F-Box Protein XREP-4 Is a New Regulator of the Oxidative Stress Response in Caenorhabditis elegans

Abstract

The transcription factor SKN-1 (Skinhead family member-1) in Caenorhabditis elegans is a homolog of the mammalian Nrf-2 protein and functions to promote oxidative stress resistance and longevity. SKN-1 mediates protection from reactive oxygen species (ROS) via the transcriptional activation of genes involved in antioxidant defense and phase II detoxification. Although many core regulators of SKN-1 have been identified, much remains unknown about this complex signaling pathway. We carried out an ethyl methanesulfonate (EMS) mutagenesis screen and isolated six independent mutants with attenuated SKN-1-dependent gene activation in response to acrylamide. All six were found to contain mutations in F46F11.6/xrep-4 (xenobiotics response pathways-4), which encodes an uncharacterized F-box protein. Loss of xrep-4 inhibits the skn-1-dependent expression of detoxification genes in response to prooxidants and decreases survival of oxidative stress, but does not shorten life span under standard culture conditions. XREP-4 interacts with the ubiquitin ligase component SKR-1 and the SKN-1 principal repressor WDR-23, and knockdown of xrep-4 increases nuclear localization of a WDR-23::GFP fusion protein. Furthermore, a missense mutation in the conserved XREP-4 F-box domain that reduces interaction with SKR-1 but not WDR-23 strongly attenuates SKN-1-dependent gene activation. These results are consistent with XREP-4 influencing the SKN-1 stress response by functioning as a bridge between WDR-23 and the ubiquitin ligase component SKR-1.

IN Caenorhabditis elegans, the reactive oxygen species (ROS) detoxification response is largely regulated by the cap “n” collar (CNC) transcription factor SKN-1 (Skinhead family member-1), which is homologous and functionally similar to the Nrf2 (NF-E2-related factor-2) protein in mammals (An and Blackwell 2003). CNC proteins are basic leucine transcription factors that are also functionally conserved in insects, fishes, and birds (Sykiotis and Bohmann 2010). In general, CNC proteins are activated when cells experience oxidative stress, and bind to antioxidant response elements (ARE) of genes encoding antioxidant and detoxification enzymes. The functional similarities between SKN-1 and Nrf-2 combined with genetic tractability have made C. elegans an attractive model to understand the regulation of this pathway. Three transcript variants of the skn-1 gene (a, b, and c) are expressed in vivo, with the most well-characterized, skn-1c, regulating transcription of phase II detoxification genes to confer redox homeostasis and resistance toward oxidative stress; in addition, skn-1c is also involved in regulation of various aspects of metabolism and plays a crucial role in promoting longevity (An and Blackwell 2003; Wang et al. 2010; Pang et al. 2014; Tang and Choe 2015). Direct roles of skn-1a and skn-1b in oxidative stress have not been reported; skn-1a was recently implicated in the regulation of proteasome homeostasis, and skn-1b functions in the ASI neurons to mediate dietary restriction-induced longevity (Bishop and Guarente 2007; Lehrbach and Ruvkun 2016). Studies to date have identified several direct regulators of SKN-1c, which include a WD40 repeat protein named WDR-23 that negatively regulates SKN-1 protein levels, the p38 MAP kinase PMK-1 that phosphorylates and activates SKN-1 nuclear localization in response to ROS, and a number of kinases that inhibit SKN-1 nuclear accumulation including glycogen synthase kinase-3 (GSK-3), protein kinase B (AKT-1/2), and serum and glucocorticoid-inducible kinase-1 (SGK-1) (An et al. 2005; Inoue et al. 2005; Tullet et al. 2008; Choe et al. 2009).

We recently showed that the ubiquitin ligase adaptor protein Skp1-related (SKR-1) is required for the SKN-1-dependent response to oxidative stress (Wu et al. 2016). SKR-1 is part of the Skp1-Cul1-F-box (SCF) E3 ubiquitin ligase complex that targets proteins for ubiquitination (Nayak et al. 2002). Skp1 (SKR-1 homolog) interacts with numerous F-box proteins that function to recruit specific protein substrates for ubiquitination (Zheng et al. 2002). In C. elegans, there are >500 genes predicted to encode the F-box domain, with only 11 found in Saccharomyces cerevisiae, 22 in Drosophila, and 38 in humans (Kipreos and Pagano 2000; Thomas 2006). Here, we performed an ethyl methanesulfonate (EMS) mutagenesis screen and identified a novel F-box-encoding gene xrep-4 (xenobiotics response pathways-4) that is required for expression of skn-1-dependent detoxification genes and survival of prooxidants in C. elegans. XREP-4 is a binding partner of SKR-1 and WDR-23, and loss of xrep-4 increases nuclear localization of WDR-23::GFP. These data, and the known functions of F-box proteins, are consistent with XREP-4 functioning to promote SKN-1 activity by recruiting WDR-23 into a complex with SKR-1.

