The conserved regulator of autophagy and innate immunity hlh-30/TFEB mediates tolerance of enterohemorrhagic Escherichia coli in Caenorhabditis elegans

Abstract

Infection with antibiotic-resistant bacteria is an emerging life-threatening issue worldwide. Enterohemorrhagic Escherichia coli O157: H7 (EHEC) causes hemorrhagic colitis and hemolytic uremic syndrome via contaminated food. Treatment of EHEC infection with antibiotics is contraindicated because of the risk of worsening the syndrome through the secreted toxins. Identifying the host factors involved in bacterial infection provides information about how to combat this pathogen. In our previous study, we showed that EHEC colonizes in the intestine of Caenorhabditis elegans. However, the host factors involved in EHEC colonization remain elusive. Thus, in this study, we aimed to identify the host factors involved in EHEC colonization. We conducted forward genetic screens to isolate mutants that enhanced EHEC colonization and named this phenotype enhanced intestinal colonization (Inc). Intriguingly, four mutants with the Inc phenotype showed significantly increased EHEC-resistant survival, which contrasts with our current knowledge. Genetic mapping and whole-genome sequencing (WGS) revealed that these mutants have loss-of-function mutations in unc-89. Furthermore, we showed that the tolerance of unc-89(wf132) to EHEC relied on HLH-30/TFEB activation. These findings suggest that hlh-30 plays a key role in pathogen tolerance in C. elegans.

Introduction

Bacterial infection is a global issue and antibiotic-resistant bacteria threaten billions of lives worldwide. The key stage for pathogenic microbes (especially for enteric bacteria) to establish successful infection is colonization. Bacterial colonization requires two factors for initiation, the receptors on the host and the ligands on the bacterium (Ribet and Cossart 2015). The host receptors are usually specific carbohydrate or peptide residues on the cell surface, but bacteria sometimes provide their own receptors to the host. The ligands (also called adhesin) are the components expressed on the bacterial cell surface that can interact with the host receptors to achieve colonization (Ribet and Cossart 2015). Colonization renders pathogens with the ability to overcome host defense clearance which facilitates the infection process.

However, animal hosts may not only be passively infected by a pathogen; they may also mount an immune response. In general, the reactions induced by microbes can be categorized into three strategies, avoidance, resistance, and tolerance to protect host organisms from infectious insults (Medzhitov et al. 2012). Avoidance is defined as the ability to reduce the chance of animals ingesting deleterious microbes and the risk of animal exposure to the pathogens. Resistance is defined as the ability to minimize the pathogen burden, either by killing the pathogen to block invasion or by clearing the invading microbes once the infection has occurred (Schneider and Ayres 2008). Tolerance is defined as the ability to alleviate the negative impact caused by pathogens directly or by the immune response damaging collateral tissues (immunopathology) without altering the pathogen burden (Schneider and Ayres 2008; Medzhitov et al. 2012). Thus, investigating how animal hosts respond to pathogen infection can assist us in developing potential therapeutic treatments.

Enterohemorrhagic Escherichia coli (EHEC) is a human enteric pathogen that can colonize in the gastrointestinal tract and cause bloody diarrhea as well as hemolytic uremic syndrome (Pennington 2010). However, the use of antibiotics to treat EHEC infection is contraindicated as the release of endotoxins exacerbates the severe symptoms (Pacheco and Sperandio 2012; Freedman et al. 2016). We have shown that EHEC accumulates and replicates in C. elegans intestinal lumen resulting in attaching and effacing lesions, a characteristic pathology of EHEC infection (Chou et al. 2013). The bacterial determinants involved in EHEC colonization of C. elegans have also been revealed in previous studies (Kuo et al. 2016, 2018b). Nevertheless, the host factors involved in EHEC colonization are still unclear. To gain insight into the molecular and cellular mechanisms of EHEC colonization from the perspective of the host, we conducted forward genetic screens and isolated C. elegans mutants that displayed enhanced colonization of EHEC in the alimentary tract. We named this phenotype enhanced intestinal colonization (Inc). Interestingly, four mutants showed enhanced EHEC colonization but became resistant to EHEC killing. This phenomenon contradicts our current knowledge about bacterial infection, which is that enhanced bacterial colonization leads to shortened host survival. We thus turned our focus to identifying these EHEC-tolerant (i.e., enhanced colonization but EHEC resistant) mutants. To clone inc mutants, we performed conventional two- as well as three-point mapping and mapped our mutants to chromosome I located from −4.29 to 3.57. SNP mapping combined with whole-genome sequencing (WGS) analysis indicated that the causative gene of inc-1 is unc-89. Furthermore, we found the avoidance behavior induced by unc-89 mutation did not affect the survival of unc-89 mutant upon EHEC challenge. However, the underlying hlh-30/TFEB, a conserved regulator of autophagy and innate immune response defended against EHEC in the unc-89 mutant. This finding suggests that hlh-30 might be more common than currently appreciated in pathogen tolerance.

Materials and methods

Bacterial and nematode strains

Caenorhabditis elegans and bacterial strains used in this study are listed in Supplementary Table S1. All worms were cultured and maintained on nematode growth medium (NGM) agar plates using the standard laboratory E. coli strain OP50 at 20°C (Brenner 1974). The E. coli O157: H7 (EHEC) clinical isolates EDL933 were from the Bioresource Collection and Research Center, Taiwan. Some C. elegans strains used in this study were obtained from the Caenorhabditis Genetics Center (CGC), which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440).

Genetic screens for mutants with enhanced EHEC-GFP expression in the intestine

The ethane methyl sulfonate (EMS) mutagenesis method was performed as described (Brenner 1974). Briefly, the wild-type strain Bristol N2 hermaphrodites in the L4 stage (P0 generation) were mutagenized by 4-h exposure to 50 mM EMS (Sigma) at room temperature. Progeny (the F1 generation) were collected and allowed to mature to gravid adults, and then their embryos were harvested using hypochlorite treatment. The progeny of the F1 generation (F2) were fed with EDL933-GFP for 1 day at 20°C in the L4 stage. Mutant animals with strong GFP signals in the intestine were isolated onto an individual NGM plate. Control wild-type N2 animals were treated identically (with the exception of the EMS treatment), in parallel, to mutagenized animals. The offspring from isolated mutants were picked randomly to re-confirm the GFP expression in the intestine at the L4 stage with feeding on EDL933-GFP for 1 day on the primary screen, and 32 alleles still showed significantly enhanced GFP expression (Table 1). L4 animals of these 32 alleles were synchronized by the hypochlorite method and infected with EDL933-GFP for 1 day for the secondary screen. Eighteen mutants still showed significantly enhanced GFP expression in the C. elegans intestinal lumen and were also examined for their susceptibility to EHEC.

Table 1.

Summary of forward genetic screens for Inc phenotype

Screen	Number of worms	Hit Chromosome	Number of candidates	Survived	Primary screen (Inc% >50%)
1st	12,500	150,000	71	38	11
2nd	12,000	144,000	43	23	5
3rd	15,400	184,800	104	45	9
4th	16,600	199,200	46	18	7
Sum	56,500	678,000	264	124	32

Enhanced intestinal colonization (Inc) phenotype assay

Escherichia coli EDL933-GFP was cultured in LB broth with 50 μg/mL carbenicillin overnight at 37°C. The overnight culture of bacteria (O.D.₆₀₀ = 2.0) was spread over the entire 6 cm NGM agar plates (30 µL per plate) and placed at room temperature to dry. Synchronized L4 animals were fed with EDL933-GFP for 24 h at 20°C, and the GFP expression was observed with an Olympus SZX16 stereomicroscope. Animals with significant GFP signal in the pharynx and the whole intestinal lumen were classified as Inc. The percentage of Inc was defined as the animals with Inc phenotype among total tested animals. Images of Inc animals were taken using a Nikon Eclipse Ti inverted microscope system with DP72 CCD camera.

Colony-forming unit assay

Experimental procedures were performed as described (Chou et al. 2013). Briefly, to quantify the bacterial colonization in the animals, synchronized L4 animals were exposed to either E. coli EDL933-GFP (ampicillin resistant) or E. coli OP50-GFP bacteria covering the entire surface of a 6 cm agar plate for 1 day and then transferred to E. coli OP50 bacterial plates for another day at 20°C. Animals were washed out from the plates, treated with 25 mM levamisole, and washed in M9 buffer 10 times. The remaining bacteria outside the worms were eliminated by incubating the worms with M9 containing 25 mM levamisole, 100 µg/mL gentamicin and 1 mg/mL ampicillin for 1 to 2 h at room temperature. These two antibiotics were eliminated by washing the worms in M9 buffer with 25 mM levamisole 3 times. The worms in the buffer were transferred to an NGM plate to dry. Ten worms were picked randomly into 100 μL M9 buffer in an Eppendorf microtube, ground for at least 1 minute using a sterile plastic pestle, and plated on LB agar containing ampicillin after serial dilution. Numbers of bacterial colonies were counted the next day and the colony-forming unit (CFU) per worm was calculated.

To clarify whether the defect of pharyngeal muscle affected the number of live bacteria entering into the worm in a short time, L4 animals were exposed to either E. coli EDL933-GFP or E. coli OP50-GFP bacteria for only 1 h (in Figure 3A) at 20°C and CFU per worms were determined as above.

Figure 3.

Avoidance behavior induced by unc-89 mutation is not involved in protecting C. elegans host from EHEC killing. (A) The number of bacteria colonized in pharynx-defect mutants, unc-89/phm-1 and phm-2, was determined by the CFU assay as animals were treated with OP50-GFP and EHEC-GFP for 1 h, respectively. Values represent the means of three independent assays. (B) Survival curves of pharynx-defect mutants infected with EHEC. unc-89(wf132), unc-89(e1460), unc-89(st85) and phm-2(ad597) mutant animals were significantly resistant to EHEC infection. (C) Analysis of the percentage of avoidance behavior of wild-type N2, unc-89(wf132), unc-89(e1460), unc-89(st85) and phm-2(ad597) on E. coli OP50 bacterial lawn. (D) Analysis of the percentage of avoidance behavior of wild-type N2, unc-89(wf132), unc-89(e1460), unc-89(st85) and phm-2(ad597) on EHEC bacterial lawn. (E) Survival curves of unc-89(wf132) and unc-89(e1460) mutants feeding on the small lawn of OP50. (F) Examination of the survival curves of unc-89(wf132) and unc-89(e1460) mutants feeding on the large lawn of OP50. (G) Survival curves of unc-89(wf132) and unc-89(e1460) mutants infected on the small lawn of EHEC bacteria. (H) Survival curves of unc-89(wf132) and unc-89(e1460) mutants infected on the large lawn of EHEC. (A, C, and D) *P < 0.05; **P < 0.01; ***P < 0.001 compared to the N2 wild-type (WT) control group by the unpaired t-test. Error bars indicate the standard deviation (SD); n, total numbers of animals tested in each group. (B and E–H) ns, not significant; **P < 0.01; ***P < 0.001 compared to N2 wild type by the Mantel–Cox log-rank test. Survival curves represent the sum of animals in a minimum of three independent experiments.

RNA interference

Overnight cultured E. coli strain HT115 transformed with RNAi plasmids or empty vector (EV) were spread on 6 cm NG-IC plates (200 μL/plate), which correspond to NGM plates with 25 μg/mL carbenicillin and 1 mM isopropyl-β-D-1-thiogalactopyranoside (IPTG) and incubated at 25°C overnight to induce dsRNA expression. E. coli HT115 with L4440, an empty vector, was used as a negative control of RNAi. We used rrf-3(pk1246) mutant animals, which are hypersensitive to RNAi (Simmer et al. 2002), in RNAi assay as wild-type animals. Synchronized L1 animals were cultured on NG-IC plates with RNAi bacteria at 20°C until the gravid adult stage. Approximately 30 to 40 gravid adult worms (P0) were transferred to fresh RNAi plates allowing them to lay eggs (F1 progeny) and then removed from plates after 4 h. L4 worms of F1 progeny were transferred to NG-IC plates spread with RNAi bacteria mixed with OP50-GFP and EDL933-GFP in a 1:1 ratio, respectively, at 20°C for scoring Inc percentage or imaging.

Genetic analyses

Dominant or recessive analysis

To determine whether four the mutant alleles, wf059, wf091, wf101, and wf132 were dominant or recessive, mutant animals were crossed with either CB224 dpy-11(e224), the recessive marker strain, or GK454 unc-119; dkIs247[Pact-5::mCherry::HA::act-5, unc119(+)], the dominant marker strain. The F1 progenies from mutants crossed with visible markers were exposed to EDL933-GFP for 1 day at 20°C to examine the presence of Inc as described above. If the allele of the mutant is recessive, the Inc percentage of F1 progeny is close to 0%.

Complementation test (cis-trans test)

To identify whether the alleles showing Inc were mutations in the same gene, we designed two strategies to conduct the complementation test. In the first strategy, one of the mutant alleles wf132 was crossed with dpy-5 marked mutant to generate YQ268 strain which is wf132 mutant allele with Dpy phenotype. Strain YQ268 was crossed with other mutant alleles and the F1 progenies without Dpy phenotype (not self-progeny from YQ268) were examined for Inc. If the percentage of Inc was close to 0% (wild-type phenotype), it would suggest that the two parental alleles are different mutant genes. In the second strategy, wf059 and wf101 hermaphrodites were first crossed with dominant marker PD9753 ccIs9753 [myo-2p::GFP + pes-10p::GFP + gut-promoter::GFP] (Mark Edgley1 1999). The males of cross F1 progeny were then crossed into YQ268, which is wf132 with Dpy phenotype. F1 progenies without GFP were examined for Inc. If the percentage of Inc was close to 0% (wild-type phenotype), it would suggest that the two parental alleles are different mutant genes.

Chromosome mapping

To identify the mutation site of the mutant alleles, we combined the results of several chromosome mapping strategies to clarify the potential region of the mutant gene. Two strategies for two-point mapping were performed. First, mutant alleles were crossed to two marker strains, MT465 [dpy-5(e61) I; bli-2(e768) II; unc-32(e189) III] and MT464 [unc-5(e53) IV; dpy-11(e224) V; lon-2(e678) X]. Hermaphrodites from F2 progenies showing the marker phenotype, Dpy, Bli, Unc, or Lon, were examined for Inc phenotype. According to Mendel’s First Law of Genetics (Law of Segregation), if the mutant allele and marker allele are unlinked, the mapping percentage of Inc phenotype will be close to one fourth. If linked, the percentage of Inc from F2 homozygous maker animals will be less than a quarter. In the second strategy, instead of picking marker phenotype from cross F2, we plated out non-marker but Inc F2 worms and scored the percentage of their F3 progeny showing makers from these non-marker and Inc F2. If the Inc allele and marker allele are unlinked, the expected outcome is two-thirds of the marker phenotype (Marroquin et al. 2000). To narrow down the mutation loci of wf132, we crossed wf132 with various markers that are located on chromosome I and examined the percentage of Inc from F2 progeny that had markers. The genetic distance between two genes is based on the frequency of meiotic recombination, and a centimorgan (cM) is a unit that describes a recombination frequency of 1% (Fay 2006). If the two genes are positioned adjacently, the possibility of recombination between them will be extremely small.

Single nucleotide polymorphism mapping

Experiments followed previously described procedures (Davis et al. 2005). In brief, hermaphrodites of wf059, wf091, wf101 and wf132 were crossed into C. elegans strain CB4856 (Hawaiian) males, respectively, to produce heterozygous F1 animals. F1 progenies without Inc phenotype (not self-progeny) were isolated to produce F2 progenies. Fifty Inc homozygotes and 50 non-Inc heterozygotes from F2 progenies were isolated for chromosome mapping. In addition, approximately 100 Inc homozygous F2 progenies from maternal wf132 were further isolated and cultured onto individual plates for interval mapping.

Single nucleotide polymorphism mapping—whole-genome sequencing

WGS combined with SNP mapping was performed based on previous research (Doitsidou et al. 2016; Lee et al. 2020). The genomic DNA of two F2 progeny alleles from wf132 and Hawaiian cross animals were extracted and subjected to WGS using the Illumina HiSeq 2500 platform in the NGS core at the Biodiversity Research Center, Academia Sinica, for 2× 150 bp sequencing. The WGS data (48× depth coverage) was uploaded and analyzed through the CloudMap pipeline as described (Minevich et al. 2012). Frameshift, synonymous, non-synonymous, splice site and stop-gain mutations from the region of interest were noted and classified as potential candidate genes as listed in Supplementary Table S2. The wf059 and wf101 alleles were sequenced with 2x 150 bp reads using WGS on the Illumina Hiseq X-ten System to confirm the mutation loci of unc-89. The WGS data (90x depth coverage) were analyzed and compared to the Bristol N2 reference genome from Genebank.

Bacterial avoidance behavior

The measurement of bacterial avoidance behavior was performed as described with slight modification (Kumar et al. 2019). NGM plates (6 cm) with a small bacterial lawn were prepared by adding 100 μL of overnight cultures of E. coli OP50 and E. coli O157: H7 EDL933 on the center. Thirty to 50 synchronized L4 animals were transferred to the top of the bacterial lawn and each animal was scored as inside or outside the lawn (Avoidance = N out/N total) after 24 h at 20°C.

Lifespan of C. elegans feeding on E. coli OP50 or EHEC

Lifespan of C. elegans fed on E. coli OP50 or E. coli O157: H7 EDL933 bacterial lawn was performed as described (Chou et al. 2013). E. coli was cultured in LB broth overnight at 37°C. On the next day, the overnight culture of bacteria was spread (30 µL per plate) to the 6 cm NGM agar plates and placed at room temperature. On the third day, about fifty synchronized L4 animals were transferred to each plate and cultured at 20°C.

To test the correlation between avoidance behavior and survival, the survival assays were carried out on the NGM agar plates with a small or large lawn by adding 30 μl of overnight cultures of E. coli OP50 or E. coli O157: H7 EDL933. The small lawn was prepared by spotting a small area of the bacteria on the center (4 to 6.25 cm², covering ∼20% surface), whereas the large lawn was prepared by spreading the bacteria over the entire surface.

Animals were transferred to fresh plates and the survival was scored daily. Nematodes that did not respond to gentle prodding and displayed no pharyngeal pumping were scored as dead. Worms that crawled off the plate were censored. The experiment was performed independently at least three times. The Mantel–Cox logrank test was used to assess statistical significance of difference in survival, and P values of less than 0.05 were considered significant.

Quantitative RT-PCR

qRT-PCR was performed as described previously (Chou et al. 2013; Chen et al. 2017). To test the mRNA expression of autophagy-related genes (lgg-1, lgg-2) and C-type lectin genes (clec-7, clec-60) in wild-type N2 and unc-89 mutant, synchronized L4 animals were seeded with either E. coli OP50 or E. coli EDL933 for 24 h at 20°C. Then, these young adults were washed and collected for RNA isolation. Total RNA was then extracted using TRI Reagent (Invitrogen), and reverse transcribed with M-MLV reverse transcriptase (Promega) using random hexamer primers. This cDNA was then subjected to qRT-PCR analysis using SYBR green detection on a 7500 fast real-time PCR system (Applied Biosystems). Relative expression between samples was normalized to the eft-2 as the reference gene. The primers used to amplify the amplicons are listed as follows: clec-7 F: 5′-GGCCGGCTTCAAATGTTTATC-3′; clec-7 R: 5′-TAGTGGACATTACCATGCAGTC-3′; clec-60 F: 5′-CTGAGCCAAGAACCACAAGA-3′, clec-60 R: 5′-GAAGTGCTGACTGACGAAAGA-3′; lgg-1 F: 5′-GAAGAAGTACTTGGTCCCATCCG-3′; lgg-1 R: 5′-CGTGATGGTCCTGGTAGAGTTG-3′; lgg-2 F: 5′-GAAGGCCATTCCATGAGCGTCA-3′; lgg-2 R: 5′-GTGGATGAAGTTGGAGGCGTC-3′. eft-2 F: 5′-TCGAAATTCAATGCCCAGAA-3′; eft-2 R: 5′-CTCCTCGAAAACGTGTCCTCTT-3′.

Quantitation of HLH-30::GFP

To quantify HLH-30::GFP expression in the wild-type and unc-89 mutant animals, L4 or young adults of OP433 and YQ502 animals were fed with E. coli OP50 or E. coli EDL933 for 24 h at 20°C, respectively. HLH-30::GFP signals in the animals were acquired using a Nikon Eclipse Ti inverted microscope system and quantified by Image J software.

Generation of double mutant in unc-89 mutant background

To generate unc-89(wf132); tph-1(mg280) double mutant, unc-89(wf132) male animals were crossed to tph-1(mg280) hermaphrodites. The L4 progenies were fed with EHEC-GFP for 1 day, and F1 animals without Inc phenotype were transferred to other NG plates to lay eggs. The progenies of F2 generation were confirmed for unc-89(wf132), tph-1(mg280), or unc-89(wf132); tph-1(mg280) by Inc assay and single worm PCR. To generate unc-89(e1460); hlh-30(tm1978) double mutant, unc-89(e1460) males were crossed to hlh-30(tm1978) hermaphrodites. The progenies of F2 generation were confirmed for unc-89(e1460), hlh-30(tm1978), or unc-89(e1460); hlh-30(tm1978) by Inc phenotype and single worm PCR. unc-89(e1460) was confirmed by Inc assay. hlh-30(tm1978) was confirmed using the primers (5′-CTTTTAGAATTGTTCCTTTGTTG-3′; 5′-GCCTTGAAAATTGATACAATAAG-3′). The generation of unc-89(e1460); hlh-30::gfp animals was similar to described above. unc-89(e1460) male animals were crossed to OP433, unc119(tm4063); wgIs433[hlh-30::TY1::EGFP::3xFLAG+unc-119(+)] hermaphrodites. hlh-30::gfp animals were confirmed by the HLH-30::GFP expression.

HLH-30 nuclear localization

The process of unc-89 RNAi knockdown in OP433, unc119(tm4063); wgIs433[hlh-30::TY1::EGFP::3xFLAG+unc-119(+)] animals was performed in the same way as for rrf-3 described above. RNAi assays were carried out on E. coli OP50 or E. coli O157: H7 EDL933 mixed with unc-89 RNAi bacteria with a 1:1 ratio (OD₆₀₀=2.0) and were spread over the entire NGM plate (60 μL/plate) at 20°C. E. coli HT115 with an empty vector (EV), L4440, was used as a negative control. The GFP signal was visualized with a Nikon Eclipse Ti Inverted Microscope System with a DP72 CCD camera or by Olympus FV3000 confocal laser scanning microscope. The ratio of HLH-30 nuclear localization in the animals was determined by counting the number of worms with increased HLH-30::GFP nuclear localization. The animals that showed more than thirty nuclear HLH-30::GFP translocation were defined as animals with increased HLH-30 nuclear localization. The images were captured by Olympus FV3000 confocal laser scanning microscope.

Data availability

Strains and plasmids are available upon request. Supplementary Figure S1 shows eighteen Inc percentage and increased survival percentage of EMS-induced mutants. Supplementary Figure S2 shows the pumping rate of four EHEC tolerance alleles. Supplementary Figure S3–S7 indicate the SNP results of the four alleles. Supplementary Table S1 shows the strains or plasmids used in this study. Supplementary Table S2 shows the gene candidates of wf132 from the WGS result. All Supplemental information including the WGS original data are available at figshare DOI: https://doi.org/10.25386/genetics.13340426.

Results

Forward genetic screens for host factors involved in EHEC tolerance

To identify the host factors involved in EHEC colonization, we performed forward genetic screens to isolate C. elegans mutants with early expression of the EHEC-GFP (GFP-labeled EHEC) signals in the intestinal lumen. In our previous study, we found that wild-type (WT) N2 worms fed with EHEC-GFP showed GFP signals in their lumens until 2 to 3 days post-infection (Chou et al. 2013). We hence isolated the mutants with increased GFP expression in the intestinal tract during EHEC-GFP infection for 1 day (Figure 1A) and named the phenotype as enhanced intestinal colonization (Inc) (Figure 1B). In EMS-mutated animals, we observed numerous mutants displaying visible marker phenotypes, e.g., bli, dpy, unc, or lon, implying our EMS treatment is valid. Two-hundred and sixty-four candidates were identified from a total of approximately 678,000 haploid genomes with four EMS screens. Among these candidates, 140 animals were excluded because of sterility or death after being isolated individually (Table 1). Surviving animals (124) were then examined for the percentage of animals with Inc from their descendants by feeding with EHEC-GFP for 1 day at 20°C. Ninety-two alleles were eliminated as the percentage of animals with Inc was less than 50% in contrast with the parental wild type and thirty-two alleles remained on this primary screen (Table 1). To examine whether enhanced colonization affects the susceptibility of C. elegans to EHEC, we analyzed the survival of these 32 Inc mutants on EHEC and re-confirmed the Inc on secondary screening. We found that 18 alleles still showed significant Inc phenotype whereas the survival of these animals upon EHEC infection was variable (Supplementary Figure S1). Intriguingly, four inc mutants displayed a prominent Inc (Figure 1B, strong GFP expression in the intestinal tracts of four mutants) while also showing significantly increased survival to EHEC (Figure 1C, Inc% > 80% and increased median survival days > 60% compared to wild type). We therefore turned our focus to mapping these four alleles that showed tolerance (i.e., enhanced intestinal colonization but pathogen resistance) to EHEC. To test whether enhanced GFP expression in C. elegans intestinal lumen represented an increased EHEC bacterial burden, we examined the colony-forming units (CFUs) of four inc mutants. Indeed, we found these four inc mutants that showed enhanced EHEC-GFP signal expression in the alimentary tract also showed increased colonization of EHEC (Figure 1D). One plausible interpretation of this result is that the increased bacterial load could be due to increased food ingestion by the four inc mutants. To test whether these inc mutants displayed increased food intake, we measured the pharyngeal pumping rate of the mutants. The pharynx is the feeding organ of C. elegans, and we found that the pumping rates of the four inc mutants were comparable to that of the wild-type N2 animals on EHEC feeding (Supplementary Figure S2), suggesting that the enhanced EHEC colonization in these mutants was not a result of increased food intake. Furthermore, survival curves showed that the four inc mutants were all resistant to EHEC infection compared to N2 (Figure 1E, All P < 0.001). An alternative explanation for the survival result is that the mutation of these alleles confers extended lifespan (longevity) to inc mutants. We therefore examined the lifespan of the four inc mutants feeding on a standard laboratory diet, E. coli OP50. As shown in Figure 1F, lifespan of inc mutants on OP50 was not prolonged compared to N2. Together, these results suggested that the four inc mutants manage the ability to tolerate EHEC infection and this tolerance is not due to an extended lifespan in general.

