ABSTRACT
Intracellular pathogen infection leads to proteotoxic stress in host organisms. Previously we described a physiological program in the nematode C. elegans called the Intracellular Pathogen Response (IPR), which promotes resistance to proteotoxic stress and appears to be distinct from canonical proteostasis pathways. The IPR is controlled by PALS-22 and PALS-25, proteins of unknown biochemical function, which regulate expression of genes induced by natural intracellular pathogens. We previously showed that pals-22 and pals-25 regulate the mRNA expression of the predicted ubiquitin ligase component cullin cul-6, which promotes thermotolerance in pals-22 mutants. Here we use co-immunoprecipitation studies to define protein-protein interactions in the IPR and show that PALS-22 and PALS-25 physically associate with each other. We also identify a previously uncharacterized RING domain protein in the TRIM family named RCS-1 as a core component that acts with CUL-6 to promote thermotolerance. Furthermore, we show that the Skp-related proteins SKR-3, SKR-4 and SKR-5 act redundantly to promote thermotolerance in pals-22 mutants. These studies describe a newly defined CUL-6/RCS-1/SKR-3,4,5 cullin-RING ligase complex that promotes thermotolerance as part of the IPR in C. elegans and provide insight into how organisms cope with proteotoxic stress.
Maintaining protein homeostasis (proteostasis) after exposure to environmental stressors is critical for organismal survival (1). Several signaling pathways have been identified that help organisms cope with stressors that perturb proteostasis. For example, elevated temperature triggers the conserved Heat Shock Response (HSR) pathway, which helps organisms survive the toxic effects of heat (2). The HSR upregulates expression of chaperones that help with refolding of misfolded proteins, to prevent the formation of protein aggregates and restore proteostasis (1). Disruptions of proteostasis and the formation of protein aggregates in humans are associated with severe neurodegenerative and age-related diseases, such as Alzheimer’s and Huntington’s diseases (1, 3–5). Given the increasing incidence of such diseases and a lack of effective treatments (6, 7), new approaches are needed to identify protective mechanisms that provide resilience against proteotoxic stress.
Pathogen infection can perturb proteostasis and several studies in the nematode Caenorhabditis elegans have demonstrated intriguing connections between immune responses to extracellular pathogens and canonical proteostasis pathways (8–11). More recently, examining the C. elegans host response to natural intracellular pathogens has uncovered a novel stress response pathway that promotes proteostasis (12, 13). Microsporidia are intracellular, fungal-like pathogens that are the most common cause of infection of C. elegans in the wild, with Nematocida parisii being the most commonly found microsporidian species in C. elegans (14). N. parisii replicates exclusively inside the C. elegans intestine, and the infection is associated with hallmarks of perturbed proteostasis in the host, such as the formation of large ubiquitin aggregates in the intestine (12). Interestingly, the host transcriptional response to this infection is very similar to the host transcriptional response to another natural intracellular pathogen of the C. elegans intestine, the Orsay virus (12, 15, 16). These molecularly distinct pathogens induce a common mRNA expression pattern in C. elegans that we termed the “Intracellular Pathogen Response” or IPR (13). Of note, IPR genes can also be induced by proteotoxic stress such as prolonged heat stress (distinct from short-term, high temperature ‘heat shock’) or proteasome blockade (12). These findings suggested that the IPR could provide protection against proteotoxic stress.
Functional insights into the IPR came from analysis of mutants that constitutively express IPR genes. Forward genetic screens identified two genes encoding proteins of unknown biochemical function called pals-22 and pals-25 that comprise an ON/OFF switch for the IPR, with pals-22 acting as a repressor of pals-25, which is an activator of the IPR (13, 17). Upregulation of IPR gene expression in pals-22 loss-of-function mutants is accompanied by a rewiring of C. elegans physiology, including increased resistance against natural pathogens like N. parisii and the Orsay virus, as well as slowed growth and shortened lifespan. pals-22 mutants also have increased proteostasis capacity characterized by improved thermotolerance and lowered levels of aggregated proteins (13). All of the pals-22 mutant phenotypes are reversed in pals-22 pals-25 loss-of-function double mutants (17).
Interestingly, these phenotypes appear to be independent of canonical proteostasis factors (13), such as the transcription factors HSF-1 and DAF-16, which mediate the HSR (18). The IPR is also independent of the proteasome bounceback response (17). This stress response is triggered by blockade of the proteasome and is mediated by the transcription factor SKN-1/Nrf, which upregulates transcription of proteasome subunits to increase degradation capacity (19–21). Instead, the IPR requires expression of a cullin gene called cul-6, which is transcriptionally regulated by pals-22/pals-25 and by infection (13, 17). Cullins are components of E3 ubiquitin ligases, which are enzymes that catalyze transfer of ubiquitin onto substrate proteins in order to alter their fate (22). Based on these findings we hypothesized that CUL-6-mediated ubiquitylation of target proteins may act as a protein quality control mechanism in the IPR to respond to proteotoxic stress.
