Collagen and actin network mediate antiviral immunity against Orsay in C. elegans intestinal cells

C. elegans is a free-living nematode that is widely used as a small animal model for studying fundamental biological processes and disease mechanisms. Since the discovery of the Orsay virus in 2011, C. elegans also holds the promise of dissecting virus-host interaction networks and innate antiviral immunity pathways in an intact animal. Orsay primarily targets the worm intestine, causing enlarged intestinal lumen as well as visible changes to infected cells such as liquefaction of cytoplasm and rearrangement of the terminal web. Previous studies of Orsay identified that C. elegans is able to mount antiviral responses by DRH-1/RIG-I mediated RNA interference and Intracellular Pathogen Response, a uridylyltransferase that destabilizes viral RNAs by 3′ end uridylation, and ubiquitin protein modifications and turnover. To comprehensively search for novel antiviral pathways in C. elegans, we performed genome-wide RNAi screens by bacterial feeding using existing bacterial RNAi libraries covering 94% of the entire genome. Out of the 106 antiviral genes identified, we investigated those in three new pathways: collagens, actin remodelers, and epigenetic regulators. By characterizing Orsay infection in RNAi and mutant worms, our results indicate that collagens likely form a physical barrier in intestine cells to inhibit viral infection by preventing Orsay entry. Furthermore, evidence suggests that the intestinal actin (act-5), which is regulated by actin remodeling proteins (unc-34, wve-1 and wsp-1), a Rho GTPase (cdc-42) and chromatin remodelers (nurf-1 and isw-1), also provides antiviral immunity against Orsay possibly through another physical barrier presented as the terminal web.


INTRODUCTION
4 microbial infections. The intestine of C. elegans is comprised of 20 large epithelial cells that are positioned as bilaterally pairs forming a long tube with a central lumen . These intestinal cells exhibits molecular polarity similar to mammalian intestines in terms of extracellular matrix and apical membrane surface with underlying terminal web (9). Importantly, genomic comparison of humans and C. elegans demonstrated that the majority of human disease genes and disease pathways are present in the worm (10). Among the available 18,452 C. elegans protein sequences, at least 83% has homologous genes in human (11). Many genes involved in defense against bacterial pathogens in C. elegans also function in humans as host defense genes (12).
Studies of Orsay infection in the past few years have uncovered several conserved antiviral defense mechanisms. These include the RNA interference pathway that degrades viral RNA and triggers antiviral gene expression (1,13,14); an inductive transcriptional response called Intracelluar Pathogen Response (15); a uridylyltransferase that destabilizes viral RNAs by 3′ end uridylation (16); and ubiquitin protein modifications and turnover (17,18). In another study, bioinformatics analyses of expressed genes during Orsay infection revealed the presence of uncharacterized anti-stress pathways (19). In addition to antiviral genes, a forward genetic screen implicated several genes and their human orthologs in endocytosis (i.e. SID-3 and WASP) to be required for viral infection (20). In spite of the progress made, the fact that only a few above-mentioned pathways have been linked to antiviral immunity shows that our understanding of the antiviral responses of C. elegans is still limited.
To comprehensively search for novel antiviral pathways in C. elegans, we performed genome-wide RNAi screens by bacterial feeding using existing bacterial RNAi libraries covering 94% of the entire genome. We identified at least 106 antiviral genes, including those belonging to known antiviral mechanisms as well as those having not been associated with C. elegans antiviral immunity. We further investigated those in three new pathways: collagens, actin remodelers, and epigenetic regulators. By characterizing Orsay infection in RNAi and mutant worms, our results indicate that collagens, including col-51, col-61, col-92, cutl-21, and sqt-2, likely form a physical barrier in intestine cells to inhibit viral infection by preventing Orsay entry. Additionally, evidence suggests that the intestinal actin (act-5), which is regulated by actin remodeling proteins (unc-34, wve-1 and wsp-1), a Rho GTPase (cdc-42) and chromatin . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 20, 2023. ; https://doi.org/10.1101/2023.04.20.537671 doi: bioRxiv preprint remodelers (nurf-1 and isw-1), also provides antiviral immunity against Orsay possibly through another physical barrier presented as the terminal web.

