The conserved Mediator subunit MDT-15 is required for oxidative stress responses in Caenorhabditis elegans

Authors

  • Grace Y. S. Goh,

    1. Graduate Program in Cell and Developmental Biology, University of British Columbia, Vancouver, BC, Canada
    2. Centre for Molecular Medicine and Therapeutics, Child & Family Research Institute, Vancouver, BC, Canada
    Search for more papers by this author
  • Katherine L. Martelli,

    1. Centre for Molecular Medicine and Therapeutics, Child & Family Research Institute, Vancouver, BC, Canada
    Search for more papers by this author
  • Kulveer S. Parhar,

    1. Centre for Molecular Medicine and Therapeutics, Child & Family Research Institute, Vancouver, BC, Canada
    Search for more papers by this author
  • Ada W. L. Kwong,

    1. Centre for Molecular Medicine and Therapeutics, Child & Family Research Institute, Vancouver, BC, Canada
    Search for more papers by this author
  • Marcus A. Wong,

    1. Centre for Molecular Medicine and Therapeutics, Child & Family Research Institute, Vancouver, BC, Canada
    Search for more papers by this author
  • Allan Mah,

    1. Centre for Molecular Medicine and Therapeutics, Child & Family Research Institute, Vancouver, BC, Canada
    Search for more papers by this author
  • Nicole S. Hou,

    1. Graduate Program in Cell and Developmental Biology, University of British Columbia, Vancouver, BC, Canada
    2. Centre for Molecular Medicine and Therapeutics, Child & Family Research Institute, Vancouver, BC, Canada
    Search for more papers by this author
  • Stefan Taubert

    Corresponding author
    1. Graduate Program in Cell and Developmental Biology, University of British Columbia, Vancouver, BC, Canada
    2. Centre for Molecular Medicine and Therapeutics, Child & Family Research Institute, Vancouver, BC, Canada
    3. Department of Medical Genetics, University of British Columbia, Vancouver, BC, Canada
    • Correspondence

      Stefan Taubert, Centre for Molecular Medicine and Therapeutics, Child & Family Research Institute, Department of Medical Genetics, Graduate Program in Cell and Developmental Biology, The University of British Columbia, Vancouver, BC, Canada. Tel: +1 604-875-3860; fax: +1 604-875-3819; e-mail: taubert@cmmt.ubc.ca

    Search for more papers by this author

Summary

Reactive oxygen species (ROS) play important signaling roles in metazoans, but also cause significant molecular damage. Animals tightly control ROS levels using sophisticated defense mechanisms, yet the transcriptional pathways that induce ROS defense remain incompletely understood. In the nematode Caenorhabditis elegans, the transcription factor SKN-1 is considered a master regulator for detoxification and oxidative stress responses. Here, we show that MDT-15, a subunit of the conserved Mediator complex, is also required for oxidative stress responses in nematodes. Specifically, mdt-15 is required to express SKN-1 targets upon chemical and genetic increase in SKN-1 activity. mdt-15 is also required to express genes in SKN-1-dependent and SKN-1-independent fashions downstream of insulin/IGF-1 signaling and for the longevity of daf-2/insulin receptor mutants. At the molecular level, MDT-15 binds SKN-1 through a region distinct from the classical transcription-factor-binding KIX-domain. Moreover, mdt-15 is essential for the transcriptional response to and survival on the organic peroxide tert-butyl-hydroperoxide (tBOOH), a largely SKN-1-independent response. The MDT-15 interacting nuclear hormone receptor, NHR-64, is specifically required for tBOOH but not arsenite resistance, but NHR-64 is dispensable for the transcriptional response to tBOOH. Hence, NHR-64 and MDT-15's mode of action remain elusive. Lastly, the role of MDT-15 in oxidative stress defense is functionally separable from its function in fatty acid metabolism, as exogenous polyunsaturated fatty acid complementation rescues developmental, but not stress sensitivity phenotypes of mdt-15 worms. Our findings reveal novel conserved players in the oxidative stress response and suggest a broad cytoprotective role for MDT-15.

Introduction

Reactive oxygen species (ROS) are ubiquitous molecules that occur as byproducts of aerobic metabolism. ROS have important biological properties and activities: on one hand, they serve as signaling molecules in regulatory circuits; on the other hand, they can damage cellular macromolecules due to their reactive nature (Hekimi et al., 2011; Back et al., 2012). Aberrant ROS accumulation causes oxidative stress, which is implicated in the development or aggravation of cancer, diabetes, and neurodegenerative diseases (Zhang et al., 2011). To protect against oxidative stress, eukaryotes possess sophisticated defense systems that cope with elevated ROS levels and promote homeostasis. Proteins that protect against high ROS levels include catalases, superoxide dismutases (SODs), and glutathione S-transferases (GSTs) (Xu et al., 2005; Lindblom & Dodd, 2006). Many such cytoprotective genes are transcriptionally induced by oxidative stress, and numerous transcription factors are required to activate overlapping, but distinct gene sets upon oxidative stress (Ma, 2010). Defining the transcriptional regulatory networks that limit cellular damage is thus an important scientific question.

The genetically tractable nematode Caenorhabditis elegans provides an excellent model to dissect cytoprotective circuits in vivo. In C. elegans, the transcriptional response to oxidative stress is mediated by conserved transcription factors such as the basic-leucine zipper type protein SKN-1 and the Forkhead BoxO (FoxO) type protein DAF-16 (An & Blackwell, 2003; Murphy et al., 2003; Ma, 2010). Both factors control the expression of detoxification enzymes and are required for resistance to heat, oxidative, and other stresses, as well as for the expression of cytoprotective genes in long-lived worm mutants (An & Blackwell, 2003; Murphy et al., 2003; Tullet et al., 2008; Oliveira et al., 2009). However, the regulatory complexes that specify the transcriptional responses to individual oxidative stressors remain poorly defined, although different types of oxidative stress activate distinct gene expression signatures (Oliveira et al., 2009; Park et al., 2009; Przybysz et al., 2009).

