Role of Ethanolamine Utilization Genes in Host Colonization during Urinary Tract Infection

ABSTRACT

Urinary tract infection (UTI) is the second most common infection in humans, making it a global health priority. Nearly half of all women will experience a symptomatic UTI, with uropathogenic Escherichia coli (UPEC) being the major causative agent of the infection. Although there has been extensive research on UPEC virulence determinants, the importance of host-specific metabolism remains understudied. We report here that UPEC upregulates the expression of ethanolamine utilization genes during uncomplicated UTIs in humans. We further show that UPEC ethanolamine metabolism is required for effective bladder colonization in the mouse model of ascending UTI and is dispensable for bladder colonization in an immunocompromised mouse model of UTI. We demonstrate that although ethanolamine metabolism mutants do not show increased susceptibility to antimicrobial responses of neutrophils, this metabolic pathway is important for surviving the innate immune system during UTI. This study reveals a novel aspect of UPEC metabolism in the host and provides evidence for an underappreciated link between bacterial metabolism and the host immune response.

INTRODUCTION

Urinary tract infection (UTI) is the second most common infection in humans after those involving the respiratory tract. The high incidence of UTIs makes these infections an important health concern that is further exacerbated by the rise of antibiotic-resistant bacterial strains, especially considering the high rates of recurrent infections (1–3). Uropathogenic Escherichia coli (UPEC) is responsible for more than 80% of uncomplicated UTIs (4). In most cases of uncomplicated UTIs, infection is initiated after the introduction of intestinal E. coli into the urinary tract. This allows bacteria to ascend the urethra and multiply in the bladder. Bacterial colonization of the bladder is characterized by robust activation of the innate immune response, including an influx of neutrophils (5).

UPEC responds to the hostile environment of the urinary tract by expressing a variety of virulence and fitness factors that facilitate bacterial colonization (2, 3, 6). In addition to these classically defined virulence factors, recent work has implicated the bacterial metabolic potential as an essential component of successful host colonization (6–8). However, as of now, there are few examples of metabolic pathways that are specific to human urinary tract colonization, and we lack understanding of how specific changes in bacterial metabolism lead to increased bacterial fitness. Defining the metabolic pathways that allow survival within the host is required to understand the environment at the site of infection and to identify potential metabolic targets for antimicrobial therapies.

Ethanolamine metabolism has been implicated as an important pathway for pathogenesis of such intestinal colonizers as enterohemorrhagic E. coli (EHEC), Listeria, and Salmonella (9–15). Ethanolamine is a component of phosphatidylethanolamine, an essential phospholipid found in both mammalian and bacterial cell membranes. Bacterial breakdown of ethanolamine, mediated by the 17-member ethanolamine utilization operon (Fig. 1A), results in production of ammonia and acetaldehyde. Ammonia can serve as a cellular source of nitrogen, while acetaldehyde can be converted into either acetyl coenzyme A or ethanol. Ethanolamine catabolism also requires the construction of bacterial microcompartments that serve to concentrate reaction intermediates and potentially limit toxic effects of acetaldehyde on the cell (16). The ethanolamine utilization operon contributes to bacterial survival via two different mechanisms. Under conditions of inflammation, Salmonella enterica serovar Typhimurium relies on ethanolamine as a supplementary carbon source used to outcompete intestinal microbiota (17). In addition, EHEC, while unable to grow on ethanolamine as a sole carbon source, outcompetes microbiota of the bovine intestinal tract by using ethanolamine breakdown products as an additional source of nitrogen (14). In contrast, during S. Typhimurium infection, ethanolamine catabolism per se is dispensable at the later stages of infection, while expression of ethanolamine utilization (eut) operon transcriptional regulator EutR is important for bacterial dissemination and survival within macrophages (12). In both Salmonella and EHEC, EutR has regulatory functions beyond the eut operon and is able to activate expression of virulence and fitness genes important for dissemination, intracellular survival, and cell adhesion (11, 15, 18).

FIG 1.

Ethanolamine utilization genes are upregulated in patients with uncomplicated UTIs. (A) Structure of ethanolamine utilization (eut) operon. The operon is composed of 17 genes that encode both enzymatic (blue) and structural components (pink) of the pathway. Transcriptional regulator gene eutR (yellow) is expressed at low levels from a secondary promoter. In the presence of ethanolamine and vitamin cofactor B₁₂, EutR drives transcription of the whole operon, initiating ethanolamine catabolism. (B) Comparison of the average eut gene expression of five clinical UPEC isolates in patients with UTI and the same strains cultured in vitro in sterile urine. Mean log₂ RPKM (reads per kilobase of transcript per million mapped reads) values for every gene in the eut operon are shown.