A genome-wide RNAi screen for antiviral genes
We designed a genome-wide RNAi screen to discover which genes are required for C. elegans antiviral immunity. We started with a strain that is resistant to the Orsay virus. Starting Figure 1. Genome-wide RNAi screen for antiviral genes. A) Schematic drawing of the design of the antiviral gene screen. A strain resistant to Orsay infection was used. We screened for RNAi that made the strain sensitive to Orsay infection. Top right image, the transparent anterior intestine phenotype was scored as infection symptom. Scale bar, 100μm. Uninfected RNAi worms were examined to ensure that the transparent intestine phenotype was caused by virus infection and not just RNAi. B) Percentage of worms with transparent intestines after RNAi of antivirus genes. rde-1 and mock RNAi were positive and negative controls, respectively. C) Fractions of worm antivirus genes with and without human orthologous genes. . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 20, 2023. ; https://doi.org/10.1101/2023.04.20.537671 doi: bioRxiv preprint from the first larval stage (L1), these worms were fed with RNAi bacteria to inactivate various worm genes, and infected with Orsay. We examined the severity of infection when the worms were day-3 adults (Fig. 1A). The worms also had the temperature-sensitive mutation glp-4(bn2ts) so that they were sterile at 20°C, allowing us to grow them to day-3 adults without the interference from any progeny. The rationale of the screen was that without RNAi, worms would be resistant to Orsay infection and asymptomatic; RNAi inactivation of an antiviral gene would make the worms sensitive to viral infection and display the infection symptom of transparent intestine (Fig. 1A). To ensure that the transparent intestine phenotype was specific to viral infection and not caused by RNAi, we also examined RNAi worms without viral infection to ensure that these worms did not have the transparency phenotype (Fig. 1A).
Bacteria containing the empty vector were used as the negative control. As expected, very few worms (< 2%) in the negative control group showed the transparency phenotype with or without the virus (Fig. 1B). rde-1(RNAi) bacteria were used as the positive control. In this group, 77.9% of infected rde-1(RNAi) worms showed the transparency phenotype while only 2.8% of uninfected rde-1(RNAi) worms did, showing a difference of 75.1% (Fig. 1B). The difference between the percentage of transparent worms in the virus-infected group and the uninfected group was used to select hit genes, and a threshold of 20% was applied. Including the positive control rde-1, a total of 106 genes were identified as hits (Fig. 1B, Table S1). RNAi of these genes reproducibly displayed elevated levels of transparent worms upon viral infection, suggesting that these genes are required for antiviral immunity.
Most of these antiviral genes have human orthologs. 69 out of these 106 antiviral genes (65%) have orthologous genes in human (WormBase W277, Fig. 1C). Since Orsay infects intestine cells, we examined how many of these orthologous genes were expressed in the human digestive system. Almost all (68/69) genes with human orthologs have orthologs expressed in the human digestive system (Fig. 1C).

Collagens mediate antiviral innate immunity
. CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made RNAi of five collagens, col-51, col-61, col-92, cutl-21, and sqt-2, significantly increased the number of worms with the symptom of transparent intestine upon Orsay infection ( Fig. 2A).
Among these five genes, three (col-51, col-61, and sqt-2) have homozygous-viable mutants . P value comparing virus vs. no virus groups of the same genotype was labeled as the first line, while p value comparing virus groups of a given genotype vs. mock or wild-type was labeled as the second line. n ≥ 6 plates for all groups. C) Viral load detected by qRT-PCR in infected worms upon various collagen RNAi. At least three independent experiments were tested with at least three technical replicates in each experiment. D) Infection dynamics as measured by Ppals-5::GFP worms showing reporter GFP. n ≥ 3 plates for all data points. E) Antiviral drug effects. A strain sensitive to Orsay infection was used. Worms were dosed with resorcinol monoacetate (RMA), bismuth subsalicylate (BSM), or the solvent DMSO, and exposed to virus using two exposure methods. Left, schematic drawing of the two exposure methods. Right, Percentage of transparent worms in all test groups. Bars and error bars indicate means and standard errors. *, p < 0.05, ** p < 0.01, ***, p<0.001, Student's t-test. publicly available. These mutants displayed the same phenotype of increased number of symptomatic worms upon viral infection (Fig. 2B), confirming that these genes mediate antiviral immunity. In addition, we applied qRT-PCR to examine the viral load in the RNAi worms. In this experiment, N2 worms without the glp-4 mutation were used to eliminate any doubts on glp-4 effects. Consistent with the phenotype of increased symptomatic worms, RNAi of these five collagens significantly increased viral load in worms (Fig. 2C). Altogether, these data demonstrated that these collagens are required for antiviral immunity.