Figure 1.

Genetic screens identifying the host factors involved in enterohemorrhagic E. coli tolerance in C. elegans. (A) Genetic screens for C. elegans mutants with enhanced colonization. Bristol N2 wild-type strain at the L4 stage were treated with mutagen, ethane methyl sulfonate (EMS), for 4 h at room temperature. L4 stage animals of the F2 generation were infected with GFP-labeled EHEC (referred to as EHEC-GFP) for 24 h at 20°C. The mutant worms with strong GFP signal expression in the intestinal tracts (bacteria colonization) were isolated onto individual plates. (B) Representative images of mutant alleles showing enhanced intestinal colonization (Inc) phenotype. Four EMS-induced mutants (wf059, wf091, wf101 and wf132) showed significantly increased GFP signal expression in the intestinal lumen when infected with EHEC-GFP for 24 h at 20°C. No or less GFP signal was observed in wild-type (WT) N2 animals after the same treatment with EHEC infection. The scale bars represent 100 µm. (C) Percentages of animals showing Inc and the increased median survival days of four inc alleles (wf059, wf091, wf101 and wf132). The increased median survival days of four inc mutants were compared to N2 animals. (D) Four inc mutants showed enhanced EHEC bacterial colonization. The number of live bacteria accumulated in the intestine was determined by CFU assay from three independent experiments. L4 animals were fed with GFP-labeled OP50 (OP50-GFP) or EHEC-GFP for 1 day, and transferred to OP50 for another day at 20°C. The accumulation of EHEC was increased in the intestine of mutants compared to the N2 wild type (WT). Values represent the means of three independent assays, and error bars indicate the standard deviation (SD). ***P < 0.001 compared to the N2 control group by t-test. (E) Survival curves of four mutants infected with EHEC at 20°C. Four mutants were significantly resistant to EHEC compared to that of wild-type N2 (all P < 0.001). ***P < 0.001 by the Mantel–Cox log-rank test. (F) Survival curves of four mutants fed with a standard laboratory diet of OP50 at 20°C. The four mutants show slightly shorter lifespan fed on OP50 compared to that of wild-type N2. *P < 0.05; ***P < 0.001 by the Mantel–Cox log-rank test. All survival curves represent the sum of animals from multiple independent experiments.

Conventional mapping of inc alleles

We then aimed to clone the mutation locus of inc mutants that showed tolerance to EHEC. First, we determined whether the inheritance of these four inc alleles was dominant or recessive. Of the four lines, all of their F1 cross progenies showed a low percentage (< 5%) of Inc (Table 2) while feeding EHEC-GFP for 1 day, suggesting that these four mutations were recessive alleles. Next, we sought to determine whether mutations of the four alleles with similar Inc represent mutations of different genes or mutations of different forms of the same alleles. We crossed two different inc alleles into one animal to determine their complementation group. All crossed F1 progenies showed a high percentage of Inc, except for the offspring from paternal wf091 (Table 3). The complementation analyses divided the four inc mutants into two groups of which three alleles fall into the sample linkage group (wf059, wf101 and wf132) and one allele (wf091) into another (Table 3). Since the complementation results indicated that the linkage of four inc alleles belongs to two groups, inc-1 and inc-2, we selected one strain (YQ139, wf132 allele) from three of inc-1 and mapped its chromosomal location as a representative allele. Our genetic analysis results placed the linkage of inc-1(wf132) to chromosome I (Table 4). Three inc-1 mutants were identified from different batches of EMS screen (wf059 and wf132 were isolated from the first screen; wf101 was from fourth screen), implying that this gene is pivotal for EHEC tolerance. Therefore, we focused on inc-1(wf132) and conducted three-point mapping to estimate the genetic distance between the locus of inc-1 allele and selected makers on chromosome I. As shown in the results in Table 5, the Inc percentage in dpy-5 marker animals was lower than that of the other marker strains. Furthermore, inc-1 mutant mated into dpy-5; unc-101 double marker strain, also showing lower Inc percentage in dpy-5 mutant than unc-101 mutant. These genetic data suggest that the mutation locus of inc-1 was adjacent to the location of dpy-5 gene. We calculated the recombination frequency of inc-1(wf132) crossed into each marker and took the mean and maximum of the percentage together. The prediction of the mutation locus of inc-1(wf132) was between the genetic positions −4.29 to 3.57 on chromosome I. The same strategy was performed to clone inc-2 alleles and we found that inc-2(wf091) was linked to chromosome V from chromosome mapping (Table 6). Furthermore, the Inc percentage of different marker strains on chromosome V were all less than a quarter, suggesting that the mutation location of inc-2 allele was on chromosome V (Table 6).

Table 2.

Dominant or recessive analysis for four mutant alleles

Marker	Alleles	(Inc%, sample size)	Dominant/recessive
dpy-11(e224)	wf132	0%, 0/66	Recessive
	wf091	3.39%, 2/59	Recessive
mCherry::Act-5	wf059	0%, 0/68	Recessive
	wf101	0%, 0/41	Recessive

F1 progeny generated from indicating parents were examined for Inc.

Table 3.

Complementation results for four mutant alleles

	wf059 ♂	wf091 ♂	wf101 ♂
wf132 ♂	Fails to complement (100%, 20/20)	Complement	Fails to complement (78%, 90/118)a
	(87%, 60/69)a	(0%, 0/85)

F1 progeny generated from the indicated parents were examined for Inc. The percentage represents the Inc percentage and the sample sizes were as indicated.

^aMapped via strategy 2.

Table 4.

Chromosome mapping result of wf132

dpy-5 I	bli-2 II	dpy-1 III	unc-5 IV	dpy-11 V	unc-2 X
8%, 8/100 0%, 0/37a	22.8%, 13/57	61.9%, 13/21a	25%, 23/92	83.3%, 10/12a	60%, 9/15a

^aMapped via strategy 2 (if not linked, 2/3 have marker alleles). Otherwise, mapping was performed via strategy 1. The percentage represents the Inc percentage and the sample sizes are as indicated.

Table 5.

Chromosome mapping result of wf091

dpy-5 I	dpy-10 II	unc-32 III	unc-5 IV	dpy-11 V	lon-3 V	lon-2 X
70%, 21/30	22.8%, 13/57	40.7%, 11/27	45.7%, 16/35	15.2%, 5/33 1.03%, 1/97 0%, 0/97	2%, 1/50	50%, 16/32

The percentage represents the Inc percentage and the sample sizes are as indicated.

Table 6.

Three-point chromosome mapping result of wf132

Allele	Chr I marker	Strategy 1	Strategy 2
		(F2 Inc%, sample size)	(F3 marker%, sample size)
wf132	unc-15	6.98%, 3/43	10.5%, 2/19
		0%, 0/38	0%, 0/30
	dpy-5	0.95%, 1/105	0%, 0/28
		0%, 0/56	0%, 0/9
	unc-54	9.375%, 3/32	33%, 3/9
		1.25%, 1/80	33%, 3/9
		16.67%, 4/24
	bli-3	10.71%, 3/28	55.56%, 5/9
		0/51, 0%
		5/22, 22.72%
wf132	dpy-5; unc-101 a	1. Dpy (4.29%, 3/70) Unc (8.33%, 1/12)
		2. Dpy (0.99%, 1/101) Unc (7.5%, 3/40)
		3. Dpy (2%, 2/100) Unc (4.76%, 2/42)

^a wf132 crossed with dpy-5; unc-101 and the Inc percentage was calculated from the F2 progeny showing Dpy or Unc. The percentage represents the Inc percentage and the sample sizes are as indicated.

