Commensal Enterobacteriaceae protect against Salmonella colonization through oxygen competition

SUMMARY

Neonates are highly susceptible to infection with enteric pathogens, but the underlying mechanisms are not resolved. We show that neonatal chick colonization with Salmonella enterica serovar (S.) Enteritidis requires a virulence factor-dependent increase in epithelial oxygenation, which drives pathogen expansion by aerobic respiration. Co-infection experiments with an Escherichia coli strain carrying an oxygen-sensitive reporter suggests S. Enteritidis competes with commensal Enterobacteriaceae for oxygen. A combination of Enterobacteriaceae and spore-forming bacteria, but not colonization with either community alone, confers colonization resistance against S. Enteritidis in neonatal chicks, phenocopying germ-free mice associated with adult chicken microbiota. Combining spore-forming bacteria with a probiotic E. coli isolate protects germ-free mice from pathogen colonization, but protection is lost when the ability to respire oxygen under microaerophilic conditions is genetically ablated in E. coli. These results suggest commensal Enterobacteriaceae contribute to colonization resistance by competing with S. Enteritidis for oxygen, a resource critical for pathogen expansion.

Graphical Abstract

eTOC blurb

Litvak, et al find neonatal chicks are highly susceptible to Salmonella infection because their microbiota is immature. Together, spore-forming bacteria and commensal Enterobacteriaceae confer colonization resistance against Salmonella. Salmonella uses its virulence factors to increase epithelial oxygenation, while Enterobacteriaceae contribute to colonization resistance by competing with Salmonella for oxygen.

INTRODUCTION

Neonates and children less than one year of age exhibit increased susceptibility to infections with enteric pathogens (Lanata et al., 2013; PrabhuDas et al., 2011). Similarly, neonatal chicks are highly susceptible to developing chronic intestinal carriage with Salmonella serovars (Sadler et al., 1969). Colonization of chickens with S. Enteritidis is of particular importance, because the association of this pathogen with chicken eggs is the leading cause of food-borne disease outbreaks in the United States (Olsen et al., 2000).

The high susceptibility of neonates to enteric infections is attributed in part to an immature gut microbiota. In the clean environment of a hatchery, neonate chicks are not exposed to microbiota from adult birds. The complexity of the neonate gut microbiota gradually increases from day 1 to day 19 of life (Crhanova et al., 2011). Whereas facultative anaerobic bacteria belonging to the phylum Proteobacteria initially dominate the neonate gut microbiota, the composition of the microbial community gradually shifts towards a dominance of obligate anaerobic bacteria belonging to the class Clostridia (phylum Firmicutes) (Crhanova et al., 2011; Mon et al., 2015). Transfer of cecal microbiota from adult chickens to neonatal chicks increases resistance to infection with Salmonella serovars (Goren et al., 1988; Nurmi and Rantala, 1973). Administration of defined cultures of avian bacterial isolates confers various degrees of niche protection against infection with Salmonella serovars (Barnes et al., 1980; Barrow and Tucker, 1986; Bielke et al., 2003; Impey et al., 1982; Soerjadi et al., 1978; Zhang et al., 2007), but the underlying mechanisms remain unknown.

Neonatal chicks infected with S. Enteritidis develop intestinal inflammation that peaks between 2 and 4 days after inoculation (Crhanova et al., 2011; Kogut et al., 2016). Infection results in robust colonization of the ceca, with little dissemination to systemic tissues such as the liver and spleen (Troxell et al., 2015). A functional invasion-associated type III secretion system (T3SS-1) is required for S. Enteritidis intestinal colonization of neonatal chick ceca (Porter and Curtiss, 1997), whereas a second type III secretion system (T3SS-2) is required for bacterial survival in tissue (Bohez et al., 2008).

Mechanisms of intestinal colonization have been studied more extensively in the mouse model of S. Typhimurium infection. In adult mice, S. Typhimurium uses its virulence factors, T3SS-1 and T3SS-2, to trigger cecal inflammation (Barthel et al., 2003; Coburn et al., 2005). The respiratory burst of phagocytes that migrate into the intestinal lumen during inflammation generates electron acceptors for anaerobic respiration, including tetrathionate (Winter et al., 2010) and nitrate (Lopez et al., 2015; Lopez et al., 2012). S. Typhimurium-induced colitis is accompanied by depletion of Clostridia from the gut microbiota and a consequent drop in the concentration of microbiota-derived butyrate (Rivera-Chavez et al., 2016). Intestinal inflammation and butyrate depletion cooperate to alter the metabolism of colonic epithelial cells from mitochondrial β-oxidation of fatty acids to anaerobic glycolysis, resulting in elevated epithelial oxygenation (Byndloss et al., 2017). Oxygen emanating from the epithelial surface together with respiratory electron acceptors generated by host phagocytes drive a luminal expansion of S. Typhimurium during colitis (Lopez et al., 2015; Lopez et al., 2012; Rivera-Chavez et al., 2016; Winter et al., 2010). However, it is not clear whether these mechanisms are necessary for colonizing the neonate gut, because the abundance of Clostridia is already low in animals with an immature microbiota.

Here we investigated how virulence factors enable S. Enteritidis to establish a replicative niche in the ceca of neonatal chicks and how bacteria from adult chicken microbiota interfere with these mechanisms to confer niche protection.

RESULTS

S. Enteritidis virulence factors promote colonization of neonates

To investigate the contribution of virulence factors to intestinal colonization we infected one day-old chicks with an S. Enteritidis wild-type strain that was marked with an antibiotic resistance cassette (YL215) or with a mutant in which the functions of T3SS-1 and T3SS-2 were genetically ablated by mutations in invA and spiB, respectively (YL144) (Table 1). By day 7 after infection, the S. Enteritidis wild type was recovered at 1,000-fold higher numbers from cecal contents than the invA spiB mutant (Fig. 1A), suggesting that virulence factors were required for intestinal colonization of neonatal chicks. Intestinal inflammation peaked at day 3 after infection with the S. Enteritidis wild type, as indicated by elevated expression of inflammatory markers in the cecal mucosa (Fig. 1B–1G) and cecal pathological changes characterized by epithelial hyperplasia and/or ulceration, diffuse infiltration of heterophils and macrophages in the lamina propria, and luminal exudates (Fig 1H and Table S1). This inflammatory response was blunted in chicks infected with an invA spiB mutant (Fig. 1B–1H and Table S1), which was consistent with the idea that virulence factors trigger intestinal inflammation (Porter and Curtiss, 1997).

