Simulated Colonic Fluid Replicates the In Vivo Growth Capabilities of Citrobacter rodentium cpxRA Mutants and Uncovers Additive Effects of Cpx-Regulated Genes on Fitness

ABSTRACT

Citrobacter rodentium is an attaching and effacing (A/E) pathogen used to model enteropathogenic and enterohemorrhagic Escherichia coli infections in mice. During colonization, C. rodentium must adapt to stresses in the gastrointestinal tract, such as antimicrobial peptides, pH changes, and bile salts. The Cpx envelope stress response (ESR) is a two-component system used by some bacteria to remediate stress by modulating gene expression, and it is necessary for C. rodentium pathogenesis in mice. Here, we utilized simulated colonic fluid (SCF) to mimic the gastrointestinal environment, which we show strongly induces the Cpx ESR and highlights a fitness defect specific to the ΔcpxRA mutant. While investigating genes in the Cpx regulon that may contribute to C. rodentium pathogenesis, we found that the absence of the Cpx ESR resulted in higher expression of the locus of enterocyte effacement (LEE) master regulator, ler, and that the genes yebE, ygiB, bssR, and htpX relied on CpxRA for proper expression. We then determined that CpxRA and select gene mutants were essential for proper growth in SCF when in the presence of extraneous stressors and in competition. Although none of the Cpx-regulated gene mutants exhibited marked virulence phenotypes in vivo, the ΔcpxRA mutant had reduced colonization and attenuated virulence, as previously determined, which replicated the in vitro growth phenotypes specific to SCF. Overall, these results indicate that the ΔcpxRA virulence defect is not due to any single Cpx regulon gene examined. Instead, attenuation may be the result of defective growth in the colonic environment resulting from the collective impact of multiple Cpx-regulated genes.

INTRODUCTION

Attaching and effacing (A/E) pathogens such as enteropathogenic and enterohemorrhagic Escherichia coli (EPEC and EHEC, respectively) are intestinal pathogens capable of causing severe diarrhea in children as well as adults (EHEC) (1). Due to the inability of EPEC and EHEC to replicate their human pathology in mice, Citrobacter rodentium, a murine A/E pathogen with genetic similarities to both EPEC and EHEC, is used for in vivo models of infection (2, 3). C. rodentium causes colonic crypt hyperplasia in mice that results in severe inflammation in the colon (2, 4–6). Infections are lethal for some mice, like strain C3H/HeJ, while C57BL/6 mice experience self-limiting infections (7, 8). EPEC, EHEC, and C. rodentium all harbor the locus of enterocyte effacement (LEE) pathogenicity island that contains five operons (LEE1-5) that encode key proteins like intimin, Tir, and a type III secretion system (T3SS), which are responsible for the hallmark lesions formed upon intimate attachment to intestinal epithelial cells (1). The LEE is regulated by numerous factors, including the master transcriptional activator Ler, encoded in the LEE1 operon, which activates the expression of LEE1-5. Ler also activates expression of the transcriptional activator GrlA and repressor GrlR, encoded in the grlRA operon, which act upon LEE1 (5). In addition to LEE-encoded regulators, numerous other regulators of the LEE that have been studied in EPEC and EHEC which activate or inhibit its expression based on environmental stressors sensed by the cell (9, 10).

A stress response that has been previously shown to be required for C. rodentium colonization and virulence is the Cpx envelope stress response (ESR) (11–13). The Cpx ESR is regulated by a two-component system that consists of the sensor histidine kinase CpxA and the response regulator CpxR (14–16). In addition, its activity can be altered with two auxiliary signaling proteins, the inhibiting periplasmic protein CpxP and the activating outer membrane lipoprotein NlpE (16–18). While the sensing mechanism used by CpxA is poorly defined, inducing cues of the Cpx ESR include alkaline pH, salt, antimicrobial peptides, and misfolded inner membrane and periplasmic proteins (14, 16, 19–23). Upon activation, CpxA autophosphorylates and transfers a phosphoryl group to CpxR, which then binds to a consensus sequence upstream of Cpx regulon members, leading to altered expression. The CpxR binding consensus sequence, 5′-GTAAA(N)_4–8GTAAA-3′, is conserved to various degrees; however, nucleotide deviations from the consensus sequence do not predict the relative expression of the downstream gene (20, 21, 24).

Previous studies investigating the role of the Cpx ESR in other pathogens have suggested mechanisms by which the Cpx ESR may impact, both negatively and positively, colonization and virulence. Some of these mechanisms include the negative regulation of perC resulting in reduced ler expression in EPEC, aiding the efficient expression of the bundle-forming pilus involved in initial host cell attachment in EPEC, and induction of genes encoding proteins required for maintaining envelope protein integrity and virulence regulation, such as DegP, PpiA, and DsbA (9, 16, 25, 26). Overall, it has been concluded that the Cpx ESR facilitates adaptation to envelope stressors, because it downregulates virulence genes and large protein complexes while upregulating envelope and protein-modifying factors, although the specific reasons for its essentiality during pathogenesis have not been definitively demonstrated in most cases (16).