Single nucleotide polymorphism mapping of inc alleles

In addition to the conventional mapping strategy, we also cloned inc mutants via a single nucleotide polymorphism (SNP) mapping method developed by Davis and colleagues (Davis et al. 2005). C. elegans Hawaiian strain (CB4856) contains 327,050 SNP from the canonical laboratory Bristol strain, N2 (Thompson et al. 2015). The SNPs acted as genetic markers to measure the segregation ratio from classical two- and three-factor mapping. We crossed three inc-1 alleles as well as inc-2(wf091) with the Hawaiian strain, CB4856, respectively, and isolated 50 Inc CB4856 cross mutants and 50 non-Inc CB4856 cross mutants from the F2 generation into separate reactions for chromosome mapping. All three inc-1 alleles were linked to the center of the linkage group (LG) I in the -6 to 5 region (Supplementary Figures S3–S5) and inc-2(wf091) was linked on LG V in the 6 to 18 region (Supplementary Figure S6), which is consistent with our conventional genetic results (Tables 4–6). We then singled out 96 F2 cross Inc mutants and conducted interval mapping for inc-1(wf132) as a representative allele for inc-1. Our genetic result placed inc-1(wf132) between -6 and -1 and was close to genetic loci -1 on LG I (Supplementary Figure S7).

inc-1(wf132) contains a premature stop codon mutation in unc-89

To identify the causative genes responsible for Inc, we employed SNP mapping combined with WGS by CloudMap (Minevich et al. 2012) and identified several gene mutations that potentially contribute to Inc in wf132. Our conventional mapping results indicated that the mutation loci of inc-1(wf132) could range from −4.29 to 3.57 and thus we focused on the genes revealed from WGS analysis of this region as our targets (Supplementary Table S2). To test whether mutation of these genes in C. elegans contributed to Inc, we knocked down these genes by RNAi and examined the percentage of Inc in an RNAi sensitive mutant, rrf-3(pk1426) (Simmer et al. 2002). rrf-3 mutant-silenced candidate genes did not show Inc except in the case of silencing one particular gene, unc-89, which showed a comparable Inc to that of inc-1(wf132) mutant (Figure 2A, both Inc% > 80%, P = 0.176). We re-confirmed our RNAi results using mutant alleles independently to avoid an off-target effect or low efficiency of RNAi that might mislead our direction. In Figure 2B, an existing unc-89(e1460) mutant showed a similar Inc phenotype to our inc-1(wf132) (P = 0.326) implying that inc-1 is unc-89. The wf132 allele contains 1 nucleoside substitution (guanosine to adenosine) of exon 25 and changes the encoded tryptophan^3,417 to a stop codon on the immunoglobulin-like domain by prediction of amino acid sequences (Figure 2C and Supplementary Figure S9). The existing allele, e1460, also causes an early stop mutation in the same exon 25 of unc-89 as 1 nucleoside alteration (cytidine to thymidine). From WGS analysis, we found that wf059 allele contains a nonsense mutation on glutamine⁶¹⁹⁹ and the wf101 allele contains a 2-nucleotide deletion on the intron potentially affecting the RNA splicing. The RNAi clone that targets unc-89 covers the region of two mutation loci of wf132 and e1460 alleles (Figure 2C and Supplementary Figure S9), conferring Inc in C. elegans as well. rrf-3 mutant RNAi unc-89 showed a strong GFP signal in their alimentary tracts while feeding on EHEC-GFP (Supplementary Figure S8A) and also had an increased percentage of Inc expression in comparison with the empty vector (EV) control (Supplementary Figure S8B, P < 0.001). In addition to RNAi knockdown of unc-89, we confirmed our result by examining the Inc in the unc-89 mutant as well. Two existing unc-89 alleles, e1460 and st85, which had been identified and studied previously (Avery 1993; Benian et al. 1996; Qadota et al. 2016) were tested for Inc. We found that the GFP signals were also robustly expressed in their intestinal lumen upon infection with EHEC-GFP for 1 day (Figure 2D) and the percentage of Inc was significantly increased in e1460 and st85 alleles compared to N2 (Figure 2E, both P < 0.001), which is similar to our wf132 allele. Of note, the wf132 mutant used was backcrossed to our laboratory wild-type N2 parents four times to minimize the potential cause of Inc being from EMS-induced background mutation. Together, these data suggest that inc-1 is unc-89. We also noticed that our inc mutants were smaller and thinner (Figure 1B), which is similar to other unc-89 mutants as previously reported (Spooner et al. 2012; Yemini et al. 2013) (Figure 2D), and this observation strengthened our interpretation that inc-1 is unc-89. To determine whether the enhanced GFP signals in the alimentary tracts represent the enhanced bacterial colonization in unc-89 mutants, we performed colony formation unit (CFU) assay. All unc-89 mutants with enhanced Inc showed increased bacterial burden of EHEC compared to that of N2 (Figure 2F, All P < 0.001). Moreover, complementation tests showed that heterozygous F1 offspring from inc-1(wf132) and unc-89(e1460) parents displayed comparable Inc percentage (> 80%) to inc-1(wf132) and unc-89(e1460) (Figure 2G, both P = 0.356) indicating that inc-1(wf132) failed to complement unc-89(e1460). Together, these results showed that inc-1 is unc-89. Another plausible explanation for the inc-1 mutants conferring tolerance to EHEC was contributed to by two distinct gene mutations; one gene mutation causes Inc phenotype and the other is responsible for EHEC resistance in C. elegans. These two genes were both located on chromosome I and should be adjacent or very close to each other and thus did not separate into two different gametes during segregation. To test our hypothesis, we examined the susceptibility of two existing unc-89 alleles to EHEC challenge. We found that wf132, and two unc-89 alleles, e1460 and st85, were all resistant to EHEC infection (Figure 3B) suggesting that mutation of unc-89 confers the Inc and tolerance to EHEC.

Figure 2.

Mutation of unc-89 is responsible for the Inc phenotype. (A) The Inc percentage of rrf-3(pk1426) RNAi silenced candidate causal genes identified from WGS analysis. rrf-3(pk1426) RNAi knockdown of unc-89 showed a similar Inc percentage to wf132 (P = 0.326) when infected with EHEC-GFP for 1 day. (B) Examination of the mutants of potential causative genes underlying Inc. (C) Diagram of different UNC-89 isoforms and the mutation loci of e1460, wf059, wf101 and wf132 alleles. The mutation site of e1460 is at 27,300 nt (cysteine to threonine), and the amino acid mutates from glutamine (Q) to a premature stop codon. The mutation site of wf059 is at 40,091 nt (cysteine to threonine), and the amino acid mutates from glutamine (Q) to a premature stop codon. wf101 allele contains two guanine deletions at 6408 to 6409 nt in the intron. The mutation site of wf132 is at 27,974 nt (glycine to alanine), and the amino acid mutates from tryptophan (W) to a premature stop codon. The RNAi targeting region is from 27,006 to 28,182 nt and is indicated by the line. SH3, Src homology-3 domain (red); DH, DbI homology domain (green); PH, Pleckstrin homology domain in (aqua blue); IG, Immunoglobulin-like domain (pink); FN3, Fibronectin type III domain in (yellow); PK, Protein kinase domain in (violet). The scale bar indicates 1000 amino acids on the top right. Graphs are illustrated by DOG 2.0. (D) Representative images of unc-89 mutant animals showing Inc. Animals were treated with EHEC-GFP for 1 day and the GFP signal expression was monitored in the intestinal lumen. The scale bar indicates 100 µm. (E) Two existing unc-89 mutant alleles, e1460 and st85, revealed comparable Inc percentage to wf132 (P = 0.437 and 0.846 by the unpaired t-test, respectively) on EHEC-GFP infection for 1 day. WT represents N2 wild-type animals. (F) The colony formation unit (CFU) of unc-89 mutants. Animals were fed with OP50-GFP and EHEC-GFP for 1 day, respectively, and then chased to normal OP50 for another day to determine the bacterial accumulation in the intestine. CFU results showed that EHEC bacteria enhanced colonization in unc-89 mutants. Values represent the means of three independent assays. (G) cis-trans test of unc-89(wf132) and unc-89(e1460). The Inc percentage of F1 cross progeny from unc-89(wf132) and unc-89(e1460) was similar to parental unc-89(wf132) and unc-89(e1460) (both P = 0.356). (A, B, and F) ***P < 0.001 compared to the control group by the unpaired t-test. Error bars indicate the standard deviation (SD); n, the total numbers of animals tested in each group.

