A non-canonical role for the C. elegans dosage compensation complex in growth and metabolic regulation downstream of TOR complex 2

The target of rapamycin (TOR) kinase regulates metabolism, development, growth and lifespan by its action in two distinct complexes: TOR complex 1 and TOR complex 2 (Bhaskar and Hay, 2007; Alessi et al., 2009; Jones et al., 2009; Soukas et al., 2009; Lamming et al., 2012). In C. elegans, loss-of-function mutations in the gene encoding the essential TOR complex 2 (TORC2) member RICT-1 leads to small worms with delayed larval development, reduced brood size and shortened lifespan with seemingly inappropriately increased fat stores (Soukas et al., 2009). These phenotypes, with the exception of increased body fat, associated with mutation of TORC2 can be phenocopied by loss of function of the downstream serum and glucocorticoid-induced kinase (SGK-1), a protein kinase A, protein kinase G, protein kinase C (AGC) family kinase.

In C. elegans, gain-of-function mutations in sgk-1 partially suppress the high fat and small body size phenotypes of the TORC2 mutant rict-1 (Jones et al., 2009). Similarly, gain-of-function mutations in Ypk2, the yeast ortholog of Sgk, suppress the lethality of Saccharomyces cerevisiae TORC2 mutations through effects on ceramide metabolism (Kamada et al., 2005; Aronova et al., 2008). Mammalian TORC2 (mTORC2) also phosphorylates and activates Sgk (García-Martínez and Alessi, 2008), though key differences between mammals and lower eukaryotes exist. In knockout mice, it is the protein kinase Akt that mediates many of the metabolic effects of mTORC2 (Bhaskar and Hay, 2007; Hagiwara et al., 2012; Lamming et al., 2012; Yuan et al., 2012). Whereas defects in glucose metabolism in mTORC2 knockout mice are the result of defective Akt signaling through FoxO (Hagiwara et al., 2012; Yuan et al., 2012), in C. elegans, the growth, reproductive and lifespan-shortening phenotypes of TORC2 mutants do not depend upon AKT signaling through FoxO (Jones et al., 2009; Soukas et al., 2009). Still, certain aspects of mammalian metabolic regulation by TORC2 cannot be fully explained by loss of Akt activation, indicating additional, unelucidated outputs of TORC2 signaling (Yuan et al., 2012).

In C. elegans, understanding of the genetics of the TORC2 pathway remains incomplete. Although the lifespan of C. elegans TORC2 mutants is shortened under standard laboratory conditions, inactivation of TORC2 by RNAi leads to lifespan extension in a manner dependent upon the Nrf ortholog SKN-1 (Hertweck et al., 2004; Soukas et al., 2009; Robida-Stubbs et al., 2012). Other phenotypes of TORC2 mutants are not suppressed by loss of skn-1 (Soukas et al., 2009). Thus, mechanistic knowledge of TORC2 and SGK-1 signaling that modulates metabolism, growth and reproduction is incomplete.

In yeast, TOR has been tied to alterations in chromatin structure. TORC1 mutations in S. pombe lead to dysregulated chromatin structure (Rohde and Cardenas, 2003; Tsang et al., 2003). Alternatively, yeast TORC2 mutants demonstrated altered heterochromatin structure with upregulation of repeated elements and subtelomeric genes resembling mutations in histone deacetylase or RSC chromatin remodeling complex members (Schonbrun et al., 2009). In C. elegans, however, studies have not yet illuminated the role of epigenetic regulation in TOR biology.

C. elegans exists as either a male (XO) or a hermaphrodite (a self-fertilizing XX female with larval stage spermatogenesis). Much like humans and other animals, C. elegans males have one X chromosome and thus have half of the X chromosomal gene dosage of female or hermaphroditic animals. X chromosomal dosage compensation (DC) has evolved in order to equalize the expression of the X-linked genes between males and females. The mechanisms by which DC occurs varies greatly between species: mammals transcriptionally mute expression of one of the two female X chromosomes (Lyon, 1961), flies double the expression of the single male X chromosome (Gelbart and Kuroda, 2009) and the C. elegans hermaphrodite downregulates gene expression from both X chromosomes by 50% (Meyer and Casson, 1986). This X chromosomal regulation is orchestrated through the dosage compensation complex (DCC).

