the fact that non-bacterial gut residents including protists and helminths can influence mucosal immune responses or cell differentiation in the epithelial

barrier,154,155,156 and it has been suggested that bacteriophage may play an important role in shaping the bacterial community.157,158 Therefore, it will be interesting to determine to what

extent these non-bacterial gut residents contribute to colonization resistance through either direct or indirect mechanisms.

A potential layer of complexity in understanding colonization resistance is that the distribution of bacteria in the gut is not homogenous,159 and, therefore, the distribution of different

effector mechanisms of colonization resistance is also likely to vary. To understand how potential pathogens try to establish themselves within the web of the microbiota, it will also be

critical to understand the spatial organization of the gut microbiota and how the pathways of metabolic competition or exclusion and antagonism are distributed across the normal biogeography

of the gut. Understanding the arrangement of both pathogens and the recovering microbiota following antibiotic treatment will likely be critical in predicting colonization outcomes. For

example, our laboratory has demonstrated that Klebsiella and VRE reside in close proximity to the mucous layer, and that despite being physically close, K. pneuomoniae and VRE occupy

different niches in the mouse gut.160 Recent work from Whitaker et al. demonstrated an elegant system for engineering Bacteroides species to express fluorescent proteins driven by promoters

with varying activities to allow for the simultaneous imaging of multiple species in the gut.161 These engineered strains can be coupled with high-throughput imaging platforms designed for

the microbiota.162 Expansion of these and similar techniques to other phyla will likely play a critical role in understanding colonization resistance in the complex spatial organization of

the gut.

The multiple mechanisms of colonization resistance can function simultaneously in the gut. As a result, evaluating the relative importance of each individual pathway is difficult in this

environment. While gene knockout strategies have been widely utilized in Gram-negative members of the microbiota, genetic manipulation of the Gram-positive taxa common in the gut remains

more difficult. Progress has been made in improving the genetic tractability of Gram-positive members of the microbiota.163,164,165 Alternatively, pathways and genes of interest from the

microbiota can also be expressed in genetically tractable species.166 Recent work from Lim et al. introduced a system that allows tunable expression of a desired protein in Bacteroides in

vivo.167 Overall, these advances in genetic manipulation of the microbiota will enable studies with microbiota compositions either lacking or overexpressing different components of

colonization resistance.

The original version of this article contained an error in the published figures, where they appeared in black and white. These have now been corrected to display in colour.

Sekirov, I. et al. Gut microbiota in health and disease. Physiol. Rev. 90, 859–904 (2010).

M.T.S. is supported by a Canadian Institute of Health Research (CIHR) Fellowship. E.G.P. has received funding from NIH grants R01 AI042135, AI095706, U01 AI124275, and P30 CA008748.

Immunology Program, Sloan Kettering Institute, Memorial Sloan-Kettering Cancer Center, New York, NY, 10065, USA

Center for Microbes, Inflammation and Cancer, Memorial Sloan-Kettering Cancer Center, New York, NY, 10065, USA

Anyone you share the following link with will be able to read this content: