Dynamic response by commensal E. coli two-component system Cpx contributes to C. elegans development

Abstract

Background

The Caenorhabditis elegans–Escherichia coli system is advantageous for studying host-microbe interactions at the single-gene level. By screening with this system, we identified that the deletion of cpxR, an E. coli transcription factor for the envelope stress response, delays C. elegans development. This finding led us to investigate how this gene regulates host development.

Results

We identified that E. coli ΔcpxR induced C. elegans developmental delay and activated the C. elegans mitochondrial unfolded protein response pathway through reactive oxygen species. It is widely accepted that the Cpx system is important for bacterial pathogenesis, and activating CpxR is regarded as an antimicrobial strategy. Moreover, we discovered that ΔcpxR cultured in LB medium, not cultured in M9 minimal medium, delayed C. elegans development, and the L-histidine-related metabolism of ΔcpxR contributed mostly to the difference. The metabolic fluctuations of commensal bacteria reveal that, rather than the activation of the E. coli Cpx response, the dynamic response of the E. coli Cpx system really contributes to C. elegans development. Furthermore, as the concentration of N-acetylcysteine increased, the phenotype of C. elegans fed ΔcpxR transitioned from developmental delay to survival resistance. The dynamic response is also indicated in the process in which commensal E. coli improves the stress tolerance of the host C. elegans to N-acetylcysteine.

Conclusions

Our results illustrate that environmental factors can shape the regulation of the E. coli Cpx response to C. elegans, providing new evidence for why Cpx-mediated virulence phenotypes are inconsistent among gram-negative species in different ecological niches.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12915-026-02513-x.

Background

Gut microbiota homeostasis critically influences host fitness and development. Microbiota-directed complementary foods effectively repair immature microbiota causally linked to undernutrition in children with moderate acute malnutrition. However, the underlying molecular mechanisms remain largely elusive [1–3]. Given the challenges in directly dissecting how human gut microbiota influence development, the Caenorhabditis elegans–Escherichia coli system offers a simplified model for deeper investigation. This system leverages several key advantages: the nematode has fundamental signaling pathways that are conserved in humans; it feeds on bacteria, and it can be monocolonized with a single E. coli strain that serves as both a food source and a gut microbe. Moreover, the E. coli Keio Knockout Collection facilitates targeted investigation of specific bacterial genes in host-microbe interactions [4, 5]. Recent studies using this interspecies model have confirmed that diverse microbiota-derived metabolites act as crucial mediators regulating host physiology (e.g., development, longevity, and metabolism) [6–10]. Indeed, these findings provide mechanistic insights into how human gut microbiota influence host physiology and diseases.

The Cpx two-component system (TCS), comprising the sensor histidine kinase CpxA (an inner membrane protein) and its cognate response regulator CpxR (a cytoplasmic protein), senses diverse extracellular and intracellular stimuli to regulate the transcription of stress response genes to maintain cell envelope integrity [11]. The Cpx system is highly conserved among Enterobacteriaceae [12]. Specifically, CpxA can phosphorylate or dephosphorylate CpxR. Upon envelope stress, CpxA auto-phosphorylates at a conserved histidine residue and transfers its phosphate group to a conserved aspartate residue in CpxR, thereby activating the system; conversely, in the absence of stress or upon its relief, CpxA functions as a net phosphatase to keep CpxR in a dephosphorylated and inactive state. Notably, CpxR phosphorylation occurs not only through CpxA but also via the Pta-AckA pathway [13–15] (Fig. 1A). Consequently, the system is inactive in the ΔcpxR mutant but remains partially active in the ΔcpxA mutant. It is well-established that once activated, CpxR contributes to envelope stress remediation by downregulating the transcription of secreted and envelope-localized proteins, including key virulence determinants. Therefore, the Cpx system is critical for bacterial pathogenesis and colonization, making it an attractive target for novel antimicrobial therapies [16–21].

Fig. 1.

ΔcpxR induces developmental delay in C. elegans. A Regulatory model of the E. coli Cpx two-component system. B Growth states of C. elegans N2 and DHS-3::GFP(mut) fed E. coli WT or ΔcpxR on NGM plates 65-h post-L1 stage. Scale bar, 1 mm. C Worm length of C. elegans N2 and DHS-3::GFP(mut) fed E. coli WT, ΔcpxR, or ΔcpxA 72-h post-L1 stage. D Worm length of C. elegans ldrIS1 and DHS-3::GFP(mut) fed E. coli WT, ΔcpxR, or ΔcpxA 60-h post-L1 stage. E Stitched microscopy images showing morphology of C. elegans N2 and DHS-3::GFP(mut) fed E. coli WT or ΔcpxR 45-h post-L1 stage. Scale bar, 100 µm. F ΔcpxR-induced irreversible developmental damage. Proportion of C. elegans DHS-3::GFP(mut) reaching adulthood following a 48-h exposure to ΔcpxR from the L1 stage and subsequent 48-h refeeding on either E. coli WT or ΔcpxR. Data from three independent experiments, error bar represented mean + SD, and n = 10. G 24-h timed egg lay. The number of eggs laid by C. elegans DHS-3::GFP(mut) within the first 24-h post-L4 stage when fed E. coli WT, ΔcpxR, or ΔcpxA. Data from three independent experiments; n ≥ 19. H Worm length of C. elegans DHS-3::GFP(mut) fed living and UV-treated E. coli WT, ΔcpxR, or ΔcpxA 50-h post-L1 stage. C, D, H Per biological replicate: ≥ 30 worms per group; cumulative n ≥ 90 across three independent experiments. Boxplot, the middle line, the upper and lower boundary, and the whisker of box represented the median, 75th percentiles and 25th percentiles, and 1.5 × interquartile range, respectively. Dots, individual measurements. Significances were displayed as *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns — not significant