Table 1:

Bacterial strains used in this study

S. Enteritidis strains
TN2	NalR derivative of S. Enteritidis isolate CDC SSU7998	(Boyd et al., 1993; Norris and Baumler, 1999)
YL28	TN2 cyxA	This study
YL55	TN2 phoN::pTH122	This study
YL56	TN2 cyxA phoN::pTH122	This study
YL57	TN2 phoN::pTH124	This study
YL86	TN2 moaA	This study
YL91	TN2 moaA phoN::pTH122	This study
YL126	TN2 spiB::KanR	This study
YL144	TN2 invA::pGP704 spiB::KanR	This study
YL170	TN2 cyoA::KanR	This study
YL180	TN2 invA::pGP704 spiB::KanR cydA::pEP185.2	This study
YL215	TN2 phoN::KanR	This study
YL216	TN2 cydAB::CarbR	This study
S. Typhimurium strains
IR715	NalR derivative of S. Typhimurium isolate ATCC14028	(Stojiljkovic et al., 1995)
AJB715	IR715 phoN::KanR	(Kingsley et al., 2003)
GTW42	IR715 cydAB::CarbR	This study
GTW46	IR715 cydAB::CarbR phoN::KanR	This study
E. coli strains
S17-1 λpir	C600::RP4 2-(Tet::Mu) (Kan::Tn7) λpir recA1 thi pro hsdR (r-m+)	(Simon et al., 1983)
YL178	Avian E. coli O62:H30, isolated from a neonatal chick	This study
E. coli Nissle 1917	E. coli strain isolated in 1917 from human feces	Ardeypharm GmbH
YL17	E. coli Nissle 1917 cydA appC	This study

Figure 1: S. Enteritidis virulence factors promote colonization of the neonatal gut.

1-day old chicks were infected with the 1×10⁹ CFU of the indicated S. Enteritidis strains. (A) S. Enteritidis colonization levels were determined in cecal contents (n = 5 for each time point) at the indicated time points. Statistically significant differences to chicks infected with the S. Enteritidis wild type (wt) are indicated. (B-G) Relative changes in transcript levels of inflammatory markers, including NOS2 (B), CXCL8 (C), IL1B (D), IL6 (E), IFNG (F), and IL22 (G), were determined by quantitative real-time PCR using RNA isolated from cecal tonsils 3 days after infection and were expressed as fold-change over samples from mock-infected animals. (A-G) data represent geometric means ± standard error. (H) Representative images of haematoxylin and eosin-stained cecal sections of chicks 3 days after infection. Scale bars = 400 μm. (I) Histopathology score for the ceca collected 3 days after infection was determined by scoring blinded sections using criteria listed in table S1. Each bar represents data from one individual animal. (B-I) The number of animals in each group (n) is indicated in panel I. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, not statistically significantly different.

In adult mice, virulence factor-dependent cecal inflammation is required for generating nitrate and tetrathionate in the intestinal lumen, two electron acceptors that drive a luminal expansion of S. Typhimurium through anaerobic respiration (Lopez et al., 2015; Lopez et al., 2012; Winter et al., 2010). We thus further investigated whether virulence factors were required for intestinal colonization of neonatal chicks (Fig. 1A) because the inflammatory responses they triggered (Fig. 1B–1H) generated electron acceptors that supported growth of S. Enteritidis by anaerobic respiration. To this end, we inactivated the moaA gene, which encodes a protein involved in the biosynthesis of molybdopterin, a cofactor required for anaerobic respiration of nitrate and tetrathionate. We then compared the fitness of the S. Enteritidis wild type (YL215) and the anaerobic respiration-deficient moaA mutant (YL86) by infecting neonatal chicks with a 1:1 mixture of both strains. Equal recovery of the wild type and moaA mutant from cecal contents 3 days after infection suggested that anaerobic respiration was not required for colonization of S. Enteritidis in neonatal chicks (Fig. 2A). Thus, the mechanism through which virulence factors promote intestinal colonization of S. Enteritidis in neonatal chicks appeared to be different from that described for cecal colonization of S. Typhimurium in adult mice.

Figure 2: Virulence factors promote pathogen expansion through aerobic respiration.

(A) 1-day old chicks (n = 4–5) were infected with a 1:1 mixture of the indicated bacterial strains. Cecal content were collected 3 days after infection to determine the competitive index. See also Figure S2. (B-G) 1-day old chicks were mock-treated or infected with the indicated bacterial strains. (B) Composition of the gut microbiota at the phyla-level (relative abundance) based on 16S rRNA gene sequencing of DNA isolated from cecal contents at the indicated time points after infection. See also Figure S1. (C-E) Total colony-forming units (CFU) in cecal contents recovered on MacConkey agar (Enterobacteriaceae, black bars) or CFU of S. Enteritidis (red bars) were determined in groups of chicks that were mock-treated (C), infected with S. Enteritidis (wt) (D) or infected with a S. Enteritidis invA spiB mutant (invA spiB) (E). Bars represent geometric means ± standard error. (F-G) Normalized abundances of members of the family Lachnospiraceae (G) and Ruminococcaceae (H) in cecal contents were determined by microbiota profiling. Bars represent mean ± standard deviation. (H-I) Representative images of pimonidazole (red, panel H) or MitoTracker (Dark orange, panel I) staining detected in cecal sections counter stained with DAPI nuclear stain (blue). (B-G) The number of animals in each group (n) is indicated in Figure S1. Scale bars = 40 μm. *, P < 0.05; ****, P < 0.0001.

Next, we investigated whether S. Enteritidis virulence factors triggered changes in the development of the neonate gut microbiota by infecting one day-old chicks with the S. Enteritidis wild type or an invA spiB mutant, followed by profiling the microbial community using 16S ribosomal RNA gene amplicon sequencing (microbiota profiling). One day after infection, Proteobacteria dominated the neonate gut microbiota (Fig. 2B and Fig. S1A). A gradual increase in the abundance of Firmicutes gave rise to a dominance of this phylum by day 7 after mock-infection (Fig. 2B and Fig. S1A). Concomitantly, we observed a decrease in the number of colony-forming units (CFU) on MacConkey agar (Fig. 2C), a selective medium for the isolation of Gram-negative enteric bacteria, commonly including members of the Enterobacteriaceae and Pseudomonaceae (phylum Proteobacteria) (MacConkey, 1905). This decrease was not observed in chicks infected with the S. Enteritidis wild type, because the pathogen colonized the ceca at high numbers and was responsible for the majority of colonies isolated on MacConkey agar (Fig. 2D). In contrast, the invA spiB mutant represented only a small fraction of CFU isolated on MacConkey agar (Fig. 2E). Thus, virulence factors drove a robust colonization of neonatal ceca by day 7 after infection (Fig. 2D and 2E). Consistent with a previous report (Mon et al., 2015), the appearance of Lachnospiraceae and Ruminococcaceae (class Clostridia) was delayed on days 3 and 7 after infection with the S. Enteritidis wild type, respectively (Fig. 2F and 2G), which correlated with decreased diversity of the microbial community (Fig. S1B). This delay in the appearance of Lachnospiraceae and Ruminococcaceae required the presence of virulence factors, because it was not observed during infection with an invA spiB mutant (Fig. 2F and 2G). However, higher colonization levels of the S. Enteritidis wild type compared to the invA spiB mutant were already observed one day after infection (Fig. 1A), when Clostridia were found in low abundance in all experimental groups (Fig. 2B, 2F and 2G), suggesting that virulence factors conferred an early fitness advantage in neonates that was independent of Clostridia depletion.