Collagen-mediated antiviral immunity functions at an early stage of viral infection
An intuitive hypothesis for the collagen antiviral function is that collagens may act as an exterior barrier blocking the Orsay virus from entering host cells. To test that, the reporter strain Ppals-5::GFP was used to monitor whether collagen inactivation changed viral infection dynamics. This reporter turns on GFP expression upon Orsay virus infection (21). We constructed strains with the reporter on either the wild-type N2 or the col-51 mutant background, cultured these worms, added the Orsay virus when they reached the L4 larval stage, and examined every two hours for the reporter GFP expression. In comparison with the wild-type N2 worms, more col-51 worms showed GFP at the same time point (Fig. 2D). Overall, the col-51 mutation shifted the infection dynamics curve to about four hours earlier (Fig. 2D), consistent with the hypothesis of collagens as a viral entry barrier.
Since inactivation of collagens reduced antiviral immunity, we asked whether enhancing collagens would increase antiviral immunity. To test that, we used the drh-1;glp-4 mutant strain, which is highly sensitive to Orsay infection. We tested the effect of resorcinol monoacetate (RMA), a chemical that crosslinks collagens (22), on these worms. Indeed, when worms were exposed to Orsay and RMA simultaneously, significantly fewer worms showed the infection symptom of transparent intestine than the group without RMA (Fig. 2E).
Interestingly, RMA was only effective when applied at an early stage of viral infection.
In a post exposure application experiment where the worms were exposed to the virus first for one day and then exposed to the drug for four days, RMA showed no protective effect (Fig. 2E).
As a control, the commonly used antidiarrheal drug bismuth subsalicylate (BSM) displayed the same protective effects in both early application and post exposure application experiments (Fig. . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 20, 2023. ; https://doi.org/10.1101/2023.04.20.537671 doi: bioRxiv preprint 2E). The early requirement for RMA effects suggested that collagen-mediated antiviral immunity functions at an early stage of viral infection, such as viral entry.

Collagens function in intestine cells to mediate antiviral immunity
Since Orsay only infects intestine cells (23), we questioned whether the site-of-action for the antiviral collagens was intestine cells. We applied collagen RNAi on a tissue-specific RNAi strain that was only sensitive to RNAi in intestine cells. This strain, MW265, had the RNAiresistant mutation of rde-1. The intestine cells were restored to be RNAi sensitive by the rescue transgene of rde-1 driven by the intestine-specific ges-1 promoter. This strain was resistant to Orsay virus infection (Fig. 3A). Inactivation of the collagens in intestine cells by RNAi significantly increased the virus sensitivity of this strain (Fig. 3A), suggesting that collagenmediated antiviral immunity was needed in intestine cells.
C. elegans has two types of collagens: cuticle collagens and type IV basement membrane collagens (24). No collagens have been reported to be synthesized in intestine cells: Cuticle collagens are believed to be synthesized in hypodermis (25), whereas the type IV collagens are synthesized in muscles and somatic gonads (26). Therefore, it was a surprising discovery that the site-of-action for these antiviral collagens was intestine cells. To confirm that these antiviral collagens were expressed in intestine cells, we constructed a translational fusion of Pcol-51::col- . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 20, 2023. 51::GFP and made a transgenic worm strain with this reporter. The reporter showed that col-51 was expressed in intestine cells in all developmental stages from L1 to adults, with the highest expression observed during early larval stages. During these early larval stages, col-51 expression was observed in all intestine cells, with the highest expression in posterior intestine cells (Fig. 3B). Altogether, these data demonstrated that there are collagens generated by intestine cells and that these intestinal collagens mediate antiviral innate immunity.

