The predicted bZIP transcription factor ZIP-1 promotes resistance to intracellular infection in Caenorhabditis elegans

Defense against intracellular infection has been extensively studied in vertebrate hosts, but less is known about invertebrate hosts. For example, almost nothing is known about the transcription factors that induce defense against intracellular intestinal infection in the model nematode Caenorhabditis elegans. Two types of intracellular pathogens that naturally infect the C. elegans intestine are the Orsay virus, which is a positive-sense RNA virus, and microsporidia, which are fungal pathogens. Surprisingly, these molecularly distinct pathogens induce a common host transcriptional response called the Intracellular Pathogen Response (IPR). Here we describe zip-1 as an IPR regulator that functions downstream of all known IPR activating and regulatory pathways. zip-1 encodes a putative bZIP transcription factor of previously unknown function, and we show how zip-1 controls induction of a subset of genes upon IPR activation. ZIP-1 protein is expressed in the nuclei of intestinal cells, and is required in the intestine to upregulate IPR gene expression. Importantly, zip-1 promotes resistance to infection by the Orsay virus and by microsporidia in intestinal cells. Altogether, our results indicate that zip-1 represents a central hub for all triggers of the IPR, and that this transcription factor plays a protective role against intracellular pathogen infection in C. elegans.


Introduction 37
Viruses and other obligate intracellular pathogens are responsible for a myriad of serious illnesses 38 (1). RNA viruses, like the single-stranded, positive-sense RNA virus SARS-CoV-2 that causes 39 COVID-19, are detected by RIG-I-like receptors (2-4). These receptors detect viral RNA 40 replication products and trigger transcriptional upregulation of interferon genes to induce anti-viral 41 defense (5). The nematode Caenorhabditis elegans provides a simple model host to 42 understanding responses to RNA viruses, as a single-stranded, positive-sense RNA virus from 43 Orsay, France infects C. elegans in the wild (6). Interestingly, natural variation in drh-1, a C. 44 elegans gene encoding a RIG-I-like receptor, was found to underlie natural variation in resistance 45 to the Orsay virus (7). Several studies indicate that detection of viral RNA by the drh-1 receptor 46 induces an anti-viral response through regulating RNA interference (RNAi) (7-9). 47 We next used qRT-PCR to assess the role of zip-1 in controlling levels of endogenous pals-5 164 mRNA, as well as mRNA of other IPR genes. Because bortezomib treatment induced the 165 strongest and most consistent IPR gene expression, we used this trigger to assess the role of zip-166 1 in mediating IPR gene induction in subsequent experiments. Here we were surprised that zip-167 1(jy13) mutants had only about a 3.5 fold reduction in pals-5 mRNA induction at 4 hours (h) after 168 bortezomib treatment compared to induction in wild-type animals (Fig. 3A). 4 h is the timepoint at 169 which zip-1 mutants were strongly defective for induction of pals-5p::GFP and pals-5p::NanoLuc 170 expression (Fig. 2E, F, G). Therefore, we considered the possibility that GFP and nanoluciferase 171 expression observed at 4 h may reflect protein synthesized from mRNA made earlier. To 172 investigate this possibility, we used qRT-PCR to measure pals-5 mRNA at 30 minutes (min) after 173 bortezomib treatment, and here we found that zip-1 was completely required for the ~300-fold 174 induction of pals-5 mRNA at this early timepoint (Fig. 3B). Thus, zip-1 is completely required for 175 pals-5 mRNA induction 30 min after bortezomib treatment, but only partially required for induction 176 at 4 h after bortezomib treatment. 177 Because zip-1 appeared to be more important at 4 h for inducing GFP and nanoluciferase 178 transcriptional reporters than for inducing pals-5 mRNA by qRT-PCR, we used smFISH as a 179 separate measure for pals-5 mRNA levels at this timepoint. Here, as in the GFP reporter studies, 180 pals-5 expression was seen in the intestine. Because it is an easily identified location, we 181 quantified pals-5 RNA levels in the first intestinal ring, which is comprised of four epithelial cells. 182 Here we found that pals-5 mRNA was induced to a lesser degree in zip-1 mutants treated with 183 bortezomib compared to wild-type animals (Fig. 3C, Fig. S4). 184 Next, to determine whether zip-1 mutants are defective in PALS-5 protein production, we raised 185 polyclonal antibodies against the PALS-5 protein. Using these antibodies for western blots, we 186 found that PALS-5 protein induction in zip-1(jy13) animals at 4 h after bortezomib treatment was 187 almost undetectable in comparison to the induction in bortezomib-treated wild-type animals ( Fig.  188 3D, Fig. S5). Therefore, zip-1 is required for high levels of PALS-5 protein production after 189 bortezomib treatment, very likely through its role in regulating induction of  Having confirmed that zip-1 is completely required for induction of pals-5 mRNA at 30 min and 191 partially required at 4 h, we examined the requirement for zip-1 in induction of other IPR genes at 192 these timepoints. We analyzed highly induced IPR genes of unknown function -F26F2. 1, 193 F26F2.3 and F26F2.4, as well as components of a cullin-ring ubiquitin ligase complexcul- 6,194 skr-3, skr-4 and skr-5, which mediates thermotolerance as part of the IPR program (18). 195 Interestingly, zip-1 was not required at either time point (30 min or 4 h after bortezomib treatment) 196 for mRNA induction of the majority of genes we analyzed, including F26F2.1 (Fig. 3 A and B). In 197 agreement with these results, zip-1 was not required for F26F2.1p::GFP expression after 198 bortezomib treatment (Fig. S6). Furthermore, zip-1 was not required for induction of the chitinase-199 like gene chil-27, which is induced by bortezomib, as well as by the natural oomycete pathogen 200 Myzocytiopsis humicola (13,23). In contrast, zip-1 was required at the 4 h timepoint for induction 201 of skr-5 RNA levels (Fig. 3B). Because the induction of skr-5 at 30 min was quite low, it was 202 difficult to assess the role of zip-1 in regulating this gene at this timepoint. Overall, these results 203 suggest that there are at least three classes of IPR genes: 1) genes that require zip-1 for early 204 but not later induction ("Early zip-1-dependent" genes like pals-5), 2) genes that require zip-1 at 205 the later timepoint ("Late zip-1-dependent" genes like skr-5), and 3) genes that do not require zip-206 1 at either timepoint for their induction ("zip-1-independent" genes like F26F2.1). 207 We also analyzed the role of zip-1 in regulating IPR gene induction upon intracellular infection. 208 Similar to the results using bortezomib as a trigger, we found that zip-1 was required for pals-5 209 induction by N. parisii infection or by Orsay virus infection, but was not required for induction of 210 F26F2.1. Interestingly, zip-1 mutation had only a partial effect on skr-5 induction following Orsay-211 virus infection, whereas skr-5 levels were not highly induced in N. parisii infected animals (Fig. 212 S7). 213