Virulence factors promote pathogen expansion through aerobic respiration

Next, we investigated whether virulence factors caused changes in the host epithelium to provide the pathogen with a luminal growth advantage in the cecal lumen. To this end we used the hypoxia marker pimonidazole, which is reduced under hypoxic conditions to hydroxylamine intermediates binding irreversibly to nucleophilic groups in proteins or DNA (Kizaka-Kondoh and Konse-Nagasawa, 2009). Strong pimonidazole staining of the cecal epithelium of mock-infected chicks suggested that the neonate cecum was lined with a hypoxic surface (Fig. 2H). Remarkably, pimonidazole staining was markedly reduced one day after S. Enteritidis infection, indicative of an increase in epithelial oxygenation. The increased in epithelial oxygenation one day after infection did not correlate with changes in the abundance of Clostridia (Fig. 2B, 2F and 2G), but required S. Enteritidis virulence factors, because it was not observed in birds infected with the invA spiB mutant (Fig. 2H). Epithelial hypoxia in the large intestine of mice is due to intracellular oxygen consumption through mitochondrial β-oxidation (Furuta et al., 2001). Consistent with this idea, epithelial hypoxia correlated with increased mitochondrial activity, as indicated by staining with MitoTracker, a dye that accumulates depending upon mitochondrial membrane potential (Fig. 2I).

Reduced epithelial hypoxia has recently been shown to drive growth of S. Typhimurium in the murine cecum through aerobic respiration, which was dependent on the cyxAB genes, encoding a cytochrome bd-II oxidase (Rivera-Chavez et al., 2016). We thus compared the fitness of the S. Enteritidis wild type (YL215) and a cytochrome bd-II oxidase-deficient cyxA mutant (YL28) by infecting neonatal chicks with a 1:1 mixture of both strains. However, equal recovery of the wild type and cyxA mutant suggested that unlike during S. Typhimurium infection in adult mice, cytochrome bd-II oxidase was neither required for cecal colonization of neonatal chicks by S. Enteritidis (Fig. 2A), nor for in vitro growth under microaerophilic conditions (Fig. S2A).

Next, we investigated whether aerobic respiration mediated by cytochrome bo₃ oxidase, encoded by the cyoAB genes, or cytochrome bd oxidase, encoded by the cydAB genes, conferred a fitness advantage in ceca of neonatal chicks. Although cytochrome bo₃ oxidase conferred a modest fitness advantage during in vitro growth in the presence of 4% oxygen (Fig. S2B), wild-type and cyoA mutant were recovered in equal numbers from the ceca of neonatal chicks infected with a 1:1 mixture of both strains, suggesting that cytochrome bo₃ oxidase was dispensable for luminal growth. In contrast, cytochrome bd oxidase conferred a striking fitness advantage during in vitro growth in the presence of 1% or 4% oxygen (Fig. S2C). Furthermore, infection of neonatal chicks with a 1:1 mixture of the S. Enteritidis wild type (YL215) and a cydA mutant (YL216) revealed that cytochrome bd oxidase conferred a marked fitness advantage (Fig. 2A). Similarly, when neonatal chicks were infected with individual bacterial strains, birds infected with the S. Enteritidis wild type exhibited significantly higher cecal pathogen burdens than birds infected with a cydA mutant (Fig. 1A). When neonatal chicks were infected with a 1:1 mixture of an invA spiB mutant (YL144) and an invA spiB cydA mutant (YL180), both strains were recovered in equal numbers from ceca (Fig. 2A), suggesting that the growth advantage conferred by cytochrome bd oxidase was virulence factor-dependent. Collectively, these data suggested that a virulence factor-dependent increase in epithelial oxygenation (Fig. 2H) was associated with a cytochrome bd oxidase-dependent expansion of S. Enteritidis in the cecal lumen of neonatal chicks through aerobic respiration (Fig. 1A and 2A).

S. Enteritidis competes with Escherichia coli for luminal oxygen

Having demonstrated how S. Enteritidis uses its virulence factors to establish an aerobic niche in the ceca of neonatal chicks, we wanted to investigate whether S. Enteritidis competes with other facultative anaerobic bacteria for oxygen. Oxygen is a limiting resource readily consumed by facultative anaerobic bacteria, which obstructs direct measurements of oxygen in the lumen. We thus developed an assay for measuring the bioavailability of oxygen, which was based on the luxCDABE operon from Photorhabdus luminescens. The proteins encoded in the operon synergistically generate bioluminescent light signals exclusive of luciferin substrate addition, but require molecular oxygen to catalyze the bioluminescent reaction (Wiles et al., 2006). Cecal contents of an adult bird were cultured on MacConkey agar and 38 colonies were further analyzed to determine the identity of species. All colonies were identified as members of the family Enterobacteriaceae, including E. coli, Klebsiella pneumoniae and Proteus mirabilis (Fig. 3A). We performed whole genome analysis of one E. coli isolate from a bird, termed YL178, which revealed that it was a commensal with the serotype O62:H30 belonging to phylogroup A (Fig. 3B). Commensal E. coli isolate YL178 was transformed with a plasmid encoding a luciferase reporter (pBR2TTS:CP25::luxCDBAE) (Table 2) and used to inoculate neonatal chicks. Bioluminescence imaging in anesthetized birds revealed bioluminescence one day after inoculation (Fig. 3C), which required active blood circulation, because bioluminescence rapidly weakened when chicks were euthanized (Fig. 3D and 3E). Bioluminescence was mainly localized to the ceca of neonatal chicks (Fig. 3F) and E. coli was recovered in high numbers from cecal contents, but not from intestinal or extraintestinal tissues (data not shown). Collectively, bioluminescence imaging suggested that E. coli had access to host circulation-derived oxygen in the cecal lumen of neonates.

Figure 3: Bioluminescence imaging of luminal E. coli requires a live host.