The urinary tract is an environment that is very different from that of the intestine, with little to no competition from other bacteria, a more severe nutritional profile, and a robust inflammatory response with which bacteria must contend (6). Therefore, whether ethanolamine metabolism plays a role in bacterial colonization of the urinary tract remains unknown. To follow up on our previous study (19), we performed a comparative transcriptome analysis of the eut operon on clinical UPEC strains and reveal increased eut gene expression in UTI patients compared to eut expression in the same strains cultured in sterile human urine. Using the well-established mouse model of ascending UTI, we demonstrated that a UPEC strain lacking the eut operon (Δeut mutant) has a severe defect in bladder colonization. The colonization defect is rescued in the mouse model of infection with a deficient innate immune response, suggesting that ethanolamine metabolism is required for surviving the host innate immune response. Moreover, UPEC lacking the ability to metabolize ethanolamine did not exhibit increased susceptibility to antimicrobial responses. Finally, the expression of eutR was necessary, but not sufficient, to complement the Δeut colonization defect. In summary, we identify ethanolamine catabolism as a novel host-specific adaptation that allows UPEC to successfully colonize the urinary tract during UTI.

RESULTS

UPEC strains increase expression of ethanolamine utilization genes during human UTI.

The ethanolamine utilization (eut) operon consists of 17 genes that are responsible for transport, compartmentalization, and breakdown of ethanolamine (Fig. 1A) (16, 20). We recently performed a study comparing the gene expression of five UPEC isolates in patients to the gene expression of the same strains cultured in vitro in sterile urine samples from healthy donors (19). In that study, we identified ethanolamine utilization as a novel host-specific fitness factor. In fact, the mean expression of every gene in the eut locus was higher in patients than during in vitro growth in urine (Fig. 1B). These data suggest that there are elevated levels of ethanolamine in the bladder and/or urine during the infection that are sufficient to induce the expression of the eut operon. Furthermore, this is the first evidence for a role of ethanolamine metabolism in bacterial pathogenesis during urinary tract infection.

Ethanolamine utilization genes are required for effective bladder colonization in mouse models of UTI.

To test our hypothesis that ethanolamine metabolism plays an important role in the establishment of UTIs, we constructed a deletion mutant lacking the eut operon (Δeut mutant) in a wild-type (WT) UPEC strain, CFT073. The growth of the Δeut mutant was indistinguishable from WT UPEC in both rich medium (i.e., Luria-Bertani [LB] medium) and filter-sterilized human urine (see Fig. S1A and B in the supplemental material). We then assessed bladder bacterial loads in C57BL/6 mice 48 h after transurethral inoculation with either the WT or the Δeut mutant. The ethanolamine metabolism mutant showed a significant defect in bladder colonization, with an ∼100-fold decrease in bacterial load in Δeut mutant-infected mice (Fig. 2A).

FIG 2.

The ethanolamine utilization mutant has a colonization defect in a mouse model of UTI. (A) C57BL/6 mice were transurethrally inoculated with 10⁸ CFU of either WT CFT073 or the ethanolamine utilization deletion (Δeut) mutant. Bacterial loads in the bladders were assessed 48 h postinoculation. P values were determined using a two-tailed Student t test. n ≥ 26 for each condition tested. Mean log(CFU/g) values and standard errors of the mean (SEM) are shown. (B) Mouse infections were performed as described in panel A. Bladder counts were evaluated at 24 h, 48 h, and 7 days postinfection. P values were determined using a two-tailed Student t test on each time point independently. n = 9 for each condition tested. Mean of log(CFU/g) values and SEM are shown.