RNA sequencing analysis reveals a genome-wide picture of zip-1-dependent genes 214
To obtain a genome-wide picture of the genes controlled by zip-1, we next performed RNA 215 sequencing (RNA-seq) analysis. Here we treated wild-type N2 or zip-1(jy13) mutant animals with 216 either bortezomib or vehicle control for either 30 min or 4 h, then collected RNA and performed 217 RNA-seq. Based on differential expression analyses, we created lists of genes upregulated in 218 each genetic background after bortezomib treatment at both analyzed timepoints. At 30 min, we 219 found that 136 and 215 genes were upregulated in wild-type and zip-1(jy13) animals, respectively, 220 with 72 genes being upregulated in both backgrounds (Fig. 4A, Table S2). Therefore, 64 genes 221 (i.e. 136 minus 72 genes) were induced only in wild-type animals, indicating that they are zip-1-222 dependent early upon proteasome blockade. Importantly, pals-5 was among these genes that 223 were only upregulated in wild-type animals and not zip-1 mutants at this timepoint, consistent with 224 our qRT-PCR analysis (Fig. 3). At 4 h, we identified many more genes that showed differential 225 expression between bortezomib and control treatments in both genetic backgrounds, with 2923 226 and 2813 genes upregulated in wild-type and zip-1(jy13) mutants, respectively (Fig. 4B, Table  227 S2). 2035 genes were upregulated in both backgrounds, meaning that 888 genes (2923 minus 228 2035 genes) were specifically upregulated in wild-type animals. 883 out of 888 genes belong to 229 the "Late zip-1-dependent" category, and include skr-5, consistent with our qRT-PCR analysis 230 (Fig. 3). Notably, five genes (ZK355.8,K02E7.10, were induced 231 only in wild-type animals at both examined timepoints, and thus we classified these genes as 232 "Completely zip-1-dependent". Therefore, 59 (64 minus 5) genes from the 30 min timepoint belong 233 to the "Early zip-1-dependent" genes category. Of note, consistent with our qRT-PCR and GFP 234 reporter analysis, the F26F2.1 gene was upregulated in both genetic backgrounds following 235 bortezomib treatment, and thus belongs to the "zip-1-independent" category. 236 We next examined the correlation between zip-1-dependent genes (separately analyzing genes 237 induced at each timepoint) and gene sets that were previously associated with IPR activation. 238 Here we found that there is a significant similarity between zip-1-dependent genes induced after 239  Table S3). In addition, there is a significant similarity between genes induced 242 at 30 min timepoint and canonical IPR genes. Similarly, there is a significant overlap between zip-243 1-dependent genes induced after 4 h bortezomib treatment and the majority of these IPR-244 associated gene-sets. Of note, there was not a significant overlap between zip-1-dependent 245 genes induced after 4 h bortezomib treatment, and genes that are upregulated at the late phases 246 of viral (96 hpi) and microsporidia infections (40 hpi and 60 hpi). These results suggest that zip-1 247 plays a more important role in the acute transcriptional response to intracellular infection, and 248 perhaps a lesser role later in infection. Furthermore, our analysis revealed significant similarity 249 between zip-1-upregulated genes and genes that are downregulated by sta-1. STA-1 is a STAT-250 related transcription factor that acts as a negative regulator of IPR gene expression. (Fig. S8A, 251 Table S3). We also found a significant overlap between zip-1-dependent genes and those induced 252 by M. humicola, a natural oomycete pathogen that infects the epidermis, although zip-1 was not 253 required for induction of the chitinase-like gene chil-27, which is a common marker for M. humicola 254 response (Fig. S8B, Table S3). Previous studies have shown connections between the IPR and 255 genes induced either by M. humicola infection, or by extract from M. humicola as part of the 256 oomycete recognition response in the epidermis (13,23,24). 257 We identified zip-1-dependent genes in our analysis here using proteasome blockade by 258 bortezomib, which has effects on transcription that are unrelated to the IPR. For example, 259 bortezomib activates the bounceback response that induces expression of proteasome subunits, 260 and it is controlled by the conserved transcription factor SKN-1/Nrf2 (25). Therefore, we compared 261 if zip-1-dependent genes (from both analyzed timepoints) have significant overlap with skn-1-262 dependent genes. Here we found no significant similarity between the majority of analyzed 263 datasets (Fig. S8C, Table S3). These results are consistent with previous IPR RNA-seq studies  264 showing a distinction between the IPR and the bounceback response, and suggest that zip-1 265 does not play a role in the bounceback response (11,13). 266 In addition, we found that zip-1 mRNA itself was strongly upregulated in wild-type animals 267 following bortezomib treatment, consistent with previous studies (13). Surprisingly however, we 268 found that zip-1 mRNA was also upregulated in zip-1 mutants. This result that was initially 269 confusing, because the zip-1 coding sequence is completely deleted in the zip-1(jy13) allele that 270 we used in RNA-seq analysis. Upon closer examination however, we found that zip-1 sequencing 271 reads in zip-1(jy13) mutant samples aligned to the region upstream of the zip-1 gene coding 272 sequence, which contains annotated 5' untranslated regions (UTRs) for several zip-1 isoforms, 273 as well as to downstream sequences that contain the zip-1 3' UTR ( Fig. S9). This finding indicates 274 that zip-1 is not required to induce its own transcription, but rather a distinct transcription factor is 275 involved in upregulation of zip-1 mRNA expression. 276 To obtain insight into other biological processes and cellular structures that may be related to zip-277 1, we performed analysis with the WormCat program, specifically designed for analysis of C. 278 elegans genomics data (26). We separately analyzed 64 genes from the early timepoint and 888 279 genes from the later timepoint that were upregulated in wild-type animals but not zip-1 mutants. 280 The only significantly overrepresented category of upregulated genes at 30 min was the stress 281 response category (Fig. 4D, Table S4). Analysis of the genes upregulated at 4 h revealed a 282 significant overrepresentation of genes implicated in mRNA function, transcription, nuclear pore, 283 signaling, development, cytoskeleton, proteolysis and DNA. 284 Finally, we analyzed and classified 80 canonical IPR genes (13) based on their expression levels 285 in our RNA-seq datasets, to place them into different categories based on their dependence on 286 zip-1. Here we found that 23 IPR genes (including pals-5) were upregulated in wild-type animals 287 but not zip-1 mutants 30 min after bortezomib treatment, but became upregulated in both genetic 288 backgrounds at 4 h (Fig. 4E). Therefore, these genes are "Early zip-1-dependent" IPR genes. 289 Notably, 11 pals genes belong to this category. Another seven IPR genes (including skr-5) were 290 not upregulated in zip-1(jy13) mutants at either timepoint analyzed, but were upregulated in wild 291 type at 4 h, and we classified these genes as "Late zip-1-dependent" IPR genes. Therefore, 292 overall, 30 IPR genes appeared to be zip-1-dependent, when including both timepoints. 42 293 canonical IPR genes were upregulated in both genetic backgrounds, and we classified them as 294 "zip-1-independent" IPR genes. Because some of these genes were not upregulated at the first 295 timepoint, we further divided this category of genes into class A that showed upregulation after 296 30 min bortezomib treatment (including F26F2.1), and class B that showed upregulation only after 297 4 h of bortezomib treatment. Of note, eight canonical IPR genes did not show significant 298 upregulation after bortezomib treatment, so we did not classify them in any category. These 299 include histone genes, which previous studies had shown to be regulated by pals-22/pals-25 and 300 N. parisii infection (and thus qualify as IPR genes), but not to be induced by bortezomib treatment 301 (11, 13). In conclusion, our RNA-seq results demonstrate that zip-1 controls RNA expression of 302 30 out of 80 IPR genes, and reveal that IPR genes can be placed into three separate classes 303 based on their regulation by zip-1. 304