Materials and Methods

C. elegans strains

All C. elegans strains were cultured at 20° using standard methods (Brenner 1974). The following strains were used: wild-type N2 Bristol, Hawaiian CB4856, CL2166 dvIs19[gst-4p::GFP], CB1370 daf-2(e1370)III, QV159 dvIs19;dvIs5[PCFJ150-pDESTttTi5605[R4-R3][gst-4p::tdTomato::unc-54 UTR], QV269 xrep-4(zj26); dvIS19, QV270 xrep-4(zj26), QV271 xrep-4(zj27); dvIs19;dvIs5, QV273 xrep-4(zj29); dvIs19;dvIs5, QV274 xrep-4(zj30); dvIs19;dvIs5, QV275 xrep-4(zj31); dvIs19, QV276 xrep-4(zj31), QV289 xrep-4(zj28), QV290 xrep-4(zj28); dvIs19;dvIs5, QV279 zjEX119 [XREP-4::GFP; myo-2p::tdTomato], QV282 xrep-4(zj26); zjEx121[xrep-4 gDNA;myo-2p::tdTomato], QV291 zjEx124[xrep-4 gDNA; myo-2p::tdTomato], QV292 zjEx125[xrep-4 gDNA; myo-2p::tdTomato], QV293 zjEx126[xrep-4 gDNA; myo-2p::tdTomato], QV294 zjEx127[xrep-4 gDNA; myo-2p::tdTomato];dvIs19, LD1 ldIs7[skn-1B/C::GFP + pRF4(rol-6(su1006))], QV297 xrep-4(zj26); ldIs7, QV212 skn-1(k1023), QV130 skn-1(k1023);dvIS19, VP579 wdr-23(tm1817); dvIs19, and KHA116 unc-119(ed3);chuIs116 [wdr-23p::wdr-23a(cDNA)::GFP, unc-119].

RNA interference (RNAi)

RNAi experiments were performed by feeding with Escherichia coli [HT115(DE3)] as described previously (Wu et al. 2016). Briefly, bacteria were inoculated overnight in Luria broth containing selective antibiotic, seeded on nematode growth media (NGM) agar plates containing 0.2% β-lactose, and then grown overnight. Synchronized L1 larvae were placed on RNAi plates until they developed into young adults (YA). RNAi clones were obtained from either the ORFeome library (Open Biosystems, Huntsville, AL) or from the MRC genomic library (Geneservice, Cambridge, UK). Clones used for experiments were sequence verified. Clone pPD129.36 (LH4440) was used as the negative control, which expresses a 202-bp double-stranded (dsRNA) that is not homologous to any C. elegans genes.

EMS screen and SNP mapping

To screen for genetic suppressors of SKN-1 activation, ∼5000 gravid animals carrying the integrated gst-4p::GFP and gst-4p::tdTomato reporters (QV159) were mutagenized with 50 mM EMS for 4 hr at room temperature and allowed to recover and propagate. F2 offspring from the mutagenized P₀ were grown on NGM agar containing 2 mM acrylamide to activate gst-4p reporter expression; when worms reached the L4/YA stage, a COPAS BioSort was used to isolate worms with low levels of gst-4p::GFP fluorescence. To avoid isolating transgene silencing mutants, a secondary screen was performed manually for worms with low gst-4p::tdTomato [mos1-mediated single copy insertion (mosSCI)]. Two of the six isolated mutants were crossed into the polymorphic Hawaiian CB4856 strain, and mutations were identified using the whole-genome sequencing (WGS) and SNP mapping strategy (Doitsidou et al. 2010). For the remaining four mutants, Sanger sequencing was used to search for mutations within the F46F11.6 locus. EMS-derived mutants were outcrossed five times to wild-type N2 to remove background mutations before being used for subsequent experiments.

GFP reporter analysis and quantitative PCR (qPCR)

To visualize activation of the gst-4p::GFP reporter, worms were treated with either 38 µM juglone for 3 hr (plus 1 hr recovery) or 7 mM acrylamide for 4 hr with both prooxidants dissolved in liquid NGM buffer. gst-4p::GFP activation was scored with a Zeiss Discovery V12 microscope (Zeiss [Carl Zeiss], Thornwood, CA) using a three-level scale; with low representing worms with minimal GFP fluorescence throughout the worm, medium representing worms with bright GFP fluorescence only at the anterior and posterior ends of the body, and high representing worms with bright GFP fluorescence throughout the body. To visualize the SKN-1::GFP reporter, worms were treated with 38 µM juglone in liquid NGM for 3 hr or 5 mM arsenite in liquid NGM for 1 hr, mounted on a 2% agarose pad containing 5 mM levamisole, and imaged using an Olympus BX60 (Center Valley, PA) microscope fitted with a Zeiss AxioCam MRm camera. Images were taken with both GFP and red fluorescent protein channels and merged into a composite using ImageJ (NIH) to allow clear differentiation of SKN-1::GFP signal from background fluorescence. Intestinal SKN-1::GFP was scored as low (no GFP nuclei), medium (GFP nuclei at the anterior and posterior region of the intestine), and high (GFP nuclei throughout the intestine) (Inoue et al. 2005). Head SKN-1::GFP was scored as negative (no GFP nuclei) or positive (GFP nuclei throughout the head region).

qPCR assays were carried out as described previously (Wu et al. 2016). For all qPCR assays, RNA samples were extracted from L4/YA worms that were fed the appropriate dsRNA starting from the synchronized L1 larvae stage. All qPCR reactions were performed in 10-µl reaction volumes using a Realplex ep gradient S Mastercycler (Eppendorf AG, Hamburg, Germany) and all data were analyzed by normalizing to rpl-2 as an internal reference. Primer sequences are available on request.

HEK293 in vitro pull-downs and western blotting

To test interactions of proteins, full-length C. elegans SKN-1c, WDR-23a, SKR-1, and XREP-4 (wild-type or mutated variant) were cloned into the glutathione-S-transferase (GST)-tag pDEST27 or the pcDNA3.1/nV5-DEST vector. Appropriate vectors were then cotransfected into HEK293 cells cultured in modified Eagle’s medium (supplemented with 10% fetal bovine serum, 4.5 g/liter glucose, 584 mg/liter L-glutamine, 100 mg/liter sodium pyruvate, and 1 U/ml penicillin) and grown for 2 days, after which cells were lysed and used for pull-down assays with glutathione–sepharose 4B beads. Beads were washed extensively with PBS + 0.1% Triton-X. For western blotting, procedures were carried out as previously described (Wu et al. 2016). Antibodies used in this experiment were mouse anti-GST MAb (1:1000; Santa Cruz Biotech, Cat #B-14) and mouse anti-V5 tag MAb (1:1000; Invitrogen, Carlsbad, CA; Cat #R960-25).

Life span and stress resistance assays

For life span assays, worms were grown on the appropriate E. coli [HT115(DE3)] bacteria agar plates until YA, after which they were transferred to agar plates containing 50 µg/ml fluorodeoxyuridine (FUDR) to prevent progeny growth and scored for survival every 2 days. Worms were considered dead if they did not respond to gentle prodding with a thin wire. For life span assays of daf-2(e1370) worms, the synchronized L1 larvae were initially grown at 16° until YA before transferring to 20° to avoid dauer formation; corresponding N2 populations were grown in the same condition for control comparisons. For juglone resistance assays, YA worms were incubated in 175 µM juglone dissolved in liquid NGM and analyzed for survival hourly. For acrylamide resistance assays, worms were grown at normal conditions from L1 larvae to YA, and then transferred onto appropriately seeded RNAi agar plates containing 7 mM acrylamide and 50 µg/ml FUDR and analyzed for survival daily.

Statistical analysis

Unless noted otherwise, statistical analysis was carried out with Prism software 5.04 (La Jolla, CA) with the Student’s t-test used when two means were compared, one-way ANOVA with Tukey’s post hoc tests when multiple comparisons were needed, and two-way ANOVA with Bonferroni post hoc tests when comparisons were made over two factors. For categorical data analysis, Chi-square tests were used. Survival and life span analyses were determined by the Log-Rank test carried out using the OASIS online statistic tool (Han et al. 2016). Statistical significance is indicated in figures as *P < 0.05, **P < 0.01, and ***P < 0.001, and “ns” indicates “not significant.”

Data availability

All strains used in this study and sequence data for mutation mapping are available upon request. Raw data are available at https://figshare.com/s/49ae49ef6565e1823093.

Results

EMS mutagenesis screening identified mutants of xrep-4, a novel F-box protein-encoding gene required for the oxidative stress response

When exposed to conditions that induce oxidative stress, SKN-1 mediates a phase II detoxification response that includes transcription of numerous GST genes, which encode enzymes that catalyze the conjugation of glutathione to toxic electrophilic compounds. One of the most strongly and consistently expressed genes upon SKN-1 activation is gst-4 (An et al. 2005; Inoue et al. 2005; Tullet et al. 2008; Choe et al. 2009). To learn more about regulation of the SKN-1 stress response, we performed an EMS screen for mutants that fail to activate gst-4p::GFP when exposed to a nontoxic level of acrylamide (Wu et al. 2016). We isolated six independent true-breeding recessive mutants. Using the WGS-SNP strategy (Doitsidou et al. 2010), we mapped two of the mutations to the same gene, F46F11.6, which encodes an uncharacterized protein with an F-box domain at amino acid position 16–53 (Figure 1, A and B). Using Sanger sequencing, we found that three of the four other mutants also contain predicted loss-of-function alleles in F46F11.6, while the other mutant contains a proline to leucine substitution at position 20 (P20L).

Figure 1.

Mutations to the xenobiotics response pathway-4 (xrep-4) gene inhibit activation of gst-4. (A) Schematic of the xrep-4 locus marked with the F-box domain and mutations identified in this study. (B) Summary of six xrep-4 EMS alleles identified. (C) Activation of the Skinhead family member-1 (SKN-1)-dependent gst-4p::GFP reporter in wild-type or xrep-4(zj26) mutant worms when exposed to prooxidants juglone or acrylamide. gst-4p::GFP scoring was as described in Materials and Methods, ***P < 0.001, Chi-square tests from 103 to 105 worms scored per condition for one trial. Representative images are shown. Bar, 100 μM.

F46F11.6 was recently identified in an RNAi screen for regulation of gcs-1p::GFP in a prdx-2 mutant, but was not characterized further (Crook-McMahon et al. 2014). While discussing nomenclature with WormBase staff and colleagues, we learned that a gene previously named xrep-4 (Hasegawa and Miwa 2010) has been subsequently cloned to the same F46F11.6 locus that contains our mutations (T. Fukushige, personal communication); therefore, we use the name xrep-4 for F46F11.6. Worms with a premature stop codon in xrep-4 have significantly reduced gst-4p::GFP fluorescence when exposed to acrylamide or juglone, another strong inducer of SKN-1 (Przybysz et al. 2009) (Figure 1C).

We next performed xrep-4(RNAi) in wild-type gst-4p::GFP worms, which phenocopied the mutants as expected (Figure 2A). To gain insight into how xrep-4 genetically interacts with the WDR-23/SKN-1 pathway, we silenced xrep-4 in two strains with constitutive activation of gst-4p::GFP, skn-1(k1023) gain-of-function and wdr-23(tm1817) loss-of-function mutants (Hasegawa and Miwa 2010; Tang and Choe 2015). As seen in Figure 2B, loss of xrep-4 had no effect on constitutive activation of gst-4p::GFP in either strain, consistent with XREP-4 functioning upstream or independently of WDR-23 and SKN-1.

