Genetic and Cellular Characterization of Caenorhabditis elegans Mutants Abnormal in the Regulation of Many Phase II Enzymes

Background The phase II detoxification enzymes execute a major protective role against xenobiotics as well as endogenous toxicants. To understand how xenobiotics regulate phase II enzyme expression, acrylamide was selected as a model xenobiotic chemical, as it induces a large number and a variety of phase II enzymes, including numerous glutathione S-transferases (GSTs) in Caenorhabditis elegans. Methodology/Principal Findings To begin dissecting genetically xenobiotics response pathways (xrep), 24 independent mutants of C. elegans that exhibited abnormal GST expression or regulation against acrylamide were isolated by screening about 3.5×105 genomes of gst::gfp transgenic strains mutagenized with ethyl methanesulfonate (EMS). Complementation testing assigned the mutants to four different genes, named xrep-1, -2, -3, and -4. One of the genes, xrep-1, encodes WDR-23, a nematode homologue of WD repeat-containing protein WDR23. Loss-of-function mutations in xrep-1 mutants resulted in constitutive expression of many GSTs and other phase II enzymes in the absence of acrylamide, and the wild-type xrep-1 allele carried on a DNA construct successfully cured the mutant phenotype of the constitutive enzyme expression. Conclusions/Significance Genetic and cellular characterization of xrep-1 mutants suggest that a large number of GSTs and other phase II enzymes induced by acrylamide are under negative regulation by XREP-1 (WDR-23), which is likely to be a functional equivalent of mammalian Keap1 and a regulator of SKN-1, a C. elegans analogue of cap-n-collar Nrf2 (nuclear factor erythroid 2-related factor 2).

Acrylamide, now recognized as a prevalent food substance, has been long known as a neurotoxin for many animals and a potential carcinogen for humans [1,14]. Previously we reported in the nematode Caenorhabditis elegans [11] that acrylamide up-regulates a large number of phase II enzymes such as GSTs and UDPglucuronosyl/glucosyl transferases (UGTs) and some phase I enzymes such as short-chain type dehydrogenases (SDRs). C. elegans offers many experimental advantages as a model for understanding diverse aspects of biology, including responses to xenobiotics [15]. In mammals, acrylamide might be detoxified mainly by GSTs and excreted in urine [16,17]. Among GSTs that acrylamide up-regulated more than two-fold, those we studied displayed spatially varied expression patterns. For all these reasons we have selected acrylamide as a representative chemical for xenobiotics exposure.
To dissect genetically a xenobiotics response pathway (xrep) from a target of xenobiotics to the final destination of phase II enzyme expression, we isolated xrep gene mutants with abnormal GST expression or response to acrylamide. Here we report on one of four genes defined by these mutants, xrep-1, that encodes a WD repeatcontaining protein, a nematode homologue of mammalian WDR23, and provide genetic and cellular evidence that the gene xrep-1 negatively regulates GSTs and some other phase II enzymes.

Isolation and mapping of mutants showing abnormal GST regulation
We used two gst::gfp transgenic animals, MJCU017 and MJCU047, to screen for mutants abnormally expressing GFP. MJCU017 and MJCU047, which have the chromosomallyintegrated gst-4::gfp and gst-30::gfp fusion genes, respectively, emitted no detectable GFP signal in the absence of acrylamide, but emitted a very strong GFP signal from the whole body when treated with acrylamide ( Fig. 1) [11,18]. Mapping with a set of conventional marker mutants located both of these fusion genes as being integrated in linkage group X (data not shown). These animals were outcrossed at least three times with the wild-type strain N2.
We mutagenized these transgenic strains with ethyl methanesulfonate (EMS) and screened about 3.5610 5 genomes to obtain 24 independent mutants. The mutants were then outcrossed more than three times with N2 wild type to reduce unwanted mutations. Complementation testing and linkage analysis assigned 16 of the mutations defined by these 24 mutants to the same gene on linkage group (LG) I (chromosome I): all of the 16 mutations were recessive, and the affected strains constitutively expressed GST in the whole body without acrylamide. Six, assigned to the same gene on LG II, were also recessive, and their mutants constitutively expressed GST in the body-wall muscle and pharynx only after they had reached the adult stage. Of the remaining two strains, one mutation on LG IV was dominant, with constitutively expressed GST throughout the whole body, whereas the other one, on LG I, was recessive with its mutant expressing no GST even when treated with acrylamide. We called these genes defined by the four complementation groups xrep-1, -2, -3, and -4, in the respective order described above.
The xrep-1 gene is wdr-23 In SNP mapping involving the Hawaiian wild-type strain CB4856 crossed with the xrep-1(k1007) mutation, one of the 16 xrep-1 alleles, we successfully assigned the gene xrep-1 to the middle of LG I between the SNP markers F21C3 and T23G11. Of the total 704 recombinants analyzed, no Hawaiian polymorphism was identified within the genomic region consisting of the cosmid clone D2030, indicating that the mutation site was in or near D2030 (Fig. 2a). We amplified 12 genes within the D2030 cosmid with T7 promoter-added primers and synthesized dsRNA for soaking RNAi. RNAi of the D2030.9 gene performed in MJCU017 resulted in the induction of GST-4 expression without acrylamide (Fig. S1). Furthermore, the DNA fragment of D2030.9 amplified from N2 genomic DNA rescued the xrep-1(k1007) mutant phenotype (data not shown).
D2030.9 encodes WDR-23, a nematode homologue of WDR23, a member of WD40 repeat-containing proteins, which are known to exist in yeast through plants and mammals [19]; and WDR-23 was recently reported to be involved in C. elegans GST-4 expression [20]. Five xrep-1 mutations sequenced so far have revealed four missense mutations and one nonsense mutation in the WD repeat domain, suggesting that this domain is important in the regulation of GST expression by keeping it from being induced in the absence of acrylamide (Fig. 2b). To avoid unnecessary confusion, we continue to use the gene name xrep-1 and its corresponding protein name XREP-1 instead of wdr-23 and WDR-23 unless necessary for clarification.

GST-4 expression is not totally but partially regulated by SKN-1
We reported previously that acrylamide-induced GST-4 expression was partially regulated by the transcription factor SKN-1 [11]. To examine whether the constitutive GST expression caused by the xrep-1 mutation was under SKN-1 control, we performed a series of knockdown experiments by feeding C. elegans RNAi constructs. As should be expected, the xrep-1(k1007) mutant constitutively emits a strong GFP signal from the whole body without acrylamide, and treating it with acrylamide did not further enhance this signal (Fig. 3). Knockdown of gfp resulted in the shutdown of both the constitutive and acrylamide-induced GFP signals except for that in the pharynx as expected (Fig. 3). The skn-1 knockdown also prevented the constitutive and acrylamideinduced GST-4 expression except for that in the pharynx and in
We then constructed an xrep-1(+)::gfp fusion gene (Fig. 4a) and introduced it into MJCU059 to obtain MJCU080 {kEx80[xrep-1(+)::gfp, pRF4]; xrep-1(k1007) I; unc-119(ed3) III, kIs15 IV}. The extrachromosomal array kEx80[xrep-1(+)::gfp, pRF4] of MJCU080 rescued the xrep-1(k1007) mutant phenotype and expressed GFP in the cytoplasm and nuclei of neurons, somatic gonads, intestine, and hypodermis over the whole body (Fig. 5). Because the extrachromosomal array of MJCU080 was mitotically unstable, Amber-colored boxes represent seven WD40 domain repeats. DWD box indicates a DDB-1 (damaged DNA binding protein) WD40 binding domain. The k1002 mutation changes CGT to CAT resulting in R to H substitution at position 342. The k1007 mutation changes TGG to TGA resulting in W to protein chain termination at position 344. The k1011 mutation changes GAT to AAT resulting in D to N at position 312. The k1012 mutation changes GGA to GAA resulting in G to E at position 331. The k1016 mutation changes TCA to TTA, resulting in S to L at position 448. doi:10.1371/journal.pone.0011194.g002 this strain often produced animals mosaic for the xrep-1(+)::gfp transgene. Interestingly, although expectedly, those cells or tissues that retained the xrep-1(+)::gfp transgene, thus keeping the xrep-1(k1007) under rescue, emitted GFP signals. In contrast, cells or tissues without the xrep-1(+)::gfp transgene emitted no GFP signal, thus displaying the xrep-1 mutant phenotype as expected; that is, they emitted the RFP fluorescence signal (Figs. 5a-f). After a 48hour acrylamide treatment, however, even those cells or tissues possessing xrep-1(+)::gfp that emitted the GFP signal were also induced to emit RFP fluorescence (Figs. 5g-l). This mosaic analysis confirmed our earlier hypothesis that XREP-1 repressed GST from being expressed in the absence of acrylamide and derepressed GST expression in its presence.