ZIP-1 is expressed in the intestine and is required in this tissue to regulate pals-5 gene 305 expression 306
To examine where ZIP-1 is expressed, we tagged the zip-1 endogenous genomic locus with gfp 307 immediately before the stop codon using CRISPR/Cas9-mediated gene editing. Here we found 308 that ZIP-1::GFP endogenous expression was not detectable in unstressed animals. Because zip-309 1 mRNA is induced by bortezomib, and bortezomib blocks protein degradation, we investigated 310 whether ZIP-1::GFP was visible after bortezomib treatment. Here we found that ZIP-1::GFP 311 expression was induced, with strongest expression found in intestinal nuclei (Fig. 5A). Nuclear 312 expression was also identified in the epidermis (Fig. 5B). Specifically, 98 % (59/60) of animals 313 showed ZIP-1::GFP expression in intestinal nuclei after 4 h bortezomib treatment, while 88 % 314 (53/60) showed expression in epidermal nuclei after 4 h bortezomib treatment. In contrast, no 315 GFP signal was observed in wild-type animals treated with bortezomib, or in zip-1::gfp mutants 316 or wild-type animals treated with DMSO control (60 analyzed animals for each condition). 317 Next we examined ZIP-1::GFP expression upon intracellular infection. First, we infected ZIP-318 1::GFP animals with N. parisii, stained with a FISH probe to label parasite cells, and quantified 319 the percentage of infected animals displaying ZIP-1::GFP nuclear expression. Here we found that 320 73.33% (44/60) of animals with parasite cells had ZIP-1::GFP expression, whereas 0% (0/60) of 321 uninfected animals had ZIP-1::GFP expression (Fig. 5C). We did note that uninfected intestinal 322 cells found adjacent to N. parisii-infected cells also displayed ZIP-1::GFP expression, suggesting 323 there may be cell-to-cell signaling from infected to uninfected cells to induce ZIP-1::GFP 324 expression, although we cannot eliminate the possibility that 'uninfected' cells have pathogen that 325 was not visible. Next we infected ZIP-1::GFP animals with Orsay virus, and stained with a FISH 326 probe to label viral RNA. Similar to studies with N. parisii, we found that 100% (60/60) virus 327 infected animals had ZIP-1::GFP expression, whereas 0% (0/60) of uninfected animals had ZIP-328 1::GFP expression (Fig. 5D). Here as well, we found evidence there may be cell-to-cell signaling, 329 as uninfected cells adjacent to viral-infected cells displayed ZIP-1::GFP expression. Importantly, 330 viral induction of ZIP-1::GFP enabled us to analyze whether the DRH-1 acts upstream of ZIP-331 1::GFP. Here we found that viral infection no longer induced ZIP-1::GFP in drh-1 mutants (0% or 332 0/60 animals) (Fig. 5E), suggesting that the DRH-1 receptor acts upstream of the ZIP-1 333 transcription factor. 334 To determine the tissue in which zip-1 acts to regulate pals-5 induction, we performed tissue-335 specific downregulation of zip-1 using RNAi, and measured the levels of pals-5 mRNA following 336 30 min bortezomib treatment. First, we used rde-1 loss-of-function mutation strains, which have 337 a rde-1 rescuing construct expressed specifically in either the intestine or in the epidermis, which 338 leads to enrichment of RNAi in these tissues. Here we observed that zip-1 RNAi in the intestinal-339 specific RNAi strain caused a block in pals-5 induction, similar to zip-1 RNAi in wild-type animals 340 ( Fig. 5F). In contrast, pals-5 induction was less compromised by zip-1(RNAi) in the epidermal-341 specific RNAi strain. Because intestinal expression of rde-1 allows generation of secondary 342 siRNAs that can spread to other tissues and silence gene expression there, we used a separate 343 tissue-specific RNAi strain, where the sid-1 transport channel is specifically expressed in the 344 intestine (27, 28). These mutants do not suffer from the problem of leakiness seen in the rde-1 345 rescue strains (28, 29). However, we did note that they suffer from the opposite problem: they 346 appear to be somewhat resistant to RNAi. To quantify this effect, we treated the intestinally 347 rescued sid-1 strain with dsRNA against act-5, which is an actin isoform expressed in the intestine, 348 but not in the epidermis, and it is essential for development (30). Here we found that act-5 RNAi 349 caused less severe effects on size in the intestinally rescued sid-1 strain compared to wild-type 350 animals ( Fig. S10A and S10B). Despite being partially resistant to RNAi, we did find that zip-1 351 RNAi in this strain caused a significant reduction in pals-5mRNA induction upon bortezomib 352 treatment compared to vector control (Fig. S10C). Taken together, our data suggest that zip-1 is 353 highly expressed in the intestinal nuclei following bortezomib treatment, and that zip-1 is important 354 in the intestine for induction of pals-5 mRNA. 355