The C. elegans DCC binds to both X chromosomes in the hermaphrodite to downregulate gene expression by half. The DCC is highly specialized and consists of a core condensin 1-like complex composed of three chromosome-associated polypeptides (DPY-26, DPY-28 and CAPG-1) and two structural maintenance proteins (MIX-1 and DPY-27) involved in chromosomal structure and segregation. Other associated members of the DCC include the SDC-1, SDC-2, SDC-3, DPY-30 and DPY-21, which, through SDC-2, SDC-3 and DPY-30, allow for the DCC to be recruited specifically to the X chromosome in the hermaphrodite. The mechanism of DCC downregulation of X chromosomal genes remains elusive; however, two recent investigations suggest that the DCC stimulates formation of the repressive epigenetic mark histone 4, lysine 20 monomethyl (H4K20me1) (Vielle et al., 2012; Wells et al., 2012). Further, the DCC is required for the enrichment of H4K20me1 on the X chromosome, as loss of dpy-21 reduces H4K20me1 (Vielle et al., 2012). The histone 4 lysine 20 (H4K20) monomethyltransferase SET-1 and di/trimethyltransferase SET-4, as well as the histone deacetylase SIR-2.1, play key roles in regulating the methylation state and acetylation state of histone H4 on the X chromosome downstream of the DCC (Wells et al., 2012).

In the present study, we identified DCC member dpy-21 through a reverse RNAi genetic screen for suppressors of the slow growth phenotype of rict-1 animals. dpy-21 inactivation suppresses multiple TORC2/rict-1 pleiotropies, restoring a near-normal brood size and reducing elevated fat mass to normal levels. Further investigation revealed that knockdown of other members of the DCC could also suppress rict-1 phenotypes, suggesting that the DCC acts downstream of TORC2/rict-1 to negatively regulate growth, reproduction and metabolism. Knockdown of dpy-21 resulted in a faster larval development for both hermaphrodites and male worms alike, implying a non-canonical mechanism of DCC functionality to suppress gene expression independent of gender. Genomic inactivation of dpy-21 negatively influenced the lifespan of rict-1 animals, suggesting that loss of dpy-21 is beneficial for larval development but results in detrimental effects on longevity in adult worms. We uncovered the histone 4 lysine 20 (H4K20) monomethyltransferase SET-1 and di/trimethyltransferase SET-4 as the major effectors of DCC functionality downstream of TORC2. Interestingly, rict-1 mutant animals are depleted of H4K20 monomethyl marks (H4K20me1 and H4K20me3), which is suppressed by abrogation of DCC activity. The data suggest a major role for epigenetic regulation in mediating TORC2 phenotypes and that a complex balance of H4K20 methylation is necessary for regulation of normal development, growth and metabolism. Finally, we find that DPY-21 protein physically associates with SGK-1, a major downstream kinase in the TORC2 pathway, suggesting that TORC2 directly regulates the DCC. The current work ties TORC2 signaling to epigenetic gene regulatory mechanisms, and invokes a new role for the DCC in development and metabolism through autosomal gene regulation. We propose that the DCC and DPY-21 are novel downstream negative regulators of the TORC2/SGK-1 pathway.

C. elegans animals were grown and maintained on Nematode Growth Media (NGM) seeded with Escherichia coli OP50 as previously described (Soukas et al., 2009). The wild-type strain was N2 Bristol. The following C. elegans mutant strains were used: MGH1 rict-1(mg451), MGH2 rict-1(mg450), MGH26 rict-1(mg451);sgk-1(mg455), MGH35 sinh-1(mg452), CB428 dpy-21(e428), MGH229 dpy-21(e428);rict-1(mg451) and MGH12 mgIs60[SGK-1::GFP pRF4].

Synchronization of animals was performed through egg preparation of a well-fed, gravid adult population of worms. Adult animals were collected into M9 media and centrifuged at 4000 rpm (3300 g) for 30 seconds. The worm pellet was resuspended in a solution of 1.3% bleach and 250 mM NaOH, and agitated until worm corpses dissolved or 5 minutes had elapsed, whichever occurred first. Eggs were quickly centrifuged at 4000 rpm for 30 seconds and the supernatant removed. Eggs were washed with sterile M9 five times. Eggs were resuspended in 6 ml M9 and allowed to hatch and synchronize to L1 larvae for 18 hours at 20°C with gentle rotation.

