Supplementary Table 1 contains a list of all C. elegans and bacterial strains used in this study. We used the C. elegans NL2099 rrf-3(pk1426) II in our assays for its increased sensitivity to RNAi (Simmer et al., 2002), but otherwise wild type behavior and life traits. Worms were maintained on nematode growth medium (NGM) agar plates at 15°C and fed with bacterial lawns of E. coli OP50 as described (Stiernagle, 2006). For experiments, C. elegans were transferred to fresh 15 cm diameter NGM plates inoculated with 3 mL E. coli OP50 and 1 mL concentrated E. coli OP50 and grown at 20°C until a sufficiently large population was obtained. Gravid C. elegans populations were washed from plates with M9 buffer (KH₂PO₄ (22 mM), Na₂HPO₄ x 2 H₂O (33.7 mM), NaCl (85.6 mM), supplemented with 1 mL/L MgSO₄ (1 M) after autoclaving) and synchronized by bleaching: 8 mL washed worms/eggs in M9 were mixed with 2 mL of a 1:1 solution of NaOH (4 M) and NaClO (12%) for 7 min followed by vortexing and inverting. Bleaching was stopped by centrifuging for 2 min at 1,550 rpm, supernatant was discarded, and pellet was washed two times with 10 mL M9 buffer and 2 times with 10 mL autoclaved MiliQ water. Only eggs survived the treatment. An extra synchronization step included incubation of bleached eggs in uninoculated 15 cm NGM plates at 20°C for 24 h until all eggs had hatched. Synchronized L1 larval stage C. elegans were subsequently used for experiments.

E. coli OP50 was grown at 37°C with shaking overnight in OP50 liquid medium (5 g Tryptone and 2.5 g of Yeast Extract per Liter of MiliQ water). Concentrated E. coli OP50 cultures for worm feeding were prepared by inoculating 1 L of Luria-Bertani (LB) broth (Invitrogen) with 10 mL of overnight E. coli OP50 culture and grown for 4 h (or until saturated) at 37°C with shaking. Saturated cultures were centrifuged at 3,500 rpm for 15 min at room temperature, supernatants were discarded and pellets of 6 liters of cultures were collected in 9 mL OP50 medium. The resulting concentrated E. coli OP50 culture was used directly to inoculate NGM plates. Enterococcus faecalis OG1RF was streaked from frozen stocks on Brain Heart Infusion (BHI) (Sigma-Aldrich) agar plates and grown at 37°C overnight. Liquid cultures were grown on BHI broth for 3 h at 37°C and used to inoculate Nematode Growth BHI plates (NGBHI) (BHI agar plates supplemented with 2 g of agar per liter to match the agar concentration of NGM). Matching E. coli OP50 controls were grown in the same way. Pseudomonas aeruginosa PA14, Staphylococcus aureus 6538 and Bacillus subtilis subsp. subtilis 168 were maintained on LB agar and broth and cultured at 37°C. Comamonas sp. B-9 MYb021 was maintained on LB agar and broth and cultured at 25°C.

To study the transcriptional response of C. elegans to microbial exposure we collected published RNAseq datasets that report differential gene expression in C. elegans when exposed to different microorganisms as a Fold Change (FC) (which were converted to Log₂FC) or Log₂FC and report an adjusted p-value. To find these publications we used the online search engine Research Rabbit,¹ which was developed to help find across the web additional publications like the publications on a user-made preselected list. The detailed list of datasets collected is summarized in Supplementary Table 2 and the collected data of differential gene expression and adjusted p-values can be found in Supplementary Table 4 reported as Log2FC and adjusted p-values. Further analyses and graphing were performed with RStudio² using the C. elegans gene functional annotation available from WormCAT (Holdorf et al., 2020). The code to produce bubble plots of lists of genes of interest with our collection of datasets is available on GitHub. As published datasets were not independently re-analyzed from raw data, or re-scaled, datasets from different studies cannot be compared quantitatively, only qualitatively.

HHLS paralogs in nematodes were identified in Wormbase Parasite by performing a cDNA TBLASTN search for distant homologs to C. elegans WRT-1, WRT-8, GRD-1, GRL-1, GRL-7 (ligands) and PTR-1, PTR-18, PTC-1, CHE-14 (receptors) followed by species per species paralog search (numbers of paralogs may be underestimated). Percentage identities of sequences were obtained using BLASTP or TBLASTN of C. elegans proteins against target species databases. For phylogenetic analyses amino acid sequences of C. elegans Hedgehog-like (HHL) pathway genes and modulators and human sonic hedgehog (SHH), patched (PTC1) and dispatched (DISP1) were downloaded using QIAGEN CLC Main Workbench 20.0.³ Sequences were aligned using the default parameters of ClustalW. Phylogenetic trees were constructed using the Neighbor-Joining method with 1,000 Bootstrap repetitions assuming the Jukes-Cantor amino acid evolution model.

Worms were grown at 20°C on 15 cm NGM plates inoculated with E. coli OP50 and concentrated E. coli OP50 until the population size was large enough for the desired experiments. The parental worm population was bleached as described above. To test the functional involvement of HHL pathway genes in resistance to stress and infection we performed target gene knockdown by feeding C. elegans NL2009 L1 larvae with the RNAi clones form the Source Bioscience Ahringer library (Kamath et al., 2003) listed in Supplementary Table 1. RNAi was performed in a 96-well plate array of NGM supplemented with 50 μg/mL Carbenicillin and 2 mM Isopropyl-β-D-thiogalactoside (IPTG) (Sigma-Aldrich) inoculated with concentrated RNAi bacteria. RNAi bacteria were cultured overnight at 37°C with shaking (200 rpm) in 96-deep well plates with 1.5 mL Terrific Broth [Tryptone (12 g), yeast extract (24 g), glycerol (100% 4 mL) in 900 mL MiliQ H₂O, autoclaved and then mixed with 17 mL KH₂PO₄ (1 M) and 72 mL K₂HPO₄ (1 M)], supplemented with 100μg/mL Ampicillin and 0.2 mM IPTG. Over 100 C. elegans larvae per well were cultured on RNAi bacteria at 20°C until reaching L4 larval stage. Subsequently they were transferred to 25°C for 24 h until they reached young adult stage and were able to take up live bacteria in their guts.

For the stress assays RNAi-fed worms were transferred to further NGM 96-well plates inoculated with 15 μL of a concentrated overnight liquid culture of a G + (Bacillus subtilis subsp. Subtilis 168) and a G- (Comamonas sp. B-9 MYb021) C. elegans commensal microbe, for 24 h at 25°C. Stress and infection assays were performed by pipetting 30 μL of M9 buffer to each well of the 96-well plates and collecting 20 μL of worms per well, which were pipetted to individual wells of a 384 black well plate with clear bottom. For stress assays each sample was done in duplicate with three independent runs and for infection assays each sample was done in tetraplicate with three independent runs and no exposure to commensal bacteria. For the heat sock assays M9 was added to a final volume of 67 μL per well and worms were treated at 42°C for 12 h. For the oxidative stress assays 40 μL of M9 and 7 μL of tert-Butyl hydroperoxide (TBHP) 70% in H₂O (Sigma-Aldrich) were added to a final volume of 67 μL and final concentration of 7.3% TBHP per well and worms were treated at 25°C for 12 h. For the infection assays RNAi-fed worms were kept at 25°C for a total of 36 h until they reached day one of adulthood and then pipetted to 384-well plates. For infections, E. faecalis OG1RF was grown for 3 h in 5 mL BHI broth, P. aeruginosa PA14 and S. aureus 6538 were grown overnight in 5 mL LB broth at 37°C with shaking. To the 384-well plate 20 μL of M9, 20 μL of OP50 medium and 7 μL of liquid pathogen culture were added to a final volume of 67 μL per well and worms were treated at 25°C for 60 h.

Median time of death was determined by Label-Free Automated Survival Scoring (LFASS) analysis (Benedetto et al., 2019) of death fluorescence recordings (Exc./Em.: 365 nm/430 nm) in Spark, Infinite Pro M200, or MNano + plate-readers (Tecan). Data were processed in MATLAB 2020a (MathWorks), exported to Excel 365 (Microsoft Office), and processed for graph plotting and statistical analyses in Prism 9.3 (GraphPad). Two-way ANOVA were performed with post hoc Tukey correction for multiple comparisons.

Our literature search identified 20 published accessible transcriptomic datasets on wild type C. elegans exposed to microbes, to which we added an unpublished RNAseq dataset of C. elegans exposed to E. faecalis OG1RF generated in our lab (publication in preparation). This collection includes differential gene expression data for C. elegans exposed to 20 different microorganisms including pathogenic and non-pathogenic strains of Gram-positive (G +) bacteria, Gram-negative (G-) bacteria, yeast, fungi, microsporidia, an oomycete, and Orsay virus (Figure 1A and Supplementary Table 2). For each different microbe we extracted a list of unique significantly differentially expressed (DE) genes (up- and down-regulated) combining all available time points and datasets, for which we performed enrichment analysis using the online tool DAVID (Figure 1A and Supplementary Table 3).

For both pathogenic and commensal microbe exposures, DE genes were enriched in genes whose function contains the keyword “signal” (Figure 1A). Genes annotated with this keyword notably include those coding for the nematode divergent and expanded families of C-type-lectin-like domain-containing proteins (CLECs) and the Hedgehog-like signaling (HHLS) pathway. DE gene lists were also enriched in genes relating to defense against stress or pathogens, innate immune function, xenobiotic metabolism (Cytochrome p450), structural proteins (collagens), and enzymes (proteases, oxidation-reduction processes, metabolism) (Figure 1A and Supplementary Table 3). Overlaps between all gene lists are shown in Supplementary Table 5. Ranking of individual genes based on the number of datasets in which there were found differentially expressed (considering each condition as a different dataset, Supplementary Figure 2 and Supplementary Table 5) confirmed the importance of lipid metabolism in host-microbe interactions. The top two genes, found differentially expressed in 35 out of 56 datasets, were the Acyl CoA Dehydrogenase acdh-1, involved in beta-oxidation (lipid metabolism), and the folate transporter family member folt-2, involved in transmembrane transport (Supplementary Figure 2 and Supplementary Table 5). While 19 out of 75 DE genes found in 26/56 datasets were expectedly associated with “stress response: Pathogens,” genes associated with lipid metabolism were the next most highly represented category (11/76, Supplementary Table 5).

For easier follow-up of the genes highlighted by our enrichment analyses, we narrowed further analyses to: (1) Candidates gene families with the potential of modulating specific interactions with microbes, (2) experimental models that are amenable to high-throughput testing. For this, we focused on genes that are DE in the context of three well-documented gut bacterial infection models (G + and G- bacteria) that are readily available, where a breadth of resources already exist, and for which high-throughput infection assays with convenient readouts have been established: Pseudomonas aeruginosa PA14 (Mahajan-Miklos et al., 1999), Staphylococcus aureus 6538 (Sifri et al., 2003), and Enterococcus faecalis OG1RF (Garsin et al., 2001). DE gene lists for these three microbes, E. faecalis OG1RF (5,951 genes), P. aeruginosa PA14 (2,715 genes), and S. aureus 6538 (1,117 genes combined from datasets from S. aureus strains RN6390 and CECT8145) highlighted a small subset of 73 shared genes predominantly belonging to the functional grouping “fatty acid metabolism” (Figure 1B and Supplementary Tables 3, 5). Collectively the results of these enrichment analyses and summary counts of gene expression data show that new microbial exposure in C. elegans broadly leads to modulation of signal molecules, immune and metabolic (lipids) genes, in line with previously published studies.

To highlight specific molecular pathways, we next characterized how the expression of genes changes in those categories during microbial exposure across all 56 datasets. From the published literature, we extracted genes of interest involved in C. elegans immune signaling and regulation (Pukkila-Worley and Ausubel, 2012; Martineau et al., 2021) and lipid metabolism (Watts and Ristow, 2017). We then used these lists to query the collected transcriptomic datasets (Supplementary Table 4) and produced bubble plots representing differential gene expression (Log₂FC) and adjusted p-values (FDR) for the relevant datasets (Figure 2). Immune signaling and modulation genes were generally upregulated across a wide variety of microbes with some exceptions, including the putative cAMP-dependent transcription factor atf-5 and the predicted signal receptor tol-1. Interestingly, the robust upregulation of the nuclear hormone receptor nhr-112 across a variety of datasets suggests a significant role for this transcription factor in interactions with G + and G-negative pathogens and commensals alike, which had not been reported. Conversely, several known defense and stress response modulators, such as the FOXO transcription factor daf-16, failed to exhibit significant changes in gene expression in either of the datasets. Despite the established role of DAF-16 in immune modulation, this was not unexpected as modulation of DAF-16 activity mostly occurs post-translationally, via regulated shuttling between the cytosol and the nucleus (Lee et al., 2001; Figure 2A).

By contrast with known conserved immune regulators, lipid metabolism genes were more extensively differentially expressed following microbial exposure (Figure 2B). Genes involved in de novo fatty acid metabolism (pod-2, fasn-1, fat-1-6, elo-1-7, pmt-1-2) or lipid catabolism (acox, maoc-1, dhs-28 and daf-22) (Watts and Ristow, 2017) were broadly down-regulated following exposure to the G + pathogens B. thuringiensis, E. faecalis and S. aureus, and the G- pathogen P. aeruginosa (Figure 2B).

Interactions between fatty acid metabolism and innate immunity (Anderson and Pukkila-Worley, 2020) or between fatty acid metabolism and Hedgehog-like signaling (Lin and Wang, 2017; Del Castillo et al., 2021) have been frequently reported in animals, including in C. elegans. Increased C. elegans fatty acid metabolism is also strongly associated with reproduction (Ezcurra et al., 2018), while a trade-off between reproduction and the mounting of effective innate immune responses has been previously established (TeKippe and Aballay, 2010). That DE fatty acid metabolism genes are mostly down-regulated upon pathogen exposure (Figure 2B), while DE innate immunity genes are up-regulated (Figure 2A) is consistent with that. However, the core immunity and lipid metabolism genes identified above represent a small fraction of the DE genes for any given microbe (Figure 1B), highlighting the fact that the transcriptional modulation of most DE genes is microbe specific. We wondered whether such specificity might be underpinned by the finely tuned expression of specific classes of signaling peptides and cognate receptors families such as the expanded HHLS network.

The top hits of our gene ontology analysis related to “Signal,” which included a high proportion of HRPs. The remarkable non-redundant expansion of the HHLS pathway ligands and Patched-related receptor (PTRs) families in nematodes (Zugasti et al., 2005; Burglin, 2008; Baker et al., 2021; Supplementary Figure 1) and the conserved and intricate roles of Hedgehog pathway genes in intercellular signaling throughout metazoa (Ingham et al., 2011) make the non-canonical (Jenkins, 2009) nematode HHLS pathway a prime candidate for enabling specificity in host-microbe interactions. We thus investigated to which extent HHLS genes and known regulators were differentially expressed in worms exposed to a diversity of microbes, and if downregulation of these genes could disrupt interactions between the worm host and cognate microbes.