(A) Cecal contents of an adult bird were cultured on MacConkey and the identity of bacterial colonies determined using an EnteroPluri assay. (B) Rooted maximum likelihood tree, visualized as a cladogram, of select E. coli isolates indicating their relatedness to the avian commensal isolate YL178. Color-coding denotes members of the same phylogroup (phylogroups A, B1, B2, D1, D2 or E1). Information on the biovar for each isolate is given in parentheses: ETEC, enterotoxigenic E. coli; EAEC, enteroaggregative E. coli; EHEC, Enterohemorrhagic E. coli; UPEC, uropathogenic E. coli; AIEC, adherent-invasive E. coli; ExPEC, etraintestinal pathogenic E. coli. (C-L) Chicks were infected one day after hatch (day 1) with E. coli strain YL178 transformed with a plasmid carrying a CP25::luxCDBAE transcriptional fusion (pBR2TTS:CP25::luxCDBAE) and bioluminescence imaging was determined one day later (day 2) in anesthetized birds (C), after euthanasia (D and E) or in the gastrointestinal tract open longitudinally (F). A representative image is displayed for each sample (n=8). (E) Time course of bioluminescence imaging (n = 6) determined before (red dot) and after (black dots) euthanasia. (C, D and F) The number of animals in each group (n) is indicated in Fig. 4A. Data are shown as geometric means ± standard error (n=6).

Table 2:

Plasmids used in this study

Designation	Relevant characteristics	Reference
pRDH10	oriR6K mobRP4 sacRB TetR CmR	(Kingsley et al., 1999)
pGP704	oriR6K mobRP4 CarbR	(Miller and Mekalanos, 1988)
pEP185.2	oriR6K mobRP4 CmR	(Kinder et al., 1993)
pYL14	pRDH10 carrying up-/downstream regions of cyxA	This study
pCAL25	pRDH10 carrying up-/downstream regions of moaA	(Faber et al., 2017)
pTH122	pGP704 carrying phoN flanking region and KanR	(Haneda et al., 2009)
pTH124	pGP704 carrying phoN flanking region and CmR	(Haneda et al., 2009)
pSPN62	pRDH10 carrying up-/downstream regions of spiB flanking KanR	(Haneda et al., 2009)
pSW127	pGP704 carrying an internal fragment of invA	(Winter et al., 2009)
pYL168	pEP185.2 carrying up-/downstream regions of cyoA flanking KanR	This study
pGW15	pRDH10 carrying up-/downstream regions of cydAB	This study
pGW18	pGW15 with CarbR inserted between the cydAB flanking regions	This study
pBR2TTS:CP25::luxCDABE	pBR322 carrying the luxCDABE operon under the constitutively active promoter CP25	(Melamed et al., 2011)
pYL9	pRDH10 carrying up-/downstream regions of cydA from E. coli Nissle 1917	This study
pYL10	pRDH10 carrying up-/downstream regions of appC from E. coli Nissle 1917	This study
pCAL61	pWSK129 carrying the omega cassette	(Spees et al., 2013)

Next, we inoculated neonate chicks with a 1:1 mixture of commensal E. coli YL178(pBR2TTS:CP25::luxCDBAE) and virulent S. Enteritidis. Imaging revealed reduced E. coli bioluminescence in birds infected with the S. Enteritidis wild type compared to mock-infected birds (Fig. 4A), suggesting that the pathogen lessened the bioavailability of oxygen for E. coli. To determine whether S. Enteritidis diminished the bioavailability of oxygen because it consumed this resource through aerobic respiration, neonatal chicks were inoculated with a 1:1 mixture of commensal E. coli YL178(pBR2TTS:CP25::luxCDBAE) and a cytochrome bd oxidase-deficient S. Enteritidis cydA mutant. Remarkably, infection with the S. Enteritidis wild type diminished E. coli bioluminescence compared to infection with a S. Enteritidis cydA mutant (Fig. 4A), thus supporting the idea that S. Enteritidis competes with E. coli for oxygen by consuming this resource through aerobic respiration. To mimic conditions under which commensal Enterobacteriaceae are already established at the time of S. Enteritidis infection, day-of-hatch birds were inoculated with E. coli YL178(pBR2TTS:CP25::luxCDBAE) and infected one day later with the S. Enteritidis wild type. Interestingly, E. coli bioluminescence did not diminish when the commensal was able to establish gut colonization one day before challenge (Fig. 4A), although the bacterial burden in the cecum was similar to that observed when birds were inoculated with a 1:1 mixture of commensal and pathogen (Fig. 4B and C). These data suggested that E. coli competes more successfully with S. Enteritidis for oxygen when the commensal establishes gut colonization prior to pathogen challenge, because priority effects determine whether oxygen is accessible.

Figure 4: S. Enteritidis competes with E. coli for luminal oxygen.

Birds were inoculated one day after hatch (day 1) with E. coli strain YL178(pBR2TTS:CP25::luxCDBAE) carrying the luxCDBAE reporter genes (E. coli (lux)) or with a 1:1 mixture of E. coli (lux) and either wild-type S. Enteritidis (SE wt) or a S. Enteritidis cydA mutant (SE cydA). Alternatively, birds were inoculated at the day of hatch (day 0) with E. coli (lux) and infected the next day (day 1) with wild-type S. Enteritidis. (A) Bioluminescence imaging was performed one day after the last inoculation (day 2). Each dot represents data from one animal. Lines indicate geometric means ± standard deviation. (B) Numbers of E. coli (lux) recovered from cecal contents. (C) Numbers of S. Enteritidis recovered from cecal contents. (B-C) Data are shown as geometric means ± standard deviation. The number of animals in each group (n) is indicated in panel A. ***, P < 0.001; ****, P < 0.0001.

Spore-forming bacteria and Enterobacteriaceae confer niche protection

To further investigate whether competition for critical resources, such as oxygen, contributes to niche protection, we chickenized germ-free Swiss Webster mice by transplanting microbiota from cecal contents of either a healthy adult (three-month-old) bird or a neonate (one-day-old) chick. Five days after the microbiota transplant, mice were challenged with wild-type S. Enteritidis and pathogen burden in the feces was determined. Germ-free mice challenged with S. Enteritidis had a pathogen burden in the feces that was similar to that of mice associated with neonate chick microbiota, but approximately 1,000-fold higher than mice associated with adult chicken microbiota (Fig. 5A), demonstrating that only adult chicken microbiota confers niche protection.

Figure 5: Enterobacteriaceae and spore-forming bacteria are required to confer niche protection.

(A and E) Germ-free Swiss Webster mice were mock-treated or inoculated with microbiota from a neonate (one-day-old) or adult (three-months-old) bird (A) or received the indicated components of adult chicken microbiota (E). 5 days later mice were challenged with S. Enteritidis (10⁸ CFU/mouse). Colonization levels of S. Enteritidis were determined in the feces two days after infection. (B-D and F) Microbial representation at the family-level was determined by 16S rRNA gene sequencing of DNA isolated from adult chicken microbiota (B), neonate chick microbiota (C), bacteria recovered from MacConkey agar seeded with adult chicken microbiota (Enterobacteriaceae component) (D) or feces of ex-germ-free Swiss Webster mice inoculated with chloroform-treated adult chicken microbiota (spore component) (F). (G) Day-of-hatch chicks were inoculated with the indicated components of adult chick microbiota and challenged the next day with S. Enteritidis (10⁹ CFU/chick). Colonization levels of S. Enteritidis were determined in the cecal contents (n = 6) one days after challenge. (H and I) Germ-free Swiss Webster mice were inoculated with chloroform-treated adult chicken microbiota (spores) and the indicated E. coli strains. 5 days later, E. coli colonization levels were determined (I) and mice were challenged with S. Enteritidis (10² CFU/mouse). Colonization levels of S. Enteritidis were determined in the feces (n = 6) two days after infection (H). (A, E, G, H and I) Bars represent geometric means ± standard error. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not statistically significantly different.