We next assessed bacterial bladder colonization at 24 h, 48 h, and 7 days postinoculation in the mouse model of UTI to investigate differences in the kinetics of infection between the WT and the Δeut mutant. Both the WT and Δeut mutant strains showed similar bacterial loads in the bladder at 24 h postinfection (Fig. 2B), suggesting that ethanolamine metabolism is not required for the initial establishment of infection. However, as shown in Fig. 2A, Δeut mutant levels dropped precipitously between 24 and 48 h (Fig. 2B). Unexpectedly, by 7 days postinoculation, mice infected with the Δeut mutant had a higher bacterial burden than mice infected with WT UPEC. This suggests that the ability to metabolize ethanolamine might play a role in later disease outcomes and/or establishment of chronic infection (Fig. 2B). The altered progression of infection observed with the Δeut mutant shows that ethanolamine metabolism is needed to maintain high bladder colonization at a specific stage of the infection. Considering the fact that Δeut bacteria are especially vulnerable between 24 and 48 h, we hypothesized that this susceptibility is associated with mobilization of the innate immune responses in the bladder.

Clearance of Δeut mutant coincides with increased levels of MPO in the bladders.

We next wanted to investigate innate immune responses and estimate neutrophil presence in the bladders during the infections by detection of myeloperoxidase (MPO; an enzyme commonly present in neutrophil granules [21, 22]). As we hypothesized, MPO levels increased dramatically between 24 and 48 h in infected bladders (Fig. 3A), suggesting mobilization of innate immune response and an increased presence of neutrophils at this point in the infection. Although there was a high level of variability in MPO levels, there were no significant differences in the bladders during WT or Δeut mutant infections (Fig. 3A). Nevertheless, activation of the innate immune response was coincident with a drop in the burden of the Δeut bacteria but not WT bacteria. This observation supports the hypothesis that ethanolamine metabolism plays a role in surviving the innate immune responses during UTI.

FIG 3.

The inflammatory response is unaffected during infection with the Δeut mutant. (A) MPO levels in the bladders of mice infected with either WT or Δeut bacteria at 24 h, 48 h, and 7 days postinfection. P values were determined using a two-tailed Student t test independently for each time point. n = 9 for each condition tested. Median values are shown. (B to D) Levels of the proinflammatory cytokines TNF (B), MIP-2 (C), and CCL2 (D) in bladders of mice after 48 h of infection with either WT or Δeut UPEC. P values were determined using a two-tailed Student t test. n ≥ 9 for each condition tested. Median values are shown.

We then looked more broadly at the inflammatory milieu in the bladders at 48 h postinfection. We measured levels of tumor necrosis factor (TNF), MIP-2, and CCL2, cytokines that are responsible for driving the inflammatory response and recruitment of neutrophils (MIP-2 and CCL2) (22). In accordance with MPO assay data, there was no difference in the levels of these proinflammatory cytokines in the bladders infected by either WT or Δeut bacteria (Fig. 3B to D). Since the ability to breakdown ethanolamine did not affect the amplitude of the inflammatory response to UTI, we conclude that ethanolamine is unlikely to be a global modulator of the immune response.

A compromised innate immune response allows efficient bladder colonization by the Δeut mutant.

We next evaluated bladder colonization by the Δeut mutant in a mouse model with compromised innate immune function. These mice were depleted primarily of neutrophils by injection of the Gr-1 antibody 24 and 6 h prior to infection (23). Although this antibody also eliminates a subset of monocytes (24), these (Gr-1^hi) monocytes were previously shown to be dispensable in the context of the UTI (25), suggesting that the phenotypic effects we observe in this model are primarily due to neutrophil depletion. The success of neutrophil depletion was assessed by measuring levels of MPO in the bladders of both treated (Gr-1) and untreated mice (Fig. 4A).

FIG 4.

Rescue of the Δeut colonization defect in mice with a compromised innate immune response. (A) Levels of MPO present in bladders of untreated mice (untreated) or mice treated with Gr-1 antibody (Gr-1) 48 h after bacterial inoculation. The P value was determined using a two-tailed Student t test. n = 10 for all conditions tested. (B) Expression of eutR in mouse bladders 48 h after bacterial inoculation of either untreated or Gr-1-injected mice was measured using qPCR. The P value was determined using a two-tailed Student t test. n = 5 for WT infections; n = 2 for Δeut mutant infections. ND, not determined. Mean values and SEM are shown. (C and D) Immunocompetent or Gr-1-treated mice (indicated with Gr-1) were infected with either WT or Δeut bacteria as already described. Bacterial loads in bladders (C) and urine samples (D) were counted at 48 h postinoculation. P values were determined using one-way analysis of variance (ANOVA). n = 10 for all conditions tested. Mean log(CFU/g) values and SEM are shown.