CUL-6 belongs to the cullin-RING Ligase (CRL) superfamily, which is found throughout eukaryotes and is the largest class of ubiquitin ligases (23). CRLs are multi-subunit enzyme complexes, a subset of which are Skp, cullin, F-box (SCF) complexes consisting of four subunits: a RING-box/RBX protein, a Skp, a cullin and an F-box protein. Due to the large number of SCF genes and their modular nature, there are many different enzyme combinations and thus a large number of different substrates that can be targeted by these enzymes. Interestingly, the SCF class of ubiquitin ligases appears to have undergone a significant expansion in the evolutionary lineage that gave rise to the nematode C. elegans (24, 25). For example, in comparison to other metazoans, C. elegans has a much larger number of F-box proteins, which serve as substrate adaptors and confer specificity to SCF ligases. There are around 520 F-box proteins in the C. elegans genome (24), in comparison to around 68 in humans (26), 22 in Drosophila and 11 in Saccharomyces cerevisiae (27). In addition, the number of core SCF components has increased in nematodes, with C. elegans having 22 Skp-related proteins in comparison to six in Drosophila, just one in S. cerevisiae, and one in humans (28). The SCF components upregulated as part of the IPR include not only cul-6 as mentioned above, but also the Skp-related proteins skr-3, skr-4 and skr-5 (12). If cul-6 were functioning as part of a ubiquitin ligase complex to promote proteostasis as part of the IPR, it should be acting with skrs and other CRL components.
Here we describe how CUL-6 promotes proteostasis as part of a novel CRL complex in C. elegans. First, we show that tissue-specific expression of cul-6 in either the intestine or in the pharynx can promote thermotolerance. Importantly, we found that CUL-6 requires a conserved neddylation site to promote thermotolerance, indicating that its activity is regulated via a canonical NEDD8 modification, which is a hallmark of bona fide CRLs (29). To identify other components in this complex we performed co-immunoprecipitation with CUL-6, as well as with PALS-22 and PALS-25, which are antagonistic paralogs that regulate expression of IPR genes (17). We found that PALS-22 and PALS-25 proteins physically associate, indicating that they interact not only genetically, but also biochemically. We also identified several CUL-6 interaction partners, including a previously uncharacterized RING domain protein called C28G1.5. We generated deletion mutations of C28G1.5 and found that they suppressed the increased thermotolerance of pals-22 mutants, just like cul-6 mutations. Testing different combinations of skr-3, skr-4 and skr-5 double mutants we found these skr genes act redundantly to promote thermotolerance in pals-22 mutants. Finally, we developed an RNAi system to test the role of other proteins that promote thermotolerance in a pals-22 mutant background. In summary, this work demonstrates that CUL-6 is part of a newly defined CRL that promotes thermotolerance as part of the IPR, and broadens our understanding of how animals survive exposure to proteotoxic stress such as heat shock.
Results
The cullin CUL-6 acts in the intestine and pharynx to promote thermotolerance
In comparison to wild type animals, pals-22 loss-of-function mutants have increased thermotolerance, which is reduced to wild-type levels in pals-22; cul-6 double mutants. Our previous results indicated that pals-22 regulates thermotolerance in the intestine, where it also regulates cul-6 mRNA expression (13). Because cul-6 is expressed in the both the intestine and the pharynx, we sought to determine in which of these tissues cul-6 acts to promote thermotolerance. Therefore, we designed tissue-specific rescue constructs with cul-6 cDNA using the Mos1-mediated Single Copy Insertion (MosSCI) system (30). With the MosSCI technique, we drove expression of GFP-tagged versions of cul-6 using intestinal (vha-6p) or pharyngeal promoters (myo-2p). We found that both the intestinal and pharyngeal rescue strains expressed GFP::CUL-6 with the expected tissue distribution pattern (Figure 1A). We then crossed these tissue-specific CUL-6 MosSCI transgenes into pals-22; cul-6 double mutants to test for rescue of thermotolerance. Here we found that expression of cul-6 in either the intestine or the pharynx was sufficient to increase the thermotolerance of pals-22; cul-6 double mutants (Figure 1B).
Our previous results demonstrated that cul-6 had a role in proteostasis only in a pals-22 mutant background, and not in a wild-type background (13). This finding suggested that cul-6 affects proteostasis only when it is overexpressed compared to wild-type animals, such as in pals-22 mutants. Consistent with this model, here we found that overexpression of CUL-6 from a multi-copy array (CUL-6 tagged at the C-terminus with GFP and 3xFLAG, surrounded by ~20kb endogenous regulatory region (31)) led to increased thermotolerance compared to wild-type animals (Fig. 1B). Furthermore, we found that either pharyngeal or intestinal expression of cul-6 cDNA in a wild type background promoted thermotolerance (Figure 1B). Importantly, transgenic strains with vha-6 and myo-2 promoters driving genes other than wild-type cul-6 did not have increased thermotolerance (see text below, and Figure 1B). Thus, increased expression of cul-6 in a wild-type background in either the intestine or the pharynx leads to increased thermotolerance.
We were curious if expressing cul-6 in tissues where it is not normally expressed would also promote thermotolerance. Therefore, we generated a myo-3p::cul-6 construct to determine whether expression in body wall muscle could promote thermotolerance. Surprisingly, we found that cul-6 expression in body wall muscle appeared to be toxic, because we failed to recover myo-3p::cul-6 transgenic animals after several rounds of germline injections. To perform a careful assessment of lethality, we injected either red fluorescent protein markers together myo-3p::cul-6, or red fluorescent protein markers alone (see Materials and Methods). Here we found none of the eggs expressing red fluorescent protein markers hatched when co-injected with myo-3p::cul-6 (0/87), while more than half of the eggs hatched when injected with the red fluorescent protein markers alone (48/84). The reason for toxicity of cul-6 expression in body wall muscle is unclear.