zip-1 regulates resistance to natural intracellular pathogens 356
Because increased IPR gene expression is correlated with increased resistance to intracellular 357 infection (10, 13), we investigated the role of zip-1 in resistance to intracellular pathogens. First, 358 we investigated Orsay virus. Here, we infected L4 animals and found that zip-1 mutants had 359 higher viral load compared to wild-type animals, as assessed by qRT-PCR (Fig. 6A). Similarly, 360 we found upon infection of L1 animals and measuring viral load with FISH staining that zip-1 361 mutants had a trend toward higher infection rate than wild-type animals (Fig. 6B). We also 362 investigated whether zip-1 might have a greater effect on viral load in a mutant background where 363 IPR genes are constitutively expressed. Indeed, we found a more pronounced role for zip-1 after 364 viral infection of pnp-1 mutants, which have constitutive expression of IPR genes, including pals-365 5 ( Fig. 6B) (15). Of note, our qRT-PCR analysis of pnp-1(jy90) animals showed that elevated 366 pals-5 mRNA levels depend on zip-1, suggesting that the IPR genes upregulated by zip-1 promote 367 resistance against viral infection (Fig. S11). Similar to what we observed after bortezomib 368 treatment, expression of highly induced IPR genes F26F2. 1, F26F2.3 and F26F2.4 in a pnp-1 369 mutant background did not require zip-1. This finding suggests that zip-1-dependent IPR genes 370 may play a more important role in Orsay virus resistance than other IPR genes. 371 Next, we examined a role for zip-1 in resistance to N. parisii infection by measuring pathogen 372 load. Here we did not see an effect of zip-1 in a wild-type background either at 3 hpi or at 30 hpi 373 (Fig. 6 C and D). However, at both timepoints, we found that loss of zip-1 significantly suppressed 374 the increased pathogen resistance (i.e. lower pathogen load) of pnp-1 mutants (Fig. 6 C and D). 375 Therefore, these experiments indicate that wild-type zip-1 promotes resistance to N. parisii 376 infection in a background where IPR genes are induced prior to infection. To further analyze the 377 role of zip-1 in response to N. parisii infection, we also performed killing assays in which we 378 analyzed survival of animals following infection. Consistent with published data, we found that 379 pnp-1 mutants survive longer than wild-type animals when infected with N. parisii, but do not 380 survive longer than wild-type animals in the absence of infection ( Fig. 6 E and F). Importantly, we 381 found that zip-1 mutations decrease survival both in a pnp-1 mutant background, as well as in a 382 wild-type background. Therefore, wild-type zip-1 promotes survival against N. parisii infection. 383 Because infections were performed by feeding pathogens to animals, it was possible that 384 differences in food intake and elimination were responsible for any differences seen in pathogen 385 load. Therefore, we measured accumulation of fluorescent beads in all tested strains and we did 386 not find any significant differences between zip-1 mutants and control animals (Fig. S12). In 387 conclusion, the increased pathogen load in zip-1 mutants is unlikely to be due to differences in 388 the exposure of intestinal cells to pathogen in these mutants. 389 Other phenotypes in pnp-1 mutants include higher sensitivity to heat shock and slightly slower 390 growth rate (15). We tested if either of these phenotypes are zip-1-dependent. First, we found 391 that zip-1(jy13) animals had a similar survival rate after heat shock compared to the control strain 392 (Fig. S13A). Similarly, we found that loss of zip-1 in a pnp-1(jy90) mutant background did not 393 significantly suppress the higher lethality observed in pnp-1(jy90) single mutants, suggesting that 394 ZIP-1 does not play a crucial role in thermotolerance regulation. Finally, we analyzed if zip-1(jy13) 395 mutants, which show a wild-type growth rate, can suppress the mild growth delay caused by a 396 pnp-1(jy90) mutation. Here, growth was assayed based on the body length measurements 44 h 397 after plating synchronized L1 animals, and we found that zip-1(jy13); pnp-1(jy90) animals were 398 still significantly smaller than control animals and zip-1(jy13) single mutants (Fig. S13B). 399 Therefore, zip-1 does not appear to be important for these non-infection related phenotypes of 400 pnp-1 mutants. Instead, it seems that zip-1 specifically plays a role in regulating immunity-related 401 IPR genes. 402