Surveying the datasets collected (Figure 3A), out of 61 C. elegans HRPs from the GRL/GRD/WRT/QUA/HOG families 31 were found differentially expressed (27 up-regulated) on B. subtilis, 29 (27 up-regulated) on P. luminescens, 25 (22 up-regulated) on Serratia marcescens, 21 (20 down-regulated) on B. thuringiensis, 18 (14 down-regulated) on Comamonas sp., 13 (11 up-regulated) on E. faecalis, 9 (8 up-regulated) on S. flexneri, 5 (3 up-regulated) on S. malitophila, 4 (3 down-regulated) on P. aeruginosa, and 4 (3 down-regulated) on Staphylococcus aureus. PTR expression followed the same trend as HRP expression (where HRP genes were mostly down-regulated, PTR genes were mainly down-regulated as well), suggesting microbe-specific purposeful activation or inhibition of HHLS altogether. G + and G- commensals and pathogen types elicited differential expression of HHLS pathway genes, with no obvious type-specific pattern beyond the fact that G + bacteria seemed to elicit more obvious shifts in HHLS gene expression in this dataset. Most changes appear to be species or condition specific. HHLS genes were also differentially expressed following exposure to pathogenic fungi such as D. coniospora (26 DE, 19 down-regulated) and Harposporrium sp. (19 DE, 11 down-regulated) and microsporidium Nematocida parisii (6 DE, 3 down-regulated), but not after viral infection by the Orsay virus (no hits—not represented in Figure 3A). Although most HHLS pathway genes did not follow that pattern, some sequence-related genes were co-regulated upon exposure to multiple microbes (Figure 3A, grd-3/grd-4, qua-1/wrt-6/wrt-4, wrt-1/wrt-10, ptr-16, and ptr-18), in line with their tissue expression patterns (Kaletsky et al., 2018; Supplementary Figure 3), which might indicate functional redundancy.

As changes in expression levels may not translate into functional changes, we next targeted HHLS pathway genes one by one from the late L1 larval stage using RNAi-expressing clones from the Ahringer library (Kamath et al., 2003) in the RNAi-sensitized strain rrf-3(pk1426)(Simmer et al., 2002), and assessed the impact of this treatment on subsequent resistance to infections by three model gut pathogens: P. aeruginosa PA14 (Mahajan-Miklos et al., 1999), S. aureus 6538 (Sifri et al., 2003), and E. faecalis OG1RF (Garsin et al., 2001; Figures 3B, 4A and Supplementary Table 6). Amongst the 73 DE genes common to all three pathogen infections, only ncr-1 [homolog of the vertebrate gene Niemann-Pick disease, type C1 (NPC1)], a sterol-sensing domain-containing protein involved in cholesterol trafficking (Patel et al., 2008) can be associated with HHLS (Figure 3B), as most gene expression changes in the HHLS pathway seem pathogen-specific. This may not be surprising when considering that together with ptr-5, ncr-1 is the most ubiquitously expressed SSD-protein in C. elegans tissues (Supplementary Figure 3). Our infection assays also indicated that RNAi against individual HHLS genes and genetic interactors differentially impacts C. elegans resistance to pathogens (Supplementary Figure 4) but revealed broader trends. Hence, inhibition of HHLS genes was generally found associated with reduced resistance to the G- pathogen P. aeruginosa, while the picture is more contrasted for the G + pathogens E. faecalis and S. aureus (Figure 3B).

Compared to S. aureus (4) and P. aeruginosa (12), worms exposed to E. faecalis exhibited more (20/87) differentially expressed HHLS ligands and receptors, most of them up-regulated (17/20). Interestingly, gene per gene comparison of expression data and RNAi-associated pathogen resistance reveals that several of these up-regulated genes (9/16 tested) are necessary for wild-type level resistance to E. faecalis infection, as RNAi against these led to increased worm susceptibility (Figure 3B). Among them are HRPs: grl-21 (previously identified as a mediator of environment-host-microbe crosstalk; Lin and Wang, 2017), grl-5, grl-7, grd-12, grd-2, and wrt-4, and PTRs: ptr-18 (required for HRP clearance; Chiyoda et al., 2021) and daf-6 (necessary for sensory organ morphogenesis together with dyf-4 (Hong et al., 2021), and thus possibly for pathogen sensing), as well as another SSD-protein expressing gene: scp-1. Similarly, RNAi inhibition of the P. aeruginosa-up-regulated ncr-1 and C27B7.7 (DSCAM) led to increased host sensitivity to infection, implicating both genes in the worm immune response to P. aeruginosa gut infections, while RNAi inhibition of ncr-1 (only gene found up-regulated in this context) did not sensitize worms to S. aureus infection. Conversely, the RNAi inhibition of several genes found downregulated upon exposure to pathogens led to increased resistance to infection. These genes include ptr-2 and the MAPKKK mom-4 and Frizzled ortholog mom-5 in P. aeruginosa infection; ptr-8 in E. faecalis infection and grl-4 and grd-14 in S. aureus infection (Figure 3B).