The multiprotein Mediator complex is an evolutionarily conserved transcriptional regulatory complex that is important for basal and activated transcription and is targeted by many transcription factors (Malik & Roeder, 2010; Conaway & Conaway, 2011). Although the complex as a whole is universally required for transcription, some Mediator subunits act in a selective fashion (Malik & Roeder, 2010; Conaway & Conaway, 2011). In C. elegans, the Mediator subunit MDT-15 is specifically required to express fatty acid metabolism genes, as well as fasting response, heavy metal detoxification, and xenobiotic detoxification genes (Taubert et al., 2006, 2011; Yang et al., 2006). MDT-15 cooperates with SBP-1, the C. elegans orthologue of sterol regulatory element binding proteins (SREBPs) and with the nuclear hormone receptor NHR-49 to regulate fatty acid metabolism genes. However, neither SBP-1, NHR-49, nor the detoxification regulator NHR-8 (Lindblom & Dodd, 2006) affect MDT-15-dependent detoxification genes (Taubert et al., 2008), suggesting that MDT-15 must interact with yet unidentified transcription factors to regulate stress response genes.

The biological functions of MDT-15 are evolutionarily conserved, as its yeast orthologue Gal11 is also required to express genes involved in lipid and drug metabolism and stress resistance (Thorpe et al., 2004; Thakur et al., 2008, 2009). To regulate lipid and drug metabolism genes, Gal11 cooperates with the zinc cluster transcription factors Pdr1 and Oaf1 (Thakur et al., 2008, 2009). Zinc cluster proteins are distantly related to metazoan nuclear hormone receptors (Naar & Thakur, 2009), suggesting that MDT-15 and its metazoan counterparts may regulate similar responses by involving such proteins.

Many detoxification genes are required for the response to oxidative stress, and we previously showed that mdt-15 is required to express many such genes (Taubert et al., 2008). Thus, we set out to test whether mdt-15 is required for oxidative stress response programs in C. elegans and to identify transcription factors that cooperate with MDT-15 to selectively induce stress responses.

Results

mdt-15 is required for survival in oxidative stress

To test whether MDT-15 is involved in oxidative stress responses, we compared MDT-15-dependent genes and oxidative stress response genes. Specifically, we looked for overlaps between genes downregulated following mdt-15 depletion and genes upregulated by exposure to hyperoxia, arsenite, tert-butyl hydroperoxide (tBOOH), or juglone. We found a statistically significant overlap in three of four cases (Table 1 and Table S1).

Table 1. Overlap between MDT-15-dependent genes and genes induced by oxidative stress, and between MDT-15- and SKN-1-dependent genes
 Down in mdt-15 RNAi (187 genes, Taubert et al., 2008)
  1. The table shows the number of genes expected to overlap between lists of relevant sizes, the number of genes that actually overlap, and P-values (Fisher's exact test) indicating significance of said overlap. The expected overlap is the fraction of MDT-15-regulated genes multiplied with the fraction of stress responsive genes.

Arsenite induced (118 genes, Oliveira et al., 2009)

Expected overlap: 2.0

Actual overlap: 11 (10 have predicted SKN-1-binding sites)

P-value 5.44 e-06

tBOOH induced (285 genes, Oliveira et al., 2009)

Expected overlap: 4.9

Actual overlap: 9

P-value 0.057

Hyperoxia (100% O2) induced (948 genes, Park et al., 2009)

Expected overlap: 16.3

Actual overlap: 37

P-value 2.17E-06

Juglone induced in L4 (103 genes, Przybysz et al., 2009)

Expected overlap: 1.8

Actual overlap: 10

P-value 1.02E-05

SKN-1 induced

(233 genes, Oliveira et al., 2009)

Expected overlap: 4.0

Actual overlap: 13 (10 have predicted SKN-1-binding sites)

P-value 1.85E-04

To test whether these gene expression changes cause oxidative stress sensitivity, we quantified population survival of wild-type worms and mdt-15(tm2182) mutants (Taubert et al., 2008; henceforth, mdt-15(rf) mutants, see below) on inorganic arsenite and on the organic peroxide tBOOH. We found that mdt-15(rf) mutants were hypersensitive to both stressors (Fig. 1, Tables S2 and S3). Depleting mdt-15 by RNA interference (RNAi) also caused tBOOH sensitivity (Fig. 4A). Thus, mdt-15 is required for normal oxidative stress resistance.

Figure 1.

mdt-15 worms are sensitive to oxidative stress. Survival plots of wild-type N2 and mdt-15(rf) worms on (A) 5 mm arsenite and (B) 6 mm tBOOH. Tables S2 and S3 show statistics and replicates.

MDT-15 is essential for the transcriptional response to arsenite

The oxidative stress sensitivity of mdt-15(rf) mutants could be due to reduced expression of stress regulators such as SKN-1 or DAF-16 (An & Blackwell, 2003; Murphy et al., 2003), changes in fatty acid metabolism (Taubert et al., 2006; Yang et al., 2006), or a requirement for MDT-15 in the induction of stress response genes. To test whether MDT-15 affects skn-1 and daf-16 expressions, we used real-time PCR (qPCR) to quantify their mRNA levels in vivo. mdt-15 depletion or mutation did not significantly alter skn-1 levels and actually increased daf-16 levels (Fig. S1A,B). The levels and the nuclear localization of DAF-16::GFP (Henderson & Johnson, 2005) and SKN-1::GFP (An & Blackwell, 2003) were also similar in control(RNAi) and mdt-15(RNAi) worms (Fig. S1C,D). Thus, the phenotypes of mdt-15(RNAi) and mdt-15(rf) worms are unlikely to originate from compromised SKN-1 or DAF-16 expression or localization.