The activity of cullin-RING ubiquitin ligases can be regulated by neddylation, which is the process of conjugating the ubiquitin-like protein Nedd8 onto a cullin protein at a conserved lysine residue (29). To determine whether CUL-6 might be regulated by neddylation, we identified the lysine residue that would likely be targeted for neddylation and mutated this residue into an arginine, which would be predicted to disrupt neddylation (Supplementary Figure 1) (32). We used the MosSCI technique to generate a strain that contains this vha-6p::cul-6(K673R) transgene (expression visualized in Figure 1A), and found that it could not rescue the thermotolerance of pals-22; cul-6 mutants (Figure 1B). Furthermore, unlike vha-6p::cul-6(wild-type), the vha-6p::cul-6(K673R) transgene did not promote thermotolerance in a wild-type background. These results suggest that CUL-6 requires neddylation to promote thermotolerance, which is consistent with CUL-6 acting as part of a ubiquitin ligase complex.
Co-immunoprecipitation/mass spectrometry analysis identifies CUL-6 binding partners
To identify the components of a CUL-6-containing ubiquitin ligase complex, we performed co-immunoprecipitation mass spectrometry (co-IP/MS) analysis to identify binding partners of CUL-6. Here we used the C. elegans strain with GFP::3xFLAG-tagged CUL-6, which is functional for thermotolerance (Figure 1). We also used similar GFP::3xFLAG-tagged strains for PALS-22 and PALS-25 (13, 17), which regulate mRNA expression of cul-6 and other IPR genes. Through analysis of binding partners for PALS-22 and PALS-25, we sought to obtain insight into their biochemical function, which is currently unknown. Two proteins, GFP::3xFLAG alone and an unrelated protein F42A10.5::GFP::3xFLAG, were added as controls for the co-IPs.
We confirmed transgene expression using immunoblotting and microscopy analysis. First, western analysis indicated that all five fusion proteins showed bands of the predicted molecular weight (Supplementary Figure 2A). Second, when we treated animals with the proteasome inhibitor Bortezomib to induce proteotoxic stress and IPR gene expression, we saw an increase in CUL-6 expression by both western and microscopy analysis, as expected from previous studies (Supplementary Figure 2A-B). As mentioned above, expression of CUL-6 is seen most strongly in the pharynx and the first two rings of the intestine. PALS-22 and PALS-25 are broadly expressed throughout animal and their expression is not affected by IPR activation (17).
We then performed three independent co-IP/MS experiments with these FLAG-tagged transgenic strains. Animals were treated with Bortezomib to induce IPR gene expression, proteins were extracted, and then subjected to IP with anti-FLAG antibodies. Mass spectrometry analysis was performed on the IP material and spectral counts were used to calculate the fold change and adjusted p-values between the experimental IPs and the control IPs (Supplementary Table 1). Proteins that were significantly more abundant in the experimental IP compared to either of the control IP’s (GFP alone control or F42A10.5 control, at adjusted p-value < 0.05 and log2 fold change > 1), were considered interacting proteins (Supplementary Table 1).
Co-IP/MS of PALS-22 identified 23 binding partners, including PALS-25 as one of the most highly enriched binding partners, as compared to co-IP/MS of control proteins (Supplementary Figure 3). PALS-22 also is physically associated with PALS-23, which is a PALS protein of unknown function, as well as with F26F2.1, which is a protein of unknown function induced by the IPR (12). Co-IP/MS of PALS-25 identified 7 binding partners, with PALS-22 being by far the most highly enriched hit when compared with co-IP/MS of either control protein. These reciprocal co-IP results indicate that PALS-22 and PALS-25 physically associate with each other, perhaps directly binding to each other.
Co-IP/MS of CUL-6 identified 26 significant binding partners. These proteins included predicted SCF ubiquitin ligase components, such as the Skp-related protein SKR-3 and the F-box proteins FBXA-75 and FBXA-158 (Figure 2A, B). Additionally, 6 subunits of the 26S proteasome were identified (RPT-3, 4 and RPN-5, 6.1, 8, 9). An SCF ubiquitin ligase complex canonically contains an RBX RING box protein, which interacts both with a cullin and with an E2 enzyme to stimulate ubiquitin chain formation on targets recognized by an F-box protein in the complex. C. elegans has two RBX proteins, RBX-1 and RBX-2, but neither of these proteins were identified as significant binding partners for CUL-6 in the co-IP/MS. Therefore, we searched for other CUL-6 binding proteins that have a RING domain. Here, we identified a single candidate, C28G1.5. Because of results described below, we renamed C28G1.5 as RING protein acting with cullin and Skr proteins (RCS-1). Interestingly, rcs-1 mRNA expression, like cul-6 mRNA expression, is higher in pals-22 mutants compared to wild-type animals (log2 FC=2.81, adjusted p-value= 4.83E-08), and in pals-22 mutants compared to pals-22 pals-25 mutants (log2 FC=2.10, adjusted p-value=3.898E-07) (17).