Across all three infection models, more RNAi tested had an adverse effect on host resistance to infection than there were up-regulated genes. The role of corresponding non-DE genes in pathogen resistance might be indirect, or regulation of their activity may primarily occur post-transcriptionally. This is expected as HRPs and PTRs activities in flies and mammals strongly rely on post-translational modifications and regulated intracellular trafficking (Carballo et al., 2018). Nevertheless, the widespread microbe-specific impact of HHLS gene inactivation on pathogen resistance suggests an active role for this pathway in response to pathogens, which may be pathogen-specific.

If the worm HHLS pathway is engaged in pathogen-specific responses, it may also be involved in the regulation of commensal-host interactions. We saw earlier (Figure 3A) that C. elegans exposure to commensal bacterium B. subtilis or to P. luminescens (which is a natural gut commensal of entomophagous nematodes; Ogier et al., 2020) leads to the up-regulation of many HHLS genes, while another commensal—Comamonas sp., leads to the down-regulation of most differentially expressed HHLS genes. We thus wondered whether these opposite effects on HRP and PTR gene expression levels are adaptive, enabling worms to tune Hedgehog-like signaling accordingly to the commensal they interact with. To test this idea, we grew worms from late L1 larval stage onto RNAi-producing dietary E. coli bacteria targeting HHLS genes, then transferred them as day 1 adults onto B. subtilis and Comamonas sp. (MYb21) isolates for 24 h, before challenging them with oxidative and thermal stresses (Figure 4B). Our expectations were that if a HHLS gene up-regulation was adaptative to a specific commensal, its down-regulation by RNAi would disrupt the relationship C. elegans normally establishes with that commensal, which would be revealed by increased frailty and sensitivity to oxidative or heat stress.

Analysis of stress assay data (Supplementary Figure 5) first revealed that oxidative stress (7% t-BHP) resistance is generally little affected by RNAi pre-treatment, while host heat-stress resistance is strongly impacted by both the RNAi applied and the nature of the microbial isolate considered. This may be related to the fact that heat-shock resistance appears to be a better proxy for adult worm health than oxidative stress resistance (Benedetto et al., 2019). Yet, RNAi toward individual HHLS genes differentially affected oxidative and heat stress resistance in a microbe-specific manner, with worms grown on different microbes exhibit strikingly different RNAi-associated stress resistance patterns. Worms exposed to B. subtilis displayed a wide-ranging up-regulation of HHLS genes, which was mirrored almost perfectly gene-by-gene by a negative impact of corresponding RNAi on heat resistance (Figure 3C). Indeed, RNAi down-regulation of 36 genes found up-regulated in worms grown on B. subtilis adversely affected the ability of worms to tolerate heat, while only 7 did not follow this rule (6 of which were found down-regulated in transcriptomics datasets and their down-regulation by RNAi did not translate into increased heat resistance). Conversely, for worms exposed to Comamonas sp. MYb21, which mostly experienced a down-regulation of HHLS genes, RNAi against these genes was as often beneficial as detrimental (9 and 8 genes, respectively). The impression emerging from these data is that while up-regulation of HHLS genes may be indicative of a “purposeful” adaptive response to a commensal, down-regulation of these genes may not be.