RNAi clones were isolated from a genome-wide E. coli RNAi library, sequence verified, and fed to animals as described (Kamath and Ahringer, 2003). RNAi feeding plates (6 cm) were prepared using a standard NGM recipe with 5 mM isopropyl-b-d-thiogalactopyranoside and 200 μg/ml carbenicillin. RNAi clones were grown for 15 hours in Luria Broth (LB) containing 200 μg/ml carbenicillin with shaking at 37°C. The stationary phase culture was then collected, concentrated through centrifugation, the supernatant discarded and the pellet resuspended in LB to 10% of the original culture volume; 300 μl of each RNAi clone concentrate was added to RNAi plates and allowed to dry no more than 48 hours prior to adding the worm embryos or animals.

Developmental timing analysis was conducted on synchronized L1 animals prepared as above. Synchronous animals were dropped onto RNAi plates containing an empty vector RNAi control or a sequence-verified target RNAi in E. coli HT115 and grown at 15°C. Time to adulthood was measured from the time when synchronized L1 animals were first exposed to food until the time at which the animal reached adulthood. Hermaphrodite animals were scored for their transition into adulthood by appearance of the vulvar slit. Male adulthood was determined by ray and fan formation (Nguyen et al., 1999). Two to three biological replicates were carried out for each condition examined and data are displayed in the supplementary material as indicated in the text.

Synchronous L1 larvae were dropped onto RNAi plates containing appropriate RNAi E. coli bacteria. After growing to young adulthood, single animals were transferred to new NGM plates containing fresh bacteria containing the same RNAi clone each day for 5 days (n=10 per group). Progeny from parents were counted and summed from each of the five transfer plates to determine brood size.

Body fat mass was measured as previously described (Pino et al., 2013). All analyses were performed on identically staged young adult worms, which were washed off of RNAi plates, fixed in 40% isopropanol for 3 minutes, and stained in 40% isopropanol containing 3 μg/ml Nile Red. Worms were washed with PBS with 0.01% Triton X-100, mounted on agarose pads, and fluorescently imaged using GFP/FITC filters. Body fat mass was scored by fluorescence detection of the Nile Red-stained C. elegans lipid droplets using MetaMorph software and expressed as the mean of at least 25 animals per group.

Body size of C. elegans mutants was determined as the maximal, longitudinal, cross-sectional area and measured as previously described (Soukas et al., 2009). Synchronous L1 larvae were exposed to RNAi plates containing the appropriate RNAi strain. Animals were processed as above for body fat measurement, imaged by brightfield microscopy, and body area calculated by MetaMorph software for a minimum of 25 animals per group.

Synchronous L1 animals were placed on RNAi plates seeded with the appropriate RNAi clone, either empty vector or dpy-21 RNAi. As day 1 adults, 30 animals per genotype were transferred in quadruplicate (120 total) to fresh RNAi plates with the corresponding RNAi supplemented with 10-50 μM 5-fluorodeoxyuridine (FUDR) solution to suppress progeny production. Three independent assays were carried out with concordant results. Live, dead and censored worms were calculated daily in the worm populations by scoring movement with gentle prodding when necessary. Data were analyzed and statistics performed through OASIS (http://sbi.postech.ac.kr/oasis/surv/). A Kaplan-Meier estimator and a log cumulative hazard plot provided by OASIS were used to estimate individual survival and mortality over the lifespan of the worms. A non-parametric Mantel-Cox log-rank test was used for comparison of different survival functions.

The mg452 mutant was obtained from an F2 forward genetic screen that has been previously described (Soukas et al., 2009). Individual isolates were backcrossed to N2 Bristol and positionally cloned based on polymorphisms between N2 and the multiply polymorphic C. elegans strain CB4586.