We next assessed whether the innate immune response alters ethanolamine levels at the site of infection by measuring expression of the ethanolamine-responsive eutR. Immunocompetent and Gr-1-injected mice were infected with either WT or Δeut bacteria, and RNA isolated from the infected bladders at 48 h postinfection was used to examine levels of eutR expression by quantitative PCR (qPCR). In infections with WT bacteria, Gr-1 treatment did not significantly alter the expression of the eutR gene (Fig. 4B). As expected, no eutR expression was detected in the Δeut mutant. These data suggest that the innate immune response does not alter the levels of ethanolamine at the site of infection.

Gr-1-injected mice, as well as untreated controls, were then infected with either WT or Δeut UPEC, and bacterial colonization was assessed 48 h postinoculation in the bladder, as well as in urine samples. Unexpectedly, we observed no increase in the WT bacterial load in the bladders or urine samples of Gr-1-treated mice despite the significant decrease in neutrophil activity as indicated by MPO assay (Fig. 4A, C, and D). Although this is surprising, UPEC strains possess a wide array of strategies that allow them to resist neutrophil responses and to maintain high colonization levels in the early stages of the infection (26–32). Importantly, in mice lacking a proper innate immune response, the ethanolamine metabolism mutant did not exhibit the colonization defect observed in immunocompetent mice in bladders and urine (Fig. 4C and D). Thus, our data show that in the absence of the innate immune response, the infectivity of the Δeut mutant is indistinguishable from that of the WT. This provides further support to the hypothesis that UPEC ethanolamine metabolism contributes to resisting the innate immune response and effectively colonizing the host bladder.

Ethanolamine metabolism does not alter bacterial susceptibility to antimicrobial defenses of neutrophils.

Neutrophils effectively eradicate invading pathogens by the generation of toxic oxygen radicals (oxidative burst) and the release of toxic granule proteins stored in their cytoplasm (22). We therefore tested whether the colonization defect of Δeut bacteria can be explained by increased susceptibility to antimicrobial responses of neutrophils.

WT and Δeut mutant bacteria, treated with various concentrations of hydrogen peroxide, showed no differences in survival (Fig. 5A), suggesting that sensitivity to oxidative stress was unchanged. Furthermore, when bacteria were treated with various concentrations of the potent antimicrobial peptide LL-37 (33), there again was no indication of increased susceptibility in the Δeut mutant (Fig. 5B). In conclusion, we found no evidence to suggest that ethanolamine metabolism increases UPEC resistance to the direct antimicrobial responses of neutrophils.

FIG 5.

The Δeut mutant does not show increased susceptibility to oxidative stress or antimicrobial peptide damage. (A) WT and Δeut bacteria were cultured to log phase in minimal medium with glucose in the presence of ethanolamine and then treated with increasing concentrations of hydrogen peroxide. Data are shown as the percent survival compared to untreated bacteria (0 mM H₂O₂). P values were determined by using a two-tailed Student t test independently for each concentration. n = 3. P values are >0.05 unless otherwise indicated. Mean values and SEM are shown. (B) WT and Δeut bacteria were cultured to log phase in minimal medium with glucose in the presence of ethanolamine and then treated with increasing concentrations of LL-37 for 1 h. Data are shown as the percent survival relative to untreated bacteria (0 μg/ml LL-37). P values were determined by using a two-tailed Student t test independently for each concentration. n = 3. P values are >0.05 unless otherwise indicated. Mean values and SEM are shown.

Transcriptional regulator EutR is necessary but not sufficient for effective bladder colonization in mice.

Recent work with EHEC and Salmonella revealed that the transcriptional regulator of the eut operon, EutR, regulates bacterial fitness and survival inside the host by activating the transcription of virulence factors outside the eut operon (12, 15). Therefore, we tested whether global EutR-mediated transcriptional regulation is required for increased survival in the presence of the innate immune response. We first constructed a eutR deletion mutant in the WT CFT073 background (ΔeutR mutant) and tested host colonization in the ascending model of UTI. Since EutR is essential for transcription of the entire eut operon, the ΔeutR mutant had a colonization defect similar to that observed with Δeut bacteria in both bladder and urine (Fig. 6). Moreover, this colonization defect was rescued by complementing bacteria with eutR [ΔeutR(pGEN-eutR)] (Fig. 6).