Notably, when MJCU080 animals were treated with acrylamide for 24 hours, the induction of GST expression was not so strong as would be expected from strains such as MJCU017 (Figs. 6a-b). Following 72 hours of acrylamide treatment, however, GST expression was well induced (Fig. 6c). We interpret the result to mean that this ''super-repression'' of acrylamide-induced GST expression was caused by over-expression from extra copies of the xrep-1(+)::gfp transgene. Its repression was eliminated when the animals were treated with xrep-1(RNAi) (Figs. 6d-f). This result further augments our hypothesis that XREP-1 controls GST expression through negative regulation, which is responsive to and inactivated or released by acrylamide. The XREP-1-mediated regulation of GST expression agrees with some functional evidence of this regulation: for instance, xrep-1(k1007) and xrep-1(RNAi) animals show more resistance to aldicarb than do their wild-type counterparts (manuscript in preparation).

Discussion
Of the phase II enzymes, GSTs are universally found in every organism from bacteria to humans and constitute a large family of enzymes that function as detoxifiers of both endogenous oxidative stress products and exogenous electrophilic chemical compounds [7][8][9][10][11]. Also very importantly, some GSTs are not just detoxifiers but multifunctional performers, as they participate in steroid and eicosanoid biosynthesis as well as in amino acid metabolism [9,12,13]. The C. elegans genome contains 52 gst-coding genes in the Alpha, Sigma, Omega, Zeta, and Pi classes (WormBase, http://www.wormbase.org/), and 18 of the GSTs were prominently up-regulated when animals were treated with acrylamide [11]. Because acrylamide is (a) found widely and abundantly in various foods as a hazardous food contaminant, (b) known as a potential carcinogen for humans, and (c) induces a large number and variety of phase II enzymes, we have selected this chemical as a model xenobiotic chemical probe that should represent a substantial number of xenobiotic compounds [11,18,21,22,23] (Fig. S5). To understand the genetic, molecular, and cellular processes of xenobiotics, we have attempted to dissect genetically a xenobiotics response pathway (xrep) in C. elegans by isolating mutants that respond abnormally to acrylamide. We have so far found four xrep genes: three genes (xrep-1 I, -2 II, and -3 IV) negatively and one gene (xrep-4 I) positively regulate GST expression.
The negatively regulating gene xrep-1 encodes XREP-1, a nematode homologue of the WD-repeat containing protein 23 or WDR23. WDR exists in a broad range of organisms from yeasts to mammals as well as plants. It participates in a variety of biochemical, cellular, and organismal processes, such as signal transduction, cytoskeletal dynamics, and RNA processing [19]. A family of WDR proteins functions as a substrate adapter for ubiquitin E3 ligase, and WDR domains are predicted to form a bpropeller structure, which acts as a dock for interaction with other proteins [19]. Another family of proteins predicted to form the bpropeller structure is a group of proteins containing Kelch-repeat domains, which are also considered to serve as substrate adaptors for ubiquitin E3 ligase [19]. One such protein called Keap1 represses the bZIP transcription factor Nrf2 via the Kelch-repeat domain for degradation through the ubiquitin-proteasome pathway in the absence of oxidative or electrophilic stresses. In the presence of such stresses, Nrf2 is freed from Keap1 into the nucleus where it induces the expression of phase II enzymes [8,24]. In C. elegans, the bZIP transcription factor SKN-1, which is necessary for mesendodermal differentiation during early embryogenesis [25], is found to function similarly to Nrf2 by inducing phase II enzyme expression in response to oxidative stresses [26][27][28][29][30], sodium arsenite [30], and acrylamide [11]. According to our transcriptome analysis, a large number of genes coding for such detoxifying enzymes as GSTs, UGTs, and SDRs are negatively regulated by xrep-1 [23]. Thus, in C. elegans, XREP-1 might control SKN-1, similarly to the mammalian Keap1 for Nfr2 operative in the so-called antioxidant response element (ARE) pathway [9,31]. An idea similar to this was recently reported [20].