Previous work in C. rodentium concluded that the Cpx ESR is activated in vivo based on the observation of increased expression levels of Cpx-regulated cpxP (11). Given that C. rodentium is an A/E pathogen that relies on the LEE and LEE-encoded T3SS for virulence, it is important to note that the absence of CpxRA does not impact the secretion of T3SS effector proteins in vitro, indicating its presence has a limited impact on overall LEE function (11, 13). In addition, inconsistent results have been reported on the ability of the ΔcpxRA mutant to grow under the virulence-inducing condition of Dulbecco’s modified Eagle medium (DMEM), which is typically used as an in vitro growth condition prior to conducting in vivo models. Thomassin et al. (11) measured an extended lag phase for cells lacking CpxRA, while Vogt et al. (13) found no growth defect in the same medium. To further investigate the role of the Cpx ESR in pathogenesis, the Cpx regulon of C. rodentium was defined by microarray, transcriptome sequencing (RNA-Seq), and stable isotope labeling by amino acids in cell culture (SILAC) proteomic analysis (12, 13). The data produced from these two studies were extensive, and the impacts of numerous genes of interest on virulence have not been investigated. Giannakopoulou et al. (12) determined that the auxiliary proteins CpxP and NlpE were not required for colonization or virulence in C. rodentium. On the other hand, Vogt et al. (13) showed that the Cpx regulon members degP and dsbA were required for C. rodentium virulence in C3H/HeJ mice; however, the Cpx regulation of these genes was not responsible for the virulence defect seen when cpxRA was absent. The Cpx regulon is extensive, with the presence of CpxRA contributing to the differential expression of over 330 transcripts in C. rodentium (12, 13). The roles of some of the more strongly upregulated genes in both studies, such as yebE, ygiB, bssR, and htpX, which are investigated in this study, have not been defined in C. rodentium.

In this study, we aimed to elucidate the role of the Cpx ESR in C. rodentium colonization and virulence by utilizing simulated colonic fluid (SCF), first developed by Beumer et al. (27). Here, we show that SCF is a preferable medium for C. rodentium growth, highlights previously unobserved growth defects in Cpx-regulated gene mutants, and more accurately predicts a mutants’ performance in vivo relative to that in LB and high-glucose DMEM (HG-DMEM). From use of luminescent reporters, generating single, double, and triple mutants, and testing strains in competition with fluorescent wild-type C. rodentium, our data suggest the Cpx ESR is required for adequate survival and proliferation in the gut environment, modeled by SCF, by ameliorating stressors and ensuring the appropriate regulation of virulence and envelope-associated genes, which contributes to the colonization and pathogenesis of C. rodentium.

RESULTS

Simulated colonic fluid induces the Cpx envelope stress response.

Previous work in the field utilized HG-DMEM to induce virulence gene expression and test mutant fitness in vitro prior to conducting in vivo experiments. To further investigate the colonization defect associated with the ΔcpxRA mutant in vitro, we used the medium SCF, meant to mimic gastrointestinal conditions, which includes relevant components like bile.

Initially, it was important to determine whether the Cpx ESR was active in SCF. A luminescent reporter for the Cpx regulon member cpxP was used to measure Cpx ESR activation in both wild-type and ΔcpxRA cells grown in LB, HG-DMEM, and SCF. Our results indicated that both HG-DMEM and SCF strongly induced the Cpx ESR in wild-type cells (Fig. 1A and B). Meanwhile, in the ΔcpxRA mutant, cpxP-lux was not detectable in any of the conditions used in this study (Fig. 1A; see also Fig. S1 in the supplemental material). Unexpectedly, the cpxP-lux expression pattern over time and, by extension, the activity of the Cpx ESR, differed depending on the medium used for growth, thus indicating condition-specific activity levels (Fig. 1B). In addition, while having a less substantial impact on the activity of the Cpx ESR, cultures that were grown in LB with shaking had higher levels of cpxP-lux activity than static cultures, whereas cultures grown statically in HG-DMEM and SCF had higher cpxP-lux activities than found in shaking cultures (Fig. 1C to E). These results suggest that numerous factors, including media, growth phase, and culture conditions, all contribute to the activity of the Cpx ESR and should be considered when investigating the activity of envelope stress responses in vitro.

FIG 1.

The Cpx ESR is induced in HG-DMEM and SCF. Luminescent assay results indicate the activity of C. rodentium cpxP-lux in LB (gray), HG-DMEM (teal), and SCF (maroon). (A) Activity of cpxP-lux promoter in wild-type (+) and ΔcpxRA (−) cells, measured 1 h post-resuspension. ***, P < 0.001; ****, P < 0.0001 (Student's t test). (B) Luminescence generated from wild-type cpxP-lux promoter over time starting from 0.5 h post-resuspension. (C to E) Activity of wild-type cpxP-lux promoter measured over time from cells grown in either shaking (filled circle) or static (open square) conditions starting from 0.5 h post-resuspension. Luminescence was measured in counts per second (CPS) and normalized to bacterial optical density (OD₆₀₀). Data points represent the means of three biological replicate cultures, with error bars indicating standard deviations.

LEE master regulator ler expression responds to SCF and HG-DMEM.

Previous studies have demonstrated that the C. rodentium ΔcpxRA mutant does not have a significantly altered T3SS secreted protein profile in vitro, suggesting the LEE-encoded T3SS remains functional (11, 13). On the other hand, in EPEC and EHEC there is reduced transcription of LEE operons upon activation of CpxRA, as well as reduced expression and secretion of translocator proteins (28, 29). Using a ler-lux reporter, we investigated whether the attenuation of the C. rodentium ΔcpxRA mutant could be in part due to mistimed or inappropriate expression levels of the LEE master regulator ler. Expression levels were measured in SCF, to mimic the colonic lumen, and tissue culture medium HG-DMEM, as a proxy for the intestinal epithelial cell layer. Interestingly, wild-type and ΔcpxRA mutant cells grown in SCF exhibited similar low expression levels of ler-lux (Fig. 2A). Meanwhile, the ΔcpxRA mutant had significantly increased ler-lux expression when grown in HG-DMEM (Fig. 2B). In addition, our results indicated that ler-lux expression is more strongly activated in HG-DMEM than in SCF, which supports the notion that LEE gene expression is dependent on the surrounding environment and likely has increased expression at the epithelial cell surface relative to the colonic lumen (Fig. 2). Experiments were repeated in static conditions as well as with low-glucose DMEM (LG-DMEM), where the increased levels of ler-lux expression could also be observed in the ΔcpxRA mutant (see Fig. S2). As a culmination, these results support the hypothesis that the Cpx ESR is an important environmental sensor that contributes to the timing and extent of expression of the LEE, which could contribute to the attenuation of the ΔcpxRA mutant.

FIG 2.