Actin remodelers mediate antiviral innate immunity
Several genes encoding actin-remodeling proteins were also identified as antiviral genes  (Fig. 4B). The discrepancy between this mutant and RNAi is likely due to the fact that the wsp-1 mutant allele gm324 only affects one of the two wsp-1 isoforms (Fig. 4C). Our wsp-1 RNAi was efficient as it reduced nearly 80% of the total wsp-1 transcript level (Fig. 4D). RNA inactivation of cdc-42, a gene encoding the ortholog of the mammalian WASP regulator CDC42, showed similar effects of increased infection symptoms (Fig. 4A), and such phenotype was observed in a heterozygous cdc-42 mutant (Fig. 4B). Our data suggested that these proteins that are known to regulate actin dynamics are crucial for antiviral immunity in C. elegans.
In addition to WSP-1 and WVE-1, UNC-34/Ena/VASP is another known actin remodeler (27). Since unc-34 was not in our RNAi library, we examined a homozygous unc-34 mutant. The mutant showed a transparent intestine phenotype even without viral infection, so we used qRT-PCR to examine the viral load and found that the viral load was significantly increased in this mutant (Fig. 4E). Therefore, similar to wsp-1 and wve-1, unc-34 also mediates antiviral immunity in C. elegans.
These results led us to hypothesize that actin is involved the C. elegans antiviral defense.
In C. elegans, the actin isoform ACT-5 is intestine specific and is an essential component of the terminal web (26). Since act-5 RNAi causes larval lethality, we diluted act-5 RNAi with control . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 20, 2023. ; https://doi.org/10.1101/2023.04.20.537671 doi: bioRxiv preprint bacteria at a ratio of 1:50 to reduce its effectiveness so that the RNAi worms could grow to adults. With this reduced RNAi, we were able to test our hypothesis. Indeed, act-5(RNAi) significantly increased the percentage of worms displaying the infection symptom (Fig. 4A),  confirming that actin mediates antiviral immunity in the intestine cells. Similar phenotype was confirmed in a heterozygous act-5 mutant (Fig. 4B).
The increased viral infection phenotype upon RNAi of act-5, wsp-1, wve-1, and cdc-42 was also confirmed by measuring viral load with qRT-PCR. As expected, RNAi of these genes caused a significant increase in viral load (Fig. 4F). Interestingly, RNAi of two other WASP regulators, nck-1 and wip-1, did not increase the viral load. Instead, they significantly reduced the viral load (Fig. 4F), suggesting that different actin remodelers have different functions during the course of viral infection.

Epigenetic regulation is involved in mediating antiviral innate immunity
Several genes encoding proteins regulating gene expression showed as hits in our antiviral genetic screen, for example, genes encoding the histone acetyltransferase MYS-1, the nucleosome remodeling factor NURF-1, and the chromatin regulator ISW-1. We examined the mutants of two of these genes, nurf-1 and isw-1. RNAi of these genes significantly increased the percentage of symptomatic animals upon infection (Fig. 5A). Similarly, homozygous mutants of these genes also showed an increased percentage of symptomatic animals (Fig. 5B).
Consistently, when we examined the viral load by qRT-PCR upon infection, these mutants had significantly higher viral load than the wild-type N2 worms (Fig. 5C). The requirement of chromatin remodelers NURF-1 and ISW-1 in antiviral defense showed that epigenetic regulation is a crucial part of the host antiviral innate immune response.