Indeed, RNAi knockdown of skn-1 prevented the xrep-1(k1007) mutants and acrylamide-treated animals from expressing GST-4 in most parts of the C. elegans body, except for the pharynx and body-wall muscle (Fig. 3). Because the RNAi knockdown of gfp did not prevent GST-4 expression in the pharynx, however, we assume that RNAi itself did not work in the pharynx. GST-4 expression was not further induced in the xrep-1(k1007) mutants when they were treated with acrylamide (Fig. 3), thus suggesting that all observed acrylamide-induced GST-4 expression was under control through XREP-1. SKN-1 seemed to control GST-4 expression downstream of XREP-1 in the xenobiotics response pathway, albeit in a tissue-specific fashion, as GST-4 was expressed in the body-wall muscle of skn-1(RNAi) animals. All five xrep-1 mutation sites so far identified are located in the WDR domain of XREP-1 (Fig. 2). This result is consistent with the idea that this domain may be important for interaction with SKN-1.
Here we have an emerging picture of a functional similarity between the two pairs of proteins XREP-1/SKN-1 in worms and Keap1/Nrf2 in mammals, as both play a key role in the oxidative and electrophilic stress pathway. At the same time, however, this leaves us with fascinating evolutionary puzzles: (a) how the two counterparts XREP-1 and Keap1, considered as phylogenetically distant relatives (19), have converged to assume essentially an identical role in the pathway critically important in defending organisms against oxidative, xenobiotic, or other life-threatening stresses, (b) why or if really the nematode C. elegans lacks a homologue of the mammalian Keap1, which exists also in the insect Drosophila (32), and (c) why no or if any mammalian WD40 repeat proteins play a role similar to that for the C. elegans XREP-1.
We obtained two cDNAs, xrep-1a and xrep-1b, which were PCR products of transcripts from the N2 wild type, and analyzed their function. The xrep-1a cDNA consisting of the exons 4 to 11, which corresponds to a C-terminus of 775 amino acid residues, was sufficient to rescue the mutation xrep-1(k1007) and regulate acrylamide-induced GST expression just as does the entire xrep-1 gene (Fig. 7). Contrarily, the xrep-1b cDNA showed none of these functions. In the present experiment we could not detect any other transcripts, as implicated in WormBase (http://www.wormbase. org/). Thus, we have yet to know what the xrep-1b or any other transcripts of the xrep-1 gene are doing.
In summary, with the aid of C. elegans genetics we have so far identified four xrep genes that regulate the GST expression in the xenobiotics response pathway, and introduced here one of them, xrep-1. The gene xrep-1 that encodes a nematode homologue of WDR23 negatively regulates a large number of phase II enzymes [23]. Currently, we are studying our three remaining xrep genes while continuing isolation of new Xrep mutants to understand the xenobiotics response pathways and more generally the exogenous/ endogenous stress response pathways and their regulation.