Figure 2.

xrep-4 is required for induction of detoxification genes upstream of wdr-23 and skn-1. (A) Knockdown of xrep-4 by RNAi phenocopies xrep-4(zj26). ***P < 0.001 Chi-square tests from 50 to 70 worms screened per condition for one trial. Representative images are shown. Bar, 100 μM. (B) Knockdown of xrep-4 does not reduce constitutive activation of gst-4p::GFP in wdr-23(tm1817) loss-of-function or skn-1(k1023) gain-of-function mutants. A total of 35–56 worms were screened per condition for one trial. (C) Fold change in mRNA levels of SKN-1 downstream target genes in N2 control worms fed xrep-4 dsRNA, xrep-4(zj26) mutants fed control dsRNA under basal conditions, or after exposure to juglone. **P < 0.01, ***P < 0.001 compared to N2;control(RNAi) as determined by two-way ANOVA with Bonferroni post hoc tests. (D) Expression of SKN-1 target genes can be rescued in the xrep-4(zj26) mutant by an extrachromosomal array carrying xrep-4 genomic DNA (zjEx121). ***P < 0.001 compared to N2;control(RNAi) as determined by two-way ANOVA with Bonferroni post hoc tests. N = 4 replicates of 200–300 worms. dsRNA, double-stranded RNA; mRNA, messenger RNA; RNAi, RNA interference; skn-1, Skinhead family member-1; xrep-4, xenobiotics response pathway-4.

We also used qPCR to measure five different skn-1-dependent detoxification genes under basal and juglone-exposed conditions in worms with loss of xrep-4 through either RNAi or mutation. The xrep-4(zj26) allele carries a premature stop codon in the third exon (Figure 1B) and has significantly reduced messenger RNA (mRNA) levels for three of five skn-1-dependent genes measured under basal conditions and four of five genes when exposed to juglone (Figure 2C). RNAi of xrep-4 had similar effects as xrep-4(zj26), albeit generally weaker. We next rescued xrep-4(zj26) with an extrachromosomal array carrying a genomic fragment covering the full length xrep-4 gene and 1200 bp upstream. Shown in Figure 2D, the xrep-4 array (zjEx121) either rescued or increased detoxification gene expression in xrep-4(zj26) worms relative to wild-type worms.

Loss of xrep-4 reduces resistance to oxidative stress

SKN-1 is required for resistance to a wide range of oxidative stressors, and is also a well-characterized regulator of aging (Tullet et al. 2008; Ewald et al. 2015; Tang and Choe 2015; Wu et al. 2016). Given that loss of xrep-4 reduces expression of skn-1-dependent genes, it may also be required for stress resistance and life span. Interestingly, the xrep-4(zj26) mutant lives just as long as N2 wild-type worms, suggesting that xrep-4 is not required for normal life span (Figure 3A). Alternatively, when exposed to a toxic concentration of juglone, xrep-4(zj26) worms had significantly reduced survival (Figure 3B and Supplemental Material, Table S1 in File S1). Similarly, xrep-4(zj26) was also sensitive to acrylamide (Figure 3C and Table S2 in File S1). Knockdown of xrep-4 by RNAi also significantly reduced acrylamide survival in wild-type worms, but not in skn-1(k1023) gain-of-function worms, which are highly resistant to acrylamide (Figure 3D and Table S2 in File S1).

Figure 3.

Loss of xrep-4 reduces C. elegans resistance to multiple stressors, but does not shorten normal life span. (A) Longevity curve for wild-type and xrep-4(zj26) mutant worms. Results for trial #1 are shown (full results and statistics are in Table S1 in File S1). (B) Survival curve of wild-type and xrep-4(zj26) mutant worms exposed to 175 µM juglone. ***P < 0.001 Log-rank test compared to N2. Results for trial #1 are shown (full results and statistics in Table S1 in File S1). (C) Survival curve of wild-type and xrep-4(zj26) mutant worms exposed to 7 mM acrylamide. ***P < 0.001 Log-rank test compared to N2. Results for trial #5 are shown (full results and statistics are in Table S2 in File S1). (D) Survival curve of wild-type and skn-1(k1023) worms fed control or xrep-4 dsRNA and exposed to 7 mM acrylamide; all conditions were P < 0.001 Log-rank test compared to N2; control(RNAi), results for trial #1 are shown (full results and statistics in Table S2 in File S1). (E) Representative images of normal and coiler worm phenotypes induced by acrylamide. Bar, 100 μM. (F) Percentage of worms displaying a coiler phenotype at days 3, 4, and 5 of adulthood after initiating exposure to 7 mM acrylamide on day 1. ***P < 0.001 two-way ANOVA with Bonferroni post hoc tests. Shown are combined results from three to six trials for each condition, with >30 worms per trial. RNAi, RNA interference; xrep-4, xenobiotics response pathway-4.