RING domain protein RCS-1 acts with CUL-6 to promote thermotolerance in pals-22 mutants
The rcs-1 gene had no previously described role in C. elegans, and has two isoforms: rcs-1b has a RING domain and a B-box domain, and rcs-1a has the RING domain only (Figure 3A). The closest potential ortholog of rcs-1 in humans is the tripartite motif-containing protein 23 (TRIM23). The TRIM family is named for having three motifs (RING finger, B box domain, and coiled coil domain), and many TRIM proteins have E3 ubiquitin ligase activity (33). However, RCS-1 and TRIM23 do not appear to be orthologous, as sequence comparison of TRIM23 against the C. elegans genome identifies other C. elegans RING proteins as closer matches than RCS-1 (see below). To investigate how RCS-1 is related to TRIM23, and other C. elegans RING and TRIM proteins, we performed phylogenetic analysis of the full-length RCS-1 (RCS-1B) using a Bayesian inference method (Supplementary Figure 4). Based on this analysis, RCS-1 seems to be part of the C. elegans TRIM protein family (34). However, several C. elegans proteins like the ADP-Ribosylation Factor related proteins (ARF-3 and ARF-6) are more closely related to TRIM23 than RCS-1 is. In addition, RCS-1 does not seem to be closely related to C. elegans RING proteins RBX-1 or RBX-2.
To investigate the role of RCS-1 in promoting thermotolerance with CUL-6, we used a CRISPR-Cas9 strategy to generate deletion alleles of rcs-1. Two full deletion alleles of rcs-1 (jy84 and jy105) were generated, and then crossed in pals-22 mutants. Here we found for both rcs-1 alleles that pals-22; rcs-1 double mutants had thermotolerance similar to wild-type animals, indicating that rcs-1 is required for the increased thermotolerance of pals-22 mutants (Figure 3B, C). If RCS-1 were acting in a CRL together with CUL-6, then loss of rcs-1 would not further lower thermotolerance in a pals-22; cul-6 mutants. Indeed, we found that pals-22; cul-6; rcs-1 triple mutants had a similar level of thermotolerance to pals-22; cul-6 mutants, as well as to pals-22; rcs-1 mutants and wild-type animals. These findings, together with CUL-6 co-IP results, are consistent with RCS-1 being the RING domain protein that acts with CUL-6 in a ubiquitin ligase complex.
SKP-related proteins SKR-3, SKR-4 and SKR-5 act redundantly to promote thermotolerance in pals-22 mutants
In addition to a cullin and a RING protein, SCF ubiquitin ligase complexes contain a Skp protein. Expression of three Skp-related proteins (skr-3, skr-4 and skr-5) is upregulated by both intracellular infection and mutation of pals-22, similar to cul-6 expression (12, 13). Thus, we speculated that these genes might act together with cul-6 in an SCF complex. However, we previously found that mutation of either skr-3, skr-4 or skr-5 alone in a pals-22 mutant background had no effect on thermotolerance (13). Therefore, we examined whether there may be redundancy among these genes. We made all of the possible combinations of skr-3, skr-4 and skr-5 as double mutants and then crossed them into a pals-22 mutant background. Here we found that pals-22; skr-3 skr-5 and pals-22; skr-5 skr-4 triple mutants had a significant reduction in thermotolerance compared to pals-22 mutants, with levels similar to wild type animals. In contrast, pals-22; skr-3 skr-4 mutants had thermotolerance similar to pals-22 mutants (Figure 4B). These results indicate that either SKR-3, SKR-4, or SKR-5 individually can act together with CUL-6 in a SCF to promote thermotolerance. However, SKR-3 and SKR-4 only seem to be important if SKR-5 has been lost, whereas SKR-5 is important if either SKR-3 or SKR-4 has been lost. Interestingly, skr-3 and skr-4 share high sequence homology, and skr-5 is the most highly induced skr gene in pals-22 mutants, suggesting that skr-5 may play a unique role (13). Importantly, none of the double mutants had an effect on thermotolerance in a wild-type background, consistent with their acting in a complex with cul-6, which also does not affect thermotolerance when mutated in a wild-type background.
In addition to regulating thermotolerance, pals-22 also regulates developmental rate of C. elegans, with pals-22 mutants being slower to develop to the fourth larval (L4) stage compared to wild-type animals. Our previous analysis indicated that while cul-6 mutations suppress the effect of pals-22 on thermotolerance, they do not suppress the effects of pals-22 on developmental rate. Similarly, we found that mutations in rcs-1, skr-3 skr-4, skr-3 skr-5, or skr-5 skr-4 do not suppress the slowed developmental rate in pals-22 mutants (Supplementary Figure 5).