Most studies of antiviral immunity in invertebrates have focused on anti-viral RNAi, and less is 405
known about transcriptional responses to intracellular infection in either of the two major 406 invertebrate model systems, Drosophila melanogaster or C. elegans (31-33). The IPR in C. 407 elegans is a common transcriptional response that is induced independently by both virus and 408 microsporidia infection, as well as by specific physiological perturbations such as proteotoxic 409 stress (11-13). Previous studies had shown that the STAT-related transcription factor sta-1 was 410 a repressor of IPR genes (34), but the activating transcription factor for the IPR was not known. 411 Here, we show that the previously uncharacterized, predicted bZIP transcription factor ZIP-1 412 functions downstream of all known IPR triggers to induce a subset of IPR genes (Fig. 7). 413 Importantly, we show that zip-1 plays a role in immunity against infection by both the Orsay virus 414 and microsporidia. Therefore, zip-1 appears to be the first transcription factor shown to promote 415 an inducible defense response against intracellular intestinal pathogens in C. elegans.  (19). What is the logic to having so many transcription factors involved in immunity 436 in C. elegans? For comparison, only one bZIP transcription factor, CrebA, has recently been 437 shown to play a role in D. melanogaster tolerance to bacterial pathogens (51). Also, a single STAT 438 transcription factora component of JAK/STAT pathway, has been shown to play a downstream 439 role in antiviral immunity, although this factor is not thought to be the first responder to viral RNA-seq analyses demonstrated that many genes induced as part of the IPR do not require ZIP-457 1 for their induction, while some require ZIP-1 only for early induction, but not late induction. 458 Future studies with screens for transcription factors that mediate induction of zip-1-independent 459 genes should enable a more complete assessment of the immune response to intracellular 460 In this study we demonstrate that ZIP-1 protein expression can be activated by different signaling 462 pathways that act in parallel to induce the IPR, including proteasome inhibition and DRH-1 463 pathway triggered by viral infection (10). While zip-1 itself appears to be transcriptionally and 464 translationally induced by infection, we believe that ZIP-1 is the immediate transcription factor that 465 activates IPR gene expression upon various triggers. zip-1 is required for IPR gene induction only 466 30 min after activation by bortezomib, which is likely too short a time for a separate transcription 467 factor to activate zip-1 transcription and translation, which would then induce IPR gene 468 expression. There is still much to be learned about how various triggers activate the IPR, although 469 a likely ligand and receptor pair have been identified for the Orsay virus, where viral RNA 470 replication products appear to be detected by the RIG-I-like receptor DRH-1 (10). As mentioned 471 earlier, C. elegans lacks the downstream factors that mediate viral/RIG-I signaling in mammals, 472 such as IRF3/7 and interferon. Therefore, we propose that ZIP-1 and the IPR may play an 473 analogous role to IRF3/7 and interferon in C. elegans defense against intracellular infection in  Table S5. 483

RNAi screens 484
RNAi screens were performed using the feeding method in liquid medium. Gravid adults were 485 bleached following a standard protocol (65), and isolated eggs were incubated in M9 medium 486 overnight to hatch into starved L1's unless stated otherwise. In particular, for the screen in the  Table S1 we also list normalized TOF values as a proxy for body 497 length, which indicates that an RNAi clone like lin-26 had low GFP/TOF, but also had low TOF, 498 indicating small size and thus potentially poor overall health, and thus was not pursued as a hit. 499 Two experimental replicates were performed for majority of RNAi clones (the transcription factor 500 RNAi library is split among five 96-well plates, three of which were tested twice and two of which 501 were tested once). 502 For the screen in which chronic heat stress was used to induce the IPR, synchronized populations 503 of 150 L1 animals carrying the jyIs8[pals-5p::gfp] transgene were transferred to S-basal medium 504 in 96-well plates. The wells were supplemented with overnight RNAi bacterial cultures, as 505 previously described for RNAi screen in pals-22(jy3) mutant background. Animals were incubated 506 in the shaker at 20°C for 48 h, and then subjected to chronic heat stress at 30°C for 18 h. 507 Subsequently, pals-5p::GFP expression was measured and standardized to the worm length 508 using TOF measurements on the COPAS Biosort machine. Three independent experimental 509 replicates were performed. 510

RNAi assays on plates 511
RNA interference assays were performed using the feeding method. Overnight cultures of HT115 512 E. coli were plated on RNAi plates (NGM plates supplemented with 5 mM IPTG and 1 mM 513 carbenicillin) and incubated at room temperature for 3 days. Synchronized L1 animals were 514 transferred to these plates and incubated at 20°C. Following 48 h incubation, specific phenotypes 515 of animals were analyzed (pals-5p::GFP expression after exposure to zip-1 RNAi; developmental 516 defects after act-5 RNAi treatment). Control RNAi experiments were carried out using a vector 517 plasmid L4440. 518

CRISPR/Cas9-mediated deletions of zip-1 and pals-5 519
Deletions of zip-1 and pals-5 were carried out using the co-CRISPR method with preassembled 520 ribonucleoproteins (66, 67). Cas9-NLS protein (27 µM final concentration) was ordered from QB3 521 Berkeley; sgRNA components and DNA primers were obtained from Integrated DNA 522 Technologies (IDT). 523 The following crRNA sequences were used to target zip-1 gene: acacaggcatctggggaccc (for 524 generating the jy13 allele), tcagcttgtgctgggcgttg (for generating the jy14 allele), 525 agcaatttgagccaagctga (for generating both jy13 and jy14 alleles). PCR screenings were 526 performed using the primers 1-4 listed in Table S6. Deletion-positive lines were backcrossed three 527 times to the N2 strain before they were used in experiments. The jy13 allele is an 8241 base pair 528 long deletion, starting 172 nucleotides upstream of the zip-1 start codon and ending at the last 529 nucleotide before the stop codon (C8069). jy14 allele is a 4108 base pair long deletion, starting 530 at nucleotide G3962 and ending at the last nucleotide before the stop codon (C8069). 531 The following crRNA sequences were used to target the pals-5 gene: aaatactcgaagcaattcag and 532 aaaacgaatagaaaatggga. PCR screenings were performed using primers 10 and 11 from the Table  533 S6. Deletion-positive lines were backcrossed three times to the N2 strain before they were used 534 in experiments. jy133 allele is a 1706 base pair long deletion, starting 128 nucleotides upstream 535 of the pals-5 start codon and ending at the 108th nucleotide after the stop codon. 536