To test whether mdt-15 is required to induce oxidative stress response genes, we grew synchronized wild-type worms to the L4 stage on control and mdt-15 RNAi, and then exposed them to 5 mm arsenite for 4 h and used qPCR to quantify oxidative stress gene expression. Arsenite reproducibly induced six genes more than twofold in control(RNAi) worms, and fold inductions (not only basal levels) of four genes were significantly reduced in mdt-15(RNAi) worms (Fig. 2A, Fig. S2A). To support the RNAi studies, we also exposed synchronized wild-type and mdt-15(rf) L4 larvae to arsenite. mdt-15(rf) mutants showed significantly impaired arsenite inductions for five genes and reduced basal levels of three genes, resembling mdt-15(RNAi) worms (Fig. S2B). Importantly, we obtained similar results with a 1-h arsenite exposure (Fig. S2C), suggesting that the compromised gene induction could cause the oxidative stress sensitivity of mdt-15(rf) worms.

Figure 2.

mdt-15 is required for the transcriptional response to arsenite. (A) Fold changes of mRNA levels (relative to untreated control(RNAi)) in L4 wild-type worms grown on control, mdt-15, or skn-1 RNAi and treated with 5 mm sodium arsenite for 4 h (n = 5). mRNA levels were normalized to act-1, ama-1, cdc-42, and tba-1; error bars represent SEM. *Untreated mdt-15(RNAi) or skn-1(RNAi) worms differ significantly from control(RNAi) worms (P < 0.05). Treated mdt-15(RNAi) or skn-1(RNAi) worms differ significantly from control(RNAi) worms (P < 0.05). (B) DIC and fluorescence micrographs show gcs-1p::gfp worms grown on control or mdt-15 RNAi after 4 h on 5 mm sodium arsenite. One of three repeats is shown. (C) Same as (B), except with gst-4p::gfp worms.

To corroborate the qPCR data, we studied worms expressing transcriptional gcs-1p::gfp or gst-4p::gfp reporters (Wang et al., 2010). In L4 stage gcs-1p::gfp worms, the induction of intestinal GFP by a 4-h arsenite exposure was severely compromised when worms were grown on mdt-15 RNAi (Fig. 2B). To ensure that this phenotype was not caused by impaired development due to mdt-15 depletion, we exposed late L4 gcs-1p::gfp worms to RNAi for 48 h and then exposed them to arsenite for 4 h. Induction of gcs-1p::gfp by arsenite remained mdt-15 dependent in this adult-only RNAi regimen (Fig. S2D). We observed a similar mdt-15 requirement in worms expressing a gst-4p::gfp reporter (Fig. 2C). Notably, mdt-15 depletion reduced intestinal gst-4p-driven fluorescence but evoked hypodermal fluorescence not seen in control(RNAi) worms, perhaps reflecting compensatory, mdt-15-independent gst-4 induction (Fig. 2C). Together, these data suggest that MDT-15 coregulates the transcriptional stress response to arsenite by affecting a subset of arsenite-responsive genes.

MDT-15 is required to induce SKN-1 targets in worms with elevated SKN-1 levels

Mediator subunits are tethered to genomic regulatory elements by transcription factors. In C. elegans, SKN-1 is a key regulator of the arsenite response (Oliveira et al., 2009). To test whether SKN-1 and MDT-15 cooperate, we compared SKN-1 and MDT-15-dependent genes and found that there is a statistically significant overlap (Table 1); predicted SKN-1-binding sites (Oliveira et al., 2009) occur in 10 of 13 shared targets (77%), similar to the approximately 80% found in ‘SKN-1-only’ targets. Moreover, qPCR revealed that several genes respond similarly to mdt-15 and skn-1 depletion (Fig. 2A), and MDT-15 primarily affects intestinal gcs-1 and gst-4 expressions (Fig. 2B,C), like SKN-1 (An & Blackwell, 2003). This suggests that MDT-15 and SKN-1 directly coregulate some arsenite-responsive genes.

To test whether MDT-15 is a SKN-1 coregulator, we studied wdr-23(tm1817) loss-of-function mutants (Choe et al., 2009). WDR-23 is part of an ubiquitin ligase complex that promotes SKN-1 degradation; thus, wdr-23(−) worms exhibit increased levels of SKN-1 and SKN-1 target genes. If mdt-15 were required to express SKN-1 targets, mdt-15 depletion should suppress SKN-1-dependent gene inductions in wdr-23(−) mutants. To test this hypothesis, we quantified mRNA levels in developmentally synchronized wild-type and wdr-23(−) worms grown on control, mdt-15, and skn-1 RNAi. We found that the inductions of five SKN-1 targets in wdr-23(−) mutants were strongly and similarly reduced by mdt-15 and skn-1 depletion (Fig. 3A, Fig. S3A). We also depleted wdr-23 in mdt-15(rf) mutants and found that mdt-15(rf) mutants exhibit significantly impaired induction of SKN-1 targets on wdr-23 RNAi (Fig. S3B). Thus, increased expression of SKN-1 targets in wdr-23 worms requires mdt-15.

Figure 3.