Chromatin remodelers NURF-1 and ISW-1 function through the actin remodeler WSP-1 to mediate antiviral innate immunity
NURF-1 and ISW-1 was reported to regulate the expression of wsp-1 to modulate axon guidance during C. elegans development (28). Since all three genes showed as antiviral genes in our screen, we tested whether wsp-1 is also the target of nurf-1 and isw-1 in antiviral defense using quantitative epistasis analysis. We performed wsp-1(RNAi) on both wild-type N2 worms and nurf-1 mutant worms, scored the percentage of symptomatic worms, and applied a quantitative epistasis model to calculate the expected phenotypic severity of two-gene inactivation based on additive effects of single gene inactivation (Fig. 5D). wsp-1(RNAi);nurf-1 . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 20, 2023. ; https://doi.org/10.1101/2023.04.20.537671 doi: bioRxiv preprint worms had significantly less severe phenotype than the expected value, suggesting a genetic interaction between wsp-1 and nurf-1 in antiviral functions (Fig. 5D). If nurf-1 and wsp-1 function independently, then the percentage of symptomatic worms should be higher in the population with two genes inactivated than the population with a single gene inactivated. In contrast, we found that wsp-1(RNAi) on nurf-1 mutants and wild-type worms had the same percentage of symptomatic worms (Fig. 5D, 44.3±2.8% vs. 44.8±2.5%, p = 0.9) despite that nurf-1 mutants had more symptomatic worms than wild-type worms (Fig. 5D, 30.8±1.7% vs. 5.8±0.9%, p < 0.0001). These data supported the hypothesis that nurf-1 regulates wsp-1 expression to mediate antiviral response. Analysis on wsp-1(RNAi) on isw-1 mutants showed similar results (Fig. 5D). As a control, analysis on wsp-1 and another chromatin remodeler mys-1 showed that wsp-1(RNAi);mys-1 worms had the same percentage of symptomatic worms as the expected value (Fig. 5D, 59.9±2.9% vs. 50.8±0.7%, p = 0.7), demonstrating that mys-1 does not was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 20, 2023. ; https://doi.org/10.1101/2023.04.20.537671 doi: bioRxiv preprint act through wsp-1. Together, these results suggest that nurf-1 and isw-1, but not mys-1, regulate wsp-1 expression to modulate antiviral immunity. To directly test the impact of nurf-1 and isw-1 mutation on wsp-1 RNA levels, we quantified wsp-1 RNA levels in these mutants by qRT-PCR under both infected and uninfected conditions. In the uninfected group, the wsp-1 RNA level difference between mutants and wildtype animals were less notable, with wsp-1 RNA being lower only in nurf-1 mutants (Fig. 5E).
Upon viral infection, wild-type animals significantly increased its wsp-1 RNA level from the uninfected level (Fig. 5E, 1±0.01 vs. 1.3±0.12, p < 0.05), but the levels of wsp-1 RNA in mutant worms remained essentially unchanged. As a result, in the virus-infected group, wsp-1 RNA levels in both nurf-1 and isw-1 mutants were significantly lower than that in wild-type animals (Fig. 5E). These data suggested that nurf-1 and isw-1 are required to up-regulate wsp-1 expression upon viral infection. While the wsp-1 RNA level changes were significant, the magnitude of changes was rather mild (Fig. 5E). This was consistent with previous report on wsp-1 RNA level changes in these mutants during C. elegans development (28). Several factors could contribute to this low magnitude of changes. For example, the wsp-1 RNA changes may be limited to intestine cells, therefore quantifying whole-body wsp-1 RNA levels would record a reduced magnitude of changes.

DISCUSSION
Our genome-wide C. elegans RNAi screen revealed 106 antiviral genes of diverse functions. The diversity of gene functions among these antiviral genes demonstrated that C. elegans has complex innate antiviral mechanisms. By implementing multiple controls and applying a stringent cutoff, our RNAi screen prioritized reducing false positives over false negatives. As a result, the screen is not saturated. For example, we did not recover known antiviral genes such as pals-22 (29) and cde-1 (16,30). More studies are needed to identify a comprehensive list of antiviral genes.
Our investigation of antiviral genes encoding collagens, actin remodelers, and epigenetic regulators suggested the following antiviral model of building physical barriers for the Orsay . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 20, 2023. ; https://doi.org/10.1101/2023.04.20.537671 doi: bioRxiv preprint virus (Fig. 6). The first barrier includes collagens generated by the intestine cells (Fig. 6). Similar to the human intestine epithelium, C. elegans intestine also has a glycocalyx layer on the apical side of microvilli within the digestive tract (26). These collagens and possibly other glycoproteins can function in the glycocalyx to form a physical barrier to block the viral entry to intestine cells. In support of this model, our data showed that intestine cells make collagens (Fig.   3). Treating worms with a collagen crosslinking drug RMA delayed the Orsay infection time course by at least four hours (Fig. 2), suggesting that these collagens mediate antiviral functions at an early stage of infection such as viral entry.