It is widely accepted that gut microbes provide the host with otherwise unobtainable dietary nutrients and equip it with the flexibility to cope with changing environments [22]. In our study, a large-scale E. coli genetic mutant screen revealed several genes whose deletion caused C. elegans to arrest development at an early stage. Among these, we were more interested in cpxR because it was involved in an active bacterial adaptation strategy, unlike passive bacterial byproducts. We wondered whether microbial stress response could modulate host stress signaling to regulate development. Indeed, we identified that ΔcpxR induced a C. elegans developmental defect through excessive reactive oxygen species (ROS), which led to the simultaneous activation of the C. elegans mitochondrial unfolded protein response (UPR^mt). Strikingly, the metabolic shift of ΔcpxR led to contrasting developmental outcomes in C. elegans, demonstrating that the Cpx response was dynamic in host-microbe interactions. Furthermore, the dynamic transition was also evident in the strategy by which C. elegans exploited E. coli to adapt to environmental stress from N-acetylcysteine (NAC). Our study expands the understanding of how the Cpx two-component system in commensal bacteria functions in host-microbe interactions.

Results

E. coli ΔcpxR delays C. elegans development

To systematically investigate the impact of gut bacteria on the host development and metabolism at the gene level, we conducted an unbiased screen. Synchronized L1-staged C. elegans were fed nearly 1000 bacterial single-gene knockout strains from the E. coli Keio Collection library [4]. We used the ldrIS1 strain carrying a lipid droplet marker (Pdhs-3p::dhs-3::GFP) to visualize C. elegans fat storage [23]. To amplify subtle changes in fat content and mitigate potential developmental confounders from impaired growth of certain E. coli mutants, 100-mM supplemental glucose was added to NGM plates. Oil-Red-O staining was used to preliminarily assess developmental stage and fat storage before microscopic examination. In the primary screen, we identified 11 E. coli deletion mutants that induced developmental delay and reduced fat storage. They were ΔyehB, Δdld, ΔcpxR, ΔfimB, ΔpepQ, ΔyfcP, ΔyccE, ΔyccJ, ΔycgK, ΔtesA, and ΔumuD (Additional file 1: Fig. S1). Among these, ΔyfcP, a mutant of a putative fimbrial protein, was also reported in a previous screen [9].

We noticed that ΔcpxR caused a severe developmental delay or arrest in worms. After feeding about 65 h since the L1 stage, most worms fed ΔcpxR only reached the L2 stage, while those fed E. coli WT reached the L4 stage (Fig. 1B). However, we later observed that wild-type C. elegans N2 fed ΔcpxR exhibited only mild developmental delay, which was significantly less severe than that in the screening strain (Fig. 1B, C). This phenotypic divergence was not attributable to the ldrIS1 locus (Pdhs-3p::dhs-3::GFP) locus, as the original ldrIS1 strain fed ΔcpxR developed similarly to N2 (Fig. 1D). We therefore hypothesized that a spontaneous enhancer mutation arose in the original ldrIS1 strain during the screen. This mutation specifically exacerbated developmental delay in worms fed ΔcpxR but not E. coli WT. We designated this strain as DHS-3::GFP(mut) hereafter. We confirmed the developmental delay in both DHS-3::GFP(mut) and N2 worms fed ΔcpxR using higher-resolution microscopy (Fig. 1E). By an intestinal bacterial colonization assay, we compared ΔcpxR colonization in DHS-3::GFP(mut) versus N2. In wild-type worm N2, E. coli WT colonized significantly better than ΔcpxR. Conversely, they showed no colonization difference in DHS-3::GFP(mut). Thus, relative to N2, ΔcpxR colonization was enhanced in DHS-3::GFP(mut) (Additional file 1: Fig. S2). It demonstrated that wild-type worm N2 employed specific mechanisms to suppress ΔcpxR over-colonization — a regulatory capacity defective in DHS-3::GFP(mut). This deficiency likely related to its mutation site, as well as linked to its severe developmental delay.