MDT-15 interacts functionally and physically with SKN-1. (A) Relative mRNA fold changes of SKN-1 targets in L4 stage wild-type N2 or wdr-23(−) worms grown on either control, mdt-15, skn-1, or mdt-6 RNAi (n = 4). mRNA levels were normalized to act-1, ama-1, cdc-42, and tba-1; error bars represent SEM. * indicates P < 0.05. (B) Relative interaction strength between Gal4DBD-MDT-15 variants and Gal4AD-SKN-1c; the schematic depicts MDT-15 fusion proteins. Values indicate average interaction strength in percent, calculated from Miller units (n > 4 per plasmid combination); error bars represent SEM. Protein expression of Gal4DBD-MDT-15 fusions was confirmed by immunoblot (Fig. S4). (C) Relative mRNA fold changes in L2 stage N2 and daf-2(e1370) worms grown on control, mdt-15, or skn-1 RNAi (n = 4). mRNA levels were normalized to act-1, ama-1, cdc-42, and tba-1; error bars represent SEM. * indicates P < 0.05.

Some Mediator subunits act in gene-specific fashion, but others are broadly required for transcription. To address subunit specificity in SKN-1 target gene transcription, we tested whether MDT-6 was required for the SKN-1-dependent inductions in wdr-23(−) mutants. Unlike mdt-15, mdt-6 RNAi did not prevent the induction of SKN-1 targets in wdr-23 mutants (Fig. 3A), although both RNAi clones delay growth and development. These data provide evidence for Mediator subunit specificity and demonstrate that developmental arrest per se is not sufficient to block SKN-1 target induction.

MDT-15 physically associates with SKN-1 via non-KIX-domain interactions

To test whether SKN-1 physically binds MDT-15, we used the yeast-two-hybrid system. MDT-15 contains an N-terminal KIX-domain that binds nuclear hormone receptors (NHRs) and the lipogenic transcription factor SBP-1 (Taubert et al., 2006; Yang et al., 2006). As this is the only recognized transcription-factor-binding domain in MDT-15, we hypothesized that the KIX-domain (aa1-124; Fig. 3B) might associate physically with SKN-1. However, an MDT-15-KIX bait did not interact with SKN-1 in our yeast-two-hybrid assays (Fig. 3B, Fig. S4A).

The yeast MDT-15 orthologue Gal11 uses multiple surfaces to bind the transcription factor Gcn4, with the KIX-domain playing a minor role (Herbig et al., 2010; Jedidi et al., 2010). Thus, we tested whether SKN-1 interacted with two longer baits, MDT-15-NT (aa 1-338), and near-full-length MDT-15-ΔCT (aa 1-600; Fig. 3B; full-length MDT-15 autoactivates and cannot be used as bait). While MDT-15-NT failed to interact, MDT-15ΔCT strongly and specifically bound SKN-1c (Fig. 3B; SKN-1a and SKN-1b were undetectable).

The KIX-domain is not sufficient for SKN-1 binding, but might be required. To test this hypothesis, we assayed binding of SKN-1c to an MDT-15ΔCT variant lacking the KIX-domain (MDT-15ΔKIXΔCT; aa 125-600). Binding of SKN-1c to MDT-15ΔKIXΔCT was as strong as binding to MDT-15ΔCT (Fig. 3B), indicating that the KIX-domain is dispensable for SKN-1c binding.

The region deleted in mdt-15 mutants is required for SKN-1 binding

As mdt-15(rf) mutants fail to activate some SKN-1 targets, we hypothesized that the MDT-15 protein they produce might be unable to bind SKN-1. Immunoblot analysis with antibodies against the MDT-15 N-terminus (Taubert et al., 2006) revealed that mdt-15(rf) worms produce an approximately 70 kDa protein instead of approximately 85 kDa wild-type MDT-15 (Fig. S5A). Sequencing the mdt-15 cDNA from mdt-15(rf) mutants (mdt-15 RNA and proteins are expressed at comparable levels in WT and mdt-15(rf) worms; Fig. S5A,B) revealed that the tm2182 allele is an in-frame deletion and insertion that produces a 618 aa protein with a predicted molecular weight of 66 kDa (Fig. S5C), reflecting the band detected by immunoblot. To study MDT-15tm2182 function, we asked whether a Gal4DBD-MDT-15tm2182 fusion protein activated a Gal4-driven β-galactosidase reporter in yeast. Gal4DBD-MDT-15tm2182 activated as well as Gal4DBD-MDT-15 (Fig. S5D), suggesting it is not generally dysfunctional. In sum, MDT-15tm2182 is expressed at wild-type levels in vivo and is transcriptionally competent. Along with the fact that the phenotypes of mdt-15(rf) worms are weaker than those of mdt-15(RNAi) worms (Taubert et al., 2008), this suggests that tm2182 is a hypomorph, not a loss-of-function allele.

Next, we addressed whether the region deleted in mdt-15(rf) worms is involved in SKN-1 binding. We found that an MDT-15tm2182ΔCT bait (aa 1-392–553-600) recapitulating the tm2182 deletion showed weak SKN-1c binding compared with MDT-15ΔCT (Fig. 3B). Thus, the tm2182 mutation compromises MDT-15 binding to SKN-1c, providing a molecular explanation for the inability of mdt-15(rf) worms to induce SKN-1 targets. We also tested whether the deleted region (MDT-15-del; aa 393-552) was sufficient for SKN-1c interaction, but it failed bind SKN-1c above background levels, suggesting that it is not (Fig. 3B).

