Search

A novel in vitro gut model was developed to better understand the interactions between Escherichia coli and the mouse cecal mucus commensal microbiota. The gut model is simple and inexpensive while providing an environment that largely replicates the nonadherent mucus layer of the mouse cecum. 16S rRNA gene profiling of the cecal microbial communities of streptomycin-treated mice colonized with E. coli MG1655 or E. coli Nissle 1917 and the gut model confirmed that the gut model properly reflected the community structure of the mouse intestine. Furthermore, the results from the in vitro gut model mimic the results of published in vivo competitive colonization experiments. The gut model is initiated by the colonization of streptomycin-treated mice, and then the community is serially transferred in microcentrifuge tubes in an anaerobic environment generated in anaerobe jars. The nutritional makeup of the cecum is simulated in the gut model by using a medium consisting of porcine mucin, mouse cecal mucus, HEPES-Hanks buffer (pH 7.2), Cleland’s reagent, and agarose. Agarose was found to be essential for maintaining the stability of the microbial community in the gut model. The outcome of competitions between E. coli strains in the in vitro gut model is readily explained by the “restaurant hypothesis” of intestinal colonization. This simple model system potentially can be used to more fully understand how different members of the microbiota interact physically and metabolically during the colonization of the intestinal mucus layer. IMPORTANCE Both commensal and pathogenic strains of Escherichia coli appear to colonize the mammalian intestine by interacting physically and metabolically with other members of the microbiota in the mucus layer that overlays the cecal and colonic epithelium. However, the use of animal models and the complexity of the mammalian gut make it difficult to isolate experimental variables that might dictate the interactions between E. coli and other members of the microbiota, such as those that are critical for successful colonization. Here, we describe a simple and relatively inexpensive in vitro gut model that largely mimics in vivo conditions and therefore can facilitate the manipulation of experimental variables for studying the interactions of E. coli with the intestinal microbiota.

Escherichia coli MG1655, a K-12 strain, uses glycolytic nutrients exclusively to colonize the intestines of streptomycin-treated mice when it is the only E. coli strain present or when it is confronted with E. coli EDL933, an O157:H7 strain. In contrast, E. coli EDL933 uses glycolytic nutrients exclusively when it is the only E. coli strain in the intestine but switches in part to gluconeogenic nutrients when it colonizes mice precolonized with E. coli MG1655 (R. L. Miranda et al., Infect Immun 72:1666–1676, 2004, http://dx.doi.org/10.1128/IAI.72.3.1666-1676.2004 ). Recently, J. W. Njoroge et al. (mBio 3:e00280-12, 2012, http://dx.doi.org/10.1128/mBio.00280-12 ) reported that E. coli 86-24, an O157:H7 strain, activates the expression of virulence genes under gluconeogenic conditions, suggesting that colonization of the intestine with a probiotic E. coli strain that outcompetes O157:H7 strains for gluconeogenic nutrients could render them nonpathogenic. Here we report that E. coli Nissle 1917, a probiotic strain, uses both glycolytic and gluconeogenic nutrients to colonize the mouse intestine between 1 and 5 days postfeeding, appears to stop using gluconeogenic nutrients thereafter in a large, long-term colonization niche, but continues to use them in a smaller niche to compete with invading E. coli EDL933. Evidence is also presented suggesting that invading E. coli EDL933 uses both glycolytic and gluconeogenic nutrients and needs the ability to perform gluconeogenesis in order to colonize mice precolonized with E. coli Nissle 1917. The data presented here therefore rule out the possibility that E. coli Nissle 1917 can starve the O157:H7 E. coli strain EDL933 of gluconeogenic nutrients, even though E. coli Nissle 1917 uses such nutrients to compete with E. coli EDL933 in the mouse intestine.

Previously we reported that the streptomycin-treated mouse intestine selected for two different Escherichia coli MG1655 mutants with improved colonizing ability: nonmotile E. coli MG1655 flhDC deletion mutants that grew 15% faster in vitro in mouse cecal mucus and motile E. coli MG1655 envZ missense mutants that grew slower in vitro in mouse cecal mucus yet were able to cocolonize with the faster-growing flhDC mutants. The E. coli MG1655 envZ gene encodes a histidine kinase that is a member of the envZ-ompR two-component signal transduction system, which regulates outer membrane protein profiles. In the present investigation, the envZ P41L gene was transferred from the intestinally selected E. coli MG1655 mutant to E. coli Nissle 1917, a human probiotic strain used to treat gastrointestinal infections. Both the E. coli MG1655 and E. coli Nissle 1917 strains containing envZ P41L produced more phosphorylated OmpR than their parents. The E. coli Nissle 1917 strain containing envZ P41L also became more resistant to bile salts and colicin V and grew 50% slower in vitro in mucus and 15% to 30% slower on several sugars present in mucus, yet it was a 10-fold better colonizer than E. coli Nissle 1917. However, E. coli Nissle 1917 envZ P41L was not better at preventing colonization by enterohemorrhagic E. coli EDL933. The data can be explained according to our “restaurant” hypothesis for commensal E. coli strains, i.e., that they colonize the intestine as sessile members of mixed biofilms, obtaining the sugars they need for growth locally, but compete for sugars with invading E. coli pathogens planktonically.

Previously, we reported that the streptomycin-treated mouse intestine selected nonmotile Escherichia coli MG1655 flhDC deletion mutants of E. coli MG1655 with improved colonizing ability that grow 15% faster in vitro in mouse cecal mucus and 15 to 30% faster on sugars present in mucus (M. P. Leatham et al., Infect. Immun. 73:8039–8049, 2005). Here, we report that the 10 to 20% remaining motile E. coli MG1655 are envZ missense mutants that are also better colonizers of the mouse intestine than E. coli MG1655. One of the flhDC mutants, E. coli MG1655 Δ flhD , and one of the envZ missense mutants, E. coli MG1655 mot-1, were studied further. E. coli MG1655 mot-1 is more resistant to bile salts and colicin V than E. coli MG1655 Δ flhD and grows ca. 15% slower in vitro in mouse cecal mucus and on several sugars present in mucus compared to E. coli MG1655 Δ flhD but grows 30% faster on galactose. Moreover, E. coli MG1655 mot-1 and E. coli MG1655 Δ flhD appear to colonize equally well in one intestinal niche, but E. coli MG1655 mot-1 appears to use galactose to colonize a second, smaller intestinal niche either not colonized or colonized poorly by E. coli MG1655 Δ flhD . Evidence is also presented that E. coli MG1655 is a minority member of mixed bacterial biofilms in the mucus layer of the streptomycin-treated mouse intestine. We offer a hypothesis, which we call the “Restaurant” hypothesis, that explains how nutrient acquisition in different biofilms comprised of different anaerobes can account for our results.