We leveraged the sensitivity of the DHS-3::GFP(mut) strain to investigate additional phenotypes associated with ΔcpxR. First, the developmental delay in C. elegans induced by a 48-h ΔcpxR exposure from the L1 stage was proven to be irreversible, even after these worms were subsequently transferred to E. coli WT (Fig. 1F). Second, synchronized L4-staged C. elegans experienced a significant delay in egg laying after being transferred from E. coli WT to ΔcpxR (Fig. 1G). These data indicated that the suppression occurred across all C. elegans developmental phases. Third, UV-treated ΔcpxR failed to cause developmental delay, demonstrating that ΔcpxR must be alive to affect the host (Fig. 1H). Fourth, because CpxA partners with CpxR, and its deletion would produce a Cpx activity opposite to ΔcpxR due to loss of its phosphatase activity and the presence of alternative phosphodonors for CpxR [13] (Fig. 1A), we reasoned that the phenotype of worms fed ΔcpxA would be informative. As expected, we found that ΔcpxA did not induce developmental delay in worms (Fig. 1C) and even exerted a slight growth-promoting effect (Fig. 1D). It implied that the activation of CpxR would not induce C. elegans developmental delay. Overall, the phenotype of worms fed ΔcpxA was similar to that of worms fed E. coli WT. The subtle phenotypic divergence between them, beyond differences in observational timing, likely resulted from varied CpxR activity between the two bacterial strains [13, 24]. Collectively, these data preliminarily reveal that activation of the E. coli Cpx response is required for C. elegans development.

E. coli ΔcpxR induces C. elegans mitochondrial stress response via excess ROS

To obtain a comprehensive assessment, we compared RNA-seq data from DHS-3::GFP(mut) worms fed ΔcpxR, ΔcpxA, or E. coli WT. Two significantly upregulated gene clusters in ΔcpxR-fed worms pointed to mitochondrial involvement. The first was enriched in xenobiotic detoxification genes, including cytochrome P450 genes, UDP-glycosyltransferase genes (UDPGT), ABC transporters, and the nuclear hormone receptor nhr-8, consistent with a mitochondrial stress response [25, 26] (Fig. 2A). The second encompassed a mitochondrial gene cluster near NADH-ubiquinone oxidoreductase subunit 4 (nduo-4), suggesting enhanced respiratory capacity and elevated ROS production [27] (Fig. 2B). Then, we used the C. elegans hsp-6p::gfp reporter strain to check whether the mitochondrial unfolded protein response (UPR^mt) was activated when worms were fed ΔcpxR and found that ΔcpxR strongly activated the hsp-6 promoter, while ΔcpxA and E. coli WT did not (Fig. 2C, D). This activation was similarly identified in an independent screen [28]. Meanwhile, the reporter strain showed a developmental delay when fed ΔcpxR and a moderate developmental acceleration when fed ΔcpxA (Fig. 2C, E). Next, we tested whether elevated ROS levels contributed to the developmental delay by supplementing worms with N-acetylcysteine (NAC), an antioxidant that reduces ROS levels [9, 29], and found that 10 mM NAC could suppress the ΔcpxR-induced developmental defect (Fig. 2F). We wondered whether excess ROS accumulation in ΔcpxR-fed worms resulted from absent protective mechanisms that decrease ROS or from failure to inhibit virulence factors inducing persistent ROS production. We tested this by feeding DHS-3::GFP(mut) worms mixtures of E. coli WT and ΔcpxR at varying ratios. Remarkably, mixtures containing as little as 10% E. coli WT substantially inhibited developmental delay, with maximal effect at ~ 20% E. coli WT (Fig. 2G). The beneficial effect of E. coli WT dominates over the detrimental impact of ΔcpxR, indicating that excess ROS results from deficient CpxR-regulated ROS mitigation mechanisms, rather than dysregulated virulence factor expression.

Fig. 2.

ΔcpxR induces a mitochondrial stress response in C. elegans through excess ROS. A Differential expression patterns of detoxification-related genes in C. elegans DHS-3::GFP(mut) fed on E. coli WT, ΔcpxR, or ΔcpxA. Columns represent individual biological replicates grouped by treatment; rows represent genes. Clustered samples indicate similar biological states. B Differential expression patterns of the mitochondrial gene cluster near nduo-4 in C. elegans DHS-3::GFP(mut) fed on E. coli WT, ΔcpxR, or ΔcpxA. Columns represent individual biological replicates grouped by treatment; rows represent genes. C Representative images of C. elegans hsp-6p::gfp reporter strain fed E. coli WT, ΔcpxR, or ΔcpxA 50-h post-L1 stage. Scale bar, 200 µm. D Quantification of C. elegans hsp-6p::gfp reporter strain fluorescence intensity from C. E Quantification of C. elegans hsp-6p::gfp reporter strain body length from C. F Worm length of C. elegans DHS-3::GFP(mut) fed living and UV-treated E. coli WT, ΔcpxR, or ΔcpxA with 10-mM NAC supplementation 50-h post-L1 stage. G Mixed bacteria feeding assay. Worm length of C. elegans DHS-3::GFP(mut) fed the mixtures of E. coli WT and ΔcpxR as indicated ratios 72-h post-L1 stage. D, E, F, G Per biological replicate: ≥ 30 worms per group; cumulative n ≥ 90 across three independent experiments. Boxplot, the middle line, the upper and lower boundary, and the whisker of box represent the median, 75th percentiles and 25th percentiles, and 1.5 × interquartile range respectively. Dots, individual measurements. Significances were displayed as *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns — not significant