Strain construction
To make reporter constructs, all PCRs were performed with KOD Plus DNA polymerase (TOYOBO, Osaka, Japan) on N2 genomic DNA. PCR primer sequences are listed in Table S1. A fragment of C. elegans unc-54 39UTR region was obtained by cutting the region with EcoRI and SpeI from the gfp vector pPD95.77 (kindly provided by A. Fire, Stanford University) and integrated into the equivalent restriction site of pDsRed-Monomer (Clontech, CA, USA) for rfp (red fluorescence protein) vector pMH06.12-Red. PCR primers for gst-2, gst-4, and gst-30 were designed to amplify predicted promoters for each gene, about 1.2-0.8 kbp upstream from the predicted start sites spanning over full coding regions without the stop codons. PCR-amplified DNA fragments were digested with the appropriate restriction enzymes (Table S1) and ligated into the gfp vector pPD95.77 (for gst-2::gfp and gst-30::gfp) or the rfp vector pMH06.12-Red (for gst-4::rfp). Each reporter construct (100 mg/mL) so obtained was co-injected with an equal concentration of pDP#MM016B into the gonadal arms of unc-119(ed3) adult hermaphrodites [34,35] [36], and transgenic animals thus obtained were outcrossed at least three times with N2 to obtain MJCU058 {unc- Linkage mapping with the conventional markers (listed in the subsection Nematode culturing and strains) located the fusion gene gst-4::rfp, gst-2::gfp of MJCU058 integrated in the linkage group IV and, similarly, the fusion gene gst-30::gfp of MJCU047 in X (data not shown). By the standard genetic methods, xrep-1(k1007) was transferred into MJCU058 {unc-119(ed3) III; kIs15 IV} to construct MJCU059 {xrep-1(k1007) I; unc-119(ed3) III; kIs15 IV}. Fluorescence expression patterns were observed with a Nikon SMZ800 dissection microscope equipped with a fluorescence filter.

Mutant isolation and mapping
We mutagenized the transgenic strains MJCU017 and MJCU047 with 50 mM ethyl methanesulfonate (EMS), following essentially the method described by Brenner [33] to obtain mutants expressing GST abnormally. Twenty-four mutants obtained independently were outcrossed at least three times with N2 wild type to reduce extraneous mutations. By complementation testing, 16 of 24 mutants were grouped as the same gene xrep-1. The allele xrep-1(k1007) was first mapped to linkage group I by using conventional markers, dpy-5(e61) I, unc-4(e120) II, dpy-18(e364) III, unc-24(e138) IV, unc-42(e270) V, lon-2(e678) X, and their positions were then narrowed by single-nucleotide polymorphism mapping [37]. We defined a genomic region between the SNP markers F21C3 and T23G11. No Hawaiian polymorphism was found within D2030 (total 704 recombinants inquired) indicating that the mutation site was nearby or within the cosmid clone D2030. We amplified 12 genes within D2030 with T7 promoter-added primers and synthesized dsRNA for soaking RNAi [38]. From the soaking RNAi results, the candidate gene D2030.9 was selected, PCR-amplified from the N2 wild-type genomic DNA, and co-injected with the pRF4 marker DNA into mutant animals following the method by Mello et al. [34].

RNAi
Gene fragments of skn-1 cDNA, xrep-1 cDNAEx4-11, or gfp were prepared by PCR amplification of C. elegans N2 cDNA, genomic DNA, or plasmid vector pPD95.77, respectively, with primers (Table S3). Each PCR fragment was digested with EcoRI and cloned into the EcoRI restriction site of the RNAi vector pPD129.36 (kindly provided by A. Fire, Stanford University). The PCR fragment-ligated plasmid or the blank vector pPD129.36 was used to transform E. coli HT115 [39].
For RNAi experiments, synchronized L1-stage animals were first cultured for 48 hours at 20uC on NGM (containing 50 mg/ mL ampicillin and 12.5 mg/mL tetracycline) plates seeded with E. coli HT115 transformed with each different RNAi plasmid. The animals were then collected and transferred onto NGM plates with or without 500 mg/L acrylamide, seeded with each different E. coli HT115 RNAi bacteria. After 24 to 48 hours of incubation, the animals were observed for GFP expression with both a Nikon SMZ800 dissection microscope equipped with a fluorescence filter and a ZEISS Axiovert 200 microscope equipped with a confocal laser-scanning module.