The impact of the Cpx ESR on expression of the major LEE regulator ler is greater in HG-DMEM than in SCF. Luminescent assay results indicate the activity of C. rodentium cpxP-lux (black) and ler-lux (pink) reporters in wild-type (closed circles) and ΔcpxRA (open squares) cells grown in SCF (A) or HG-DMEM (B). Activity was measured over time starting from 0.5 h post-resuspension. Luminescence was measured in counts per second (CPS) and normalized to bacterial optical density (OD₆₀₀). Data points represent the means of three biological replicate cultures, with error bars indicating standard deviations.

Identification and confirmation of Cpx regulon members with potential virulence contributions.

Research done by two independent groups collected proteomic, RNA-Seq, and microarray data to identify genes that were differentially expressed in the absence of CpxRA (12, 13). Using the data collected, the genes listed in Table 1 were selected for further study based on their predicted function, expression levels, and/or lack of previous investigation. htpX and yebE showed some of the highest changes in transcript abundance when the cpxRA locus was ablated besides cpxP and yccA, which have both been previously investigated in C. rodentium (13). The SILAC data for htpX were insignificant due to the detection of only 1 peptide, while yebE had a 5.36-fold-higher peptide count in wild-type cells, with a P value of 0.08. ygiB had significantly higher transcript abundances in both transcriptomic studies as well as peptide counts in the presence of CpxRA (Table 1, values in bold). bssR is involved in biofilm regulation and had significantly higher transcript abundance in wild-type cells in both transcriptomic studies, despite an absence of detection in the SILAC data (Table 1).

TABLE 1.

Mined RNA-Seq and SILAC data from Vogt et al. (13) and microarray data collected by Giannakopoulou et al. (12)^a

Gene	Fold change data from Vogt et al. based on:		Fold change data from Giannakopoulou et al. based on ΔcpxRA-WT microarrayc	Protein functiond
	RNA-Seqb	SILACb
htpX	12.10	10.77	−0.93	Membrane-localized protease
yebE	24.52	5.36	−1.92	Inner membrane protein with transmembrane domain
ygiB	3.54	2.24	−0.99	Outer membrane protein
bssR	3.72	ND	−3.36	Biofilm regulator involved in catabolite repression and stress response

^aValues in bold indicate P_adj < 0.05 (RNA-Seq), false discovery rate of <0.05 (SILAC), or P < 0.05 (microarray). WT, wild type; ND, not determined.

^bWild type versus ΔcpxRA.

^cNegative values indicate higher expression in wild-type DBS100.

^dProtein function as described in UniProtKB for E. coli K-12.

Using the lux reporter plasmid pNLP10, four reporters were constructed and tested in LB broth, the virulence-inducing medium HG-DMEM, and SCF (21, 30, 31). While some of these genes, namely yebE and htpX, have shown Cpx-dependent expression confirmed experimentally in previous studies using other bacterial strains, their expression has never been studied in C. rodentium DBS100 (20, 21, 32, 33). Here, we showed that the expression of yebE, ygiB, bssR, and htpX relied on the Cpx ESR in LB, HG-DMEM, and SCF (Fig. 3A to D). Furthermore, all four reporters had significantly increased expression in HG-DMEM and SCF relative to the expression levels measured in LB in a Cpx-dependent manner (Fig. 3A to D). When comparing the expression of the four reporters in the ΔcpxRA mutant background, only htpX-lux indicated differences between the media tested (see Fig. S3). The yebE-lux reporter indicated the greatest level of induction, as promoter activity was upregulated 59- and 29-fold in HG-DMEM and SCF, respectively, relative to the measured expression in LB (Fig. 3A). Finally, each reporter gene, except for yebE, had increased luminescence in wild-type cells when grown on solid LB agar relative to that of ΔcpxRA mutant cells (see Fig. S4A).

FIG 3.

Cpx-regulated genes have increased expression in HG-DMEM and SCF, while the activity of the Cpx ESR is only impacted in ΔhtpX cells. (A to D) Luminescent assay results indicated the promoter activity of C. rodentium genes of interest in LB (gray), HG-DMEM (teal), and SCF (maroon). Activity of yebE-lux (A), ygiB-lux (B), bssR-lux (C), and htpX-lux (D) promoters in wild-type (+, solid bars) and ΔcpxRA (−, striped bars) cells measured 1 h post-resuspension. Luminescence was measured in counts per second (CPS) and normalized to bacterial optical density (OD₆₀₀). Bars represent the means of three biological replicate cultures, with error bars indicating standard deviations. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001 (Student's t test). (E to G) Mutant strains harboring cpxP-lux reporter plasmids were grown in LB (E), HG-DMEM (F), and SCF (G). Bars represent the means of three biological replicate cultures, with error bars indicating standard deviations. The asterisks indicate a statistically significant difference between the mutant strain relative to wild-type cells (*, P < 0.05; ****, P < 0.0001; one-way ANOVA with Dunnett's multiple-comparison test).

Given that our evidence indicates that yebE, ygiB, bssR, and htpX rely on the presence of CpxRA for proper expression, we questioned whether the absence of these genes would impact the activity of the Cpx ESR. Using allelic exchange, C. rodentium knockout mutants were generated for yebE, ygiB, bssR, and htpX and transformed with the cpxP-lux reporter to measure Cpx activity in LB, HG-DMEM, and SCF. Our results indicated that only ΔhtpX cells had significantly increased expression of cpxP in LB, HG-DMEM, and SCF, supporting previously reported findings observed in E. coli strain K-12 MC4100 (Fig. 3E to G) (32). Increased cpxP-lux expression was also seen on LB agar plates (see Fig. S4B).

Cpx-regulated genes impact C. rodentium fitness in simulated colonic fluid.

