NHR‐49/HNF4 integrates regulation of fatty acid metabolism with a protective transcriptional response to oxidative stress and fasting

Summary Endogenous and exogenous stresses elicit transcriptional responses that limit damage and promote cell/organismal survival. Like its mammalian counterparts, hepatocyte nuclear factor 4 (HNF4) and peroxisome proliferator‐activated receptor α (PPARα), Caenorhabditis elegans NHR‐49 is a well‐established regulator of lipid metabolism. Here, we reveal that NHR‐49 is essential to activate a transcriptional response common to organic peroxide and fasting, which includes the pro‐longevity gene fmo‐2/flavin‐containing monooxygenase. These NHR‐49‐dependent, stress‐responsive genes are also upregulated in long‐lived glp‐1/notch receptor mutants, with two of them making critical contributions to the oxidative stress resistance of wild‐type and long‐lived glp‐1 mutants worms. Similar to its role in lipid metabolism, NHR‐49 requires the mediator subunit mdt‐15 to promote stress‐induced gene expression. However, NHR‐49 acts independently from the transcription factor hlh‐30/TFEB that also promotes fmo‐2 expression. We show that activation of the p38 MAPK, PMK‐1, which is important for adaptation to a variety of stresses, is also important for peroxide‐induced expression of a subset of NHR‐49‐dependent genes that includes fmo‐2. However, organic peroxide increases NHR‐49 protein levels, by a posttranscriptional mechanism that does not require PMK‐1 activation. Together, these findings establish a new role for the HNF4/PPARα‐related NHR‐49 as a stress‐activated regulator of cytoprotective gene expression.


| INTRODUCTION
The ability to respond to acute stress conditions is vital to prevent or limit organismal damage. For instance, transcriptional responses promote adaptation and survival under stress conditions. Impaired stress responses contribute to human age-related diseases such as diabetes and neurodegenerative disorders (Hetz, Chevet & Harding, 2013;Lin & Beal, 2006), and likely also contribute to aging (Hekimi, Lapointe & Wen, 2011;Shore & Ruvkun, 2013). However, these adaptive responses also protect pathogens and cancer cells against cytotoxic drugs and the immune system (Leprivier, Rotblat, Khan, Jan & Sorensen, 2015;Rankin & Giaccia, 2016). Accordingly, there is widespread interest in defining the mechanisms involved in coordinating these transcriptional responses.
Transcription factors of the nuclear factor (erythroid-derived 2)like 2 (Nrf2) family are required to activate many cytoprotective genes in animals. In the nematode Caenorhabditis elegans, the Nrf2 ortholog SKiNhead (SKN-1) implements a wide range of homeostatic and stress responses in response to internal and external stimuli (Blackwell, Steinbaugh, Hourihan, Ewald & Isik, 2015). Although the transcription factors DAF-16, HSF-1, HLH-30, and HIF-1 activate a variety of stress/pathogen-protective responses, SKN-1 is especially important to maintain redox balance and defend against oxidative stress caused by xenobiotics that target GSH and cause protein unfolding Miranda-Vizuete & Veal, 2016).
Here, we have used fmo-2 as a marker to investigate SKN-1independent stress response mechanisms by determining how this pro-longevity gene is activated in response to organic peroxide and fasting. We found that the nuclear hormone receptor (NHR)-49, an HNF4/PPARa ortholog and established regulator of C. elegans lipid metabolism and longevity (Pathare, Lin, Bornfeldt, Taubert & Van Gilst, 2012;Ratnappan et al., 2014;Van Gilst, Hadjivassiliou, Jolly & Yamamoto, 2005;, is also activated by and required for oxidative stress resistance. We show that NHR-49 and MDT-15 are both required for the increased expression of fmo-2 as part of a stress response program activated by organic peroxide, fasting, and in long-lived, germline-less glp-1 mutants. Together, our data suggest that the function of NHR-49 is not confined to regulating lipid metabolism, but also involves activation of a stress-protective transcriptional program.