The intestinal terminal web composed mostly of actin and intermediate filaments likely
constitutes another physical barrier for the Orsay virus. Based on our results, we propose that actin remodelers UNC-34, WSP-1, WVE-1 reorganize ACT-5 in terminal web to mediate such antiviral defense (Fig. 6). It was previously shown that CDC42, a Rho GTPase, regulates the  was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 20, 2023. ; https://doi.org/10.1101/2023.04.20.537671 doi: bioRxiv preprint membrane localization of WSP-1 which then promotes the polymerization of F-actin (31).
Likewise, here we demonstrated that WSP-1 activity is regulated by CDC-42 in C. elegans.
Additionally, WSP-1 transcription is upregulated by chromatin remodelers NURF-1 and ISW-1 (Fig. 6). All these actin-regulating factors are required for antiviral immunity (Figs. 4, 5), suggesting that this is a highly coordinated activity. Although our data strongly implicate actin/actin modelers in antiviral immunity, we cannot yet determine whether the antiviral effect occurs during viral entry, viral release, or at other stages of the infection life cycle. Because our transparency assay and RT-qPCR analysis for scoring Orsay infection were performed at five and two days after infection, respectively, multiple rounds of replication could have taken place and therefore antiviral effects at either entry or exit step could be manifested in our results.
It is important to note that actin may functions at multiple stages of Orsay infection.  (35). Consistent with this report, we also found that nck-1(RNAi) reduced viral load (Fig. 4F). RNAi of another WASP regulator wip-1 also reduced viral load (Fig. 4F), suggesting that WIP-1 shares a similar function with NCK-1 in promoting viral infection. Our results demonstrate the complexity of host-virus interactions and provide a framework to unravel these interactions.
. CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 20, 2023. ; https://doi.org/10.1101/2023.04.20.537671 doi: bioRxiv preprint

Strains
C. elegans strains were maintained on standard nematode growth medium (NGM) seeded with E. coli OP50 as described (36). All strains were maintained at 20°C except those with glp-4(bn2ts) mutation, which were maintained at 15°C.
The following strains were obtained from the Caenorhabditis Genetics Center (CGC):  (Fire Lab C. elegans vector kit; Addgene) and microinjecting to N2 worms.

Viral filtration
Small amount of viral filtration was prepared as previously described (23). To scale up, two 6 cm NGM agar plates of infected rde-1(ne219) worms were cultured by adding 40 μl of the viral filtration to each plate. These worms, together with 2 ml of viral filtration, were then used to seed 50 ml liquid culture following the standard liquid culture protocol (36). Worms were cultured in a 20°C incubating shaker for 5 days. The culture was then cleared by centrifugation at . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 20, 2023. ; https://doi.org/10.1101/2023.04.20.537671 doi: bioRxiv preprint 20,000g for 10 minutes at 4°C. The supernatant was passed through a 0.22 μm filter, and aliquots were kept at -80°C.

RNAi screen
RNAi by feeding was performed using a method modified from standard protocol (38).
Bacteria from published RNAi library (39) were cultured in 2.4 ml of L-broth containing 50 μg/ml carbenicillin on 24-well deep-well plates on a 37°C incubator shaker overnight. 1 mM Isopropyl β-d-1-thiogalactopyranoside (IPTG) was then added to the culture to induce bacteria.
After two hours of induction, bacteria were pelleted by centrifugation, and culture media were discarded. The cells were then resuspended in 60 μl S medium (36) containing 50 μg/ml carbenicillin and 1 mM IPTG.
Synchronized L1 larvae from the strain SS104 were obtained by bleaching (36). ~100 L1s were placed in each well of a 96-well plate with 30 μl of RNAi bacteria suspension, and the volume was brought up to 100 μl with S medium with 50 μg/ml carbenicillin and 1 mM IPTG.
For the virus-infected group, 1 μl of Orsay viral filtration was added to each well. The plates were parafilmed and placed on a 20°C incubator shaker for five days till the worms reached day-3 adults. Worms were then transferred to unseeded scanning plates (modified NGM plates that do not contain peptone or cholesterol), killed by sodium azide, and observed under a highcontrast stereoscope for the transparent intestine phenotype.
A quick manual inspection was applied to select RNAi wells that showed qualitative differences between infected and uninfected groups. The worms on these wells were then quantitatively scored. The RNAi effects of hit genes were then confirmed in at least three independent experiments. The identity of hit genes in the RNAi bacterial clones was confirmed by Sanger sequencing.