Previous research showed differing results with regard to the effect of removing the Cpx ESR on C. rodentium growth in vitro. Vogt et al. (13) showed no growth defects associated with a ΔcpxRA mutant in shaking LB broth or static high-glucose DMEM with 5% CO₂. On the other hand, Thomassin et al. (11) found that C. rodentium ΔcpxRA cultures had a longer lag phase in DMEM but eventually grew to a comparable optical density (OD) to the wild-type and complemented strains. In this study, all the mutants tested grew comparably in LB (Fig. 4A). In HG-DMEM, the mutants grew slower and to a lower OD than wild-type cells (Fig. 4B). One predominant issue to note with C. rodentium growth in HG-DMEM is that wild-type cells can neither grow quickly nor to a high density, as seen by the exponential growth phase starting at approximately 3.5 h and ending at 10.5 h while only reaching an OD at 600 nm (OD₆₀₀) maximum of 0.2 (Fig. 4B). Contrary to this, when grown in SCF, C. rodentium wild-type cells grew rapidly and reached an OD of 0.7 at 6.5 h (Fig. 4C). Interestingly, only the ΔcpxRA mutant had a marked growth defect when grown in SCF relative to the other strains (Fig. 4C). Furthermore, when SCF was not buffered and therefore unable to maintain a pH of 7 over time, the ΔcpxRA mutant could not grow, suggesting that unstable pH is toxic to these cells (Fig. 4D). A cpxRA-complemented strain showed restored growth and Cpx activity, measured by expression of a cpxP-lux reporter, in all conditions tested (see Fig. S5). While not exhibiting growth defects in buffered SCF (Fig. 4C), in unbuffered SCF both the ΔygiB and ΔhtpX mutants experienced an extended lag phase as well as increased variability between biological replicates, highlighting growth defects associated with these mutants (Fig. 4D).

FIG 4.

SCF promotes C. rodentium growth and highlights fitness defects in ΔcpxRA, ΔygiB, and ΔhtpX cells. Overnight cultures were standardized to an OD₆₀₀ of 1.0 and then subcultured 1:100 into LB (A), HG-DMEM (B), SCF (C), and SCF without MOPS buffer (D). Data points represent the means of three biological replicate cultures, with error bars indicating standard deviations.

Beyond unstable pH, another stressor relevant to the gastrointestinal tract and C. rodentium proliferation is the presence of oxidative stress. To test this physiologically relevant stressor, we grew the wild-type and mutant strains in morpholinepropanesulfonic acid (MOPS)-buffered LB and SCF with a subinhibitory level of hydrogen peroxide. In LB, the ΔcpxRA mutant had an extended lag phase relative to the other strains, although it reached a comparable OD after 9 h growth (Fig. 5A). Interestingly, similar to the growth in unbuffered SCF shown in Fig. 4D, the ΔcpxRA mutant was unable to grow in buffered SCF containing 1 mM hydrogen peroxide (Fig. 5B). Growth was restored to wild-type levels in the cpxRA-complemented strain (see Fig. S5E). In addition, the mutants ΔyebE, ΔygiB, and ΔhtpX all exhibited various degrees of susceptibility to subinhibitory levels of hydrogen peroxide in SCF, with ΔygiB and ΔhtpX mutants having the most severe growth defects (Fig. 5C to F). Finally, a consistent reduction in OD indicating cell lysis was also observed in SCF as cells transitioned from exponential to stationary phase (Fig. 4C and 5B to F; see also Fig. S5C and E). The cause of this reduction requires further study. These results demonstrate that SCF is a suitable growth medium for C. rodentium as indicated by the observed rapid exponential phase, high OD₆₀₀, and stable stationary phase. In addition, the ΔcpxRA mutant is extremely sensitive to physiologically relevant stressors present in the gastrointestinal environment, like unstable pH and oxidative stress.

FIG 5.

SCF with subinhibitory levels of oxidative stress inhibits ΔcpxRA mutant growth and highlights various fitness defects in ΔyebE, ΔygiB, and ΔhtpX cells. Overnight cultures were standardized to an OD₆₀₀ of 1.0 and then subcultured 1:100 in LB with 1 mM H₂O₂ (A) or SCF with 1 mM H₂O₂ (B to F), both buffered with 0.1 M MOPS. Data points represent the means of three biological replicate cultures, with error bars indicating standard deviations.

Following these observations, we hypothesized that a contributing factor to the inability of the ΔcpxRA mutant to grow in SCF with an unstable pH could be attributed to the additive effects of reducing the expression of Cpx-regulated genes, including yebE, ygiB, and htpX. To test this, double and triple mutants were generated using allelic exchange and grown in LB, SCF, and SCF with no MOPS (Fig. 6). These mutants exhibited no aberrant growth phenotypes in LB (Fig. 6A). In SCF, the double mutants containing ΔygiB as well as the triple mutant had extended lag phases like that of ΔcpxRA (Fig. 6B). These growth defects were exacerbated in unbuffered SCF where the only mutant that could successfully reach stationary phase was ΔyebE ΔhtpX, albeit with an extended lag phase relative to wild-type cells (Fig. 6C). The ΔygiB ΔhtpX double mutant as well as the triple mutant could not grow in unbuffered SCF (Fig. 6C). These results support the hypothesis that the growth defects of the ΔcpxRA mutant in both buffered and unbuffered SCF can be attributed to the cumulative reduction in expression of Cpx-regulated genes, with ygiB, yebE, and htpX being especially important.

FIG 6.

C. rodentium double and triple mutants of Cpx-regulated genes exhibit fitness defects similar to ΔcpxRA cells. Overnight cultures were standardized to an OD₆₀₀ of 1.0 and then subcultured 1:100 into LB (A), SCF (B), and SCF without MOPS buffer (C). Data points represent the means of three biological replicate cultures, with error bars indicating standard deviations.