| nhr-49 is required to induce an oxidative stress and fasting response program
To screen for factors cooperating with MDT-15 in the tBOOHinduced expression of fmo-2, we generated a transgenic strain expressing a transcriptional fmo-2p::gfp reporter ( Figure S1a). This reporter was constitutively expressed in some neurons and weakly active in the intestine of well-fed animals, but strongly induced in the intestine, hypodermis, and pharynx following exposure to tBOOH (Figure 1a,b). The fmo-2p::gfp reporter was also activated by DTT and H 2 O 2 , but not by arsenite and cadmium ( Figure S1b). Thus, consistent with other studies, our transcriptional reporter was strongly activated by tBOOH but not by stimuli that activate SKN-1-dependent gene expression (Goh et al., 2014;Oliveira et al., 2009).
The discovery that nhr-49 was required to induce fmo-2 in response to either tBOOH or fasting raised the possibility that these stresses activate a common biological response. Indeed, meta-analysis revealed a significant overlap between genes upregulated by tBOOH and by fasting (Figure 2a), with qPCR analysis confirming that 20 of these genes are upregulated in response to either stress (Figure 1c,d, data not shown). Notably, the tBOOHand fasting-induced expression of seven of these genes was impaired in nhr-49(nr2041) worms, indicating that NHR-49 is important for the shared transcriptional response to these stresses (Figure 1c,d). nhr-49 is also required for metabolic remodeling and increased lifespan due to loss of glp-1/notch receptor activity in a glp-1(e2141) mutant (Ratnappan et al., 2014). Notably, meta-analysis of transcriptome data revealed significant overlaps between genes induced in glp-1(bn18) mutants  and induced by either tBOOH or fasting ( Figure 2a). Notably, qPCR analysis showed that the expression of fmo-2 and other tBOOHand fasting-induced genes was also increased in long-lived glp-1 (e2141) mutants and that this was abrogated by loss of NHR-49 ( Figure 1e).
We also used RNA-seq to compare the transcriptomes of wild-type and nhr-49(nr2041) worms before and following exposure to tBOOH (Figure 2b; Tables S3 and S4). We found that tBOOH induced 250 genes more than fourfold in wild-type. Consistent with our meta-analysis with published datasets (Figure 2a), there was a significant overlap between these 250 tBOOHresponsive genes and previously published tBOOH, fasting, and glp-1-regulated genes (Table S5). Notably, of the 250 genes, the induction of 75 was compromised by loss of nhr-49, including fmo-2 and nlp-25 (FDR > 0.05 and >twofold reduced induction in nhr-49 mutants vs. wild-type); in contrast, these genes were not significantly (FDR < 0.05) deregulated by loss of nhr-49 in the absence of tBOOH (Table S4). Together, these data suggest that nhr-49 is required for increased expression of a shared set of genes following oxidative stress, fasting, and in at least one longlived mutant.
mdt-15 is an essential coactivator for SKN-1 in the arsenite response and for NHR-49 in the regulation of lipid metabolism genes (Goh et al., 2014;Taubert et al., 2006). ). This could reflect specific tBOOHinduced increased levels of one of the larger NHR-49::GFP isoforms ( Figure S4a), but is also consistent with the predominant isoform(s) becoming posttranslationally modified, for example, by phosphorylation.
2.5 | Stress activation of the p38 MAPK, PMK-1, is required for the tBOOH-induced expression of some NHR-49-dependent genes but dispensable for increase in NHR-49 protein levels The conserved p38 mitogen-activated protein kinase (MAPK) PMK-1 is activated by phosphorylation by the SEK-1 MAPKK in response to many stresses, including tBOOH. PMK-1-dependent phosphorylation of SKN-1 is vital for stress-induced increases in its activity Inoue et al., 2005). Hence, we hypothesized that these kinases might be required for tBOOHinduced increases in NHR-49 levels and/or fmo-2 expression.