Microbiota profiling revealed that obligate anaerobic bacteria belonging to the phyla Firmicutes and Tenericutes dominated the adult chicken microbiota transplant (Fig. 5B), whereas the neonate chick microbiota transplant was dominated by Enterobacteriaceae (Fig. 5C). The adult chicken microbiota transplant was cultured on MacConkey agar, which resulted in the isolation of Enterobacteriaceae as revealed by microbiota profiling of cultivable bacteria (Fig. 5D). We then colonized germ-free mice with this microbial consortium and challenged the animals five days later with S. Enteritidis. Colonization with only a consortium of Enterobacteriaceae isolated from an adult bird did not confer niche protection against S. Enteritidis (Fig. 5E).

A recent study shows that acquisition of Clostridia species by neonatal mice confers niche protection against S. Typhimurium infection (Kim et al., 2017). Since Clostridia belong to the phylum Firmicutes, a group of spore-forming bacteria, we killed vegetative bacterial cells in the adult chicken microbiota transplant using chloroform treatment (Velazquez et al., 2017). The chloroform-treated microbiota transplant was then used to inoculate germ-free mice and microbiota profiling was performed seven days later on cecal contents to determine its composition. The chloroform-treated microbiota transplant was composed of Firmicutes and Tenericutes, two closely related groups of Gram-positive spore-forming bacteria (Fig. 5F). S. Enteritidis challenge of mice inoculated with a chloroform-treated microbiota transplant showed that spore-forming bacteria alone conferred little protection against pathogen colonization (Fig. 5E).

We reasoned that Clostridia might confer niche protection in neonatal mice (Kim et al., 2017), because Proteobacteria colonize early in life (Dupont et al., 2016) and both taxa contribute to this trait. We thus explored the hypothesis that Enterobacteriaceae and spore-forming bacteria are both necessary for niche protection. To test this idea, germ-free mice received Enterobacteriaceae isolated from an adult bird (Fig. 5D) together with a chloroform-treated microbiota transplant (Fig. 5F) and were challenged five days later with S. Enteritidis. Interestingly, the combination of Enterobacteriaceae and spore-forming bacteria conferred niche protection at levels similar to those observed in mice receiving a microbiota transplant from an adult bird (Fig. 5E). Similar results were obtained when day-of-hatch chicks received components of adult chicken microbiota and were challenged one day later with the S. Enteritidis wild type (Fig. 5G). Collectively, these results suggested that colonization resistance required the presence of spore-forming bacteria, which are highly abundant in the adult chicken microbiota, along with a low-abundance keystone taxon, the Enterobacteriaceae (Fig. 5B).

E. coli confers niche protection against S. Enteritidis by competing for oxygen

We then wanted to test the hypothesis that Enterobacteriaceae conferred niche protection by competing with S. Enteritidis for a critical resource, oxygen. First we determined whether niche protection could be modeled using a single isolate of commensal Enterobacteriaceae. To this end, germ-free mice received a chloroform-treated microbiota transplant (Fig. 5F) together with the probiotic E. coli Nissle 1917 (Fig. 3B) and were challenged five days later with S. Enteritidis. Strikingly, mixing spore-forming bacteria with E. coli Nissle 1917 conferred niche protection against low-dose S. Enteritidis challenge (Fig. 5H), suggesting that E. coli could be used to study the mechanism of colonization resistance. Next, we genetically ablated the ability of E. coli Nissle 1917 to respire oxygen under microaerophilic conditions by inactivating cydA, encoding a subunit of cytochrome bd oxidase, and appC, encoding a subunit of cytochrome bd-II oxidase. To determine whether the ability of E. coli to respire oxygen contributed to colonization resistance, germ-free mice received a chloroform-treated microbiota transplant (Fig. 5F) together with an E. coli cydA appC mutant (YL17) and were challenged five days later with S. Enteritidis. Interestingly, an E. coli cydA appC mutant no longer conferred niche protection against S. Enteritidis (Fig. 5H), although E. coli Nissle 1917 and its isogenic cydA appC mutant colonized the intestine of germ-free mice at similar levels (Fig. 5I), thus supporting the idea that the ability of E. coli to respire aerobically impedes intestinal colonization of S. Enteritidis.

DISCUSSION

Elevated susceptibility to S. Typhimurium infection of neonatal mice compared to adult animals (Zhang et al., 2014) has been attributed to an absence of Clostridia species from the neonatal microbiota (Kim et al., 2017). Similarly, depletion of Clostridia from a mature gut microbiota drives a luminal expansion of S. Typhimurium in the murine large intestine (Rivera-Chavez et al., 2016). A mechanism by which Clostridia confer niche protection is the production of butyrate (Rivera-Chavez et al., 2016), a short-chain fatty acid sensed by the epithelial butyrate sensor PPAR-γ (peroxisome proliferator-activated receptor-gamma) (Byndloss et al., 2017). PPAR-γ-signaling drives the metabolism of epithelial cells in the large bowel towards mitochondrial β-oxidation (Byndloss et al., 2017), thereby maintaining epithelial hypoxia (Furuta et al., 2001) to limit oxygen availability in the gut lumen (Rivera-Chavez et al., 2016). However, a lack of epithelial PPAR-γ signaling alone is not sufficient to increase epithelial oxygenation, which in addition requires an inflammatory signal generated by S. Typhimurium virulence factors (Byndloss et al., 2017). Consistent with these previous reports, the absence of Clostridia from the neonatal microbiota was not sufficient to eliminate epithelial hypoxia, but elevating epithelial oxygenation also required inflammation triggered by S. Enteritidis virulence factors (Fig. 1 and 2H).

Whereas Clostridia are necessary for colonization resistance against S. Typhimurium (Kim et al., 2017; Rivera-Chavez et al., 2016), studies with defined microbial communities suggest Clostridia are not sufficient for niche protection (Brugiroux et al., 2016). Specifically, assembly of a defined microbial community conferring colonization resistance against S. Typhimurium in mice requires inclusion of facultative anaerobic bacteria, including E. coli, Streptococcus danieliae (class Bacilli) and Staphylococcus xylosus (class Bacilli) (Brugiroux et al., 2016). Here we show colonization resistance against S. Enteritidis in chicks required the presence of both obligate anaerobic spore forming bacteria and facultative anaerobic Enterobacteriaceae (Fig. 5G).