Five thousand synchronous mid-L4 stage animals were collected for each genotype and knockdown. Collected worms were washed three times in M9 media and snap frozen in liquid nitrogen. Worms were thawed on ice by adding two pellet volumes of 1% SDS, 50 mM Tris, pH 7.4, and 100 mM NaCl, heated to 80°C on a dry heat block for 5 minutes, and sonicated in a BioRuptor XL for 10 minutes on maximum intensity (30 second pulses with 30 seconds rest between pulses) to obtain lysates. Lysates were cleared by centrifugation at 21,000 g for 15 minutes. Sixty micrograms of total worm protein was electrophoresed by SDS-PAGE, and transferred to nitrocellulose membrane. Even transfer was confirmed by Ponceau S staining. Membranes were blocked with TBST (TBS + 0.1% Tween20) containing 5% nonfat milk, washed twice with TBST, and blotted overnight at 4°C with anti-H4K20me1 (Abcam, ab9051) rabbit polyclonal antibody at 1:2000, anti-H4K20me3 (Abcam, ab78517) mouse monoclonal antibody at 1:500, or total histone H4 (Abcam, ab10158) rabbit polyclonal antibody at 1:1000 in TBST and 5% BSA. Anti-rabbit HRP secondary antibody (Thermo Pierce) was used at 1:10,000 (H4K20me1 and total histone H4) or anti-mouse HRP secondary antibody (Thermo Pierce) at 1:10,000 (H4K20me3), all in TBST at room temperature for 2 hours. Thermo SuperSignal West Pico Chemiluminescent Substrate Kit was used for detection.

The C-terminus of the C. elegans dpy-21 gene was cloned into the pGEX-KG vector under the control of the tac promoter. The 1626 base pair fragment was amplified from wild-type (WT) worm cDNA using the following primers: PF, TACGTCGACTCGAGCAGTTGATGTCGAGGAAG; PR, ACCAAGCTTCTATTCAGTTGATTCACGCACTTG. The fragment and vector was digested with HindIII and Sal1 and individually gel purified. The fragments were ligated to produce pGST-DPY-21(920-1461). GST-DPY-21(920-1461) was expressed in E. coli BL21(DE3)plysS from pGEX-KG plasmids and purified with Pierce GST Spin Purification Kit (Thermo Scientific, Prod# 16106). Five micrograms purified GST and GST-DPY-21(920-1461) proteins were re-immobilized with pre-equilibrated 20 μl GST beads in lysis buffer.

Three milliliters of packed, mixed-stage SGK-1::GFP transgenic worms were collected and washed three times in M9 buffer and stored at -80°C. Worms were freeze ground and lysed in lysis buffer [1 × PBS, 1 mM EDTA, 0.5% NP-40, 50 mM β-glycerophosphate, 0.1 mM NaVO4, 50 mM NaF and protease inhibitor cocktail (EDTA-free, Roche)]. The worm extract was further lysed twice through French Pressure cell (Thermo Scientific) and centrifuged for 30 minutes at 38,000 rpm (178,000 g) at 4°C in a Beckman Ti-41 rotor. The supernatant was filtered through a 0.45 μm filter and protein concentration determined with Pierce BCA Protein Assay Kit (Thermo Scientific) before the pull down assay. Three hundred milliliters of 5 μg/ml SGK-1::GFP lysate was incubated with the immobilized beads at 4°C for 1.5 hours. After washing beads three times with lysis buffer, samples were analyzed by western blot using rat anti-GFP antibody (JFP-J5, Riken, 1:2000).

Our previous work shows that mutations in rict-1 (ortholog of human Rictor), a critical subunit of the heteromeric kinase TOR complex 2 (TORC2) in C. elegans, lead to developmental delay (Soukas et al., 2009). In order to identify genes acting downstream of TORC2/SGK-1 regulating the essential process of development, we performed a reverse genetic RNA interference (RNAi) screen to genes encoding proteins that contain the potential target phosphorylation sequence for SGK-1 (RXRXXS/T-p) (Bodenmiller et al., 2008).

RNAi to only one of these genes, dpy-21, a member of the DCC (Meyer, 2005), significantly suppressed the developmental delay evident in rict-1 mutants in multiple replicate experiments. In order to quantify the suppression of rict-1 developmental delay, we studied the developmental timing profile of both wild type and rict-1 mutants exposed to dpy-21 RNAi. We measured the time from the L1 larval stage to development of the mature vulva (indicative of the young adult stage), and found that dpy-21 RNAi significantly shortened the average time to adulthood of rict-1 mutants (Fig. 1A; supplementary material Table S1). We were surprised to see that dpy-21 RNAi also significantly, albeit to a much lesser extent, shortened developmental timing in wild-type animals (Fig. 1A; supplementary material Table S1). Data from triplicate analyses were consistent with these findings (supplementary material Table S1). This indicates that reduction in function of dpy-21 accelerates development and does so downstream of TORC2.