UNC-89, known as PHM-1, is a giant multi-domain protein expressed in pharyngeal muscle, body wall muscle and intestinal muscle in C. elegans (Benian et al. 1996; Small et al. 2004). The homolog of UNC-89 to human is obscurin, which is required for the muscle cell architecture (Manring et al. 2017). The largest isoform, UNC-89-B, is composed of 53 immunoglobulin (Ig) domains in the middle, a triplet of Src homology 3 (SH3), Dbl homology (DH), and pleckstrin homology (PH) domains at its N-terminus, and two fibronectin type 3 (Fn3) domains and two protein kinase domains (PK1 and PK2), at the C-terminus (Figure 2C) (Spooner et al. 2012; Manring et al. 2017). UNC-89/obscurin is located at the sarcomeric M-line in C. elegans muscle and interacts with protein paramyosin (an invertebrate-specific coiled-coil dimer protein) as well as phosphatase 2 A (PP2A) to maintain muscle structure (Manring et al. 2017; Qadota et al. 2018). Loss of function of unc-89 contributes to abnormal organization of thick filaments in C. elegans muscular cell (Spooner et al. 2012; Wilson et al. 2012).

unc-89 mutant displays avoidance behavior to EHEC

UNC-89 has not been previously linked to immunity in C. elegans. Hence, we aimed to investigate the mechanism of unc-89 mutation conferring C. elegans increased tolerance to EHEC. Recent studies showed that in pharynx grinder defect mutant, phm-2, bacterial colonization and consequent bloating in the intestinal lumen activated animals’ protective avoidance behavior as well as the immune response to improve organism survival (Kumar et al. 2019; Singh and Aballay 2019), which closely resembles our observation of wf132 allele. Also, deficiency of phm-2 and unc-89/phm-1 leads to abnormal pharyngeal function (Avery 1993). We examined whether these two pharyngeal-defective mutants, phm-2 and unc-89/phm-1, increase bacterial accumulation after exposure to bacteria for a short period of time (1 h). From CFU analysis, we found that mutation of unc-89 and phm-2 enhanced the bacterial accumulation compared to N2 when fed with either OP50 or EHEC (Figure 3A) indicating that bacteria accumulates in the two pharyngeal-defective mutants rapidly. We thus examined the survival of these mutants against EHEC infection as both mutation of unc-89 and phm-2 enhanced bacterial colonization. Survival curves of unc-89 and phm-2 mutants feeding on EHEC were both extended (Figure 3B) implying that the mechanism by which unc-89 mutant tolerates EHEC infection might be similar to that of phm-2 alleles. Consistent with the phm-2 mutant, enhanced colonization in the unc-89 allele caused lumen dilation (Figures 1B and 2D) and might stimulate the aversion behavior as well as the innate immune pathway resulting in tolerance to EHEC. Therefore, we next examined whether unc-89 mutation induces avoidance behavior in C. elegans. unc-89 mutants displayed increased avoidance behavior to OP50 and EHEC with more than 85% of animals lingering outside the bacterial lawn whereas fewer than 10% of the N2 animals were outside the OP50 and EHEC bacterial lawns (Figure 3, C and D). Aversion behavior has been reported to affect the susceptibility of C. elegans to the pathogen as the animals that moved off the lawn of bacteria had reduced ingestion of pathogenic bacteria leading to increased survival (Styer et al. 2008; Reddy et al. 2009). To determine whether the enhanced avoidance behavior induced by unc-89 mutation affects the susceptibility of C. elegans to EHEC infection and its lifespan on OP50, we used two different types of agar plates that contain a small lawn or large lawn of bacteria and examined the survival of unc-89 mutants on them. For the small lawn agar plates, the bacteria were seeded as a spot on the center of the agar plates. In contrast, on the large lawn plates, the bacterial lawn was spread to the edges of the agar plates covering the entire surface of the agar plates in such a manner that the animals could not avoid the bacteria. Survival of unc-89 mutants cultured on the small lawn of the OP50 plate was equivalent or slightly increased compared to N2 (Figure 3E). However, when cultured on the large lawn, both unc-89(wf132) and unc-89(e1460) had slightly shorter lifespans (Figure 3F). unc-89 mutants cultured on the small lawn of EHEC showed extended survival (Figure 3G), which might be caused by avoidance behavior contributed to by the unc-89 mutation. Intriguingly, unc-89 mutants were still significantly resistant to EHEC infection when cultured on the large lawn of EHEC (Figure 3H, both P < 0.001 compared to N2). These data suggest that aversion behavior of unc-89 mutants is not involved in EHEC tolerance.

tph-1 activity is involved in the Inc and avoidance behavior in unc-89 mutant

To further elucidate the mechanism of action of unc-89 in EHEC tolerance, we generated and analyzed unc-89; tph-1 double-mutant strain for which tph-1 had been reported to be necessary to mediate bacterial accumulation-induced avoidance behavior in C. elegans (Kumar et al. 2019). tph-1 encodes a tryptophan hydrolase gene, which is essential for serotonin biosynthesis, and loss of tph-1 causes a defect in serotonin signaling in C. elegans (Sze et al. 2000). We first examined whether loss of tph-1 influences the Inc phenotype in the unc-89 mutant. unc-89(wf132); tph-1(mg280) fed with EHEC-GFP for 1 day showed a lower percentage of Inc than unc-89(wf132) (Figure 4A). Similarly, the bacterial burden was significantly reduced in unc-89(wf132); tph-1(mg280) compared to that of unc-89(wf132) (Figure 4B). Together, these results indicated that loss of tph-1 suppresses the EHEC colonization in unc-89 mutant. To test whether tph-1 is involved in the aversion behavior of unc-89 mutant, we examined the avoidance of unc-89(wf132); tph-1(mg280). The percentage of unc-89(wf132); tph-1(mg280) escaping from either OP50 or EHEC was significantly less than that of unc-89(wf132) (Figure 4, C and D), suggesting that serotonin signaling is required for unc-89(wf132) in sensing bacteria. Furthermore, tph-1 mutation suppressed the prolonged survival of unc-89(wf132) on the small and large lawn of EHEC (Figure 4, E and F, respectively), indicating that avoidance behavior is partly involved in EHEC tolerance. However, the shortened survival of unc-89(wf132); tph-1(mg280) due to the defect of tph-1 activity did not restore it to WT level, suggesting that there are other effectors downstream mediating EHEC tolerance in unc-89 mutant.

Figure 4.

Mutation of tph-1 suppresses the Inc and avoidance in unc-89 mutant. (A) The Inc percentage of WT, tph-1(mg280), unc-89(wf132) and unc-89(wf132); tph-1(mg280). Animals were infected with EHEC-GFP for 1 day and were examined for Inc. (B) The colony formation unit (CFU) of WT, tph-1(mg280), unc-89(wf132) and unc-89(wf132); tph-1(mg280). Animals were fed with OP50-GFP and EHEC-GFP for 1 day, respectively, and then chased to normal OP50 for another day to determine the bacterial accumulation in the intestine. (C) Analysis of the percentage of avoidance behavior of WT, tph-1(mg280), unc-89(wf132) and unc-89(wf132); tph-1(mg280) on the OP50 bacterial lawn. (D) Analysis of the percentage of avoidance behavior of WT, tph-1(mg280), unc-89(wf132) and unc-89(wf132); tph-1(mg280) on the EHEC bacterial lawn. (E) Survival curves of WT, tph-1(mg280), unc-89(wf132) and unc-89(wf132); tph-1(mg280) infected on the small lawn of EHEC bacteria. (F) Survival of curves WT, tph-1(mg280), unc-89(wf132) and unc-89(wf132); tph-1(mg280) infected on the large lawn of EHEC bacteria. (A–D) ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001 compared to the N2 WT control group by the unpaired t-test. Error bars indicate the standard deviation (SD). The total numbers of animals tested in each group are indicated by n. (E and F) **P < 0.01; ***P < 0.001 compared to N2 WT by the Mantel–Cox log-rank test. Survival curves represent the sum of animals from a minimum of two independent experiments.