FIG 6.

eutR expression does not restore the colonization defect in the Δeut mutant. (A and B) Mice were infected as described above with the WT, ΔeutR mutant, complemented ΔeutR mutant [ΔeutR(pGEN-eutR) mutant], and Δeut mutant complemented with eutR [Δeut(pGEN-eutR) mutant]. Bladder (A) and urine (B) bacterial loads were measured at 48 h postinoculation. P values were determined using one-way ANOVA analysis. n ≥ 10 for all conditions tested. P values are >0.05 unless otherwise indicated.

We next expressed eutR in a Δeut background [Δeut(pGEN-eutR)]. Expression of eutR without the rest of the operon did not alter growth in LB and urine (see Fig. S1C and D in the supplemental material). However, mice infected with Δeut(pGEN-eutR) showed lower bacterial loads than WT mice in both urine and bladder (Fig. 6), with bacterial loads that were comparable to those of the ΔeutR mutant, as well as the ΔeutR and Δeut mutants complemented with an empty vector (see Fig. S1E and F in the supplemental material). Thus, eutR expression is necessary but not sufficient for effective bladder colonization. These data strongly suggest that the Δeut colonization defect is due to bacterial inability to break down ethanolamine rather than to a failure in the activation of other virulence pathways.

DISCUSSION

Despite intensive research into UPEC pathogenesis, we have yet to fully comprehend the fitness requirements for establishing a UTI. The focus has shifted recently from studies of specific virulence factors toward the appreciation of bacterial metabolic state induced by the host (6). We previously identified ethanolamine metabolism genes to be specifically induced in UPEC strains during the infection of human hosts, but not during in vitro growth in sterile urine (19). Although ethanolamine metabolism has been studied in the context of intestinal models of infection (9, 12, 14), we provide here the first evidence of its importance during infection of the urinary tract. Using the well-established mouse model of ascending UTI, we demonstrate a novel role for ethanolamine metabolism in surviving the innate immune response and establishing a successful infection. Ethanolamine operon mutants displayed a drop in numbers compared to WT bacteria that correlated with the time frame for the activation of the innate immune response in the bladder. Moreover, the bacterial burdens of the Δeut mutant and WT UPEC were indistinguishable in mice with a compromised innate immune response, supporting our hypothesis for the role of ethanolamine metabolism in survival of the host innate immune response. Unexpectedly, we saw a consistent, although not significant, change in the baseline bacterial burdens between untreated and immunocompromised mice infected with WT bacteria. Interestingly, the increased susceptibility of the Δeut mutant to neutrophils observed in the mouse model did not result from the bacteria's increased sensitivity to oxidative stress or antimicrobial peptide damage.

Recent work has implicated ethanolamine as a signaling molecule, driving bacteria to adopt gene expression suitable for survival within the host. Specifically, in both EHEC and Salmonella species, the presence of ethanolamine triggered EutR-mediated expression of virulence genes important for bacterial cell adhesion, intracellular survival, and dissemination (12, 15, 18, 34). In contrast, we show that eutR expression is essential for proper bladder colonization by driving the transcription of the eut operon. This lends support to a model in which ethanolamine catabolism is important for bladder and urine colonization during UTI.

To our knowledge, this is the first work suggesting a link between ethanolamine metabolism and the innate immune response during infection. Although the actual mechanism remains unclear and requires further investigation, there are a number of intriguing possibilities. Since we saw no changes in the neutrophil activity or cytokine profiles during the infections with WT and Δeut bacteria, ethanolamine involvement in modulation of the global inflammatory responses appears to be unlikely. Furthermore, we saw no differences in the levels of eutR expression (and by extension, the ethanolamine levels) between immunocompetent and Gr-1-treated mice. Thus, it does not appear that the innate immune response mediates an increase in ethanolamine levels to serve as an additional nutrient source for the bacteria, although we cannot completely rule out that ethanolamine metabolism confers a growth advantage. An alternative hypothesis is that ethanolamine could serve as a local “danger” signal by increasing neutrophil chemotaxis and/or bactericidal potential. In this scenario, ethanolamine breakdown by UPEC could generate “pockets” in the bladder with diminished innate immune response and facilitate increased bacterial survival.