While conducting the acrylamide survival assays, we noticed that some worms exhibit a coiler-like phenotype after 3 days of exposure (Figure 3E), and we scored the percentage of coiling up to day 5 (Figure 3F). A similar coiler phenotype was observed in worms that carry a mutation in unc-8, which encodes for a degenerin/epithelial sodium channel (DEG/ENaC) cation-selective channel subunit that is involved in a wide range of mechanotransduction functions (Tavernarakis et al. 1997; Mano and Driscoll 1999). Worms carrying the xrep-4(zj26) mutation have a higher incidence of coiling from day 4 onwards, with >80% of the worms exhibiting the coiler phenotype by day 5 compared to ∼30% in wild-type; conversely, <5% of worms with the skn-1(k1023) gain-of-function allele have the coiler phenotype at day 5, consistent with SKN-1 protecting against acrylamide toxicity. Acrylamide is a known neurotoxin and in C. elegans has been shown to induce neuronal degeneration (Li et al. 2016). It is possible that a decrease in detoxification capacity of the xrep-4 mutants may exacerbate the toxicity of acrylamide and promote this abnormal locomotive behavior. Taken together, the results shown in Figure 3 are consistent with xrep-4 playing a key regulatory role in the skn-1-dependent stress response to juglone and acrylamide.

Overexpression of xrep-4 is sufficient to activate the oxidative stress response

In our rescue experiment in Figure 2D, we observed that xrep-4(zj26) mutant worms carrying the xrep-4 rescue array had significantly higher detoxification gene expression than wild-type worms, suggesting that xrep-4 overexpression might activate SKN-1-dependent gene expression. To test this without an xrep-4 mutation, the xrep-4 array was introduced into a wild-type background. As shown in Figure 4A, worms carrying the xrep-4 array had elevated xrep-4 mRNA levels, which we note as xrep-4 OE (overexpression); RNA samples for this qPCR analysis were treated with DNAse to avoid amplification of the genomic rescue array. xrep-4 OE worms had increased basal levels of gst-4p::GFP fluorescence compared to wild-type worms (Figure 4B), and qPCR analysis confirmed that basal levels of two skn-1-dependent genes were also elevated (Figure 4C); three of the five genes also had further activation in the presence of juglone. The xrep-4 OE strain was also resistant to toxic levels of juglone, but had no effect on the normal life span (Figure 4, D and E). Similar effects of xrep-4 OE on skn-1-dependent gene expression, juglone resistance, and longevity were observed in two other independent xrep-4 OE lines (Figure S1 in File S1). Taken together, these results demonstrate that overexpression of xrep-4 alone can activate transcription of SKN-1 target genes.

Figure 4.

Overexpression of xrep-4 can activate SKN-1 downstream genes and increase oxidative stress resistance. ***P < 0.001 compared to N2 as determined by Student's t-test. (A) Levels of xrep-4 mRNA in a strain carrying an extrachromosomal array expressing the full length xrep-4 genomic DNA. (B) Overexpression of xrep-4 in a wild-type background increases basal levels of gst-4p::GFP. myo-2p::tdTomato was used as an injection comarker; representative images are shown. Bar, 100 μM. (C) Fold change in mRNA levels of SKN-1 downstream target genes under basal and juglone conditions in wild-type and xrep-4 overexpression strains. *P < 0.05 and ***P < 0.001 compared to N2 as determined by two-way ANOVA with Bonferroni post hoc tests. N = 4 replicates of 200–300 worms. (D) Survival curve of wild-type and xrep-4 OE worms exposed to 175 μM of juglone. Results from trial #1 are shown, ***P < 0.001 Log-rank test compared to N2. (E) Longevity curve of wild-type and xrep-4 OE worms. Results for trial #3 are shown. (D and E) Results for all trials are in Table S3 in File S1. Two other independent lines were generated for the xrep-4 OE strain that showed similar results (Figure S1 in File S1). mRNA, messenger RNA; OE, overexpressing; RNAi, RNA interference; SKN-1, Skinhead family member-1; xrep-4, xenobiotics response pathway-4.

xrep-4 is epistatic to many pathways upstream of skn-1

To begin defining mechanisms of xrep-4 function, we first used a SKN-1::GFP reporter to determine if loss of xrep-4 prevents nuclear accumulation of SKN-1 during oxidative stress. SKN-1::GFP can accumulate in the intestinal nuclei upon exposure to arsenite, and we previously reported that SKN-1::GFP can also accumulate in the head region after exposure to juglone (Wu et al. 2016). In an xrep-4 mutant, normal SKN-1::GFP accumulation in the head and intestinal nuclei was not affected after exposure to either juglone or arsenite compared to wild-type, suggesting that xrep-4 may regulate SKN-1 independently of nuclear localization (Figure S2 in File S1). Discordances between SKN-1 nuclear localization and transcriptional activity have been shown by previous studies, suggesting that additional and yet to be identified regulatory mechanisms are required for SKN-1 transcriptional activity (Kahn et al. 2008; Paek et al. 2012; Glover-Cutter et al. 2013).

SKN-1 has been shown to be negatively regulated by a variety of upstream signals that include insulin/IGF-1 signaling (IIS), target of rapamycin (TOR) pathway, glycogen synthase kinase-3 (GSK-3), and the proteasome; disruption of these pathways results in different degrees of constitutive activation of SKN-1-dependent genes (Blackwell et al. 2015). We used RNAi to knockdown different SKN-1 upstream regulators that have previously been shown to result in the activation gst-4p::GFP (An et al. 2005; Kahn et al. 2008; Li et al. 2011; Robida-Stubbs et al. 2012). RNAi knockdown of the proteasome alpha subunits (pas-5 and pas-6; components of the 20S proteasome complex), chaperonin-containing TCP-1 (cct-1; cytosolic chaperonin), glycogen synthase kinase-3 (gsk-3), and Ras-related GTP-binding protein C-1 (ragc-1; TORC1 pathway activator) all induced gst-4p::GFP, and this response was largely suppressed in the xrep-4(zj26) mutant with the exception of pas-5 and pas-6 (Figure 5A). SKN-1 has also been shown to be directly negatively regulated by the IIS pathway, and mRNA of gst-4 is overexpressed in the daf-2(e1370) mutant in a skn-1-dependent manner (Tullet et al. 2008). In our experiments, the daf-2(e1370) mutant also had significantly elevated gst-4 mRNA compared to the wild-type, and this was dependent on xrep-4 (Figure 5B). Interestingly, knockdown of xrep-4 had no effect on the long-lived phenotype of the daf-2(e1370) mutant (Figure 5C and Table S4 in File S1). Taken together, these results suggest that xrep-4 is epistatic to daf-2, ragc-1, and gsk-3 in regulation of gst-4.