The Cpx response is involved in the adaptive strategy that C. elegans exploits E. coli to cope with the reductive stress from N-acetylcysteine

From the NAC concentration-gradient supplementation assay, we identified three critical phenomena. First, the developmental state of E. coli WT-fed worms resembled ΔcpxA-fed worms at low NAC (≤ 10 mM) but resembled ΔcpxR-fed worms at high NAC (≥ 15 mM) (Fig. 3A). Second, at low NAC, ΔcpxR-fed worms arrested at an early developmental stage, while at high NAC, ΔcpxA-fed worms arrested at an early developmental stage (Fig. 3A). While E. coli WT-fed worms consistently exhibited optimal growth, ΔcpxR-fed worms and ΔcpxA-fed worms exhibited converse responses to rising NAC levels: the growth of ΔcpxR-fed worms improved, while that of ΔcpxA-fed worms worsened. Third, compared to live E. coli WT, at any NAC concentration, UV-treated E. coli WT failed to support worms’ normal development (Fig. 2F, and 3B, C). Together, these data indicate that live E. coli enables NAC resistance in C. elegans through Cpx-mediated sensing.

Fig. 3.

C. elegans adaptation to N-acetylcysteine requires functional E. coli Cpx response. A Worm length of C. elegans DHS-3::GFP(mut) fed indicated E. coli strains 50-h post-L1 stage with graded NAC concentrations (0–20 mM). B Worm length of C. elegans DHS-3::GFP(mut) fed living and UV-treated E. coli WT, ΔcpxR, or ΔcpxA with 15-mM NAC supplementation 50-h post-L1 stage. C Worm length of C. elegans DHS-3::GFP(mut) fed living and UV-treated E. coli WT, ΔcpxR, or ΔcpxA with 20-mM NAC supplementation 50-h post-L1 stage. D Representative images of C. elegans hsp-6p::gfp reporter strain fed indicated E. coli strains 50-h post-L1 stage with graded NAC concentrations (0–20 mM). Scale bar, 200 µm. E Quantification of C. elegans hsp-6p::gfp reporter strain body length from D. F Quantification of C. elegans hsp-6p::gfp reporter strain fluorescence intensity from D. A, B, C, E and F Per biological replicate: ≥ 30 worms per group; cumulative n ≥ 90 across three independent experiments. Boxplot, the middle line, the upper and lower boundaries, and the whiskers of the box represent the median, the 75th and 25th percentiles, and 1.5 × the interquartile range, respectively. Dots, individual measurements. Significances were displayed as *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns — not significant

Additionally, in this assay, we noticed that although worms fed on UV-treated E. coli could not develop normally, there was a consensus that UV-treated ΔcpxA-fed worms consistently showed the most severe developmental defect, while UV-treated ΔcpxR-fed worms could maintain viability comparable to UV-treated E. coli WT-fed worms (Fig. 2F, Fig. 3B, C). It suggested that UV-treated ΔcpxA did not induce more ROS, which may not confer the resistance to NAC as UV-treated ΔcpxR. Simultaneously, we monitored mitochondrial stress response under these conditions using the hsp-6p::gfp reporter strain. The hsp-6p::gfp reporter strain reproduced the developmental phenotypes observed in the DHS-3::GFP(mut) strain (Fig. 3D, E). ΔcpxR-induced UPR^mt activation and developmental defect could be suppressed by 10 mM NAC (Fig. 3D, E, F). However, at high NAC, the developmental defect in C. elegans induced by ΔcpxA was not associated with UPR^mt activation, and ΔcpxA-fed worms exhibited reporter fluorescence intensity even lower than that of E. coli WT-fed worms, suggesting an impairment of ROS-mediated beneficial functions (Fig. 3D, E, F). Consequently, the early developmental arrests observed in ΔcpxR-fed worms without NAC and in ΔcpxA-fed worms at high NAC were likely due to too high ROS and too low ROS, respectively. These data imply that ROS play an underlying role in the adaptive strategy.

The metabolic fluctuations of E. coli ΔcpxR influence its effect on the development of C. elegans