Orsay virus infections 537
Orsay virus isolate was prepared as previously described (11). For pals-5p::GFP expression 538 analysis and FISH staining for infection level quantification, L1 animals were exposed to a mixture 539 of OP50-1 bacteria and Orsay virus for 12 h at 20°C, whereas animals used for ZIP-1::GFP 540 analysis were infected with a high dose of virus for 9 h at 20°C. pals-5p::GFP reporter expression 541 was analyzed in animals that were anesthetized with 10 mM levamisole. For FISH analysis, 542 animals were collected and fixed in 4% paraformaldehyde for 15 to 45 min depending on the 543 assay. Fixed worms were stained at 46°C overnight using FISH probes conjugated to the red Cal 544 Fluor 610 fluorophore, targeting Orsay virus RNA1. GFP imaging and FISH analysis were 545 performed using Zeiss AxioImager M1 compound microscope. For qRT-PCR analyses, 546 synchronized L4 animals were exposed to a mixture of OP50-1 bacteria and Orsay virus for 24 h 547 at 20°C. RNA isolation and qRT-PCR analysis were performed as described below. ZIP-1::GFP 548 expression was analyzed and imaged on a Zeiss LSM700 confocal microscope run by ZEN2010 549 software. 550

Microsporidia infections 551
N. parisii spores were prepared as previously described (63). Spores were mixed with food and 552 L1 synchronized animals (a dose of 8 million spores per plate was used for ZIP-1::GFP expression 553 analyses, a dose of 0.5 million spores per plate was in all other assays). Animals were incubated 554 at 25°C for 3 h (for sporoplasm counting and ZIP-1::GFP analysis), 24 h (for qRT-PCR analysis 555 of IPR gene expression) or 30 h (for pathogen load analysis). For pals-5p::GFP expression 556 analysis, animals were anesthetized with 10 µM levamisole and imaged using Zeiss AxioImager 557 M1 compound microscope. For FISH analysis, animals were collected and fixed in 4% 558 paraformaldehyde for 15 to 45 min depending on the assay. Fixed worms were stained at 46°C 559 for 6 h (for ZIP-1::GFP analysis) or overnight (for pathogen load analyses) using FISH probes 560 conjugated to the red Cal Fluor 610 fluorophore, targeting ribosomal RNA. 3 hpi samples were 561 analyzed using Zeiss AxioImager M1 compound microscope; 30 hpi samples were imaged using 562 ImageXpress automated imaging system Nano imager (Molecular Devices, LLC), and 563 fluorescence levels were analyzed using FIJI program. ZIP-1::GFP expression was analyzed and 564 imaged on a Zeiss LSM700 confocal microscope run by ZEN2010 software. 565

Bortezomib treatments 566
Proteasome inhibition was performed using bortezomib (Selleckchem, catalog number S1013) as 567 previously described (18,22). Synchronized L1 animals were plated on 10 cm (for RNA 568 extraction) or 6 cm NGM plates (for phenotypic analyses and transgene expression 569 measurements), and grown for 44 h or 48 h at 20°C depending on the assay. 10 mM stock solution 570 of bortezomib in DMSO was added to reach a final concentration of 20 µM per plate. The same 571 volume of DMSO was added to the control plates. Plates were dried and worms incubated for 30 572 minutes, 4, 21 or 25 hours at 20°C. Imaging was performed using Zeiss AxioImager M1 compound 573 microscope or ImageXpress automated imaging system Nano imager (Molecular Devices, LLC), 574 and analyzed using FIJI program. For RNA extraction, animals were washed off the plates using 575 M9, washed with M9 and collected in TRI reagent (Molecular Research Center, Inc.). ZIP-1::GFP 576 expression was analyzed and imaged on a Zeiss LSM700 confocal microscope run by ZEN2010 577 software. 578

Fluorescence measurements 579
Fluorescence measurements shown in Fig. 1A and B and Fig. 2D were performed using the 580 COPAS Biosort machine (Union Biometrica). The fluorescent signal was normalized to TOF, as 581 a proxy for worm length. Fluorescence measurements shown in Fig. 2F, Fig. 6D and Fig. S12  582 were performed by imaging animals using ImageXpress automated imaging system Nano imager 583 (Molecular Devices, LLC), followed by image analysis in FIJI. Mean gray value (as a ratio of 584 integrated density and analyzed area) was measured for each animal and normalized to the 585 background fluorescence. 586

Bioluminescence measurements 587
Synchronized L1 animals were grown at 20°C for 44 h and then treated with bortezomib or DMSO 588 for 4 h. Sample preparation and nanoluciferase bioluminescence measurements were performed 589 as previously described (22). In brief, animals were collected and disrupted using silicon carbide 590 beads in lysis buffer (50 mM HEPES pH 7.4, 1 mM EGTA, 1 mM MgCl2, 100 mM KCl, 10% 591 glycerol, 0.05% NP40, 0.5 mM DTT, protease inhibitor cOmplete (Sigma, catalog number 592 11836170001)). The lysates were centrifuged and the supernatants were collected and stored at 593 -80°C until bioluminescence was measured. Nano-Glo Luciferase Assay System reagent 594 (Promega, catalog number N1110) was added to the worm lysate supernatant before analysis, 595 and incubated at room temperature for 10 minutes. Analysis was performed on a NOVOstar plate 596 reader. The results were normalized to blank controls. 597 smFISH analysis 598 smFISH experiments were performed as previously described (12). In brief, L4 animals were 599 treated with bortezomib or DMSO for 4 h at 20°C. Animals were washed off the plates, fixed in 600 4% paraformaldehyde in phosphate-buffered saline + 0.1% Tween 20 (PBST) at room 601 temperature for 30 min, and incubated in 70% ethanol overnight at 4°C. Staining was performed 602 with 1 µM Cal Fluor 610 conjugated pals-5 smFISH probes (Biosearch Technologies) in smFISH 603 hybridization buffer (10% formamide, 2X SSC, 10% dextran sulfate, 2 mM vanadyl ribonucleoside 604 complex, 0.02% RNase free BSA, 50 μg E. coli tRNA) at 30°C in the dark overnight. Samples 605 were incubated in the wash buffer (10% formamide, 2X SSC) at 30°C in the dark for 30 min. 606 Vectashield + DAPI was added to each sample, and stained worms were transferred to 607 microscope slides and covered with glass coverslips. Z-stacks of the body region containing 608 anterior part of the intestine was performed using Zeiss AxioImager M1 compound microscope 609 with a 63X oil immersion objective. Image processing was performed using FIJI. smFISH spot 610 quantification was performed using StarSearch program 611 (http://rajlab.seas.upenn.edu/StarSearch/launch.html). When selecting the region of interest, the 612 anterior boundary of the first four intestinal cells was determined based on the prominent border 613 between pharynx and intestine, which is visible in the DIC channel. The posterior boundary was 614 set at the middle distance between DAPI-stained nuclei of the first and the second intestinal rings. 615