Figure 5.

Epistatic analysis of xrep-4 with known regulators of SKN-1. (A) Activation of gst-4p::GFP upon knockdown of SKN-1 upstream regulators in wild-type or xrep-4(zj26) mutant backgrounds. Shown are representative images and categorical scoring of each RNAi. Bar, 100 μM. *P < 0.05 and ***P < 0.001 as determined by the Chi-square test. Shown are results from one trial with a total of 91–226 worms scored for each condition. (B) Relative levels of gst-4 mRNA in N2 and daf-2(e1370) mutants fed with control or xrep-4 dsRNA. ***P < 0.001 as determined by one-way ANOVA and Tukey’s post hoc test. N = 4 replicates of 200–300 worms per condition. (C) Longevity curve of wild-type and daf-2(e1370) worms fed with either control or xrep-4 dsRNA. Knockdown of xrep-4 had no effect on life span in either the daf-2(e1037) or N2 wild-type strain. Trial #3 is shown for both strains, with statistics for all trials shown in Table S4 in File S1. dsRNA, double-stranded RNA; mRNA, messenger RNA; ns, not significant; RNAi, RNA interference; SKN-1, Skinhead family member-1; xrep-4, xenobiotics response pathway-4.

XREP-4 interacts with SKR-1

As mentioned earlier, xrep-4 encodes a predicted F-box protein, which is a substrate recognition component of the SCF ubiquitin ligase complex. We generated a transgenic worm carrying a XREP-4::GFP translational fusion protein, and found that this protein was widely expressed in multiple tissues including the intestine, neurons, muscle cells, and pharynx (Figure S3 in File S1). A previous study reported that F46F11.6 (XREP-4) interacts with SKR-1 in a yeast two-hybrid screen (Boxem et al. 2008), and that SKR-1 is one of six C. elegans orthologs of the mammalian Skp-1 protein (Killian et al. 2008). To test this interaction using full-length proteins, we cloned the C. elegans xrep-4 and skr-1 complementary DNA into mammalian GST-tag and V5-tag expression vectors, respectively, and cotransfected into HEK293 cells. Shown in Figure 6A, SKR-1 interacted strongly with XREP-4, as evident by the western blot using an anti-V5 antibody on lysates obtained from glutathione–sepharose pull-downs. XREP-4 also interacted with C. elegans WDR-23a, but not SKN-1c (Figure 6A).

Figure 6.

XREP-4 interacts with SKR-1 and WDR-23. (A) XREP-4 interacts with WDR-23 and SKR-1 when coexpressed in HEK293 cells. (B) Knockdown of skr-1 by RNAi reduces expression of SKN-1 downstream genes in response to juglone in N2 wild-type worms. All conditions are significantly different from N2; control(RNAi) with ***P < 0.001 as determined by ANOVA Bonferroni test; xrep-4(zj26);control(RNAi) and xrep-4(zj26);skr-1(RNAi) are not significantly different. N = 4 replicates of 200–300 worms. ns, not significant; RNAi, RNA interference; SKN-1, Skinhead family member-1; SKR, Skp1-related; xrep-4, xenobiotics response pathway-4.

We recently identified skr-1 as a novel regulator of the skn-1 transcriptional response, and the positive interaction between SKR-1 and XREP-4 might suggest that these two proteins function together to regulate SKN-1 (Wu et al. 2016). If so, then skr-1 and xrep-4 should not additively affect detoxification gene expression. To test this, we silenced skr-1 in wild-type and xrep-4(zj26) worms during exposure to juglone. In wild-type worms, knockdown of skr-1 significantly reduced three skn-1-dependent genes, in agreement with our recent findings (Figure 6B). Knockdown of skr-1 in the xrep-4(zj26) mutant showed no further reduction in expression of skn-1-dependent genes (Figure 6B).

To further explore interactions between XREP-4 and SKR-1, we characterized another xrep-4 mutant from the EMS screen, xrep-4(zj28), which has a C to T mutation that is predicted to cause a proline to leucine substitution at position 20 within the F-box motif (Figure 7A). The F-box motif functions to mediate protein–protein interactions, and in yeast and human this domain is required for its interaction with the Skp1 protein (Bai et al. 1996). The mutated proline in xrep-4(zj28) is a highly conserved residue within the F-box motif that is found in 92% of the 234 known F-box proteins in C. elegans (Kipreos and Pagano 2000). When we tested the P20L variant of XREP-4 for its interaction with SKR-1 and WDR23, we found that XREP-4(P20L) strongly reduced pull-down with SKR-1, but still pulled-down well with WDR-23a (Figure 7B). Worms carrying the xrep-4(zj28) mutation showed a similar inhibition of SKN-1-dependent gene activation as the premature stop codon xrep-4(zj26) allele under basal and juglone-exposed conditions (Figure 7, C and D).