To confirm the phenotypes observed using the skr mutant allele and to more easily analyze factors that may be functionally redundant, we sought to use RNA interference (RNAi) in lieu of generating mutants and performing crosses. However, we found that pals-22 mutants do not have increased thermotolerance compared to wild type animals when fed on the standard E. coli bacteria used for feeding RNAi (HT115) (Supplementary Figure 6A). This effect is likely due to dietary differences between HT115 and the OP50 strain used for thermotolerance experiments described above (35). Therefore, we tested thermotolerance on the OP50 strain (R)OP50 that was modified to enable feeding RNAi studies (36). Here we found that pals-22 mutant animals fed on these RNAi-competent OP50 bacteria (transformed with empty vector L4440) have increased thermotolerance compared to wild type animals, which is restored back to wild type levels in pals-22; cul-6 mutants (Supplementary Figure 6A). This system thus enables RNAi analysis of factors that promote thermotolerance in a pals-22 mutant background. Using this system, we were able to successfully transfer the cul-6 RNAi clone into this strain, and we confirmed that RNAi knocked down expression of cul-6, by measuring reduction of expression from a CUL-6::GFP::3xFLAG transgene (Supplementary Figure 6B, C). We also found that cul-6 RNAi suppressed the enhanced thermotolerance of pals-22 mutants (Supplementary Figure 6D). With this system, we confirmed that skr-3 and skr-5 act redundantly to promote thermotolerance, as RNAi against skr-3 in a skr-5 mutant background suppressed thermotolerance of pals-22 mutants (Figure 4C). Thus, this improved RNAi system provides an independent means of assessing the role of factors that regulate thermotolerance in a pals-22 mutant background. Altogether, these studies indicate that a RCS-1/CUL-6/SKR-3,4,5 ubiquitin ligase complex promotes thermotolerance in C. elegans.
Discussion
Thermal stress is one of many types of proteotoxic stress that can impair organismal health and survival. Here we used a combination of genetics and biochemistry to broaden our understanding of a recently identified proteostasis pathway called the IPR, which enables animals to survive exposure to thermal stress in a manner distinct from the canonical heat shock response (HSR). Specifically, we demonstrate that the IPR involves the CUL-6/cullin acting as part of newly described ubiquitin ligase that appears to function in either the intestine or the pharynx of C. elegans. This CUL-6-containing CRL includes a previously uncharacterized RING protein we named RCS-1, as well as the Skp-related proteins SKR-3, SKR-4, and SKR-5 (Figure 5). We propose that this RCS-1/CUL-6/SKR-3,4,5 ubiquitin ligase complex is able to target proteins for ubiquitin-mediated proteasomal degradation and that this activity is a critical part of the IPR program. Consistent with this model, our co-IP/MS identified several proteasomal subunits that interact with CUL-6.
We also investigated protein-protein interactions of the PALS-22 and PALS-25 proteins, which comprise an ON/OFF switch in the IPR that regulates mRNA expression of cul-6, skr-3,4,5 and rcs-1. Previous studies indicated that pals-22 and pals-25 are in the same operon and interact genetically, and our co-IP/MS studies indicate they also interact biochemically. Notably, the pals gene family has expanded in the C. elegans genome, with 39 genes in C. elegans, in comparison to only 1 pals gene each in mouse and human (37). The only PALS proteins with identified phenotypes are PALS-22 and PALS-25, and it will be interesting to determine the phenotypes of the other pals genes in C. elegans, as well as the role of the single pals homolog in mammals. Although the divergent ‘pals’ protein signature that defines PALS proteins is of unknown function, PALS-22 does have weak homology with F-box proteins (37), leading to the speculative idea that PALS proteins function as adaptor proteins in ubiquitin ligase complexes. Biochemical studies of PALS-22 and PALS-25 will help elucidate how these antagonistic paralogs control expression of IPR genes, including SCF ligase components like cul-6, skr-3,4,5 and rcs-1.
C. elegans has a greatly expanded repertoire of SCF ligase components. This expansion has been suggested to reflect the results of an arms race against intracellular pathogens, as SCF components are among the most rapidly diversifying genes in the C. elegans genome (25). The most dramatically expanded class of SCF components includes ~520 F-box adaptor proteins in C. elegans (24). C. elegans also has 22 SKR proteins, in comparison to only 1 Skp in humans, indicating there has been an expansion of core SCF components as well. Here we identified SKR-3 as a binding partner for CUL-6, which is consistent with previous 2-hybrid results (28). Interestingly, we found redundancy in the role of SKRs at the phenotypical level: either SKR-3, SKR-4 or SKR-5 appear capable of acting with CUL-6 to promote thermotolerance.
Canonical CRLs contain an RBX protein as the RING domain protein (23). C. elegans has only two RBX proteins, RBX-1 and RBX-2, suggesting that there was not expansion of these core components in C. elegans. However, our co-IP/MS studies with CUL-6 did not identify these proteins as interacting partners for CUL-6, but rather identified RCS-1. Given our genetic and biochemical results, we propose that RCS-1 plays the same role as an RBX protein would in a canonical SCF complex (Figure 5). RCS-1 does not have a clear human ortholog, but the closest human protein is TRIM23 (38). Interestingly, the TRIM family contains 68 genes in humans, many of which encode single subunit E3 ubiquitin ligases, including those that restrict viral infection, and regulate inflammatory signaling (33). C. elegans has 18 TRIM proteins, and they appear to have a simpler structure than human TRIM proteins, in that they do not have additional motifs in the C-terminal domains, which are found in the majority of mammalian TRIM proteins (34). If these other C. elegans TRIM proteins can act in SCF ligases like RCS-1, it indicates there may also be an expansion of the RING core SCF components in C. elegans, in addition to the previously described expansion of SKRs and adaptor proteins.