Previous work demonstrated that E. coli undergoes growth-related Cpx activation when cultured to stationary phase in LB medium [13], while another study found that cpxP (a CpxR-regulated gene) was significantly less expressed in M9 minimal medium than in LB medium [30]. These observations suggest that shifts in bacterial metabolism alter the demand for the Cpx response, and that envelope stress is minimal in M9-cultured bacteria. We then reasoned that the deletion of cpxR would have minimal impact on E. coli cultured in M9 minimal medium, since CpxR activity was inherently low under this condition. We wondered whether altering bacterial CpxR demand via M9 cultivation would modify host dependence on CpxR activation. Moreover, it was reported that C. elegans raised on M9-cultured E. coli accumulated more fat than those fed LB-cultured bacteria, which not only evidenced that bacterial metabolic state could modulate host physiology but also supported the feasibility of this approach [7]. As expected, we found that M9-cultured ΔcpxR did not induce worm developmental delay (Fig. 4A). It means that CpxR activation is conditionally dispensable. Given our previous finding, we accordingly assessed mitochondrial stress response under M9 conditions using the hsp-6p::gfp reporter strain. Consistent with DHS-3::GFP(mut) strain, M9-cultured ΔcpxR neither induced developmental delay, nor activated UPR^mt in this strain (Fig. 4B, C, D). These data demonstrate that the developmental defect in worms is due to excess ROS caused by Cpx dysregulation, especially when proper Cpx function is required, and lies in this dysregulation, not in the absence of constitutive CpxR activation. Bacteria cultured in LB and M9 media represent distinct states of envelope homeostasis and consequently have different demands for CpxR activation. The host’s differential dependence on CpxR activation between these two conditions therefore demonstrates that bacterial envelope homeostasis is critical for host development. Moreover, we still observed that compared to UV-treated LB-cultured E. coli, UV-treated M9-cultured E. coli failed to support normal C. elegans development (Fig. 4A). It reminded that the influence of bacterial metabolic differences should be considered. Given fundamental differences in amino acid biosynthesis between minimal and rich media [31–33], we supplemented M9-cultured ΔcpxR with individual amino acids to assess their effects on host development (Additional file 1: Fig. S3). Finally, we found that supplementation with 50-mM L-histidine moderately delayed development in the DHS-3::GFP(mut) strain (Fig. 4E). Furthermore, this supplementation also induced moderate developmental delay and UPR^mt activation in the hsp-6p::gfp reporter strain (Fig. 4F, G, H). These findings demonstrate that histidine-involved metabolic reprogramming reverses the effect of M9-cultured ΔcpxR on host development. Collectively, the Cpx system orchestrates context-dependent responses to metabolic fluctuations, fine-tuning its activity as needed to maintain bacterial envelope homeostasis and thereby contribute to host development.

Fig. 4.

The effect of ΔcpxR on C. elegans development varies with culture medium. A Worm length of C. elegans DHS-3::GFP(mut) fed living vs UV-treated E. coli (E. coli WT, ΔcpxR, ΔcpxA) cultured in LB or M9 medium 50-h post-L1 stage. B Representative images of C. elegans hsp-6p::gfp reporter strain fed indicated E. coli strains cultured in LB or M9 medium 50-h post-L1 stage. Scale bar, 200 µm. C Quantification of C. elegans hsp-6p::gfp reporter strain fluorescence intensity from B. D Quantification of C. elegans hsp-6p::gfp reporter strain body length from B. E Worm length of C. elegans DHS-3::GFP(mut) fed M9-cultured E. coli with 50-mM L-histidine 50-h post-L1 stage. F Representative images of C. elegans hsp-6p::gfp reporter strain fed M9-cultured E. coli with 50-mM L-histidine 50-h post-L1 stage. Scale bar, 200 µm. G Quantification of C. elegans hsp-6p::gfp reporter strain fluorescence intensity from F. H Quantification of C. elegans hsp-6p::gfp reporter strain body length from F. A, C and D, E, G and H Per biological replicate: ≥ 30 worms per group; cumulative n ≥ 90 across three independent experiments. Boxplot, the middle line, the upper and lower boundary, and the whisker of box represented the median, 75th percentiles and 25th percentiles, and 1.5 × interquartile range, respectively. Dots, individual measurements. Significances were displayed as *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns — not significant

Discussion

Our findings identify that the ΔcpxR mutant induces mitochondrial stress and developmental delay in C. elegans only when it compromises E. coli envelope homeostasis. Moreover, 244 E. coli mutants from a prior screen exhibited similar phenotypes [9], with ΔcyoA, ΔsdhC, ΔycbB, and ΔfepG independently validated [10, 28, 34]. Interestingly, these confirmed genes are involved in Cpx-regulated processes: the electron transport chain (CyoA), the peptidoglycan structure (YcbB), and the efflux of the siderophore enterobactin (FepG) [29, 30, 35, 36]. We therefore reasoned that compromised envelope stability might be the unifying feature across these 244 mutants. To test this, we analyzed 2 key characteristics of these 244 mutant genes: Cpx regulation and envelope localization. Strikingly, based on data from [15], ~ 71% (174/244) of genes are Cpx-regulated targets, while based on data from [37], ~ 49% (120/244) of genes encode envelope-localized proteins. This significant overlap strongly supports our hypothesis that bacterial envelope homeostasis is essential for host development.

The Cpx response is widely recognized as a regulator of virulence gene expression in gram-negative pathogens and represents a promising antimicrobial target [38–40]. Nevertheless, significant variations in Cpx-related virulence exist. For example, in Haemophilus ducreyi, the cpxA mutant is avirulent in humans, whereas the cpxR mutant is fully virulent [41]. In contrast, Citrobacter rodentium ΔcpxRA displays colonization defects during infection [17]. Further complexity is observed in enterohemorrhagic Escherichia coli, where adhesion can significantly alter Cpx-regulated virulence according to conflicting reports [39, 42, 43]. From this perspective, E. coli is a mild pathogen in C. elegans [44], and the developmental delay caused by the LB-cultured ΔcpxR parallels a virulence enhancement in the C. elegans–E. coli interaction. Notably, environmental factors, such as NAC supplementation and M9 cultivation, can easily influence the virulence of ΔcpxR. Thus, Cpx-mediated virulence is highly context dependent. After all, under natural conditions, bacteria occupy different ecological niches with diverse metabolic profiles. Our work provides a possible explanation for the inconsistent virulence phenotypes observed across pathogens. Furthermore, it underscores that any therapeutic application targeting the Cpx system requires careful contextual evaluation.