We next measured other pleiotropies of rict-1 mutants when exposed to dpy-21 RNAi. rict-1 mutants at young adulthood show a 30-100% increase in fat mass as assessed by fixation-based lipid staining or quantitative lipid biochemistry (Fig. 1B) (Soukas et al., 2009). RNAi to dpy-21 fully rescued the increased fat mass of rict-1 mutant animals to a level equivalent to that of wild-type animals (Fig. 1B). dpy-21 knockdown also reduced the fat mass of wild-type animals; however, the effect was much less powerful than in rict-1 mutants.

rict-1 mutants have a brood size that is ∼40% of wild-type animals, and deliver that brood over a much longer time period than do wild type (Fig. 1C,D; supplementary material Table S2) (Soukas et al., 2009). When treated with RNAi to dpy-21, brood size was returned nearly to normal for rict-1 mutants, and the progeny was delivered over a much shorter timespan than the control mutant animals (Fig. 1C,D). This effect was consistent between triplicate analyses of rict-1 brood with dpy-21 RNAi (supplementary material Table S2). dpy-21 RNAi had no effect on the brood size of wild-type animals.

We next analyzed dpy-21 loss-of-function mutants. Unlike RNAi, nonsense mutations of dpy-21 lead to dramatically slowed developmental rate and reduced brood size, and to further delayed development and reduced brood size in dpy-21;rict-1 double mutants (supplementary material Fig. S1A-C). This suggested that partial reduction of DPY-21 function or a spatiotemporal loss of function that was not recapitulated by nonsense mutation in dpy-21 was needed to suppress rict-1 mutant phenotypes. Although body fat mass was decreased in dpy-21;rict-1 double mutants (supplementary material Fig. S1D), we believe this to be a consequence of the poor fitness of the double mutant animal, as body size in dpy-21 and dpy-21;rict-1 double mutants was substantially decreased (supplementary material Fig. S1E). We confirmed that dpy-21 mRNA was efficiently knocked down (88.8±9.7%, P<0.02) by dpy-21 RNAi (data not shown). Thus, RNAi was used for all subsequent analyses.

In spite of dpy-21 RNAi suppressing multiple rict-1 mutant phenotypes, we noted negative effects when lifespan and body size were examined. dpy-21 RNAi significantly shortened the lifespan of rict-1 mutants from an average of 14.1±0.28 days (mean ± s.e.m.) (vector RNAi) to 9.36±0.37 days (dpy-21 RNAi) (Fig. 2A; triplicate analysis in supplementary material Table S3) at 20°C. dpy-21 RNAi also shortened lifespan in wild-type animals to an equivalent extent, suggesting that, unlike for growth rate, fat mass and brood size, dpy-21 does not act in the same pathway as rict-1 when modulating lifespan. In a similar fashion, rict-1 mutants, which are normally 75% of the area of wild-type worms at a similar developmental stage, when treated with dpy-21 RNAi showed a further 15% decrease in body size (Fig. 2B; supplementary material Fig. S2A). Thus, RNAi to dpy-21 seems to positively affect larval growth rate, reduce fat mass and boost brood size at the expense of shortened lifespan and reduced overall body size.

In order to establish that dpy-21 regulates growth specifically in the TORC2 pathway, we returned to the forward genetic screen in which we had previously isolated two loss-of-function alleles in rict-1 (mg450 and mg451) and two loss-of-function mutations in sgk-1 (mg455 and mg456) that phenocopy rict-1 (Soukas et al., 2009). In this screen, we had isolated an additional mutant with slow growth and small body size that did not contain mutations in sgk-1 or rict-1 and mapped to LG II, albeit closer to the centromere than rict-1. Using the polymorphic strain CB4856 we refined the map location of this mutant, mg452, to a region containing another well-described essential subunit of TORC2, sinh-1, the C. elegans ortholog of mSin1 (Mapkap1 - Mouse Genome Informatics) (Fig. 3A) (Hansen et al., 2005). mg452 contains an early stop mutation in the second to the last coding exon of sinh-1, predicted to truncate 139 of the total 642 amino acids from the C-terminus of SINH-1 (Fig. 3B).