Mutation of unc-89 leads to HLH-30/TFEB nuclear translocation

We next aimed to identify the effectors downstream of unc-89 that confer EHEC tolerance. Kumar and colleagues recently showed that enhanced bacterial colonization induces nuclear localization of a transcription factor HLH-30/TFEB in the mutant with defective pharynx grinder, phm-2, and thus activates innate immune genes to extend the lifespan of C. elegans (Kumar et al. 2019). We therefore addressed whether mutation of unc-89 induced HLH-30 nuclear localization in C. elegans. When HLH-30/TFEB is activated, HLH-30/TFEB protein accumulates in the nucleus to trigger the upregulation of the downstream signaling pathways (Lapierre et al. 2013; Visvikis et al. 2014). We found that WT HLH-30::GFP C. elegans infected with EHEC for 1 day induced HLH-30 nuclear localization (Figure 5, A and C) and the percentage of HLH-30 nuclear localization in EHEC-infected animals (> 50%) was significantly higher than that of the OP50 control group (Figure 5, B and D). Moreover, RNAi knockdown of unc-89 induced HLH-30 translocation into the nucleus robustly in the intestinal and muscle cells both when the animals were fed on OP50 and EHEC (Figure 5, A and B), suggesting that loss of unc-89 enhanced HLH-30 activity. In order to reconfirm our RNAi results independently, we crossed unc-89(e1460) into HLH-30::GFP transgenic animals to observe the nuclear localization of HLH-30 in the unc-89(e1460) background. Similar to RNAi results, HLH-30 nuclear translocation was significantly increased in the intestinal and muscle tissues of unc-89(e1460) fed on OP50 or EHEC (Figure 5, C and D). Notably, the total HLH-30::GFP signal intensity was not upregulated in unc-89(e1460) animals (Figure 5E). Furthermore, the HLH-30-dependent genes including antimicrobial proteins (clec-7 and clec-60) as well as autophagy-related genes (lgg-1 and lgg-2) were significantly upregulated in unc-89 mutant upon EHEC infection or feeding on OP50 compared to that of WT (Figure 5F and Supplementary Figure S10). Together, these data showed that EHEC infection and unc-89 mutation (bacterial colonization) induced HLH-30 activation in C. elegans. We next examined whether the activation of HLH-30 affected the bacterial colonization in unc-89 mutants. We found that unc-89(e1460); hlh-30(tm1978) still showed enhanced Inc and bacterial colonization (Figure 5, G and H) although unc-89(e1460); hlh-30(tm1978) showed a slight reduction in EHEC colonization compared to unc-89(e1460) (Figure 5H). Together, these data indicated that hlh-30 has a minor effect on EHEC accumulation of unc-89(e1460).

Figure 5.

unc-89 mutation confers HLH-30/TFEB nuclear localization. (A) Representative images of HLH-30::GFP transgenic animals cultured on unc-89 RNAi bacteria and OP50 or EHEC for 24 h at 20°C. HLH-30 nuclear localization is indicated by white arrows. Enlarged images (top right) are from the yellow rectangular on the images. Scale bars represent 20 and 5 µm (insets). (B) Quantification of HLH-30 nuclear localization of HLH-30::GFP animals feeding on unc-89 RNAi bacteria and OP50 or EHEC. EV represents the empty vector. (C) Representative images of wild-type (WT) HLH-30::GFP transgenic animals or HLH-30::GFP transgenic animals with unc-89(e1460) feeding on OP50 and EHEC, respectively. HLH-30 nuclear localization is indicated by white arrows. Enlarged images (top right) are from the yellow rectangular on the images. Scale bars represent 20 µm and 5 µm (insets). (D) Quantification of HLH-30 nuclear localization of HLH-30::GFP wild-type and unc-89(e1460) animals feeding on OP50 and EHEC, respectively. (E) Quantification of the HLH-30::GFP signal in wild-type and unc-89(e1460) animals exposed to OP50 or EHEC at 20°C for 24 h. The fold-change of GFP expression level is relative to wild-type animals feeding on OP50. (F) qRT-PCR analysis of the expression of clec-7, clec-60, lgg-1 and lgg-2. cDNA from wild-type and unc-89 mutant animals feeding on EHEC for 24 h at 20°C were analyzed. Results were normalized to the expression level of the eft-2 control gene. Expression is relative to wild-type N2 animals. (G) The Inc percentage of wild-type (WT), hlh-30(tm1978), unc-89(e1460) and unc-89(e1460); hlh-30(tm1978). Animals were infected with EHEC-GFP for 1 day and the Inc was examined. (H) The number of bacteria colonized in wild-type (WT), hlh-30(tm1978), unc-89(e1460) and unc-89(e1460); hlh-30(tm1978) feeding with OP50-GFP or EHEC-GFP for 1 day and then chased to normal OP50 for another day. (B, D, E–H) ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001 compared to the control group by the unpaired t-test. Error bars represent SD; n, the total numbers of animals tested in each group.

unc-89 mutation-induced EHEC tolerance requires HLH-30/TFEB

Next, we aimed to dissect whether HLH-30 activation is required for the EHEC tolerance in unc-89 mutant. To address this question, we first examined whether hlh-30 is involved in avoidance behavior induced by unc-89 mutation. As shown, hlh-30(tm1978) maintained a low percentage of avoidance behavior on OP50 and EHEC, which was comparable to the N2 wild type (Figure 6, A and B). However, mutation of hlh-30 in unc-89(e1460) did not suppress the aversion behavior of unc-89(e1460) either feeding on OP50 or EHEC (Figure 6, A and B). This suggested that hlh-30 is not involved in the avoidance behavior induced by unc-89 mutation. Strikingly, when the mutant animals were placed on agar plates seeded with the small lawn of EHEC bacteria, unc-89(e1460); hlh-30(tm1978) was hypersusceptible to EHEC despite unc-89(e1460); hlh-30(tm1978) escaping from the EHEC pathogen (Figure 6C). A similar result was shown with the large lawn of EHEC; unc-89(e1460); hlh-30(tm1978) was hypersensitive to EHEC (Figure 6D). These results suggested that hlh-30 is required for protecting unc-89 mutant from EHEC killing. Notably, unc-89(e1460); hlh-30(tm1978) double mutant had a shorter survival under EHEC infection compared to hlh-30(tm1978) on both the small and large lawn of EHEC (Figure 6, C and D, P  < 0.001) and together with the data in Figure 5, G and H, unc-89(e1460); hlh-30(tm1978) displayed a high bacterial burden, suggesting that EHEC tolerance of unc-89 mutant is dependent on hlh-30.

Figure 6.

HLH-30/TFEB is required for EHEC tolerance in unc-89 mutant. (A) Analysis of the percentage of avoidance behavior of wild-type N2, unc-89(wf132), unc-89(e1460), hlh-30(tm1978) and unc-89(e1460); hlh-30(tm1978) on OP50 bacterial lawn. (B) Analysis of the percentage of avoidance behavior of wild-type N2, unc-89(wf132), unc-89(e1460), hlh-30(tm1978) and unc-89(e1460); hlh-30(tm1978) on EHEC bacterial lawn. (C) Survival curves of wild-type N2, unc-89(wf132), unc-89(e1460), hlh-30(tm1978) and unc-89(e1460); hlh-30(tm1978) mutants infected on the small lawn of EHEC bacteria. (D) Survival curves of wild-type N2, unc-89(wf132), unc-89(e1460), hlh-30(tm1978) and unc-89(e1460); hlh-30(tm1978) mutants infected on the large lawn of EHEC bacteria. (A and B) ns, not significant; ***P < 0.001 compared to the N2 control group by the unpaired t-test. Error bars represent SD. The total numbers of animals tested in each group are indicated in the column. (C and D) ns, not significant; **P < 0.01; ***P < 0.001 compared to N2 wild type by the Mantel–Cox log-rank test. Survival curves represent the sum of animals from three independent experiments.

Discussion

We isolated four EHEC tolerance alleles from forward genetic screens and mapped three of these four mutants via SNP mapping combined with WGS analysis. The causative gene underlying EHEC tolerance was found to be unc-89. Mutation of unc-89 resulted in enhanced bacterial colonization which triggered hlh-30-dependent immune responses to tolerate EHEC infection in C. elegans (Figure 7).

Figure 7.

Schematic illustrating HLH-30-mediated tolerance to EHEC infection in C. elegans. In wild-type C. elegans, EHEC colonized in the intestine and slightly induced HLH-30 nuclear localization in C. elegans. By contrast, in the mutants of defective pharynx (e.g., unc-89 mutant), EHEC accumulated in C. elegans intestine rapidly. When facing this challenge, C. elegans selects to tolerate the EHEC infection instead of clearing EHEC bacteria in intestine through HLH-30 nuclear translocation activating the hlh-30-mediated immune responses.