To further analyze the fitness defects associated with the single-deletion mutants, in vitro competitive assays with a fluorescent wild-type C. rodentium strain, DBS100-AC, were conducted in LB and SCF. Of the five mutants tested, the ΔcpxRA and ΔygiB mutants were significantly outcompeted by DBS100-AC after 24 h growth in SCF (P < 0.05) (Fig. 7). Importantly, this competitive disadvantage was not seen in LB, further supporting the use of SCF to investigate mutant phenotypes in C. rodentium. Unexpectedly, the ΔcpxRA mutant significantly outcompeted DBS100-AC in LB after 24 and 48 h of growth (P < 0.01 and P < 0.05) (Fig. 7B). This result was repeated in multiple experiments using different ΔcpxRA mutant glycerol stocks to confirm validity. The reason for the opposite results of competitions in LB and SCF requires further investigation. In addition, while not statistically significant, the proportion of the ΔhtpX mutant in SCF was lower than that of DBS100-AC, and this trend was repeated in distinct experiments (Fig. 7F). Therefore, the in vitro data collected thus far indicate that the ΔcpxRA mutant, along with mutants carrying deletions of the Cpx-regulated genes ΔygiB and ΔhtpX, have growth defects and a competitive disadvantage in SCF.

FIG 7.

ΔcpxRA, ΔygiB, and ΔhtpX mutants exhibit fitness defects in competition assays. Wild-type and mutant strains of C. rodentium were competed against a fluorescent C. rodentium strain expressing amCyan from a tetracycline-inducible promoter (DBS100-AC). Strains were inoculated 1:1 into either LB or SCF, and cultures were enumerated at 0, 24, and 48 h on LB agar containing 0.2 μg/mL anyhrotetracycline. Bars represent the mean relative proportions of three biological replicate cultures, with error bars indicating standard deviations. Significance was calculated using arcsine-transformed data for 0 versus 24 h and 0 versus 48 h in both LB and SCF. *, P < 0.05; **, P < 0.01; ***, P < 0.001; n.s., not significant (Student's t test).

The in vivo colonization levels replicate growth trends observed in SCF.

With yebE, ygiB, bssR, and htpX expression confirmed to be upregulated by the presence of CpxRA, we then tested whether these genes were required for the colonization or virulence of C. rodentium, as this could provide a possible explanation for why the removal of cpxRA is detrimental to pathogenesis (11–13). C57BL/6J mice, which experience a self-limiting form of disease, were used to determine whether colonization was negatively impacted by any of the mutants (7). Only the ΔcpxRA mutant cells could not consistently colonize to the same level as wild-type cells and the other mutant strains (day 4. P < 0.05; day 9 and 12, P < 0.05 and P < 0.01; Mann-Whitney U test) (Fig. 8A; see also Fig. S6). While the ΔyebE mutant exhibited a slight lag in colonization levels on day 9 (P < 0.05) and the ΔbssR mutant showed a moderate increase in colonization of the colon on day 12 (P < 0.05), these minor statistically significant effects are not reflected in the degree of disease state measured using a splenic index (Fig. 8B; see also Fig. S6). All strains caused a similar level of disease relative to the wild type except for ΔcpxRA, which had a significantly lower splenic index indicating attenuated virulence (Fig. 8B), thus confirming the results of previous studies (11, 12). Similarly, in C3H/HeJ mice, used for testing disease progression and survival, only the ΔcpxRA mutant exhibited a colonization defect, as seen by an approximate 2-fold reduction in CFU for three mice and undetectable levels in two mice on day 4 postinfection (Fig. 8C). In addition, all five mice infected with the ΔcpxRA mutant survived until day 30, indicating a significant attenuation of virulence (Fig. 8D). Importantly, the in vivo results shown here reflect similar trends for the growth phenotypes of the C. rodentium strains observed in SCF, further indicating the importance of using physiologically relevant growth conditions in a laboratory setting prior to conducting animal trials.

FIG 8.

Genes of interest do not significantly contribute to C. rodentium colonization or virulence in C57BL/6J or C3H/HeJ mice. (A to C) Colonization measured in CFU per gram of feces at 4 days postinfection (A) and spleen index (B) in C57BL/6J mice infected with C. rodentium mutants. Colonization of C3H/HeJ mice infected with C. rodentium mutants measured in CFU per gram of feces at 4 days postinfection (C). Horizontal lines indicate the medians of n = 5 mice, and the asterisks indicate a statistically significant difference between the mutant strains and wild-type. *, P < 0.05; **, P < 0.01 (Mann-Whitney U Test). (D) Percent survival of C3H/HeJ mice infected with C. rodentium mutants. Mice were euthanized if any one of the following critical endpoints was reached: 20% body weight loss, hunching and shaking, inactivity, or body condition score of <2.

DISCUSSION

A frequently used gastrointestinal fluid is simulated gastric fluid, which has demonstrated the acid tolerance of pathogens like EPEC, EHEC, Vibrio cholerae, Listeria monocytogenes, and Salmonella spp. (34–37). Meanwhile, SCF is rarely, if ever, used in individual pathogen fitness studies, as it has primarily been used for determining drug solubility and delivery systems which involve looking at the microbiota’s effect on drug release (38, 39). On the other hand, HG-DMEM is more commonly used in pathogen studies and has been used to induce virulence gene expression in C. rodentium prior to conducting animal model experiments (30, 31). In this study, we chose to investigate the role of the Cpx ESR and Cpx-regulated genes in a medium relevant to the conditions present in the colon by using SCF, with the goal of identifying factors contributing to the ΔcpxRA mutants’ attenuation in vivo (11).

Throughout this study, we identified numerous growth conditions which influenced the activity of the Cpx ESR. The Cpx ESR is associated with monitoring proper membrane protein biogenesis, the repression of virulence factors, and maintaining cell wall integrity (16). A previous study published microarray data indicating that HG-DMEM moderately induced cpxP gene expression in EPEC overexpressing NlpE, though they did not comment on this result being DMEM dependent (33). While DMEM has been used in previous Cpx-related studies to induce virulence gene expression, to our knowledge this is the first study to distinctly demonstrate a strong induction of the Cpx ESR in HG-DMEM and SCF relative to LB (13, 25, 26, 28, 29, 33). The activation of the Cpx ESR in HG-DMEM and SCF highlights the importance of stringent regulation of envelope functions under virulence factor-expressing and physiologically relevant conditions.