We reasoned that if RNAi of dpy-21 was acting to suppress rict-1 phenotypes through activity in the TORC2 pathway, it should also suppress sinh-1 mutant growth delay. Similar to rict-1 mutants, sinh-1(mg452) mutants show a pronounced delay in reaching adulthood (Fig. 3C). RNAi of dpy-21 in sinh-1(mg452) mutants led to suppression of growth delay (Fig. 3C; duplicate analysis in supplementary material Table S4).

TORC2 acts in a genetic pathway with sgk-1 regulating growth rate, body size, reproduction, lifespan and fat mass (Jones et al., 2009; Soukas et al., 2009). sgk-1 loss-of-function mutants show a similar developmental delay to rict-1 animals (Fig. 3D), so we investigated whether dpy-21 RNAi acted in a manner dependent upon SGK-1 or whether DPY-21 acts downstream of SGK-1. As dpy-21 suppressed rict-1;sgk-1 mutant growth delay, we reasoned that DPY-21 was acting downstream of TORC2 and SGK-1 to regulate growth (Fig. 3D; duplicate analysis in supplementary material Table S4).

In C. elegans, DPY-21 acts as a member of the DCC. The DCC is responsible for binding to X chromosomes in somatic cells of hermaphrodite (XX) animals, thereby enacting changes that reduce X-linked gene expression by half (Meyer, 2005). Although the DCC functions canonically in hermaphrodite animals, it remains unknown how the specific DCC components interact with or affect other condensin complexes and cellular processes in the cell. In order to test whether dpy-21 was functioning to suppress growth delay and reproductive defects in rict-1 mutants through its role in the DCC, we inactivated six of the nine additional members of the DCC by RNAi in rict-1 mutants. These data indicated that all six of these RNAi either suppressed growth delay or low brood size in rict-1 mutants (Fig. 4). RNAi of the core DC component dpy-27 (Chuang et al., 1994) had the strongest suppressive effect on rict-1 developmental delay (Fig. 4A; supplementary material Table S5), followed by dpy-30, sdc-1, sdc-2 and capg-1 (Fig. 4C-F; supplementary material Table S5). Triplicate analyses showed consistent results (supplementary material Table S5).

Although RNAi of sdc-2 and dpy-28 had only minor effects on rict-1 developmental rate (Fig. 4B,E; supplementary material Table S5), both RNAi significantly increased both the total brood size of rict-1 animals (Fig. 4G,H). RNAi to both dpy-28 and sdc-2 reduced the number of live progeny per animal for wild type, but had an exact opposite effect on rict-1 mutants.

RNAi to dpy-21 suppressed the elevated fat mass of rict-1 mutants, so we determined next what effect knockdown of other DCC component genes had on fat mass. Much like dpy-21, RNAi to dpy-27, dpy-28, dpy-30, sdc-1, sdc-2 and capg-1 all reduced fat mass of rict-1 mutants and had little effect in wild-type animals (Fig. 4). Further supporting the idea that knockdown of the DCC does not suppress the small body size pleiotropy of rict-1 mutants, we saw no change or further decrease in body size with knockdown of DCC genes (supplementary material Fig. S2B).

The major reported role for the DCC is the reduction in X chromosomal gene expression in hermaphrodites. Given that male C. elegans carry a genotype of XO, the DCC does not perform X chromosomal dosage compensation in males. If RNAi to dpy-21 was suppressing rict-1 phenotypes by its canonical role in X chromosomal gene dosage, rict-1 male growth delay should not be suppressed. We noted reproducible suppression of rict-1 male growth delay by dpy-21 RNAi, albeit to a lesser extent than in hermaphrodites (Fig. 5; supplementary material Table S6). Results were qualitatively similar in triplicate repeat experiments (supplementary material Table S6).