The role of CUL-6 in promoting thermotolerance was first demonstrated in pals-22 mutants, where there is overexpression of CUL-6, as well as several other SCF components, including skr-3,4,5 and rcs-1 (13, 17). However, here we found that animals overexpressing only CUL-6, without the other components of the SCF, have increased thermotolerance. One explanation for this result is that CUL-6 is the limiting factor in a SCF that promotes thermotolerance in the IPR. Consistent with this idea, other components of the complex like SKR-3,4,5 are functionally redundant for thermotolerance, so even basal expression level might be sufficient to build a functional SCF, once CUL-6 expression increases past a certain threshold level. Of note, we have yet to find a phenotype for cul-6 mutants in a wild-type background using a standard heat shock resistance assay. This assay uses an acute heat shock, a stressor that is not sufficient to upregulate cul-6 expression: previous studies showed that prolonged heat stress is required to induce cul-6 and other IPR gene expression (12, 13, 39). Perhaps through a combination of prolonged heat stress to induce cul-6 expression, followed by acute heat shock to induce lethality, we may find a role for cul-6 mutants in a wild-type background. If so, upregulation of a RCS-1/CUL-6/SKR-3,4,5 ubiquitin ligase complex might be part of a chronic heat stress response that follows the acute response, mediated by the canonical HSR.
What substrate adaptor protein(s) acts in the RCS-1/CUL-6/SKR-3,4,5 ubiquitin ligase complex, and what substrate(s) does it target? It is possible that the effects of this complex are mediated through targeting a single regulatory protein for ubiquitylation and degradation. For example, ubiquitylation of the DAF-2 insulin receptor by the ubiquitin ligase CHIP can alter DAF-2 trafficking, and it appears that ubiquitylation of just this one target has significant effects on proteostasis in C. elegans (40). In contrast, CRL2 and CRL4 complexes in humans have recently been shown to target the C-termini of a large number of proteins for degradation as part of a newly identified protein quality control system (41, 42), Given the modular nature of the SCF ligase family, the large number of F-box substrate adaptor proteins in C. elegans, and the redundancy we found in the SKR-3,4,5 proteins, it seems possible that there are multiple substrates and adaptors used in the IPR. An exciting possibility is that the IPR involves distinct CRLs that ubiquitylate several different targets to improve proteostasis and tolerance against environmental stressors, including infections. Identifying these factors will be the subject of future studies.
Experimental Procedures
Cloning and generation of cul-6 tissue-specific rescue strains
A full list of constructs and strains used in this study is in Supplementary Table 2. To generate the vha-6p::CUL-6 transgene (pET499), vha-6p::SBP::3XFLAG, cul-6 cDNA, and the unc-54 3’ UTR were assembled in pCFJ150 using Gateway LR. To generate myo-2p::CUL-6 (pET686) and myo-3p::CUL-6 (pET687) constructs, the promoters myo-2p, myo-3p and the pET499 linearized backbone without vha-6p promoter were amplified by PCR from pCFJ90, pCFJ104 and pET499, respectively and assembled by Gibson recombination (43). To generate the cul-6(K673R) construct (pET688), a 86 bp single strand oligonucleotide and a linearized pET499 backbone that had been amplified by PCR were assembled by Gibson recombination. To generate the vha-6p::3XFLAG::GFP (pET554) transgene, vha-6p::3XFLAG::GFP and let-858 3’ UTR were assembled in pCFJ150 using Gateway LR. Mos1-mediated Single Copy Insertion (MosSCI) was performed as described previously (30). Briefly, the plasmid of interest (25 ng/μl) was injected with pCFJ601 (50 ng/μl), pMA122 (10 ng/μl), pGH8 (10 ng/μl), pCFJ90 (2.5 ng/μl), and pCFJ104 (5 ng/μl) into the EG6699 strain. Injected animals were incubated at 25°C for 7 days and then subjected to heat shock for 2h at 34°C. After 5 h non-Unc animals were selected and the presence of the insertion was verified by PCR and sequencing.
Transgenic strains with TransgeneOme fosmids (Supplementary Table 2) were generated as extrachromosomal arrays (31) by injecting into ttTi5605; unc-119(ed3) worms (strain EG6699) and then selecting non-Unc worms.
Lethality scoring of myo-3p::cul-6 tissue-specific rescue strains
To score the lethality of myo-3p::cul-6 expression, worms were injected with a complete MosSCI mix (see above) containing either myo-3p::cul-6 as the plasmid of interest, or water. Injected animals were incubated at 25°C and after 1 day, eggs expressing red fluorescence were transferred onto new plate at 25°C for 24 h. The hatching ratio of transferred eggs was then scored for both conditions. The assay was repeated two independent times.
Thermotolerance assays
Animals were grown on standard NGM plates at 20°C. L4 stage animals were transferred onto fresh NGM plates seeded with OP50-1 and then subjected to heat shock for 2 h at 37.5°C. The plates were then placed in a single layer on a benchtop at room temperature for 30 min, and then transferred to a 20°C incubator. Then, 24 h later the survival was scored in a blinded manner. Worms not responding to touch were scored as dead, and 30 worms were scored per plate. Three replicate plates were scored for each strain per experiment, and each experiment was performed at least three independent times. Statistical significance was tested using one-way ANOVA and Tukey’s HSD for post-hoc multiple comparisons.