Although the redox environment constructed by NAC supplementation is artificial, it mimics redox potential fluctuations in humid tropical soils [45]. We propose a new environment–gut microbe–host axis in which C. elegans exploits E. coli to adapt to environmental redox stress, and we confirm that the Cpx system in E. coli functions in this axis. Through graded NAC exposure, we analyzed host adaptation to infer the Cpx response in E. coli, thereby revealing a phenomenon that was not detected in studies of the bacteria alone. Using E. coli WT-fed worms, which always maintain optimal growth, as a reference, ΔcpxR-fed worms transition from defective to normal growth, whereas ΔcpxA-fed worms exhibit the opposite phenotype. This inversion demonstrates that constitutive Cpx activation is not required, but context-dependent dynamic regulation is necessary. However, further details are still needed to refine the model of how microbial adaptation enhances host resistance. For instance, although our data provide some clues that CpxR activity declines as NAC concentration increases and that the primary function of CpxA is to inactivate CpxR under high NAC conditions, how the E. coli Cpx system responds to NAC remains poorly characterized [46, 47].

In our opinion, ROS play an important role in Cpx-mediated host development. It has been reported that ROS cause bacterial protein carbonylation that can be transferred to the host intestine, subsequently directly activating host mitochondrial stress response [34]. Given that in the NAC supplementation assay UV-treated LB-cultured ΔcpxR makes worms more resistant than UV-treated LB-cultured ΔcpxA, it suggests that LB-cultured ΔcpxR correspondingly has higher levels of carbonylated protein than LB-cultured ΔcpxA. Nevertheless, such carbonylation levels are not enough to threaten worm development, because all the UV-treated LB-cultured E. coli support normal worm development in the absence of NAC. Critically, by comparing phenotypes of live and UV-treated LB-cultured ΔcpxR, we further reason that a ROS burst occurs during host-microbe interaction, whose magnitude correlates with the intrinsic ROS levels of the bacteria. This burst likely results either from E. coli producing ROS in the adverse gut environment or from C. elegans producing ROS as an immune response to the bacteria. Moreover, it has also been reported that decreasing ROS levels in ΔcyoA through another mutant suppresses the ΔcyoA-induced C. elegans developmental delay [34]. We thus hypothesize that LB-cultured ΔcpxR, possessing elevated ROS levels, readily initiates more ROS generation through a “domino effect,” which ultimately drives host developmental defects.

Conclusions

To maintain a symbiotic relationship with their host, microbes need to provide precise regulatory signals and beneficial metabolites. Using the C. elegans–E. coli model system, we investigated bacterial traits critical for host-microbe adaptation, with a specific focus on the Cpx two-component system, a bacterial envelope stress response (Fig. 5). While Cpx activation is recognized as an antimicrobial strategy in pathogenesis, our study demonstrates that dynamic modulation of Cpx activity, rather than its constitutive activation, drives context-dependent bacterial adaptations that promote host development and enhance host stress tolerance. These findings reveal a previously unrecognized role of the Cpx system in host-microbe interactions. Although Cpx activity is dynamic, its ultimate function is to maintain envelope homeostasis. Furthermore, our results support the crucial role of bacterial envelope stability in symbiosis.

Fig. 5.

Summary of the effects of E. coli Cpx mutants on C. elegans development under various conditions. When C. elegans and E. coli are subjected together to graded NAC exposure, C. elegans fed E. coli WT are resistant to NAC, while those fed UV-treated E. coli WT are not. This confirms that a subpopulation of live E. coli functions as a gut microbe that provides the C. elegans host with enhanced tolerance. C. elegans fed E. coli WT could survive under both conditions, but those fed ΔcpxR tend to survive with NAC, while those fed ΔcpxA are more adaptable without NAC. This indicates that the dynamic bacterial Cpx response contributes to host adaptation. When C. elegans are fed E. coli cultured in different growth media, the development of those fed E. coli WT is unaffected by the media, while the development of those fed ΔcpxR is influenced. C. elegans development is delayed when fed LB-cultured ΔcpxR. In contrast, C. elegans development is normal when fed M9-cultured ΔcpxR. Nevertheless, normal development is disrupted by supplementation with L-histidine. This reflects that the requirement for bacterial Cpx response activation for C. elegans development is context dependent. Moreover, LB-cultured ΔcpxR and M9-cultured ΔcpxR supplemented with L-histidine both induce developmental delay as well as UPR^mt activation in C. elegans. This suggests that ROS are involved in the Cpx-mediated regulation of host development. The primary responsibility of the bacterial Cpx response, although Cpx activity is variable, is to maintain bacterial envelope homeostasis, which also benefits their host. Finally, we show how the bacterial Cpx response works in the environment–gut microbe–host axis

Methods

C. elegans strain

The C. elegans strains used in this study included wild-type Bristol N2, SJ4100 (zcIs13[hsp-6p::gfp]), ldrIS1 (Pdhs-3p::dhs-3::GFP), and DHS-3::GFP(mut). Worms were maintained on nematode growth medium (NGM) plates seeded with E. coli OP50 at 20 °C [48]. Worms were non-starved for at least two generations prior to experiments.