In embryos, the DCC controls methylation of histone H4 at lysine 20 (H4K20) such that in states of increased DCC activity, increased SET-1 activity leads to higher H4K20 monomethylation (H4K20me1) (Vielle et al., 2012). SET-4 activity, which di- and trimethylates H4K20, is correspondingly decreased during embryonic dosage compensation (Vielle et al., 2012), but SET-4 has been reported to mediate some of the effects of the DCC (Wells et al., 2012). H4K20me1 likely contributes to the chromosomal condensation (Nishioka et al., 2002; Rice et al., 2002), restricting expression of X chromosomal genes during dosage compensation. The role of H4K20me1 and H4K20me3 and their connection to the DCC post-embryonically has not been well established.

We hypothesized that altered gene expression as a consequence of modulations in chromosome condensation might be responsible for rict-1 mutant developmental delay, decreased brood size and increased fat mass phenotypes. Given that H4K20me1 is responsible for mediating effects downstream of the DCC, we inactivated set-1 in rict-1 mutants by RNAi. set-1 RNAi strongly suppressed developmental delay of rict-1 mutants (Fig. 6A; supplementary material Table S7). Further, RNAi to set-4 also potently suppressed rict-1 developmental delay (Fig. 6B; supplementary material Table S7). Triplicate analyses of set-1 and set-4 RNAi were consistent (supplementary material Table S7). sir-2.1, which has been reported to act to deacetylate lysine 16 of H4 (H4K16Ac) downstream of the DCC (Wells et al., 2012), also modestly suppressed developmental delay in rict-1 mutants (Fig. 6C; supplementary material Table S7).

The elevated fat mass evident in rict-1 mutants was reduced significantly by RNAi to set-1 and set-4 (Fig. 6D). Much like knockdown of dpy-21, RNAi to set-1 and set-4 did not suppress the small body size or shortened lifespan of rict-1 mutants (Fig. 6E,F; supplementary material Table S8). This is further evidence that these H4K20 methyltransferases act as effectors of the DCC in the TORC2 pathway.

Although RNAi to the DCC component genes set-1 and set-4 can suppress TORC2 mutant phenotypes, it was not clear whether DCC, SET-1 and SET-4 activity are altered in TORC2 mutants, and whether this could be responsible for mutant phenotypes. The full extent of how epigenetic modifications affect biological phenotypes is not known; however, in yeast, loss of TOR signaling has been suggested to increase nucleolar condensation (Tsang et al., 2003). Thus, we next sought to determine whether TORC2/rict-1 mutants have altered H4K20me1 or H4K20me3 marks. Despite our hypothesis that increased DCC activity in rict-1 mutants would be positively correlated with H4K20me1 and H4K20me3, we found both marks to be decreased overall in rict-1 mutants (Fig. 7A-D). This suggests that the DCC, at the post-embryonic developmental stage studied, is negatively correlated with SET-1 and SET-4 activity. Consistent with this, knockdown of dpy-21 by RNAi led to elevation of H4K20me1 and H4K20me3 in both wild-type and rict-1 mutant animals (Fig. 7A-D). Knockdown of set-1 led to reductions in H4K20me1 and H4K20me3 (Fig. 7A,B), as previously published for mutant set-1 animals (Vielle et al., 2012). Alternatively, knockdown of set-4 led to the largest increase in H4K20me1 levels (Fig. 7A,C) and only a slight trend towards lower H4K20me3 levels (Fig. 7B,D).

In order to address possible mechanisms of regulation of the DCC by the TORC2 pathway, we tested whether DPY-21 protein could physically associate with SGK-1. DPY-21 protein carries two potential phosphorylation sites for the TORC2 pathway kinase SGK-1, and we reasoned that the two proteins might associate. We found that purified C-terminal DPY-21 fused to glutathione-s-transferase (GST) was able to associate with and pull down SGK-1 tagged with GFP from whole-worm extract (Fig. 7E). The interaction was reproducible and specific, as free GFP did not associate with DPY-21-GST and SGK-1-GFP did not associate with GST alone (Fig. 7E).