Figure 7.

A mutation that disrupts XREP-4 and SKR-1 interaction can inhibit the oxidative stress response. (A) Schematic and DNA chromatogram of the xrep-4(zj28) mutation that causes a predicted proline to leucine single amino acid substitution at position 20 in XREP-4. (B) XREP-4 carrying the P20L mutation has normal interactions with WDR-23, but reduced binding to SKR-1 when coexpressed in HEK293 cells. (C) The xrep-4(zj28) mutation inhibits activation of gst-4p::GFP upon exposure to juglone or acrylamide. ***P < 0.001 Chi-square test with a total of 99–109 worms scored per condition for one trial. Representative images are shown. Bar, 100 μM. (D) Fold change in SKN-1 downstream target mRNA levels in the xrep-4(zj28) mutant compared to N2 wild-type under basal and juglone exposure, P < 0.05 and ***P < 0.001 compared to N2 as determined by two-way ANOVA with Bonferroni post hoc tests. N = 4 replicates of 200–300 worms. (E) SKR-1, XREP-4, and WDR-23 form a protein complex when coexpressed in HEK293 cells. Coexpression of XREP-4 enhances the interaction between SKR-1 and WDR-23, and this interaction is reduced by substituting the wild-type XREP-4 protein with the P20L mutated variant. mRNA, messenger RNA; SKN-1, Skinhead family member-1; SKR, Skp1-related; xrep-4, xenobiotics response pathway-4.

To test if SKR-1, XREP-4, and WDR-23 can form a protein complex, we pulled-down SKR-1 and tested its interaction with WDR-23 while altering the expression or variant of the XREP-4 protein. We had previously reported that SKR-1 can weakly interact with WDR-23 (Wu et al. 2016), and here we show that this interaction is markedly enhanced when XREP-4 is coexpressed (Figure 7E); both WDR-23 and XREP-4 were detected in lysates pulled-down for SKR-1, suggesting that these three proteins can form a complex. In addition, we found that coexpressing the P20L variant of the XREP-4 protein weakened the interaction between SKR-1 and WDR-23 compared to the wild-type XREP-4 (Figure 7E); this experiment also confirmed the reduced interaction between SKR-1 and XREP-4 (P20L) observed in Figure 7B. Overall, these results suggest that XREP-4 and SKR-1 likely function together in a complex with WDR-23 to promote expression of SKN-1-dependent detoxification genes.

xrep-4 regulates WDR-23::GFP localization

Given that XREP-4 interacts strongly with WDR-23, we further explored this interaction using a transgenic worm carrying an integrated WDR-23::GFP translational fusion protein. WDR-23 is a strong suppressor of SKN-1 in C. elegans, and is thought to control SKN-1 by promoting its constitutive degradation under basal conditions via the CUL4/DDB1 ubiquitin ligase (Choe et al. 2009). As shown in Figure 8A, worms fed xrep-4 dsRNA had a ∼1.3-fold increase in total fluorescence of the WDR-23::GFP fusion protein (Figure 8, A and B). Upon closer examination, knockdown of xrep-4 also significantly increased visible nuclear accumulation of WDR-23::GFP protein within the hypodermis of the worm (Figure 8, C and D); intestinal WDR-23::GFP was less obvious in this strain. Taken together, these results suggest that xrep-4 functions to negatively regulate WDR-23 accumulation in the nuclei (Figure 8E).

Figure 8.

xrep-4 negatively regulates the expression level and nuclear localization of the WDR-23::GFP fusion protein. (A and B) Knockdown of xrep-4 by RNAi increases the overall GFP fluorescence of worms carrying the integrated WDR-23::GFP translational fusion protein; ***P < 0.001 compared to control as determined by Student’s t-test. N = 7 slides containing a total of 69–70 worms; representative images are shown. Bar, 100 μM. (C and D) Knockdown of xrep-4 increased the total number of worms with visible nuclear accumulation of the WDR-23::GFP protein compared to worms fed with control dsRNA. Arrowheads indicates WDR-23::GFP in the hypodermal nuclei. A total of 69–70 worms were scored for each condition from two trials, ***P < 0.001 Chi-square test. Representative images are shown. Bar, 25 μM. (E) Proposed model of XREP-4 regulation of the SKN-1 oxidative stress response. XREP-4 serves as an F-box protein that together with ubiquitin ligase adapter SKR-1 regulates WDR-23, which is a principal suppressor of SKN-1. Loss of either SKR-1 (Wu et al. 2016) or XREP-4 attenuates the SKN-1-dependent response to oxidative stress. XREP-4 appears to regulate SKN-1 downstream of DAF-2, TORC1, GSK-3, and CCT-1, but could function upstream or independent of the proteasome genes pas-5 and pas-6. CCT-1, chaperonin-containing TCP-1; TORC1, target of rapamycin complex 1; dsRNA, double-stranded RNA; GSK-3, glycogen synthase kinase-3; RNAi, RNA interference; Skinhead family member-1; SKR, Skp1-related; xrep-4, xenobiotics response pathway-4.

Discussion

C. elegans is an excellent system to study regulation of stress responses, and in our genetic screen, we identified a role for the newly characterized F-box protein XREP-4 in the regulation of the SKN-1-dependent oxidative stress response. We show that loss-of-function or knockdown of the xrep-4 gene significantly attenuates expression of SKN-1-dependent detoxification genes (Figure 1 and Figure 2), and reduces oxidative resistance (Figure 3). Overexpression of xrep-4 increases expression of SKN-1 target genes and enhances stress resistance (Figure 4). Our genetic interaction and protein–protein interaction data support a model where XREP-4 functions with the ubiquitin ligase adapter SKR-1 (Figure 6 and Figure 7) to regulate WDR-23, a direct repressor of SKN-1 (Figure 8).