Rosetta (DE3) cells (Novagen) for protein expression. For expression, LB with carbenicillin/ 621
chloramphenicol was inoculated with Rosetta (DE3)/pBEL2159 and grown at 37°C with shaking 622 at 200rpm. The overnight culture was diluted 1:50 in LB+carbenicillin/chloramphenicol and then 623 induced by adding IPTG to a final concentration of 1mM at 16°C, and allowed to shake overnight. 624 Cells were harvested by centrifugation and resuspended in lysis buffer (50mM Tris pH8, 300mM 625 NaCl, 10mM Imidazole, 10% Glycerol, 1mM phenylmethylsulfonyl fluoride (PMSF)). Cells were 626 lysed using the Emulsiflex-C3 cell disruptor (Avestin) and then centrifuged at 4°C, 12,000g to 627 pellet cell debris. The pellet, containing a large amount of insoluble PALS-5, was resuspended in 628 urea lysis buffer (100 mM NaH2PO4/10 mM Tris base, 10 mM Imidazole, 8 M Urea [titrated to 629 pH8 by NaOH]). The solubilized pellet was centrifuged at 4000g, and the supernatant collected. Samples were boiled at 100°C for 10 min and stored at -30°C. Proteins were separated on a 10% 654 sodium dodecyl sulfate-polyacrylamide gel electrophoresis precast gel (Bio-Rad), and transferred 655 onto polyvinylidene difluoride (PVDF) membrane. 5% nonfat dry milk in PBST was used to block 656 for nonspecific binding for 2 h at room temperature. The membranes were incubated with primary 657 antibodies overnight at 4°C (rabbit anti-PALS-5 diluted 1:1,000 and mouse anti-tubulin (Sigma, 658 catalog number T9026) diluted 1:3000 in blocking buffer). Next, the membranes were washed 659 five times in PBST, and then incubated in horseradish peroxidase-conjugated secondary 660 antibodies at room temperature for 2 h (goat anti-rabbit (MilliporeSigma, catalog number 401315) 661 and goat anti-mouse (MilliporeSigma, catalog number 401215) diluted 1:10,000 in blocking 662 buffer). After five washes in PBST, the membranes were treated with enhanced 663 chemiluminescence (ECL) reagent (Amersham) for 5 min, and imaged using a Chemidoc XRS+ 664 with Image Lab software (Bio-Rad). Quantification of band intensities in 3 Western blot replicates 665 was performed using Image Lab software (Bio-Rad). PALS-5 band intensities for each sample 666 were normalized to the ratio of the tubulin expression levels between N2 DMSO control and a 667 given sample. 668

RNA isolation 669
Total RNA isolation was performed as previously described (15). Animals were washed off plates 670 using M9, then washed with M9 and collected in TRI reagent (Molecular Research Center, Inc.). 671 RNA was isolated using BCP phase separation reagent, followed by isopropanol and ethanol 672 washes. For RNA seq analysis, samples were additionally purified using RNeasy Mini kit from 673 Qiagen. 674

qRT-PCR analyses 675
qRT-PCR analysis was performed as previously described (15). In brief, cDNA was synthesized 676 from total RNA using iScript cDNA synthesis kit (Bio-Rad). qPCR was performed using iQ SYBR 677 Green Supermix (Bio-Rad) with the CFX Connect Real-Time PCR Detection System (Bio-Rad). 678 At least three independent experimental replicates were performed for each qRT-PCR analysis.

Comparisons of differentially expressed genes from different datasets 697
An R studio package GeneOverlap was used for RNA-seq datasets comparative analyses. 698 Differentially expressed genes from RNA-seq analyses from this study were compared with 699 relevant previously published datasets (11,13,15,24,25,34,(70)(71)(72). Statistical similarity 700 between datasets was determined using Fisher's exact test. The odds ratios, Jaccard indexes 701 and p-values were calculated. Total number of genes was set to 46902. Data are represented in 702 the contingency tables in which odds ratio and Jaccard index values are shown in the heat map 703 format, whereas p-values are indicated numerically. 704

Tissue-specific RNAi analysis 714
Tissue-specific RNAi analysis was performed using the feeding method. E. coli OP50-1 strain 715 was modified to enable zip-1 RNAi or control RNAi (L4440). Bacterial overnight cultures were 716 plated on NGM plates supplemented with 2.2 mM IPTG and 1 mM carbenicillin, and incubated at 717 room temperature for 3 or 4 days. 3000 synchronized L1 animals were transferred to prepared 718 plates and grown at 20°C for 48 h. Animals were then treated with bortezomib or DMSO as 719 described earlier. VP303 (rde-1) and MGH167 (sid-1) strains were used for intestinal RNAi; 720 NR222 (rde-1) strain was used for epidermal RNAi. Replicates that were included into analysis of 721 sid-1 mutants had at least 50-fold increase in pals-5 expression levels on control RNAi plates 722 following bortezomib treatment. This threshold allowed detection of any substantial decrease in 723 pals-5 induction in zip-1(RNAi) samples. 724

Killing assays 725
For N. parisii killing assays, about 150 L1 worms were mixed with 50 μl of a 10X concentration of 726 OP50-1 E. coli and 1 million N. parisii spores, and placed onto a 3.5 cm tissue culture-treated 727 NGM plate (3 plates for each strain). After 66 h of infection at 25°C, alive animals were transferred 728 onto new NGM plates containing only OP50-1 E. coli food (30 animals per plate, 3 plates per 729 worm strain). Animals were scored daily and alive animals were transferred to fresh NGM plates. 730 Data from 3 experimental replicates were merged and analyzed using Survival function in 731 GraphPad Prism 9; log-rank (Mantel-Cox) test was used for statistical analyses. 732