MDT-15 regulates gene expression downstream of DAF-2

Besides activating stress response genes, skn-1 is required for gene inductions and longevity in worms carrying a mutation in the insulin/IGF-1 receptor gene daf-2 (Kenyon et al., 1993; Tullet et al., 2008). To test whether MDT-15 coregulates SKN-1 in this context, we quantified gene expression in wild-type and daf-2(e1370) worms grown on control, skn-1, and mdt-15 RNAi by qPCR. Tullet et al. found several gst genes to be upregulated in daf-2(e1370) mutants in a skn-1-dependent fashion; we found that mdt-15 depletion also significantly reduced these inductions (Fig. 3C, Fig. S6A). mdt-15 depletion additionally affected SKN-1-independent genes such as fat-5 and fat-6 (Taubert et al., 2006; Fig. 3C, Fig. S6A). Thus, MDT-15 regulates stress response and lipid metabolism genes in SKN-1-dependent and SKN-1-independent fashion in daf-2 mutants.

The transcription factor DAF-16 modulates gene expression downstream of daf-2 and hence might cooperate with MDT-15 to regulate SKN-1-independent genes. We found that mRNA induction of the direct DAF-16 target sod-3 was slightly reduced in daf-2(e1370) mutants grown on mdt-15 RNAi (Fig. 3C). Similarly, daf-2(e1370) worms expressing a transgenic sod-3p::gfp reporter (Libina et al., 2003) showed reduced intestinal GFP fluorescence on mdt-15 RNAi (Fig. S6B). However, DAF-16 did not bind MDT-15 in our yeast-two-hybrid assays (data not shown), and the effect of MDT-15 depletion on sod-3 mRNA levels was relatively weak, suggesting that MDT-15 only plays a minor role in DAF-16-dependent transcription.

As MDT-15 is required for gene expression in daf-2 mutants (Fig. 3C), we examined two phenotypes linked to daf-2, longevity and dauer larvae formation (Fielenbach & Antebi, 2008). As published by Zhang et al. (2013), adult-only mdt-15 RNAi reduced the lifespan of daf-2(e1370) and daf-2(e1368) mutants, but also reduced wild-type lifespan (Fig. S6B,C). To test whether mdt-15 is required for dauer formation, we exposed wild-type and mdt-15(rf) worms to dauer-inducing ascaroside #2 (Butcher et al., 2007), and found that they entered and exited dauer with wild-type kinetics and frequency (data not shown). Thus, the role of mdt-15 in insulin signaling resembles that of skn-1, which is required for longevity but dispensable for dauer formation (Tullet et al., 2008).

Altered fatty acid metabolism in mdt-15 worms does not cause tBOOH sensitivity

Our data indicate that MDT-15 and SKN-1 coregulate some arsenite response genes. However, unlike mdt-15, skn-1 is largely dispensable for the tBOOH response (Oliveira et al., 2009), implicating SKN-1-independent mechanisms for MDT-15 in this context. One possibility is that the altered fatty acid profiles of mdt-15 worms (Taubert et al., 2006; Yang et al., 2006) underlie their tBOOH sensitivity. The fatty acid desaturases fat-6 and fat-7 are MDT-15 targets (Taubert et al., 2006; Yang et al., 2006), and RNAi against either enzyme causes sensitivity against the oxidative stressor paraquat (Horikawa & Sakamoto, 2009). To test whether fatty acid desaturases are required for tBOOH resistance, we exposed young adult fat-6(tm331); fat-7(wa36) double mutants (Brock et al., 2007) to tBOOH. Surprisingly, these worms were less tBOOH sensitive than WT (Fig. S7A), perhaps because their unusual C18 polyunsaturated fatty acids (PUFAs) substitute for normal C20 PUFAs. fat-6 RNAi also failed to evoke tBOOH sensitivity, despite delaying development (Fig. 4A, Fig. S7B). This suggests that reduced fat-6 and fat-7 expressions do not cause tBOOH sensitivity in mdt-15 worms. To corroborate these data, we allowed wild-type and mdt-15(rf) worms to complete development in the presence of exogenous PUFAs and then exposed these rescued worms to tBOOH. Although fertility, mobility, and development were improved in mdt-15(rf) mutants, they remained fully tBOOH sensitive (Fig. 4B). Thus, reduced PUFA levels are unlikely to cause the tBOOH sensitivity of mdt-15(rf) mutants. In fact, we observed that wild-type worms were slightly tBOOH hypersensitive in the presence of exogenous PUFAs (Fig. 4B). Furthermore, fat-5, fat-6, and fat-7 mRNA levels decreased following tBOOH exposure (Fig. 4C). Taken together, these data argue that altered fatty acid profiles are unlikely to cause the tBOOH sensitivity of mdt-15 worms.

Figure 4.

MDT-15 regulates oxidative stress responses independently of fatty acid desaturation. (A) Survival plots of control(RNAi), mdt-15(RNAi), and fat-6(RNAi) worms on 6 mm tBOOH. Table S4 shows statistics and replicates. (B) Survival plots of wild-type N2 and mdt-15(tm2182) worms on 6 mm tBOOH following development on PUFAs (300 μm oleic acid, 300 μm linoleic acid, and 100 μm eicosapentaenoic acid). Table S5 shows statistics and replicates. (C) Relative mRNA fold changes of fat-5, -6, and -7 in wild-type worms in unstressed conditions and after 4 h on 7.5 mm tBOOH (n = 4). mRNA levels were normalized to act-1, tba-1, and ubc-2; error bars represent SEM. *indicates P < 0.05.