Co-immunoprecipitation
Each sample for co-IP/MS was prepared in 3 independent experimental replicates. For each sample, 200,000 synchronized L1 animals were transferred onto NGM plates and grown for 48 h at 20°C. Bortezomib was added to reach a final concentration of 22 μM or the equivalent volume of DMSO for the control plates. After 6 h at 20°C worms were washed off of the plates with M9, washed twice with M9, resuspended in 500 μl of ice-cold lysis buffer (50mM HEPES, pH7.4, 1mM EGTA, 1mM MgCl2, 100mM KCl, 1% glycerol, 0.05% NP40, 0.5mM DTT, 1x protease inhibitor tablet) and immediately frozen dropwise in liquid N2. Frozen pellets were ground into powder with a pre-chilled mortar and pestle. Protein extracts were spun for 15 min 21,000g at 4°C and supernatants were filtered on 0. 45 μm filters (WHATMAN). Protein concentration was determined using Pierce 660nm protein assay and adjusted to 1 μg/μl with fresh lysis buffer. 1 mg of each sample was mixed with 25 μl of ANTI-FLAG M2 Affinity Gel (Sigma) and incubated at 4°C with rotation (12 rpm) for 1h. The resin was washed twice with 1 ml lysis buffer, twice with 1 ml lysis buffer for 5min, twice with 1 ml wash buffer (50mM HEPES, pH7.4, 1mM MgCl2, 100mM KCl) and 20min with 1 ml wash buffer with rotation. The liquid was removed and the beads were then stored at −80°C.
Trypsin digestion
The immunoprecipitated proteins bound to the beads were digested overnight in 400 ng trypsin (Sigma, V511A) in 25 mM ammonium bicarbonate (Sigma) at 37°C. Samples were then reduced with 1mM final concentration of Dithiothreitol (DTT, Acros Organics) for 30min and alkylated with 5mM final concentration of Iodoacetamide (IAA, MP Biomedicals, LLC) for 30min in dark. The peptides were extracted from the beads by adding 50 μL of 5% Formic acid (Sigma). The extraction was repeated one more time and the eluted peptides were combined. Digested peptides were desalted using Stage-Tip, C18 peptide cleanup method. The eluates were vacuum dried, and peptides were reconstituted in 15 μL of 5% Formic acid, 5% Acetonitrile solution for LC-MS-MS analysis.
LC-MS-MS Analysis
Samples were analyzed in triplicate by LC-MS-MS using an EASY-nLC 1000 HPLC (Thermo Scientific) and Q-Exactive mass spectrometer (Thermo Scientific, San Jose, CA) as described previously (44) with the following modifications. The peptides were eluted using a 60min acetonitrile gradient (45 min 2%-30% ACN gradient, followed by 5 min 30-60% ACN gradient, a 2min 60-95% ACN gradient, and a final 8min isocratic column equilibration step at 0% ACN) at 250nL/minute flow rate. All the gradient mobile phases contained 0.1% formic acid. The data dependent analysis (DDA) was done using top 10 method with a positive polarity, scan range 400-1800 m/z, 70,000 resolution, and an AGC target of 3e6. A dynamic exclusion time of 20 s was implemented and unassigned, singly charged and charge states above 6 were excluded for the data dependent MS/MS scans. The MS2 scans were triggered with a minimum AGC target threshold of 1e5 and with maximum injection time of 60 ms. The peptides were fragmented using the normalized collision energy (NCE) setting of 25. Apex trigger and peptide match settings were disabled.
Peptide and Protein Identification and Quantification
The RAW files obtained from the MS instrument were converted into mzXML format. The SEQUEST search algorithm was used to search MS/MS spectra against the concatenated target decoy database comprised of forward and reverse, reviewed C. elegans FASTA sequences from Uniprot (downloaded on 6/8/2015) along with GFP and E. coli sequences appended in the same file. The search parameters used were as follows: 20 ppm peptide mass tolerance; 0.01 Da fragment ion tolerance; Trypsin (11 KR P) was set as the enzyme; maximum 2 missed cleavages were allowed; Oxidation on methionine (15.99491 Da) and n-term acetylation (42.01056 Da) were set as differential modifications; static modification (57.02146 Da) was set on cysteine for alkyl modification. Peptide matches were filtered to a peptide false discovery rate (FDR) of 2% using the linear discrimination analysis. The protein level matches were filtered at 2% FDR using the protein sieve analysis. The spectral counts from the triplicates were then summed and used for the data analysis.
Analysis of mass spectrometry data
Peptide spectral counts were used to calculate fold change ratio, p-value and adjusted p-value between sample IPs and control IPs (GFP and F42A10.5) using the DEP package in R (45). Briefly the data were filtered to keep only the peptides present in at least two replicates in one condition. Filtered data were normalized using Variance Stabilizing Normalization. Missing values were imputed using the MiniProb method from the DEP package by randomly selecting values from a Gaussian distribution centered on a minimal value of the dataset. Fold change ratio and adjusted p-values were calculated. Proteins were considered as significant with adjusted p-value < 0.05 and log2 fold change > 1. Any protein with a negative log2 fold change with one control or the other (i.e. more affinity to the control protein than the tested bait) was considered as non-significant.