Small-scale experiments for scoring infection symptom of transparent intestine
Standard 6-cm NGM agar plates seeded with OP50 were prepared as described (36). For RNAi experiments, RNAi plates (38) were used. To make RNAi plates, RNAi bacteria were cultured in L-broth containing 50 μg/ml carbenicillin at 37°C overnight. Bacteria were then used to seed 6-cm NGM agar plates that contained 50 μg/ml carbenicillin and 1 mM IPTG. act-5 . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 20, 2023. ; https://doi.org/10.1101/2023.04.20.537671 doi: bioRxiv preprint RNAi bacteria was diluted 1:25 with control empty vector bacteria before seeding. The seeded RNAi plates were left at room temperature overnight. ~100 synchronized L1 larvae were dropped onto each plate. For the infected group, 40 μl of viral filtration was added to each plate. The worms were cultured in a 20°C incubator for five days until they were day-3 adults and scored. For strains that did not have the glp-4(bn2ts) mutation, adult worms were manually picked and transferred to a fresh RNAi plate everyday once they reached the day-1 adult stage to avoid interference from progeny.

RNA quantification by qRT-PCR
For standard culture, ~500 synchronized L1 larvae were dropped onto each 6-cm NGM agar plates seeded with OP50 (36). 40 μl of viral filtration was added to each plate in the infected group.
For RNAi worms, 10 L4 animals were placed on each well of a 6-well RNAi plate seeded with RNAi bacteria, and cultured in a 20°C incubator for 24 hours. The worms were then removed to keep only synchronized eggs on the plates. 20 μl of viral filtration was added to each well for the infected group.
The worms were cultured in a 20°C incubator for two days till they reached L4 stage. L4 worms were collected and washed four times with 10 ml S basal medium (36). Total RNA from worms was extracted by using TRIzol (Invitrogen), digested with DNase (Invitrogen), and reverse transcribed to cDNA using RETROscript (Thermo). qRT-PCR was performed by using When the worms reached day-3 adults, they were transferred to unseeded scanning plates, killed by sodium azide, and observed under a high-contrast stereoscope for the transparent intestine phenotype.

Infection dynamics
About 100 synchronized L1 larvae of ERT54 or WWZ369 were placed on a 3-cm NGM agar plate seeded with OP50, and cultured at a 20°C incubator for 48 hours till they reached L4 stage. A mixture of 15 μl water and 5 μl orsay was dropped on the plate and the timer was set as time point 0. Worms were observed for GFP under a Zeiss SteReo Discovery V20 stereoscope starting from time point 14 hours to time point 40 hours. GFP-positive worms were counted and removed from plates. A control group of uninfected worms was always used as a negative control to ensure that no GFP was observed in those worms.

Microscopy
. CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 20, 2023. ; https://doi.org/10.1101/2023.04.20.537671 doi: bioRxiv preprint Epifluorescent images were taken using a Zeiss AxioImager M2m microscope equipped with a Zeiss AxioCam MRm camera and AxioVision software 4.8.

Quantitative Epistasis
Quantitative epistasis analysis was performed using math model as previously described (40). At least three independent trials were performed for each gene pair. In each trial, at least two plates were tested for each genotype.      . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 20, 2023. ; https://doi.org/10.1101/2023.04.20.537671 doi: bioRxiv preprint . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 20, 2023. ; https://doi.org/10.1101/2023.04.20.537671 doi: bioRxiv preprint . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 20, 2023. ; https://doi.org/10.1101/2023.04.20.537671 doi: bioRxiv preprint . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 20, 2023. ; https://doi.org/10.1101/2023.04.20.537671 doi: bioRxiv preprint . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 20, 2023. ; https://doi.org/10.1101/2023.04.20.537671 doi: bioRxiv preprint