MDT-15-binding NHRs are required for normal oxidative stress resistance

To test whether mdt-15 is required for the transcriptional tBOOH response, we exposed L4 stage control(RNAi), mdt-15(RNAi), and skn-1(RNAi) worms to 7.5 mm tBOOH for 4 h. mdt-15, but not skn-1, was required to induce some, but not all genes in response to tBOOH (Fig. 5A, Fig. S8A), again affecting both basal levels and fold inductions. We obtained similar data with mdt-15(rf) mutants, with three of six genes displaying MDT-15-dependent regulation in tBOOH (Fig. S8B,C; some genes show increased basal expression in mdt-15(rf) mutants). A 1-h tBOOH exposure also showed early mdt-15 dependence for two genes (Fig. S8C; some genes were barely induced by 1 h). Thus, mdt-15 is selectively required for the transcriptional response to two compounds evoking distinct oxidative stress signatures.

Figure 5.

MDT-15-interacting NHRs are required for oxidative stress resistance. (A) Fold changes of mRNA levels in L4 wild-type N2 worms grown on control, mdt-15, or skn-1 RNAi and treated with 7.5 mm tBOOH for 4 h, relative to untreated control(RNAi) worms (n = 4). mRNA levels were normalized to act-1, ama-1, cdc-42, and tba-1; error bars represent SEM. *Untreated mdt-15(RNAi) or skn-1(RNAi) worms differ significantly from control(RNAi) (P < 0.05). Treated mdt-15(RNAi) or skn-1(RNAi) worms differ significantly from control(RNAi) worms (P < 0.05). (B, C) Survival plots of wild-type N2 and nhr-64(−) worms on 5 mm arsenite and 6 mm tBOOH. (D, E) Survival plots of wild-type N2 and nhr-49(−) worms on 5 mm arsenite and 6 mm tBOOH. For B–E, Tables S2 and S3 show statistics and replicates. (F) mRNA fold changes in L4 stage N2 and nhr-64(−) worms treated with 7.5 mm tBOOH for 4 h, relative to untreated N2s (n = 4). mRNA levels were normalized to act-1, tba-1, and ubc-2; error bars represent SEM. *Untreated nhr-64(−) worms differ significantly from N2 worms (< 0.05). (G) Relative interaction strength between Gal4DBD-MDT-15 variants and Gal4AD-NHR-64. Values indicate average interaction strength in percent, calculated from Miller units (n > 4 per plasmid combination); error bars represent SEM.

SKN-1 is largely dispensable for tBOOH-induced transcription, whereas MDT-15 is necessary (Fig. 5A). MDT-15-binding-transcription factors other than SKN-1 must therefore confer transcriptional tBOOH responses and tBOOH resistance. To test this hypothesis, we quantified the tBOOH sensitivity of previously characterized nhr-64(ok1957) and nhr-49(nr2041) null mutants (Van Gilst et al., 2005; Liang et al., 2010); both genes encode MDT-15-binding-transcription factors (Taubert et al., 2006). We found that nhr-64(−) mutants were sensitive to tBOOH but not to arsenite, whereas nhr-49(−) worms were sensitive to both molecules (Fig. 5B–E). nhr-49 RNAi also causes paraquat sensitivity (Horikawa & Sakamoto, 2009), but prior microarray studies of nhr-49(−) mutants revealed no link to stress responses (Pathare et al., 2012), and nhr-49(RNAi) worms normally induce xenobiotic response genes upon toxin exposure (Taubert et al., 2008). Perhaps the broad oxidative stress sensitivity in nhr-49 worms originates from their mitochondrial defects (Pathare et al., 2012).

NHR-64 regulates fat metabolism genes, but oxidative stress response genes were not studied (Liang et al., 2010). Thus, we quantified tBOOH responsive genes in wild-type worms and nhr-64(−) mutants by qPCR. Tested genes were only mildly affected in L4 stage nhr-64(−) mutants, and induction by tBOOH was not significantly compromised (Fig. 5F; note that basal levels were increased in nhr-64(−) mutants). Further work is required to define the molecular cause of tBOOH sensitivity in these mutants.

Aside from the KIX-domain, the MDT-15 surfaces involved in NHR-64 binding have not been comprehensively tested. We found that NHR-64 interacted with MDT-15-KIX, -NT, and -ΔCT, as expected, because they all contain the KIX-domain (Fig. 5G). Unlike SKN-1, NHR-64 failed to bind MDT-15-ΔKIXΔCT, demonstrating that the KIX-domain is sufficient and necessary for NHR-64 binding (Fig. 5G). Like SKN-1, NHR-64 only weakly bound to MDT-15tm2182-ΔCT, suggesting that the region deleted in mdt-15(rf) worms is involved in both NHR-64 and SKN-1 interactions (Fig. 5G). Thus, two separable regions in MDT-15 are required for NHR-64 binding, although the KIX-domain is sufficient for partial binding. Reduced binding of MDT-15 to NHR-64 (or other factors required for the tBOOH response) might be responsible for the tBOOH sensitivity of mdt-15(rf) mutants.

Discussion

Reactive oxygen species possess both beneficial and detrimental properties, making tight control of their levels necessary. Here, we report a novel role for the C. elegans Mediator subunit MDT-15 in the oxidative stress response, involving a distinct functional region required for interactions with at least two transcription factors, including the well-characterized stress regulator SKN-1.