Phylogenic analysis of RCS-1
Amino acid sequences of 15 proteins were aligned using MUSCLE (version 3.7) and trimmed with trimAI (version 1.3) (46) using phylemon2 online platform (47). Bayesian Markov chain Monte Carlo inference (LG + I + G + F) was performed using BEAST (version 1. 10.4) (48). Analysis was run using a Yule model tree prior, an uncorrelated relaxed clock (lognormal distribution) and a chain length of 10 million states sampled every 1,000 iterations. Results were assessed with Tracer (version 1. 7.1), maximum clade credibility tree was built after a 25% burn-in. Posterior probability values greater than 0.5 are marked on branch labels.
CRISPR deletion of rcs-1
To generate the 2.18 kb deletion alleles of rcs-1, a co-CRISPR strategy was used, adapted from the IDT proposed method for C. elegans. We designed two crRNA encompassing the whole rcs-1 locus (crRNA1: 5’-GTTTGTTGAAGGAAATGCACAGG-3’, crRNA2: 5’-GGTTTCCTATAGCTGTGACACGG-3’).
These two oligonucleotides were synthetized by IDT, and used with a crRNA targeting the dpy-10 gene (dpy-10 crRNA3:5’-GCTACCATAGGCACCACGAG-3’) and assembled with commercial tracrRNA (50 μM crRNA1 and crRNA2, 25 μM dpy-10 crRNA, and 40 μM tracrRNA). After an annealing step for 5 min at 95°C, the resulting guide RNA mixture was added to CAS9-NLS protein (27 μM final – ordered from QB3 Berkeley) and microinjected into N2 worms. F1 Dpy progeny were screened by PCR for a deletion and positive lines were backcrossed 4 times to the N2 background before testing.
RNA interference assays
RNA interference assays were performed using the feeding method. Overnight cultures of HT115 or OP50 strain (R)OP50 modified to enable RNAi (36) (gift from Meng Wang lab, Baylor College of Medicine) were plated on RNAi plates (NGM plates supplemented with 5 mM IPTG, 1 mM carbenicillin) and incubated at 20°C for 3 days. Gravid adults were transferred to these plates, and their F1 progeny (L4 stage) were transferred onto new RNAi plates before being tested for thermotolerance as previously described.
Western blot analysis
For each strain, 1500 synchronized L1 worms were placed on NGM plates seeded with OP50 bacteria and incubated at 20°C for 48 h. These animals were then treated with Bortezomib (22 μM final) or control DMSO for 6 h and washed off the plate with M9. Proteins were extracted in 200μl lysis buffer (50 mM HEPES, pH7.4, 1 mM EGTA, 1 mM MgCl2, 100 mM KCl, 1% glycerol, 0.05% NP40, 0.5 mM DTT, 1x protease inhibitor tablet) as previously described (13). Protein levels were determined using the Pierce 660nm assay. Equal amount of proteins (5 μg) were boiled in protein loading buffer, separated on a 5-20% SDS-PAGE precast gel (Bio-Rad) and transferred onto PVDF membrane. Nonspecific binding was blocked using 5% nonfat dry milk in PBS-Tween (0.2%) for 1 hour at room temperature. The membranes were incubated with primary antibodies overnight at 4°C (mouse anti-FLAG diluted 1:1,000 and mouse anti-Tubulin diluted 1:7,500), washed 5 times in PBS-Tween and blotted in horseradish peroxidase-conjugated secondary antibodies at room temperature for 2 h (Goat anti-mouse diluted 1:10,000). Membranes were then washed 5 times in PBS-Tween, treated with ECL reagent (Amersham), and imaged using a Chemidoc XRS+ with Image Lab software (Bio-Rad).
Florescence microscopy
For cul-6 tissue-specific expression lines shown in Figure 1, images were taken using a Zeiss LSM700 confocal microscope with 10X and 40X objectives. For RNAi knock-down in Supplementary Figure 6, L4 stage F1 progeny of ERT422 worms fed with OP50 expressing cul-6 RNAi were anesthetized using 10 μM levamisole in M9 buffer and mounted on 2% agarose pads for imaging with a Zeiss LSM700 confocal. Using ImageJ software (version 1.52e), GFP signal in the pharynx and the first intestinal cells ring was measured, as well as three adjacent background regions. The total corrected fluorescence (TCF) was calculated as TCF = integrated density – (area of selected cell × mean fluorescence of background readings). For each condition, 30 worms were imaged. Significance was assay with Student’s t-test.
Measurement of developmental rate
40 – 50 gravid adults were transferred onto standard 10 cm NGM plate and incubated at 20°C to lay eggs for 2h before being removed. These plates were incubated at 20°C and the proportion of eggs that hatched and developed into L4’s was scored at 48 h, 64 h, and 72 h by scoring 100 animals each replicate.
Conflict of interest
The authors declare that they have no conflicts of interest with the contents of this Article.
The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Author contributions
JP, ERT and EJB designed the experiments. JP, KCR, AF, RJL and ERT performed the experiments and analyzed the results. JP wrote the paper with ERT.
Acknowledgements
We thank Spencer Gang, Vladimir Lazetic, Ivana Sfarcic, Jessica Sowa, and Eillen Tecle for comments on the manuscript. This work was supported by NIH under R01 AG052622 to ERT and EJB, and a Burroughs Wellcome Fund Investigators in the Pathogenesis of Infectious Diseases fellowship to ERT.
Footnotes
↵* Email: etroemel{at}ucsd.edu, Address: 9500 Gilman Dr #0349, La Jolla, CA 92093