Enterobacteriaceae compete with S. Typhimurium for siderophore-bound iron (Deriu et al., 2013; Raffatellu et al., 2009) and elaborate colicins or microcins to inhibit growth of closely related bacteria (Nedialkova et al., 2014; Sassone-Corsi et al., 2016). However, genes encoding antimicrobial activities are highly variable, even between different isolates belonging to the same species (Riley and Wertz, 2002), and other taxa, such as Firmicutes (Grandchamp et al., 2017) and Bacteroides (Rocha and Krykunivsky, 2017), also compete for siderophore-bound iron. Our results identify oxygen as a critical resource S. Enteritidis requires for establishing robust colonization within neonatal ceca (Fig. 1A and 2A) and suggest that E. coli contributes to niche protection by consuming this critical resource (Fig. 5H). Importantly, the ability to respire oxygen is conserved among all members of the facultative anaerobic Enterobacteriaceae and sets this keystone taxon apart from obligate anaerobic bacteria, which dominate mature gut-associated microbial communities. Aerobic respiration drives an expansion of Salmonella serovars because it enables these pathogens to consume fermentation products, such as formate (Hughes et al., 2017), lactate (Gillis et al., 2018) or 1,2-propanediol (Faber et al., 2017). However, competition for oxygen does not reduce growth of S. Enteritidis (Fig. 5G) or commensal Enterobacteriaceae (Fig. 4C) when obligate anaerobic spore-forming bacteria are absent, presumably because this resource only becomes limiting when butyrate-producing Clostridia reduce oxygen availability in the gut lumen by maintaining epithelial hypoxia (Byndloss et al., 2017).

The picture emerging from these studies is that consumption of oxygen by commensal Enterobacteriaceae and maintenance of epithelial hypoxia by Clostridia are both required to inhibit growth of facultative anaerobic enteric pathogens by aerobic respiration (Fig. 6). Consistent with the idea that a mature gut microbiota confers niche protection by limiting the bioavailability of oxygen, aerobic respiration does not provide a luminal growth benefit for S. Enteritidis in neonatal chicks colonized with gut microbiota from adult chickens (Barrow et al., 2015).

Figure 6: Model for cooperation between Enterobacteriaceae and Clostridia to confer colonization resistance against S. Enteritidis.

Clostridia in the large intestine break down complex carbohydrates from the diet (fiber) into fermentation products, such as butyrate. Butyrate helps to polarize the epithelial metabolism towards oxidative phosphorylation in the mitochondria, resulting in high oxygen (O₂) consumption. In turn, high oxygen consumption renders the epithelial surface hypoxic, thereby limiting the amount of oxygen diffusing from the mucosal surface into the intestinal lumen. The small amount of oxygen that emanates from the hypoxic epithelial surface is consumed by commensal Enterobacteriaceae through aerobic respiration. Epithelial hypoxia and aerobic respiration by Enterobacteriaceae cooperate to maintain anaerobiosis in the lumen. Anaerobiosis confers colonization resistance because it blocks the access to oxygen for the facultative anaerobic S. Enteritidis (Salmonella), thereby curbing pathogen growth. The resulting decline in pathogen numbers can lead to an extinction, an outcome that becomes more likely when the host is exposed to a low infectious dose.

STAR METHODS

Contact for reagent and resource sharing

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Andreas J. Bäumler (ajbaumler@ucdavis.edu).

Experimental model and subject details

Animal Experiments.

All experiments in this study were approved by the Institutional Animal Care and Use Committee at the University of California at Davis.

Chicken experiments.

Male and female highly inbred Leghorn layer chickens from line UCD003 were obtained on the day of hatch from Hopkins Avian Facility at the University of California, Davis. The birds were then transferred to Meyer Hall Avian Facility and housed in temperature-controlled chambers with ad libitum access to water and commercial feed without antibiotic treatment. Unless otherwise indicated, 1 day after hatch the birds were orally inoculated with 1×10⁹ CFU of S. Enteritidis in Luria-Burtani (LB) broth (BD Biosciences #244620). Dosage of S. Enteritidis was confirmed by serial dilution and plating of the inoculum. At each indicated time point, groups of birds were euthanized by carbon dioxide asphyxiation. For administration of microbiota or microbiota components, day-of-hatch chicks were orally inoculated with the indicated microbiota fractions, and 1 day after hatch the birds were infected with S. Enteritidis, as described above. The Institutional Animal Care and Use Committee at the University of California, Davis, approved all experiments in this study.

Mouse experiments.

Male and female germ-free Swiss Webster mice were bred and housed at the SPF Teaching and Research Animal Care Services facility at UC Davis. Mice were bred and housed in germ-free isolators (Park Bioservices). At 5–8 weeks of age mice were removed from the isolators into sterilized cages (ISOcage P Bioexclusion System, Techniplast) and inoculated with microbiota components by oral gavage. 5 days after inoculation, the mice were mock-treated with LB broth or infected with S. Enteritidis in LB broth.

Bacterial strains.

Bacterial strains used in this study are listed in Table 1. Bacterial strains were routinely grown in LB broth or on LB-agar plates unless otherwise indicated. Antibiotics were used at the following concentrations: carbenicillin (Carb), 0.1 mg/ml; chloramphenicol (Cm), 0.015mg/ml; kanamycin (Kan), 0.1 mg/ml; tetracycline (Tet), 0.02 mg/ml; nalidixic acid (Nal), 0.05 mg/ml. Suicide plasmids were propagated in E. coli DH5α λpir. Phage P22 HT int-105 was used for generalized transduction. Transductants were cleaned from phage contamination on Evans Blue-Uranine (EBU) plates and tested for phage sensitivity by cross-streaking against lytic phage P22 H5. For determining CFU, the cecal content was weighed, homogenized in PBS and plated on selective MacConkey agar plates.

The avian E. coli strain was isolated from the cecal content of a neonate chick by plating on MacConkey agar plates and the species identity was verified using the EnteroPluri test (Liofilchem). The strain was sequenced and its genome annotated by the DNA Technologies Core Facility and the Bioinformatics Core Facility at UC Davis.

Method details

Construction of bacterial mutants.