One of our unc-89 alleles, wf101 is a mutation containing a 2 guanine deletion on the intron (6408–6409) between exon 5 and 6. Although the WGS analysis did not show that this deletion directly affects the splicing site of unc-89, we found a predicted donor splicing site at the 5' end of the intron from 6427 to 6428 on the negative strand of unc-89 (predicted by http://www.cbs.dtu.dk/services/NetGene2/), which is close to our wf101 mutation, implying RNA splicing might be affected. Moreover, our SNP results showed that wf101 mutation was on chromosome I (Supplementary Figure S5) and complementation test indicated that wf132 and wf101 were in the same group (Table 3). Together, these results suggested that wf101 is an unc-89 mutation.

We found that unc-89(wf132) and unc-89(e1460) cultured on the large lawn of OP50 showed slightly shorter lifespans (Figure 3F) but had an extended lifespan on the small lawn (Figure 3E). One explanation for this result could be that OP50 is a weak pathogen to C. elegans (Chou et al. 2013; Win et al. 2013; Revtovich et al. 2019) and avoidance behavior toward OP50 leads to dietary restriction in C. elegans (Kumar et al. 2019). Therefore, it is reasonable that these enhanced bacterial accumulation mutants had shorter lifespans on the large lawn of OP50 plates and prolonged lifespans on the small lawn of OP50.

tph-1 encodes a gene involved in serotonin biosynthesis named tryptophan hydrolase that mediates bacterial avoidance behavior in C. elegans (Melo and Ruvkun 2012). Since unc-89 and tph-1 mutations alter the behavior of C. elegans, we performed Inc analysis and CFU assay by plating animals on the large lawn to force bacterial exposure and minimize the possibility of behavioral changes that might influence our experiments. Although tph-1(mg280) did not show aversion behavior (Figure 4, C and D), tph-1(mg280) cultured on the large lawn showed compatible Inc percentage and bacteria burden to that of the wild type (Figure 4, A and B). These data suggested that EHEC bacteria are not adequate to colonize in tph-1(mg280) with 1 day feeding. Furthermore, unc-89(wf132); tph-1(mg280) displayed lower Inc and CFU than unc-89(wf132) (Figure 4, A and B) even though the robust lawn-leaving behavior was abolished by lack of tph-1 (Figure 4, C and D), indicating that the bacteria entering and accumulating in the intestine is mainly due to the pharyngeal defect. We noticed that the Inc percentage of tph-1(mg280) did not show a high correlation to the CFU assay. This could be due to the limitation of detection and quantification from diverse experimental methods. Nevertheless, the percentage of worms with GFP and EHEC accumulation in tph-1(mg280) were both similar to wild type (Figure 4, A and B).

The pharyngeal grinder dysfunction in phm-2 mutant causes increased E. coli. OP50 bacterial colonization activating innate immunity pathways and inducing aversion behavior that led to dietary restriction and extended lifespan (Kumar et al. 2019). unc-89, also known as phm-1, is expressed in the C. elegans pharynx muscle as well as the body wall and intestine, and unc-89 mutation also contributes to dysfunction of the pharynx (Avery 1993). We found that unc-89 mutant enhanced EHEC bacterial colonization activating aversion behavior and hlh-30 nuclear localization. However, increased survival upon EHEC infection (EHEC tolerance) of unc-89 alleles did not rely on the avoidance behavior but heavily depended on the hlh-30 function. Although EHEC induces HLH-30 nuclear translocation, the function of hlh-30 seems less crucial for defending EHEC in wild-type C. elegans since the survival of hlh-30 mutant infected with EHEC is comparable to that of WT. It could be interpreted that hlh-30-dependent immune responses are relatively pivotal for C. elegans tolerance of EHEC while bacteria rapidly and massively colonize in the host (e.g., pharyngeal defect mutants, Figure 3A). Therefore, hlh-30 loss of function renders unc-89 mutant significantly hypersusceptible to EHEC (Figure 6, C and D).

The helix-loop-helix transcription factor HLH-30/TFEB, is a key regulator of lysosome biogenesis and autophagy required for lifespan extension in a multiple longevity model in C. elegans (Lapierre et al.2013). Moreover, HLH-30/TFEB mediates innate immune response including antibacterial and autophagy gene expression to promote organism fitness during infection in C. elegans (Visvikis et al. 2014; Chen et al. 2017). Nuclear localization of HLH-30/TFEB is controlled by phosphorylation. LET-363/mTOR directly phosphorylates and retains HLH-30/TFEB in the cytosol inhibiting HLH-30/TFEB nuclear translocation (Martina et al. 2012). By contrast, dephosphorylated HLH-30/TFEB is imported into the nucleus to induce the transcription of target genes. UNC-89 contains two Ser/Thr kinase motifs at the C-terminus (Figure 2C). We found that HLH-30/TFEB nuclear translocation was significantly enhanced in the unc-89 mutant (Figure 5, A–D). It may be interesting to test whether UNC-89 directly or indirectly regulates HLH-30 nuclear translocation.

Autophagy is a cellular process that can degrade large endogenous material (e.g., damaged organelles) and exogenous materials such as intracellular pathogens, a process which is called ‘xenophagy’ (Kuo et al. 2018a). Upon Staphylococcus aureus infection, C. elegans triggers hlh-30 activation to induce autophagy, which acts as a mechanism of tolerance to infection (Visvikis et al. 2014). Moreover, the autophagic cycle is able to sequester and degrade bacterial pore-forming toxins (PFT), Cry5B produced by the extracellular gram-positive bacterium Bacillus thuringiensis, to protect C. elegans from PFT killing (Chen et al. 2017). Together, these studies suggested that instead of targeting bacteria per se, autophagy may target and eliminate bacterial products resulting in tolerance to infection. Several toxins have been reported to be secreted by EHEC to damage host tissues e.g., Shiga toxin (Stxs). Our previous study showed that Shiga toxin 1 is crucial for the full pathogenicity of EHEC toward C. elegans (Chou et al. 2013). Given that EHEC is not an invasive pathogen, one potential interpretation of increased tolerance in unc-89 mutant could be that it is due to enhanced autophagic activation that degrades toxins produced by EHEC instead of the bacterium itself leading to higher pathogen load but extended C. elegans survival.

Another alternative explanation for unc-89 mutant tolerating EHEC infection is that instead of targeting the deleterious substances produced by EHEC, the animal hosts may alleviate or repair the damage caused by EHEC. C. elegans hosts can sense the physiological consequences of pathogenic assaults and upregulate the immune signaling to respond to the attack (also known as effector-triggered immunity, ETI) (Dunbar et al. 2012; McEwan et al. 2012). Exotoxin A secreted from Pseudomonas aeruginosa inhibits host translation and this translational inhibition triggers C. elegans immune gene expression through the activation of ZIP-2, ATF-7 and CEBP-2 transcription factors to protect nematodes from P. aeruginosa killing (Dunbar et al. 2012; McEwan et al. 2012; Reddy et al. 2016). EHEC produces Stxs which also block the host protein translation by targeting the 28S rRNA and induce the ribotoxic response to hosts (Mohawk and O'Brien 2011). Disruption of core cellular activities (e.g., translation, respiration, and protein turnover) stimulates behavioral avoidance and expression of detoxification as well as innate immune effectors in C. elegans (Melo and Ruvkun 2012). Furthermore, the autophagic cycle pathway has been reported to be involved in the process of repairing the membrane pore induced by PFT, Cry5B (Chen et al. 2017). Therefore, it is possible that unc-89 mutant showed enhanced EHEC tolerance due to amelioration of worms’ health through diminishing physiological damage induced by immunopathology or pathogen insults.

In light of these results, the question arises as to why C. elegans operates immune responses tolerating EHEC rather than clearing the pathogenic microbes. Although hlh-30 regulates antimicrobial peptides (AMPs) that could assist nematodes in eliminating pathogen loads (Visvikis et al. 2014) when facing EHEC, the LPS architecture might hinder the AMPs targeting the bacterial cells (Kuo et al. 2016) constraining the function of AMPs. In mutant animals showing abnormal pharynx grinder morphology, such as unc-89 and phm-2, live bacteria can enter the intestinal lumen without being crushed by the pharynx implying that the undamaged LPS architecture of EHEC bacteria remains. Also, due to a massive number of microbes rapidly accumulating in the unc-89 and phm-2 mutants (Figure 3A), animal hosts might not be able to remove the bacterial burden in a timely manner but select a strategy for neutralizing the detrimental substances from the pathogen or repairing the damage to improve the organism fitness.