Not only was the Cpx ESR activated in HG-DMEM and SCF, but it also had unique activity patterns demonstrated over time as well as patterns differentially impacted by shaking versus static conditions in a medium-dependent manner. In Salmonella enterica, it has been shown that the expression of hilA is CpxA dependent in low pH, and its expression is increased when cultures are grown statically (40). In uropathogenic E. coli, the agglutination titer was opposite for the wild type and strains lacking the small RNA RyfA when grown with shaking or statically in LB and human urine (41). Our observations of Cpx-regulated gene expression over time in different media indicate the dynamic nature of the Cpx ESR and how the integration of numerous, likely intrinsic and extrinsic signals can alter its activity. While single time points can indicate activation or repression by the Cpx ESR, collecting data over time allows for a further understanding of the reliance of gene expression on the Cpx ESR throughout the various phases of growth. Therefore, our study highlights the importance of culture conditions on the activity of the Cpx ESR and prompts questions as to the nature of the envelope stressors in the media tested and, by extension, how those conditions could impact the physiology of growing cells.

Presence of the Cpx ESR negatively impacts expression of master LEE regulator ler.

Previous studies have shown that the Cpx ESR negatively impacts the expression of virulence factors. In EPEC, overexpression of the response regulator CpxR reduced the activation of LEE1, LEE4, and LEE5, while the removal of CpxR resulted in increased expression of all five LEE operons when in a nonpathogenic E. coli background that lacked ler, indicating that the regulation of LEE by the Cpx response was likely ler independent (29). In EHEC, it has also been shown that the Cpx ESR negatively regulates virulence factors, including those that are LEE-encoded through sigma factor 32 and the Lon protease (28). On the other hand, a recent study by Kumar et al. (42) proposed a model for EHEC suggesting that CpxR upregulates the expression of ler directly and that serotonin is an inhibitor of the Cpx ESR that results in reduced transcription of LEE. Those authors used reverse transcription-quantitative PCR (qRT-PCR) and growth in LG-DMEM to show that ler was reduced in the absence of CpxR in EHEC and was similarly impacted in a C. rodentium ΔcpxA mutant. Due to these findings, we wished to verify the impact of the Cpx ESR on the expression of ler to determine whether the avirulence associated with the ΔcpxRA mutant could be due to reduced expression of the LEE in C. rodentium. Unlike Kumar et al. (42), our data indicated that in static SCF, as well as both high- or low-glucose DMEM, in static and shaking conditions, the expression of ler is consistently higher in the ΔcpxRA mutant (Fig. 2; see also Fig. S2 in the supplemental material). Therefore, our data suggest that the colonization and virulence phenotypes observed for the ΔcpxRA mutant are not due to reduced expression of ler but perhaps could be from the overexpression of ler, which may contribute to reduced fitness and inappropriately timed virulence mechanisms in vivo.

SCF is beneficial for determining susceptibilities of mutants to stressors and predicting colonization defects in vivo.

In this study, we showed that simulating the conditions that C. rodentium cells face during colonization of the colon by using SCF was able to uncover growth defects and susceptibilities that would have been unidentified in LB or HG-DMEM. Of note, wild-type C. rodentium grew to a significantly higher OD in SCF than in HG-DMEM and had an increased growth rate relative to that in LB, suggesting that this medium simulates an environment to which this gastrointestinal pathogen has adapted. Furthermore, mutants bearing mutations that ablated part or all of the Cpx ESR grown in SCF were more susceptible to extraneous stressors than wild-type C. rodentium, and these phenotypes were exacerbated in double and triple mutants, as well as in competition assays. Thus, employing conditions that more closely emulated those of the intestinal lumen allowed us to uncover phenotypes for mutations in genes that likely play roles in vivo that might be missed in less sensitive infection models.

Interestingly, the ΔcpxRA mutant had an increased lag phase when grown in the presence of oxidative stress in LB, while growth was abolished in SCF with H₂O₂ (Fig. 5A and B). As detailed thoroughly in recent reviews, it is understood that C. rodentium utilizes aerobic respiration to outcompete host microbiota during colonization (6, 43, 44). A previous study found that deletion of the cydAB genes in C. rodentium, which are required for aerobic respiration, resulted in a severe reduction of growth in vivo (45). Following this, it was determined that disruption to the mitochondrial respiration of intestinal epithelial cells is largely responsible for C. rodentium infection causing oxygenation of the mucosal surface, as opposed to solely colonic crypt hyperplasia (46, 47). The Cpx ESR has been implicated in the regulation of aerobic respiration in EPEC, where it has been shown that removal of cpxRA reduced the oxygen consumption capabilities of cells, which was attributed to problems with cytochrome bo₃ oxidase biogenesis or function (48). Similar conclusions have been made for Salmonella enterica serovar Typhimurium, which also utilizes aerobic respiration to expand in the gastrointestinal tract and experiences colonization defects in the absence of CpxRA (49, 50).

A follow-up study to that by Lopez et al. (45) found that prior to expansion by aerobic respiration, C. rodentium utilizes host-derived H₂O₂ as an electron acceptor during anaerobic respiration via cytochrome c peroxidase (ccp) (51). Both wild-type and Δccp mutant cells survived in LB in the presence of millimolar concentrations of H₂O₂, though in the absence of ccp there was increased expression of the catalase peroxidase katG, suggesting that cells were experiencing higher levels of oxidative stress. The RNA-Seq data mined in this study also showed katG expression was significantly induced in the ΔcpxRA mutant grown in HG-DMEM, which suggests the cells were experiencing elevated levels of oxidative stress (13). Given that rapid expansion by aerobic respiration is a proposed mechanism of C. rodentium pathogenesis in overcoming colonization resistance and that the production of reactive oxygen species by intestinal epithelial cells is an important defense mechanism, our results support the hypothesis that the Cpx ESR is required to successfully adapt to oxidative stressors during colonization in the gastrointestinal tract (45, 52–54).