Longevity assays 733
For longevity assays, about 75 L1 worms were mixed with 50 μl of a 10X concentration of OP50-734 1 E. coli, and placed onto a 3.5 cm tissue culture-treated NGM plate (3 plates for each strain). 735 After 66 h incubation at 25°C, animals were transferred to new NGM plates supplemented with 736 OP50-1 E. coli food source (30 animals per plate, 3 plates per strain). Animals were scored daily 737 and alive animals were transferred to fresh NGM plates. Data from 3 experimental replicates were 738 merged and analyzed using Survival function in GraphPad Prism 9; log-rank (Mantel-Cox) test 739 was used for statistical analyses. 740

Thermotolerance assay 749
Animals were grown on NGM plates at 20°C until L4 stage. L4 animals were transferred to new 750 plates and exposed to heat shock at 37.5°C for 2 h. Recovery was performed at room temperature 751 for 1 h on a single layer, followed by 24 h incubation at 20°C. After this time, animals were scored 752 for viability based on their ability to move after touch. Three plates with 30 animals per plate were 753 analyzed for each strain. Three experimental replicates were performed. 754

Body length measurements 755
For body length analysis of wild-type and sid-1(-); vha-6p::sid-1 mutant strains (Fig. S10B), 756 synchronized L1 animals were placed on control or act-5 RNAi plates and allowed to grow at 20°C 757 for 48 h. For analysis of wild type, zip-1(jy13) and pnp-1(-) mutants (Fig. S13B), synchronized L1 758 animals were plated on NGM plates and allowed to grow at 20°C for 44 h. Animals were washed 759 off the plates with M9 and fixed in 4% paraformaldehyde (Fig. S10B) or anesthetized with 10 µM 760 levamisole (Fig. S13B). Animals were imaged using ImageXpress automated imaging system 761 Nano imager (Molecular Devices, LLC) in 96-well plates. Length of each animal was measured 762 using FIJI program. 50 animals were analyzed for each strain, in each of three experimental 763 replicates. 764

Data availability 765
RNA-seq reads were uploaded to the NCBI GEO database with Accession number GSE183361. 766 All data supporting this manuscript is available from the corresponding author upon request. 767 one-tailed t-test was used to calculate p-values; black asterisks represent significant difference 817 between the labeled sample and the wild-type DMSO control; red asterisks represent significant 818 difference between wild-type (WT) N2 and zip-1(jy13) bortezomib treated samples; **** p < 819 0.0001; *** p < 0.001; ** p < 0.01; * 0.01 < p < 0.05; p-values higher than 0.05 are not labeled. was observed in animals exposed to DMSO control, or in the non-transgenic control strain N2. 857 Composite images consist of merged fluorescent (GFP and autofluorescence) and DIC channels. replicates were analyzed, the values for each replicate are indicated with circles. Error bars 959 represent standard deviations. A one-tailed t-test was used to calculate p-values; black asterisks 960 represent significant difference between the labeled sample and the uninfected wild-type control; 961 red asterisks represent significant difference between infected wild-type and infected zip-1(jy13) 962 samples; **** p < 0.0001; *** p < 0.001; ** p < 0.01; * 0.01 < p < 0.05; p-values higher than 0.05 963 are not labeled. 964 965 966 Fig. S8. Correlation between zip-1-dependent genes and sta-1-regulated, ORR and skn-1-967 regulated genes. (A-C) Statistical similarity between zip-1-dependent gene set and genes 968 downregulated in sta-1(ok587) mutants (A), ORR genes (B) and genes upregulated following skn-969 1 downregulation (C). Fisher's exact test was used to calculate odds ratios and p-values. If odds 970 ratio is greater than one, two data sets are positively corelated. Jaccard index measures similarity 971 between two sets, with the range 0-1 (0no similarity, 1same datasets). For approximate 972 quantification, the odds ratio and Jaccard index color keys are indicated on the right side of each 973  pnp-1(-) and zip-1(jy13); pnp-1(-) animals. The results are shown as the fold change in gene 999 expression relative to control strain. All strains are in jyIs8  strain 1000 background. Three independent experimental replicates were analyzed, the values for each 1001 replicate are indicated with circles. Error bars represent standard deviations. A one-tailed t-test 1002 was used to calculate p-values; black asterisks represent significant difference between the 1003 labeled sample and the wild-type control; red asterisks represent significant difference between 1004 pnp-1(jy90) and zip-1(jy13); pnp-1(jy90) backgrounds; **** p < 0.0001; *** p < 0.001; ** p < 0.01; 1005 * 0.01 < p < 0.05; p-values higher than 0.05 are not labeled. 1006 Fig. S12. zip-1(jy13) and pnp-1(jy90) single and double mutants have similar accumulation 1009 of fluorescent beads. Quantification of fluorescent bead accumulation in the control strains, zip-1010 1(jy13); jyIs8, pnp-1(jy90); jyIs8 and zip-1(jy13); pnp-1(jy90); jyIs8 mutants. Mean fluorescence 1011 was measured in 150 animals per genotype; background fluorescence was subtracted. In the 1012 box-and-whisker plot, each box represents 50% of the data closest to the median value (line in 1013 the box). Whiskers span the values outside of the box. AUarbitrary units. A Kruskal-Wallis test 1014 was used to calculate p-values; **** p < 0.0001; ns indicates nonsignificant difference (p > 0.05). 1015 1029   Table S1. Results of RNAi screens. The expression of PALS-5::GFP reporter was analyzed in 1030 pals-22(jy3) mutant background. Expression of pals-5p::GFP reporter was analyzed in animals 1031 exposed to prolonged heat stress. The values of GFP intensity were normalized to the length of 1032 worms (TOF). 1033 Table S2. An overview of differentially expressed genes in animals treated with bortezomib 1034 and DMSO. Differentially expressed genes with adjusted p-value lower than 0.05 are listed for 1035 wild-type (N2) animals and zip-1(jy13) mutants. 1036  Table S7. RNA-seq statistics. Numbers of total and mapped reads are given for each sample 1047 and each replicate. R1, R2 and R3 represent replicate 1, 2 and 3 respectively. 1048 Table S8. Normalized counts for all mapped genes and samples from RNA-seq analysis. 1049