Bacteria strain

E. coli OP50 and BW25113 were maintained on LB agar at 37 °C. E. coli knockout mutants were maintained on LB agar supplemented with 50 µg/mL kanamycin. The E. coli WT strain used in the experiments was E. coli BW25113, the parental strain of E. coli knockout mutants. Each knockout mutant colony was checked by PCR as described [49]. ΔcpxR was checked by primers (F: GCAAACATGCGTCAGGGGGT R: GCTTGGGTAACATCAAAACC); ΔcpxA was checked by primers (F: CAACCTGCGTCGTAAACTGC R: AGTGTAGGCCTGATAAGACG). Kanamycin gene was checked by primers (F: CAGTCATAGCCGAATAGCCT R: CGGTGCCCTGAATGAACTGC). PCR products were size-verified by electrophoresis and confirmed by sequencing. To ensure consistency with the E. coli WT strain, ΔcpxR and ΔcpxA mutants were cultured without kanamycin during experiments.

Culture media, plates, and supplementations

LB-cultured bacteria were prepared by seeding overnight LB cultures onto NGM plates. M9-cultured bacteria were prepared by concentrating overnight M9 cultures approximately 10 × via centrifugation (3000 rpm, 10 min) and then seeding onto peptone-free customized M9 plates. These plates were formulated as standard NGM without peptone to avoid interference. The M9 minimal medium composition was: 0.4% glucose, 12.8 g/L Na₂HPO₄ ·7H₂O, 3 g/L KH₂PO₄, 0.5 g/L NaCl, 1 g/L NH₄Cl, 2 mM MgSO₄, and 0.1 mM CaCl₂.

Stock solutions were prepared as follows: A 1-M N-acetylcysteine (Macklin, Shanghai) solution in ddH₂O, filter-sterilized (0.22 µm) and stored at − 20 °C, and a 4-M glucose (Macklin, Shanghai) solution in ddH₂O, filter-sterilized (0.22 µm) and stored at room temperature. These stock solutions were diluted in NGM agar to the final concentration before plate pouring.

The following amino acids (Sigma-Aldrich, Shanghai) were individually dissolved in ddH₂O to prepare 100-mM stock solutions, filter-sterilized (0.22 µm), and stored at − 20 °C: L-isoleucine, L-leucine, L-valine, L-lysine monohydrochloride, L-histidine, L-arginine, L-tryptophan, L-methionine, L-threonine, L-phenylalanine, L-tyrosine, L-alanine, L-aspartic acid, L-cysteine hydrochloride monohydrate, L-glutamine, L-glutamic acid monosodium salt monohydrate, L-glycine, L-proline, and L-serine. Amino acid stock solutions were diluted to a final concentration in M9-cultured E. coli suspensions before seeding on customized M9 plates.

All experiment plates with bacteria were prepared fresh on worm transfer days. Plates were dried before worm transfer. For UV-treated bacterial lawns, seeded plates without lids were exposed twice in a UV crosslinker (9999 × 100 µJ/cm² per exposure).

C. elegans developmental phenotype assay

In the screening experiment, the E. coli mutants were cultured overnight at 37 °C in LB medium containing 25 µg/mL kanamycin in 96-well plates. Subsequently, 60 µl of each mutant culture was seeded into eight 12-well plates containing NGM supplemented with 100-mM glucose. Synchronized L1 larvae were obtained by bleaching adult hermaphrodites and incubating the resulting eggs in M9 buffer for approximately 22 h. About 60 synchronized L1 larvae were added to each well and cultured for 48 h at 20 °C. Worms were then washed with M9 buffer and transferred to fresh 96-well plates for Oil-Red-O staining, adapted from a published protocol to stain around 100 worms per condition [50].

Bacterial colonization assay

To assess bacterial colonization, synchronized L4 worms were transferred to NGM plates seeded with E. coli WT, ΔcpxA, or ΔcpxR. After 24 h, 20 young adults were picked and incubated in a 1/50-diluted bleach solution at room temperature for 10 min to remove surface bacteria. Animals were washed four times with cold M9 buffer, with the final wash saved for contamination control. Worms were lysed using glass beads in a bead beater (3 min at maximum speed). Lysates were diluted in M9 buffer and plated on LB agar. Bacterial colonies were counted after overnight incubation at 37 °C to calculate CFU/worm. The absence of external bacterial contamination was confirmed by plating the final wash solution on LB agar [51].