Previous research using variations of simulated gut media have indicated specific bacterial responses to conditions emulating the gastrointestinal tract. Musken et al. (55) used simulated intestinal fluids mimicking the ileum and colon to demonstrate differential expression of a major fimbrial subunit in sorbitol-fermenting EHEC, while Polzin et al. (56) found that EHEC proteins involved in nucleotide biosynthesis and the expression of Shiga toxins were increased in simulated ileal and colonic environments. In addition, it was found that outer membrane vesicle (OMV) production and cytotoxicity, which is an important virulence factor, as well as OMV-associated Shiga toxin 2a in EHEC was increased in both simulated ileal and colonic environments (57). On the other hand, OMV cytotoxicity and OMV-associated Shiga toxin 2a was not increased in DMEM, which was confirmed with qRT-PCR for stx_2a expression (57). One study in Salmonella demonstrated differences in cell viability upon exposure to gastrointestinal fluids, where the investigators found that gastric juice with a pH of 4 or 5 in conjunction with bile salts from simulated intestinal juice reduced cell viability more than acidic pH alone (37). These studies, in conjunction with the data presented here, highlight the importance of using a physiologically relevant environment when it comes to monitoring gene and protein expression in cells during gastrointestinal survival, colonization, and virulence.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Bacterial strains used in this study are listed in Table S1 in the supplemental material. Unless otherwise indicated, cells were grown in lysogeny broth (LB; 10 g/liter tryptone, 5 g/liter yeast extract, and 5 g/liter NaCl), with appropriate antibiotics at 37°C and 225 rpm. Cells were also grown on LB with 1.5% agar plates and incubated at 37°C for 16 to 18 h. When required, media were supplemented with 30 μg/mL kanamycin, 100 μg/mL ampicillin, 25 μg/mL chloramphenicol, 0.3 mM diaminopimelic acid, 5% sucrose, 0.2 μg/mL anhydrotetracycline.

Deletion mutant construction.

All deletion mutants were generated using allelic exchange and the methods described by Vogt et al. (13). In summary, in-frame deletion constructs were generated using overlap extension PCR and the primers listed in Table S2. Amplicons were digested using the XbaI and SphI/PaeI, ligated into pUC18, and transformed into OneShot TOP10 chemically competent cells (Invitrogen, USA). Plasmids were mini-prepped and sent for Sanger sequencing for amplicon confirmation. Once confirmed, the deletion construct was digested out of pUC18 and ligated into the suicide vector pRE112, where it was transformed by electroporation into MFDpir cells (58). MFDpir cells containing the deletion construct underwent conjugation with a recipient strain of C. rodentium DBS100, and single-crossover colonies were selected with chloramphenicol. Single-crossover colonies were confirmed by PCR using primers designed to flank the deletion site by ~75 bp on each side (Table S2). Loss of pRE112 was determined by plating on LB -NaCl with 5% sucrose (filter sterilized) and growing at room temperature for 2 days. Colonies that were sucrose resistant and chloramphenicol sensitive were screened by PCR to confirm the intended deletion.

Luminescent reporter construction.

Luminescent reporters were constructed using the pNLP10 lux reporter plasmid (21). Primers listed in Table S2 were designed to amplify ~500 bp upstream and 50 bp downstream of the translation start site apart from ygiB, where the amplified promoter was ~300 bp. The restriction enzymes EcoRI and BamHI were incorporated into the forward and reverse primers, respectively, unless otherwise indicated. Promoter regions were amplified, cloned into pNLP10, and transformed into OneShot TOP10 chemically competent cells (Invitrogen, USA). TOP10 colonies harboring the pNLP10 plasmid were confirmed for insert presence using colony PCR with primers flanking the pNLP10 multiple-cloning site as well as with Sanger sequencing. Plasmids with the correct insert were mini-prepped and transformed into electrocompetent C. rodentium wild-type or mutant cells.

Luminescence assays.

Strains grown overnight in biological triplicate were subcultured 1:100 and grown to an OD₆₀₀ of 0.4 to 0.6. Following this, subcultures were divided into 1-mL aliquots and centrifuged, and the subculturing medium was removed. Cells were then resuspended in 1 mL of fresh medium containing 30 μg/mL kanamycin (T = 0). Resuspension medium included LB, HG-DMEM with no phenol (Gibco catalog number 31053028), and SCF, using ox bile in lieu of porcine bile (6.25 g/liter proteose-peptone, 2.6 g/liter glucose, 0.88 g/liter NaCl, 0.43 g/liter KH₂PO₄, 1.7 g/liter NaHCO₃, 2.7 g/liter KHCO₃, 4.0 g/liter ox bile, and 0.1 M MOPS) (27). HG-DMEM and SCF were buffered with 0.1 M MOPS to maintain physiological pH levels of 7.4 and 7.0, respectively, unless otherwise indicated. SCF was prepared fresh for each experiment and filter sterilized. For luminescent assays with subinhibitory levels of oxidative stress, H₂O₂ was added to a final concentration of 1 mM in LB and SCF, both of which were buffered with 0.1 M MOPS.

A 200-μL aliquot of induced cells was then transferred into a black-walled clear-bottom 96-well plate and incubated at 37°C with shaking, unless otherwise indicated. Empty wells were left between strains to prevent contaminating luminescence from adjacent wells. The OD₆₀₀ and luminescence measurements were taken either 1 h postresuspension or over 6 h starting at 0.5 h postresuspension, using the Victor X3 2030 multilabel plate reader (Perkin Elmer). Luminescence was measured in counts per second (cps) and normalized using the measured OD₆₀₀ of the same well to accommodate for differences in cell number between cultures. Prior to calculating the cps/OD₆₀₀ ratio, the OD and cps values measured for blank wells containing each induction medium were subtracted from experimental wells. Biological triplicates were averaged to determine the mean luminescence in each condition, and statistical significance was calculated using a Student’s t test. For cpxP-lux activity in mutant strains, luminescence was compared to wild-type levels of cpxP-lux by using a one-way analysis of variance (ANOVA) with Dunnett’s multiple-comparison test. Luminescence assays were repeated at least twice, with the data from one representative experiment shown.