Caenorhabditis elegans MDT-15 is required for at least two distinct oxidative stress responses

Our data show that mdt-15 is required for two oxidative stress responses. Specifically, its mutation or depletion prevents normal gene inductions by, and renders worms sensitive to arsenite and tBOOH. The actions of MDT-15 are specific and not a consequence of sickness or impaired development because: (i) mdt-15 RNAi in fully developed adults causes a defective arsenite response (Fig. S2D); (ii) PUFA complementation rescues the development, fertility, and mobility of mdt-15 worms (Taubert et al., 2006; Yang et al., 2006), but does not rescue tBOOH sensitivity (Fig. 4B); (iii) fat-6(RNAi) worms and fat-6; fat-7 double mutants are not susceptible to tBOOH, despite pleiotropic phenotypes resembling mdt-15 worms (Fig. 4A, Fig. S7A); (iv) mdt-15 depletion or mutation impairs the transcriptional response and causes sensitivity to oxidative stress (Figs 2 and 5) but does not block the transcriptional heat-shock response or affect thermotolerance (Taubert et al., 2008); and (v) mdt-6 RNAi causes larval arrest yet fails to block SKN-1-dependent gene inductions in wdr-23(−) worms, unlike mdt-15 RNAi (Fig. 3A). Moreover, MDT-15's yeast orthologue Gal11 was identified in a screen for genes involved in oxidative stress sensitivity (Thorpe et al., 2004), suggesting that this is an evolutionarily ancient role for MED15 proteins. It will be interesting to test whether mammalian MED15 is required for antioxidant responses, perhaps by coregulating the SKN-1 orthologue Nrf2.

Arsenite attacks the thiol groups of glutathione and other peptides and promotes ROS production, whereas tBOOH directly damages proteins and lipids. Worms mount distinct defenses against each molecule: SKN-1 is required to induce arsenite response genes, but plays only a minor role in the tBOOH response (Oliveira et al., 2009). MDT-15 is essential for both responses, suggesting a broad role in cytoprotective pathways. To implement specific responses, MDT-15 interacts with SKN-1 to combat arsenite exposure and may act with NHR-64 or other NHRs to defend against tBOOH (Fig. 6). Functional RNAi screens using promoter::GFP reporter fusions will be useful to identify the genes required to induce tBOOH responsive genes.

Figure 6.

Model of MDT-15 action in the oxidative stress response. In the arsenite response and downstream of daf-2, MDT-15 interacts with SKN-1 to regulate some stress response genes such as gsts. MDT-15 is also required for the SKN-1-independent tBOOH response, presumably involving NHRs; nhr-49 might be broadly sensitive to oxidative stress due to its impact on mitochondria.

MDT-15 physically binds SKN-1; to our knowledge, this is the first report of a positive coregulator for SKN-1. The interaction is KIX-domain independent, as MDT-15 instead associates with SKN-1c via a region partially deleted in mdt-15(rf) mutants. This region is also required for NHR-64 binding, although NHR-64 also requires the KIX-domain. We favor the view that the tm2182 region is directly involved in protein binding rather than causing MDT-15 misfolding, because MDT-15tm2182 is transcriptionally active in yeast suggesting that it adopts functional conformation, is expressed at wild-type levels in vivo, and tm2182 causes hypomorph phenotypes that are less severe than mdt-15 depletion. A multisurface interaction between MDT-15 and NHR-64 might thus resemble the interaction between Gcn4 and Gal11 in yeast (Herbig et al., 2010; Jedidi et al., 2010).

Depleting the fatty acid desaturases fat-6 or fat-7 causes paraquat sensitivity (Horikawa & Sakamoto, 2009). We found that mdt-15(rf) and nhr-49(−) mutants, which strongly downregulate these genes and show altered fatty acid profiles (Van Gilst et al., 2005; Taubert et al., 2006; Yang et al., 2006), are also sensitive to oxidative stress. This initially suggested that fatty acid imbalance might cause oxidative stress susceptibility. However, fat-6(RNAi) and fat-6(−); fat-7(−) worms were insensitive to tBOOH despite pleiotropic phenotypes, and PUFA complementation failed to protect mdt-15(rf) mutants from tBOOH. Thus, the altered fatty acid composition of mdt-15 worms is unlikely to cause oxidative stress sensitivity, although it impacts growth, fertility, and mobility. Regulation of lipid biology and oxidative stress responses are thus separable MDT-15-regulated processes. Viewed more broadly, our data show that even severe changes in lipid composition do not necessarily disrupt stress responses.

Despite not being required for tBOOH resistance, we found that fatty acid desaturases were downregulated by oxidative stress. Synthesizing PUFAs in an oxidizing environment may be undesirable as they are peroxidation targets. Accordingly, we observed that exogenous PUFAs caused slight tBOOH sensitivity in wild-type animals. Alternatively, factors such as MDT-15, which is required to transcribe fat-5, fat-6, and fat-7, may be required to transcribe cytoprotective pathways rather than promote anabolic metabolism in a pro-oxidative environment.

MDT-15's role in aging

The view of ROS in aging has evolved in recent years (Hekimi et al., 2011; Back et al., 2012). Low levels of oxidative stress have been proposed to elicit beneficial stress responses early in life, which may defend against oxidative damage later in life (Hekimi et al., 2011; Back et al., 2012). Yet, as organisms age, cellular damage accumulates, and the induction of cytoprotective pathways is likely critical to combat age-related molecular damage (Shore et al., 2012). Factors such as MDT-15 should thus contribute to the longevity of certain mutants (Hou & Taubert, 2012). Indeed, mdt-15 is at least partially required for the long lifespan of insulin receptor mutants, dietary-restriction-mimicking eat-2 mutants, the translation pathway mutants ifg-1 and rsks-1, the nutrient sensor mutant aak-2, and the germline-less glp-1 mutant (Rogers et al., 2011; McCormick et al., 2011; Zhang et al., 2013; and this study). However, mdt-15 mutation or depletion also shortens the lifespan of wild-type worms (Taubert et al., 2006). Thus, MDT-15 regulated processes likely counteract aging in many genotypes.