| DISCUSSION
NHR-49 is a critical regulator of lipid metabolism and longevity (Burkewitz et al., 2015;Chamoli, Singh, Malik & Mukhopadhyay, 2014;Folick et al., 2015;Khan et al., 2013;Ratnappan et al., 2014;Seah et al., 2016), while related nuclear receptors, HNF4 and PPARa, are therapeutic targets that share at least some of these functions in mammals. Here, we show that NHR-49 is also vital for an adaptive transcriptional response to the organic peroxide tBOOH. This response includes the induction of the flavin-containing monooxygenase fmo-2, which is important for dietary restriction-induced longevity and resistance to several stresses (Leiser et al., 2015), and two other genes, K05B2.4 and sodh-1, that we reveal here to be important for resistance to organic peroxide. Moreover, we find that nhr-49 is also required for the induction of tBOOH-induced genes by acute fasting. Collectively, our data suggest that, in addition to its role in regulating lipid metabolism, NHR-49 participates in a cytoprotective acute stress response program that functions parallel to and a b c d e f F I G U R E 5 nhr-49 is required and sufficient for stress resistance. (a) Survival plots of wild-type N2, nhr-49(nr2041), and fmo-2(ok2147) worms on 6 mM tBOOH. (e, f) Survival plots of wild-type (e) and glp-1(e2141) (f) worms grown on control RNAi or RNAi clones targeting six different NHR-49-regulated genes while exposed to 6 mM tBOOH. Tables S10 and S11 show statistics and replicates independently of SKN-1/Nrf2 and HLH-30/TFEB signaling Lapierre et al., 2013). This raises the possibility that NHR-49's pro-longevity function, for example in glp-1 mutants (Ratnappan et al., 2014), may involve modulating both lipid metabolism and stress defenses ( Figure 6).
NHR-49 regulates transcription in response to fasting, to drive changes in lipid metabolism . Here, we report nhr-49-dependent expression of several fasting-induced genes with pro-survival/longevity functions that are not involved in lipid metabolism, including the glyoxylate cycle enzyme icl-1, the oxidoreductase sodh-1, and the flavin-containing monooxygenase fmo-2. Consistent with these findings, nhr-49-dependent induction of predicted detoxification genes was observed in another long-lived state (Chamoli et al., 2014). This was ascribed to NHR-49driven lipid metabolic reprogramming, but our findings suggest that these genes may contribute to NHR-49's stress-protective function. where skn-1 is important, but nhr-49 dispensable for the increase in lifespan (Ratnappan et al., 2014;Tullet et al., 2008), the requirement for mdt-15 may exclusively reflect MDT-15's function as a SKN-1 coregulator (Goh et al., 2014).
However, although fmo-2 is induced by some DR feeding regimens (Leiser et al., 2015), it is not upregulated in eat-2 mutants, and nhr-49 is dispensable for the increased longevity of this mutant (Heestand et al., 2013). These studies suggest that NHR-62 and NHR-49 may act in parallel pro-longevity pathways that are both regulated by nutrient availability and provide further evidence that different fasting/DR regimes evoke different pro-longevity transcriptional responses involving different transcriptional regulators to extend lifespan (Greer & Brunet, 2009).
Our discovery that fasting and peroxide stresses activate a similar, NHR-49-dependent transcriptional response raised the possibility that in both cases, NHR-49 might be regulated by a reactive oxygen metabolite. Hydrogen peroxide is likely increased upon fasting, potentially as a by-product of increased fatty acid oxidation.
Although it may cause oxidative damage, peroxide is also an established signaling molecule, regulating the activity of target proteins by oxidizing specific redox-sensitive protein thiols (Veal, Day & Morgan, 2007). However, using the oxidation status of the peroxide-sensitive peroxiredoxin-2 as a readout (Ol ahov a et al., 2008), we could find no evidence that fasting increases peroxide levels (data not shown).
Alternatively, lipid-derived ligand(s), including oxidized molecules, may increase NHR-49 activity, with exposure to organic peroxide or fasting both increasing the levels of such a lipid. A recent paper showed that linolenic acid, and an oxidized derivative thereof, regulate lifespan and gene expression in an nhr-49-dependent fashion (Qi et al., 2017). Although we note that fmo-2 was not among the responsive genes (Qi et al., 2017), this nevertheless raises the possibility that similar molecules may be involved in the nhr-49-dependent stress adaptation mechanisms identified here.