Mixed bacteria feeding assay

Overnight bacterial cultures were adjusted to an OD600 = 0.3 in LB medium. E. coli WT and ΔcpxR suspensions at identical OD were mixed at volume ratios of 1:5, 1:10, and 1:100. Mixed cultures were seeded equally onto NGM plates. Plates were dried in a biosafety cabinet before worm transfer. About 100 synchronized L1 larvae were placed on each bacterial lawn and cultured at 20 °C for about 72 h, after which worms were collected for body length measurement.

A 24-h timed egg lay assay

Two synchronized L4 worms were transferred to individual NGM plates seeded with E. coli WT, ΔcpxA, or ΔcpxR and were fed for 24 h at 20 °C. Parental worms were then removed, and progeny were counted at the L3/L4 stage. Ten replicate plates per condition were analyzed in each experiment, with three biological replicates performed.

ΔcpxR-induced irreversible damage assay

Synchronized L1 larvae were cultured for 48 h at 20 °C on NGM plates seeded with ΔcpxR. Ten worms were then transferred to individual NGM plates seeded with E. coli WT or ΔcpxR for an additional 48 h of culture. The proportion of adult worms per plate was calculated. Five replicate plates per condition were analyzed per experiment, with three biological replicates performed.

Microscopy

Worms were immobilized by placing them in M9 buffer containing 5-mM levamisole hydrochloride on 2% agarose pads. The samples were then covered with a glass coverslip. Bright-field and fluorescence images were acquired using either an OLYMPUS BX53 microscope equipped with a PHOTOMETRICS Prime camera or a ZEISS Axio Imager.Z2 microscope equipped with an Axiocam 506 mono camera. Worm length and fluorescence intensity were quantified from raw images using Fiji software [52]. Worm length was measured by drawing segmented lines spanning the head to tail of the worms. Fluorescence intensity was quantified over the entire worm body area. At least 30 worms were measured per condition in each experiment.

RNA isolation and transcriptomic analysis

Synchronized L1 larvae were fed for 12 h on NGM plates seeded with LB-cultured ΔcpxA, ΔcpxR, or E. coli WT. Worms were then collected, washed three times with M9 buffer, flash-frozen in liquid nitrogen, and stored at − 80 °C. Total RNA was extracted from three biological replicates per condition using TRIzol reagent (Invitrogen) according to the manufacturer’s protocol. Transcriptome sequencing was performed by Novogene Co., Ltd., using the Illumina NovaSeq PE150 platform (paired-end, 150-bp read length).

The resulting clean reads were aligned to the Caenorhabditis elegans reference genome (Wbcel235) using HISAT2 (version 2.0.5) [53]. Read counts per gene were generated with featureCounts (version 1.5.0-p3) [54]. Differential expression analysis was performed using the R package DESeq2 (version 1.16.1) [55]. Genes with an adjusted p-value < 0.05 were considered differentially expressed. Heatmap analysis was performed based on FPKM values of differentially expressed genes. The data underwent log₂ transformation followed by Z-score normalization per gene. Visualization was implemented using the R package pheatmap (version 1.0.13).

Statistical analyses

All statistical analyses were performed with R (version 4.0.5) implemented in RStudio (version 1.3.1093). Visualization of data was carried out using the R package ggplot2 (version 3.3.5). Data represent at least three independent biological replicates. For comparisons between two groups, significance was assessed using two-tailed Student’s t-tests. For multigroup comparisons, one-way ANOVA followed by post hoc pairwise t-tests with Bonferroni correction were applied. Results with p-values or adjusted p-values < 0.05 were considered statistically significant.

Abbreviations

TCS

Two-component system

NGM

Nematode growth medium

NAC

N-Acetylcysteine

UPR^mt

Mitochondrial unfolded protein response

UV

Ultraviolet

ROS

Reactive oxygen species

LB

Luria broth

WT

Wild type

CFU

Colony-forming unit

FPKM

Expected number of Fragments Per Kilobase of transcript sequence per Millions base pairs sequenced

NHR

Nuclear hormone receptor

UDPGT

UDP glycosyltransferase

Authors’ contributions

H.X. and H.Z. conceived the study. H.X. and H.Z. designed the experiments. H.X. and B.H. performed the experiments. H.X. analyzed the data. H.X. wrote the manuscript. H.X. and H.Z. critically revised the manuscript. H.Z. and B.H. got the funding. All authors read and approved the final manuscript.

Funding

This work was supported by the National Key R&D Program of China (2021YFA0804701) (H. Z.), the National Natural Science Foundation of China (32170837 and 32500604) (H. Z. and B. H.), the China Association for Science and Technology (CAST) Young Scientists Cultivation Program — PhD Student Special Project 2025 (B. H.), the Ministry of Science and Technology of China (2019YFA0802804) (H. Z.), the Recruitment Program of Global Experts of China (Youth) (H. Z.), and the ShanghaiTech Startup program (H. Z.).

Data availability

The data obtained in the study are included in the article. Further inquiries can be directed to the corresponding author. Next generation sequencing data are available from the Sequence Read Archive database under the accession number PRJNA1138473 [56].

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.