Mutant genotypes were routinely confirmed by PCR. Plasmids and primers used for cloning are listed in Tables 2 and S2, respectively. To construct the E. coli cydA appC mutant, upstream and downstream regions of approximately 0.5kb flanking the cydA and the appC genes were amplified by PCR from the E. coli Nissle 1917 genome. The pRDH10 suicide plasmid was digested with SalI and assembled with the fragments of each gene using the Gibson Assembly Master Mix (NEB) to form plasmids pYL9 (pRDH10-ΔcydA EcN) and pYL10 (pRDH10-ΔappC EcN). Plasmid pYL9 was transformed into E. coli S17-1λpir and conjugated into E. coli Nissle 1917 using E. coli S17-1λpir as the donor strain. Clones that had integrated the suicide plasmid were subjected to sucrose counter-selection and a colony that was sucrose resistant and Cm^S was verified by PCR to be E. coli Nissle 1917 cydA. Then, Plasmid pYL10 was transformed into E. coli S17-1λpir and conjugated into E. coli Nissle 1917 cydA using E. coli S17-1λpir as the donor strain. Clones that had integrated the suicide plasmid were subjected to sucrose counter-selection and a colony that was sucrose resistant and Cm^S was verified by PCR to be E. coli Nissle 1917 cydA appC (YL17). Plasmid pCAL61, conferring spectinomycin and kanamycin resistance, was transformed into E. coli Nissle 1917 and YL17 to facilitate recovery from feces.

To facilitate isolation from chicken feces, the S. Enteritidis wild type (TN2) was marked with a kanamycin resistance cassette by transducing the phoN::Kan^R marker from S. Typhimurium strain AJB715 using P22 transduction. A colony carrying the marker was termed YL215.

To construct a S. Enteritidis cyxA mutant, upstream and downstream regions of approximately 0.5kb flanking the cyxA gene were amplified by PCR from the S. Enteritidis (TN2) genome. The pRDH10 suicide plasmid was digested with SalI and assembled with the fragments using the Gibson Assembly Master Mix (NEB) to form plasmid pYL14. Plasmid pYL14 was transformed into E. coli S17-1λpir. Plasmid pYL14 was conjugated into S. Enteritidis using E. coli S17-1λpir as a donor strain and plated on LB+Nal+Cm agar to select for clones that had integrated the suicide plasmid. Sucrose counter-selection was performed and a colony that was sucrose resistant and Cm^S was verified by PCR and named YL28. TN2 and YL28 were then conjugated with E. coli S17-1λpir(pTH122) and plated on LB+Nal+Kan to give rise to YL55 and YL56, respectively.

To construct a moaA mutant, S. Enteritidis (TN2) was conjugated with E. coli S17-1λpir(pCAL25) and plated on LB+Nal+Cm agar to select for clones that had integrated the suicide plasmid. Sucrose counter-selection was performed and a colony that was sucrose resistant and Cm^S was verified by PCR and named YL86. Strain YL86 was then conjugated with E. coli S17-1λpir/pTH122 and plated on LB+Nal+Kan to give rise to YL91.

To construct a S. Enteritidis invA spiB mutant, S. Enteritidis (TN2) was conjugated with E. coli S17-1λpir(pSPN62) and plated on LB+Nal+Kn. A colony that was sucrose resistant, Nal^R, Kan^R and Cm^S was verified by PCR and named YL126. YL126 was then conjugated with E. coli S17-1λpir(pSW127) and plated on LB+Carb+Kan to give rise to YL144.

To construct a S. Enteritidis cyoA mutant, upstream and downstream regions of approximately 0.5kb flanking the cyoA gene were amplified by PCR from the S. Enteritidis (TN2) genome. The pEP185.2 suicide plasmid and the kanamycin resistance cassette were each amplified by PCR and assembled with the cyoA fragments so that these were flanking the resistance cassette using the Gibson Assembly Master Mix (NEB) to form plasmid pYL168. Plasmid pYL168 was transformed into E. coli S17-1λpir and conjugated into S. Enteritidis. A Nal^R, Kan^R and Cm^S colony was selected and named YL170.

To construct S. Enteritidis and S. Typhimurium cydAB mutants, regions upstream and downstream flanking the cydA and cydB genes were PCR amplified from genomic DNA of S. Typhimurium IR715 with primers cydAB_FR1_Fwd and cydAB_FR1_Rev or cydAB_FR2_Fwd and cydAB_FR2_Rev and cloned into pRDH10 using Gibson Assembly Master Mix (NEB) yielding the plasmid pGW15. The Carbenicillin cassette of pUC4 KSAC was PCR amplified with primers KSAC_bla_Fwd and KSAC_bla_Rev and cloned into the XbaI site between the cydAB flanking regions in plasmid pGW15 to generate plasmid pGW18. Plasmid pGW18 was conjugated into S. Typhimurium strain AJB715 using E. coli S17-1λpir as a donor strain. A colony that underwent 2 crossover events and was Nal^R, Kn^R, Cm^S was selected and named GTW46. A S. Enteritidis cydAB mutant was constructed by P22 transduction using lysate from GTW46.

In vitro growth under different oxygen concentrations.

For growth under different aerobic, microaerobic, or anaerobic conditions, minimal medium containing 0.4% glycerol and 0.1% casamino acids was inoculated with the indicated strain mixtures and incubated at 37°C in a conventional incubator (21% oxygen), in a hypoxia chamber (set at 1% or 4% oxygen), or in an anaerobe chamber (0% oxygen).

Histopathology.

Cecal tissue was fixed in 10% phosphate-buffered formalin and 5μm sections of the tissue were stained with hematoxylin and eosin. Scoring of blinded tissue sections was performed by a veterinary pathologist based on the criteria listed in Fig S1. Representative images were taken using an Olympus BX41 microscope.

Fluorescence microcopy.

Imaging For detection of hypoxia, birds were treated with 60mg/kg of pimonidazole HCl (Hypoxyprobe^™−1 kit, Hypoxyprobe) via intraperitoneal injection one hour prior to euthanasia. Cecal samples were fixed in 10% buffered formalin and paraffin-embedded tissue was probed with mouse anti-pimonidazole monoclonal antibody MAb1 (Hypoxyprobe) followed by a secondary goat anti-mouse IgG antibody (Alexafluor 546, Life Technologies) and DAPI counter-staining. Representative images were taken using an Zeiss Axiovert 200 M fluorescent microscope.

For mitochondria staining, cecal tissue was frozen in Optimal Cutting Temperature (OCT) compound (Fisher HealthCare) and cut in transverse sections to a thickness of 7μm. Sections were thawed and incubated in 250nM MitoTracker Red CMXRos (Molecular Probes, M7512) in PBS for 30 minutes in room temperature, counter stained with DAPI (Sigma, D9542), fixed in 4% paraformaldehyde and mounted (ImmuMount, Thermo Scientific, 9990402). Representative images were taken using a Zeiss Axiovert 200 M fluorescent microscope.

RNA isolation and Quantitative real-time PCR.

For chicken RNA isolation, cecal tonsil tissue was homogenized in a Mini-Beadbeater (BioSpec Products, Bartlesville, OK) and RNA was isolated by the TRI-Reagent method (Molecular Research Center, Inc.) following the manufacturer’s protocol. Contaminating DNA was removed using the DNA-free kit (Applied Biosystems) and RNA was stored at −80°C. Isolated RNA was rever se transcribed using random hexamers and Moloney murine leukemia virus reverse transcriptase (Applied Biosystems). Quantitative real-time PCR was performed using SYBR green (Applied Biosystems) PCR mix and the appropriate primer sets (Table S2).