F-box proteins in C. elegans

The xrep-4 gene encodes for a Caenorhabditis-specific F-box protein. The lack of a similar F-box in other organisms is not surprising as C. elegans has an unusually large number of F-box proteins (>500 predicted) compared to other organisms (Thomas 2006). F-box proteins function as adapters to recruit protein substrates for polyubiquitination via linkage to an ubiquitin ligase including SKR-1 (Kipreos and Pagano 2000). Many F-box proteins contain the Skp1 (SKR-1)-interacting F-box motif at the N-terminus. The remainder of the proteins often contain other motifs that are involved in substrate binding such as the WD-40 repeat, kelch repeat, leucine-rich repeat, FOG-2 homology (FTH) domain, or F-box associated (FBA) domain (Thomas 2006). XREP-4 possesses the signature F-box domain, but none of these other known domains.

It has been speculated that the large number of F-box proteins in C. elegans could function as part of the nematode innate immune system by targeting toxic and pathogenic bacterial or viral proteins for degradation (Thomas 2006). In this hypothesis, the large numbers of F-box proteins are a result of an “evolutionary arms race,” where the continuous evolution of host defense proteins is driven by the selective pressures of evolving pathogens (Thomas 2006). However, few studies have successfully assigned functions to F-box proteins in C. elegans. FSN-1 regulates germline apoptosis and synapse development (Liao et al. 2004; Gao et al. 2008), DRE-1 regulates larvae development (Fielenbach et al. 2007), LIN-23 regulates cell cycle progression (Kipreos et al. 2000), MED-15 regulates touch receptor neuron development (Bounoutas et al. 2009), and SEL-10 promotes hermaphrodite development (Jäger et al. 2004). To our knowledge, XREP-4 is the first F-box shown to regulate a stress response in C. elegans.

Ubiquitin ligase and the proteasome in regulation of oxidative stress

SKN-1 is regulated by the proteasome, presumably in part via WDR-23, which functions with CUL-4 and DDB-1 to target SKN-1 for degradation (An et al. 2005; Inoue et al. 2005; Tullet et al. 2008; Choe et al. 2009). This proposed mechanism is analogous to the Keap1-Nrf-2 degradation model in mammals (Choe et al. 2012). Worms with a wdr-23 loss-of-function mutation have constitutive SKN-1 activation in the absence of stress and are hyper-resistant to oxidative stress (Tang and Choe 2015). Our protein–protein and genetic interaction results suggest that WDR-23 may be recruited to the SCF ubiquitin ligase with XREP-4 functioning as the F-box substrate adaptor (Figure 6 and Figure 7); a model is depicted in Figure 8E.

Polyubiquitination by E3 ubiquitin ligases typically targets proteins for degradation by the proteasome. If XREP-4 regulates WDR-23 in this manner, then the relationship between proteasome function and SKN-1 activity is likely to be complex. Previous studies have shown that disruption of the proteasome regulatory subunits by RNAi activates SKN-1, the opposite effect of loss of xrep-4 (Kahn et al. 2008; Li et al. 2011). Proteasome disruption increases mRNA levels of multiple skn-1-dependent proteasome-related genes in a compensatory response to restore proteasome function (Kahn et al. 2008; Li et al. 2011). It is also possible that SKN-1 is directly degraded by the proteasome (Choe et al. 2009). In our epistasis analysis, knockdown of pas-5 and pas-6, genes encoding proteins of the 20S α-ring of the 26S proteasome complex, activated gst-4p::GFP, in agreement with previous findings (Kahn et al. 2008; Li et al. 2011); and this activation still took place in the xrep-4(zj26) mutant (Figure 5A). This result is consistent with proteasome disruption activating SKN-1 by mechanisms largely independent or downstream from XREP-4 (Figure 8E).

xrep-4 and life span

SKN-1 is a well-characterized regulator of life span and aging in C. elegans (Tullet et al. 2008; Tang and Choe 2015). Although loss of xrep-4 significantly decreases resistance to acrylamide and juglone, it did not shorten life span under nonstressed conditions. Similar results were also previously shown for skr-1 (Ghazi et al. 2007; Wu et al. 2016). Interestingly, skr-1 is partially required for the life span extension of the daf-2(e1370) mutant (Ghazi et al. 2007), whereas xrep-4 is not (Figure 5C). Although skn-1 has been implicated in regulating daf-2-dependent longevity, it appears to be context-dependent (Ewald et al. 2015). While SKR-1 and XREP-4 may function together to regulate the oxidative stress response, this mechanism may not be involved in regulation of daf-2(e1370) longevity, and Ghazi et al. (2007) have suggested a role for daf-16 in regulating the requirement for skr-1 in daf-2(e1370) longevity.

In summary, our study identified a novel regulator in the skn-1-dependent oxidative stress response in C. elegans. The newly characterized F-box protein XREP-4 functions together with the SKR-1 ubiquitin ligase adaptor to antagonize nuclear localization of the SKN-1 inhibitor protein WDR-23.

Supplementary Material

Supplemental material is available online at www.genetics.org/lookup/suppl/doi:10.1534/genetics.117.200592/-/DC1.