F I G U R E 6 Model for NHR-49 action in stress defense and longevity
Notably, our data suggest that NHR-49 activity is increased in response to tBOOH by increasing NHR-49 protein levels. In contrast to the increased NHR-49 levels in glp-1 mutants, which reflect DAF-16-and TCER-1-dependent upregulation of nhr-49 mRNA (Ratnappan et al., 2014), our data suggest that this increase does not reflect tBOOH-induced increases in nhr-49 mRNA levels ( Figure S4d). It remains to be determined whether this is due to a general increase in NHR-49 protein stability or specific increases in a particular isoform. However, the tBOOH-induced increase in levels of a lower mobility NHR-49 isoform suggests that increases in NHR-49 levels/ activity may involve tBOOH-induced posttranslational modification (s). We note that NHR-49 contains several candidate phosphorylation sites, some of which are phosphorylated in vivo (Bodenmiller et al., 2008). Moreover, phosphorylation regulates the activity of functionally related mammalian NHRs, for example PPARa (Burns & Vanden Heuvel, 2007). Hence, it is possible that tBOOH-induced phosphorylation might stabilize NHR-49 or direct its activity toward activating the expression of fmo-2 and other stress-induced genes.
The p38 MAPK, PMK-1, plays an important role in defending against oxidative stress and infectious microbes. For example, PMK-1dependent phosphorylation is vital for the stress-induced nuclear accumulation of SKN-1 and activation of SKN-1-dependent gene expression (Inoue et al., 2005). Moreover, PMK-1 is rapidly activated by phosphorylation by the SEK-1 MAPKK in response to tBOOH (Inoue et al., 2005). However, although loss of sek-1 and pmk-1

| Nematode strains and growth conditions
We cultured C. elegans strains using standard techniques. To avoid background effects, each mutant was crossed into our N2 strain; original mutants were backcrossed to N2 at least six times. E. coli OP50 was the food source in all qPCR and RNA-seq experiments except in Figure 3b, all population stress resistance experiments (tBOOH and starvation assays; Figure 5), and for the initial characterization of the fmo-2p::gfp reporter (Figures 1b and S1b); for all other experiments, we used E. coli HT115 as food source. All experiments were carried out at 20°C, except experiments with strains harboring the glp-1(e2141) mutation; these strains and pertinent controls were grown at 25°C until the L4 stage to induce germline loss, and then downshifted to 20°C. Worm strains used in this study are listed in Table S13.
We used standard genetic crossing techniques to attempt to construct the nhr-49(nr2041); hlh-30(tm1978) double mutant; genotyping the clonal progeny of 100 individual candidate F2 worms did not reveal any homozygous double mutants, suggesting that these null mutations cause synthetic lethality.
For experiments on media containing exogenous stressors, sodium meta-arsenite (Sigma 71287) and tert-butyl hydroperoxide solution (tBOOH; Sigma 458139), DTT, and H 2 O 2 were added at indicated concentrations. Plates containing tBOOH were made fresh on the day of use. For the RNAi candidate screen, we used 10 mM tBOOH and a 3-hr exposure as these conditions resulted in strong and uniform GFP induction throughout the worm population. For qPCR and RNA-seq experiments, we used 7.5 mM tBOOH exposure for 4 hr, consistent with our previous study (Goh et al., 2014). For stress resistance assays, we found that resolution of population death events was improved using 6 mM tBOOH and hence used this concentration. ll of pRF4 rol-6(su1006) co-injection marker, and 50 ng/ll pBluescript carrier DNA. Two independently generated lines were selected using the roller phenotype. Both lines showed similar expression patterns and induction by tBOOH (data not shown).
Starvation survival assays were adapted from (Lee & Ashrafi, 2008). Worms were maintained for at least two generations in a nonstarved state and synchronized to obtain day 1 adults. Adults were then bleached to obtain embryos that were hatched in S-basal medium without cholesterol overnight on a rotator. Synchronized L1s were transferred to S-basal medium without cholesterol containing an antibiotic-antimycotic mix (Gibco 15240062) at a concentration of~1 worm/ll. To assess viability, 200-300 animals were transferred to seeded NGM plates every 2-3 days and assessed for growth to the L4 stage after 48 hr. Average survival and standard error of four independent biological repeats are shown in Figure 5b; statistical significance was calculated using area under the curve and Student's t test.

ACKNOWLEDG MENTS
Some strains were provided by the CGC, which is funded by NIH