Growth curves.

Strains were grown in biological triplicate overnight, washed with 1× phosphate-buffered saline (PBS), and standardized to an OD₆₀₀ of 1.0. Cells were inoculated 1:100 into medium aliquoted in a clear 96-well plate. Growth media included LB, HG-DMEM, SCF, or SCF with no MOPS. Growth assays conducted in the presence of oxidative stress were prepared in the same manner, with the addition of hydrogen peroxide to a final concentration of 1 mM. For these assays, both the LB and SCF was buffered with 0.1 M MOPS to control for pH changes in the medium. Plates were read in an Epoch2 microplate reader (Biotek, USA) set to 37°C with continuous orbital shaking at a frequency of 237 cpm (4 mm) at slow speed. Blank wells were subtracted from corresponding culture wells prior to calculations. Biological triplicates were averaged, with data points indicating the mean and the standard deviation indicated by error bars. Growth curves were completed at least twice, with the data from one experiment shown.

Construction of fluorescent amCyan C. rodentium.

A C. rodentium strain containing a chromosomal, tetracycline-inducible amCyan gene was constructed via two sequential rounds of allelic exchange. The purpose of the first round of allelic exchange was to integrate tetRA genes into the C. rodentium gene xylE. xylE has been used in previous studies as an integration site for constitutive lux expression as well as tetracycline-induced expression of tccP, and its interruption has been shown to not impair C. rodentium fitness (59, 60). Using the primers listed in Table S2, fragments 1 and 3 constituting the intended flanking regions of xylE were amplified from C. rodentium DBS100 genomic DNA based off the primers designed by Girard et al. (60), while fragment 2 containing the tetRA genes was amplified from the E. coli commensal MP7 (61), which contains a chromosomal tetracycline-inducible mcherry gene. The three fragments were joined 1-2-3 using overlap-extension PCR and amplified with primers containing restriction sites for SphI and SacI, creating the fragment xylE-tetRA-xylE. Confirmation of the fragment, generation of the donor strain, and conjugation were then carried out as previously described for construction of deletion mutants. Sucrose-resistant, tetracycline-resistant, and chloramphenicol-sensitive C. rodentium colonies were PCR screened to confirm insertion of tetRA into xylE (DBS100-tetRA).

To insert the amCyan gene downstream of tetRA and using the primers listed in Table S2, fragment 1 containing the 3′-end of tetA was amplified from MP7 (61), fragment 2 containing amCyan was amplified from the plasmid pFU78 (62), and fragment 3 containing the portion of xylE homologous to that which is downstream of the tetA gene in the strain DBS100-tetRA was amplified from DBS100 genomic DNA. The three fragments were fused 1-2-3 using overlap extension PCR to generate the new fragment, tetA-amCyan-xylE, which was then sequenced in pUC18, subcloned into pRE112, and transformed into MFDpir for conjugation with DBS100-tetRA. Successful transconjugants were selected on LB -NaCl with 5% sucrose plates containing tetracycline. After 2 days of growth at room temperature, plates were imaged using the ChemiDoc system, using Cy2 for visualization of fluorescent amCyan C. rodentium colonies (DBS100-AC), which were then selected for further confirmation via PCR.

Competitive assays.

The in vitro competitive assays between DBS100-AC and the generated mutant strains were conducted in both LB and SCF following the protocol previously described by Pick et al. (63), with minor changes. In summary, overnight cultures of each strain grown in biological triplicate were standardized to an OD₆₀₀ of 1.0. The ΔcpxRA mutant was standardized to an OD₆₀₀ of 1.4, as previous enumeration indicated overnight cultures had fewer viable cells relative to the other strains (data not shown). Ten microliters of DBS100-AC and 10 μL of the competitor strain were then inoculated into 2 mL of either LB or SCF. Cocultures were vortexed, and a 100-μL aliquot was removed for serial dilution, which was plated using the track dilution method on LB agar containing 0.2 μg/mL anhydrotetracycline (64). The cocultures were immediately incubated at 37°C for 24 h following aliquot removal, while plates with track dilutions were incubated at 30°C overnight. Plates were imaged using the ChemiDoc system, with Cy2 for visualization of fluorescent DBS100-AC. The number of competitor (nonfluorescent) colonies was determined by subtracting the number of fluorescent colonies from the total colony count. At 24 h, cocultures were subcultured 1:100 into fresh medium, and an aliquot from the mature coculture was serially diluted and plated for enumeration. At 48 h, the mature coculture was serially diluted and plated for enumeration.

C57BL/6J and C3H/HeJ mouse infections.

All animal experiments were performed under conditions specified by the Canadian Council on Animal Care and were approved by the McGill University Animal Care Committee. C57BL/6J and C3H/HeJ mice were purchased from the Jackson Laboratory (Bar Harbor, ME, USA). Five-week-old female mice were infected by oral gavage with 0.1 mL of LB medium containing 2 × 10⁸ to 3 × 10⁸ CFU of bacteria. The infectious dose was verified by plating serial dilutions of the inoculum onto MacConkey agar (Difco). For survival analysis of C3H/HeJ mice, the mice were monitored daily and were killed if they met any of the following clinical endpoints: 20% body weight loss, hunching and shaking, inactivity, or body condition score of <2 (65). To monitor bacterial colonization, fecal pellets or the terminal centimeter of the colon were homogenized in PBS, serially diluted, and plated on MacConkey agar. Plates containing between 30 and 300 colonies were counted. Spleens were removed and weighed, and splenic indexes were calculated as the square root of [(weight of spleen × 100)/(weight of mouse)].