Environments of the gut and of the urinary tract are considered to be distinct from one another, suggesting that bacterial metabolic requirements should also vary between the two. However, our present study implicates ethanolamine metabolism as an important pathway for bacterial pathogens in both the intestine and the urinary tract, albeit potentially via different mechanisms. Although the expression of eut genes in intestinal pathogens may primarily confer a nutritional advantage, expression of the same pathway by bacteria in the urinary tract might also help counteract the innate immune response. This study highlights the underappreciated role of bacterial metabolic states in resistance to host immune responses.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

E. coli CFT073 was isolated from the blood and urine of a patient hospitalized with acute pyelonephritis (35, 36). Clinical strains of E. coli were isolated from the urine of women presenting with symptoms of lower UTI at the University of Michigan Health Service Clinic in Ann Arbor, MI, in 2012 (19). All human subject protocols were approved by the institutional review boards of the University of Michigan Medical School.

Strains to be tested in the mouse model of ascending UTI were cultured overnight in the absence of antibiotic selection, resuspended in phosphate-buffered saline, and adjusted to 2.0 × 10⁹ CFU/ml. All strains were assessed for the ability to grow in LB broth and in filter-sterilized human urine pooled from three to five healthy donors. Strains were cultured overnight in LB broth, and the LB broth or urine was inoculated with 2 × 10⁷ CFU/ml of bacterial suspension. The optical density at 600 nm was measured at specified time points.

Construction of Δeut and ΔeutR mutants.

Δeut mutant construction was performed as previously described (19). The ΔeutR mutant of E. coli CFT073 was constructed using the lambda red recombinase system (37). For the in vivo complementation experiments, the eutR gene was cloned into the pGEN-MCS vector (38) under the control of the native promoter. The primers used in this study can be found in Table S1 in the supplemental material. All mutants and their complements were tested for expression of the eutR and eut operon using reverse transcription-PCR (data not shown).

RNA sequencing and analysis of clinical UPEC isolates.

Sample collection, sequencing, and analysis have been previously described (19). Briefly, clean-catch midstream urine samples from participants were used for RNA isolation and sequencing. UPEC isolates HM26, HM27, HM46, HM65, and HM69 were cultured statically in sterile human urine at 37°C to midexponential phase, conditions that were previously established as equivalent to the growth phase and oxygenation levels encountered by UPEC within human (natural) and murine (experimental) hosts. Illumina reads were analyzed using an automated pipeline (39) at the Institute for Genome Sciences.

Murine model of ascending UTI.

An adaptation (40) of the mouse model of ascending UTI, originally developed by Hagberg et al. (41), was used to assess the virulence of E. coli CFT073 and the deletion constructs. Briefly, female C57BL/6 mice (6 to 8 weeks old) were inoculated transurethrally with 50 μl of a bacterial suspension, delivering 10⁸ CFU/mouse. At 24 h, 48 h, and 7 days postinoculation, urine samples and bladders were collected and plated on LB agar to determine colonization counts. All animal protocols were approved by the University Committee on Use and Care of Animals at the University of Michigan Medical School.

Neutrophil depletion.

Neutrophil depletion from mice was achieved by double intraperitoneal injections of 100 μl of sterile saline containing 250 μg of the Gr-1-specific monoclonal antibody (BioXCell, catalog no. BE0075) 24 and 6 h prior to infection.

MPO and cytokine measurements.

Bladder homogenates from mice infected for 24 h, 48 h, or 7 days were used to measure MPO and cytokine levels. Quantitative measurements were performed using enzyme-linked immunosorbent assay kits from BD Biosciences.

RNA extraction and measurement of in vivo eutR expression.

Bladders (n ≥ 5) were harvested 48 h postinoculation and homogenized in 1 ml of TRIzol. RNA was isolated according to standard molecular biology procedures. RNA was converted to cDNA by using a Bio-Rad iScript Select cDNA synthesis kit (catalog no. 1708896). qPCR was carried out using Agilent Brilliant III Ultra-Fast SYBR QPCR master mix (catalog no. 600882). The primers used for qPCR can be found in Table S1 in the supplemental material.

Oxidative stress and antimicrobial peptide assays.

Midlogarithmic E. coli was cultured in MOPS (3-morpholinopropanesulfonic acid) minimal medium with glucose in the presence of ethanolamine (5 mM ethanolamine and 150 nM vitamin B₁₂ [cyanocobalamin; Sigma]). Bacteria (2 × 10⁷ cells) were exposed to the indicated concentrations of hydrogen peroxide for 15 min (catalase was then added to a final concentration of 10 μg/ml) or LL-37 (InvivoGen) for 1 h. Survival was calculated by enumerating the CFU at that time point and normalized to untreated controls to yield the percent survival for each concentration.