Isolation and fractionation of chicken microbiota.

For collection of chicken microbiota, the cecal contents of a 1 day old (neonate) and 3 month-old (adult) bird was collected, moved into an anaerobe chamber, diluted in PBS and glycerol, aliquoted and frozen.

The spore component was produced by chloroform treatment as described previously (Velazquez et al., 2017). Briefly, the cecal content was incubated for 30 minutes with 3% chloroform. Then, the chloroform was allowed to settle and the supernatant was carefully removed. two germ-free Swiss-Webster mice were orally inoculated with the supernatant. Seven days after inoculation, the mice were euthanized, and their ceca content collected, moved into an anaerobe chamber, diluted in PBS and glycerol, aliquoted and frozen.

The Enterobacteriaceae component was prepared freshly by spreading the adult chicken microbiota on MacConkey-agar plates. The plates were incubated aerobically in 37°C for 16 hours and then bacterial colonies were scraped and collected in LB broth.

In vivo bioluminescence.

For in vivo bioluminescence (BLI), animals were anesthetized using isoflurane in an induction chamber. For imaging of deceased animals, cervical dislocation was performed after anesthesia. Sedated or deceased animals were transferred to an optically clear XIC-3 isolation chamber (PerkinElmer) and positioned on their ventral surface. For ex-vivo BLI, the gastrointestinal tract was removed, placed in a petri dish, sliced open with a scalpel and imaged. Bioluminescence was imaged using an IVIS Spectrum (PerkinElmer) with no emission filter. Photons were quantified using an ultrasensitive charge-coupled device (CCD) camera and the resulting heat maps of bioluminescent photon emission intensity were overlaid on still images of anesthetized animals. Region of interest (ROI) analysis was used to quantify bioluminescence (Living Image). BLI data were quantified as total flux (photons/second) for exposure time-independent quantification of signal intensity.

Library preparation for 16S rRNA gene sequencing.

PCR amplification targeting the V4 hypervariable regions of the 16S rDNA was performed with isolated DNA samples as template, forward primer, F515 (5’NNNNNNNNGTGTGCCAGCMGCCGCGGTAA3’) and reverse primer, R806 (5’GGACTACHVGGGTWTCTAAT3’) (primers kindly provided by Dr. Elizabeth Maga). Forward primer was modified to contain the linker region (GT) for sequencing on the Illumina Miseq platform and also contain a unique 8 basepair barcode sequence (N) to identify each unique samples to be sequence. PCR recipe contained 12.5 μl of 2x GoTaq Green Master Mix (Promega, Madison, WI, USA), 0.5 μl of forward primer, 0.5 μl of reverse primers, 2 μl of DNA template and 9.5 μl nuclease-free water to make up total reaction volume of 25 μl. PCR program consisted of the following steps: initial denaturation at 94°C for 3 minutes, 35 cycles at 94°C for 45 sec, 50°C for 1 m inute, 72°C for 1 minute 30 second and final extension step at 72°C for 10 minu tes. Samples were amplified in triplicate and combined afterward. PCR products were then visualized on a 1% agarose gel stained with SYBR safe (Life Technologies, CA, USA). Purification of PCR products were performed with QIAquick PCR Purification kit (Qiagen, Valencia, CA, USA) following the manufacturer’s instruction. Purified pooled amplicon was then submitted to UC Davis Genome Center, DNA Technology Core Facility for generation of 250 bp paired-end reads on the Illumina Miseq platform.

QIIME data processing.

The Quantitative Insights into Microbial Ecology (QIIME) version 1.9.1 was used to analyze the sequencing data generated. Using the QIIME default settings, raw data were de-multiplexed and quality-filtered (Caporaso et al., 2010). The 250-bp reads were truncated at any site of more than three sequential bases receiving a quality score <Q10 and any read containing ambiguous base calls or barcode/primer errors were discarded as were reads with <75% (of total read length) consecutive high-quality base calls. OTU were clustered against GreenGenes 16S rRNA reference database version 13_8 at 97% identity.

Phylogenetic analysis.

Draft genome assembly was completed using Spades (Bankevich et al., 2012) and putative plasmids were generated using plasmid spades. Provisional genome annotation was performed using prokka (Seemann, 2014), using the genome annotation of E. coli strain AZ162 (Genbank number CP019015.1) as a guide.

Representative genomes with known phylogroup assignments were chosen from the literature (Chaudhuri and Henderson, 2012; Clermont et al., 2013). The sequences were then downloaded from the NCBI RefSeq database and re-annotated using an identical command in prokka (Stamatakis, 2014), to reduce batch effects. Thirteen amino acid sequences were chosen from the literature for their low mutation and recombination rates. These proteins were encoded by the following genes: aes, adk, fumC, gyrB, icd, mdh, pabB, polB, purA, putP, recA, trpA, trpB. The amino acid sequences were gathered from the prokka generated faa files, and subsequently aligned using the clustal omega command line tool (Sievers and Higgins, 2014). Protein alignments were then manually inspected for congruency, and concatenated into a single 6636 character-length alignment. A maximum likelihood tree was generated from the concatenated alignment using RaxML (Stamatakis, 2014). Tree visualization was performed using the R package ggtree (Team, 2011; Yu et al., 2017).

Quantification and statistical analysis

Data for our experiments displaying bacterial numbers (CFU/g) or luminescence (photons/sec) as bar graphs represent the geometric mean and the standard error of the mean. Unless otherwise indicate in the figures, data collected from a minimum of 4 animals are presented. For most experiments, data points were first log transformed and differences between experimental groups were determined on the transformed data using a Student’s T-test (for comparing two groups). Fold-changes of ratios (bacterial numbers or mRNA levels) were transformed logarithmically prior to statistical analysis. An unpaired Student’s t-test was used to determine whether differences in fold-changes between groups were statistically significant. A P-value of less than 0.05 was considered significant.

For QIIME data processing, a Mann-Whitney U test was used to detect the statistical differences in alpha diversity metric, Shannon’s diversity, which accounts for both abundance and evenness in the distribution of microbial community members.

For phylogenetic analysis, twenty phylogenetic trees on distinct starting trees were created, and one hundred bootstrap replicates were performed using rapid bootstrapping. The bootstraps were then used to draw bipartitions on the maximum likelihood tree.

Significance of differences in histopathology scores was determined by a one-tailed non-parametric test (Mann-Whitney). A P-value of less than 0.05 was considered significant.

Highlights.

S. Enteritidis virulence factors increase epithelial oxygenation in neonatal chick guts

S. Enteritidis expands in the neonatal chick gut through aerobic respiration

Spore-forming bacteria and commensal Enterobacteriaceae confer niche protection

E. coli requires aerobic respiration to block S. Enteritidis colonization