In daf-2 mutants, DAF-16-driven intestinal mdt-15 induction was proposed to promote downstream lipid signaling to distant tissues not expressing mdt-15, thus coordinating longevity throughout the worm (Zhang et al., 2013). However, we observed no mdt-15 induction in daf-2(e1370) worms (Fig. S6D), and Zhang et al. found unaltered intestinal expression of mdt-15p::gfp in daf-2 mutants. Even if not induced, MDT-15 could be essential for downstream processes. We propose that MDT-15's function in ROS defense may also underlie its lifespan promoting role, perhaps in parallel to mdt-15-dependent lipid signaling.

Together, our data provide evidence that MDT-15 is required for the defense against two distinct exogenous oxidative stressors. This role of MDT-15 is separable from its impact on fatty acid metabolism, which suggests that altered lipid composition does not strictly correlate with an inability to combat oxidative stress. The broad cytoprotective activities of MDT-15 involve multiple transcription factors and highlight this conserved Mediator subunit as an important contributor to ROS homeostasis.

Experimental procedures

Nematode strains, growth conditions, and RNAi

We cultured C. elegans strains using standard techniques (Brenner, 1974) and E. coli OP50 as food, except for RNAi. For worm strains, see Table S8. To avoid background effects, each mutant was crossed into our N2 strain; original mutants were backcrossed to N2 at least six times. All experiments were carried out at 20 °C, except those with temperature sensitive daf-2 mutants, which were grown at 16 °C to prevent dauer entry, and shifted to 25 °C for experiments.

RNAi was performed on nematode growth media (NGM) plates with 25 μg mL−1 carbenicillin, 1 mm IPTG, and 12.5 μg mL−1 tetracycline, and seeded with appropriate HT115 RNAi bacteria. The mdt-15, skn-1, and wdr-23 RNAi clones are from the Ahringer library (Kamath et al., 2003) and were sequenced prior to use. Sodium meta-arsenite (Sigma 71287), tBOOH (Sigma 458139), oleic acid (S-1120), linoleic acid (S-1127), and eicosapentaenoic acid (S-1144) (Nu-Chek Prep) were added at indicated concentrations.

For synchronized worm growths, bleached embryos were hatched overnight on unseeded NGM plates; hatched, synchronized L1 larvae were then grown to the desired stage, as indicated, and growths were adapted to ensure developmental synchronicity of slow-growing wdr-23(−) and mdt-15(rf) mutants.

For stress response and lifespan assays, animals were grown to the late L4 stage and then transferred to normal or oxidant-containing plates. We used GraphPad prism 5 (GraphPad Software, Inc., La Jolla, CA, USA) to generate survival curves and calculated statistical significance using the log-rank (Mantel–Cox) test.

RNA isolation and quantitative PCR analysis

RNA isolation was performed as described (Taubert et al., 2006), with an added sonication step of the Trizol suspension to improve RNA yield. 2 μg total RNA was used to generate cDNA with superscript II reverse transcriptase (Invitrogen 18064-014), random primers (Invitrogen 48190-011), dNTPs (Fermentas R0186), and RNAseOUT (Invitrogen 10777-019). qPCR was performed in 30 μL reactions using Taq (Invitrogen 18038-240) and an Applied Biosystems StepOnePlus machine. We analyzed data with the math formula method. For each sample, we calculated normalization factors by averaging the (sample expression)/(reference expression) ratios of three or four normalization genes, as indicated; the reference was control(RNAi), WT, or untreated, as appropriate. We used nonparametric Mann–Whitney tests to calculate statistical significance of gene expression changes. Primers were tested on serial cDNA dilutions and analyzed for PCR efficiency (sequences in Table S9).

DIC and fluorescence microscopy

Worms were transferred onto 2% (w/v) agarose pads for microscopy. We captured images on a CoolSnap HQ camera (Photometrics, Tucson, AZ, USA) attached to a Zeiss Axioplan 2 compound microscope (Carl Zeiss Microscopy, Thornwood, NY, USA) and used MetaMorph Imaging Software with Autoquant 3D (Molecular Devices, LLC, Sunnyvale, CA, USA) digital deconvolution for image acquisition.

Yeast-two-hybrid assays and immunoblots

To estimate relative interaction strength, we transformed plasmid pairs into strain Y187 (Clontech, Mountain View, CA, USA) and performed liquid β-galactosidase assays as described (Taubert et al., 2006) using an OmegaStar plate reader (BMG Labtech, Ortenberg, Germany). Each assay included at least four technical replicates and was repeated three or more times. Western blots to detect protein expression in yeast and in C. elegans were carried out as described (Taubert et al., 2006) using Myc (Santa Cruz sc-40, Santa Cruz, CA, USA), Gal4AD (Clontech 630402, Mountain View, CA, USA), β-actin (Cell Signaling Technologies, Danvers, MA, USA), and MDT-15 (Taubert et al., 2006) antibodies.

Acknowledgments

Some strains were provided by the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440). We thank Dr. K. P. Choe for providing wdr-23(tm1817) mutants and SKN-1 plasmids, Dr. T. K. Blackwell for discussions, Dr. J. D. Ward for comments on the manuscript, and the Taubert lab for input. ST holds a Canada Research Chair and obtains support from CIHR (MOP-93713), CFI, UBC, CMMT, and CFRI. GG was supported by a UBC scholarship, KP by UBC and CFRI scholarships, and AM by a CFRI fellowship.

Author contributions

GG and ST conceived and designed the experiments. GG, KM, KP, AK, MW, AM, and ST performed experiments and analyzed data, NH provided intellectual input, GG and ST wrote the paper.

Conflict of interest

The authors declare no conflict of interests.

Ancillary