In the current study, we identified dpy-21 and the dosage compensation complex (DCC) as negative regulators of development acting downstream of the TORC2/SGK-1 pathway. dpy-21 suppressed rict-1 developmental delay in a reverse genetic, RNAi screen of candidate SGK-1 targets. dpy-21 has been broadly characterized in C. elegans as a member of the DCC, which is involved in downregulation of the hermaphroditic X chromosome by 50% in order to prevent overexpression of potentially toxic X-linked genes (Yonker and Meyer, 2003; Meyer, 2005). We find in this study that wild-type dpy-21 operates downstream of the TORC2 pathway to regulate developmental rate in larval stage worms, to increase body fat and to reduce brood size. Thus, reduction in function of dpy-21 or other members of the DCC by RNAi in rict-1 mutants reverses the slow developmental rate, lowers body fat and raises brood size. RNAi to dpy-21, however, has negative effects on rict-1 mutant body size and longevity, resulting in a reduced lifespan and a decrease in body size. These antagonistic results suggest divergence of the TORC2 pathway upstream of DPY-21 and the DCC. The TORC2 effector kinase SGK-1 physically associates with DPY-21 protein, suggesting that physical associations between SGK-1 and the DCC are responsible for regulating its activity. We further show that the DCC is likely to be hyperactive in rict-1 mutants, and is associated with decreased levels of the silencing epigenetic marks H4K20me1 and H4K20me3. Our work demonstrates that the DCC acts at least partly in a non-canonical fashion to suppress TORC2 mutant phenotypes as dpy-21 RNAi also suppresses developmental delay of male rict-1 mutants, in which classical dosage compensation does not occur. Finally, we can suppress rict-1 developmental delay by RNAi to the H4K20 monomethyltransferase set-1 and di/trimethyltransferase set-4, indicating that these epigenetic marks are crucial for TORC2 mutant phenotypes.

TOR is a serine/threonine kinase that serves as a governor of cellular energetics, metabolism and growth. TOR functions as a biologically conserved core for the functionally distinct TORC1 and TORC2 complexes (Helliwell et al., 1994; Bhaskar and Hay, 2007). The TORC2/Rictor pathway has been implicated in not only cellular structure and dynamics but also energy metabolism, feeding behavior, growth, reproduction, lifespan and numerous disease states (Jacinto et al., 2004; Sarbassov et al., 2005; Bhaskar and Hay, 2007; Guertin and Sabatini, 2007; Guertin et al., 2009; Jones et al., 2009; Soukas et al., 2009). By identifying DPY-21 and the DCC as downstream effectors of TORC2, we add a new dimension to our understanding of biological outputs of TORC2.

In mammals, three AGC family kinase targets of mTORC2, SGK, AKT (protein kinase B) and protein kinase Cα are subject to activation via TORC2-mediated phosphorylation on a conserved Ser/Thr residue within a C-terminal hydrophobic motif (HM) (Sarbassov et al., 2005; García-Martínez and Alessi, 2008; Ikenoue et al., 2008). In C. elegans, the only known targets of SGK-1 and AKT-1 are the FoxO transcription factor DAF-16 and the stress responsive ortholog of the mammalian NRF1/2/3, SKN-1 (Paradis and Ruvkun, 1998; Brunet et al., 2001; Hertweck et al., 2004; Greer and Brunet, 2008; Tullet et al., 2008). We previously found the TORC2 C. elegans ortholog of Rictor, RICT-1, acts through SGK-1 to regulate reproduction, growth, feeding behavior and lifespan, and through AKT and SGK-1 to regulate fat metabolism (Soukas et al., 2009). Here, we report similarly that mutations in sinh-1, the C. elegans ortholog of Sin1 (S. pombe stress activated protein kinase interactor), a conserved member of TORC2 signaling pathway, led to comparable pleiotropies to rict-1 and sgk-1 mutant animals.

Dosage compensation (DC) is an evolutionarily conserved mechanism between mammals, flies and invertebrates for balancing sex chromosome expression between genders (Lyon, 1961; Meyer and Casson, 1986). In C. elegans, five proteins are involved in the core DCC, which is structurally similar to the condensin 1 complex: DPY-26, DPY-27, DPY-28, MIX-1 and CAPG-1 (Csankovszki et al., 2004; Meyer, 2005). The additional five associated members are SDC-1, SDC-2, SDC-3, DPY-30 and DPY-21 (Hodgkin, 1987; Hsu and Meyer, 1994; Lieb et al., 1996; Yonker and Meyer, 2003). In this work, we show that DPY-21 is most likely to be working via its role in the DCC as RNAi directed towards each DCC member tested had suppressive effects on rict-1 phenotypes. These data comprehensively suggest that reduced function of the DCC results in beneficial effects specifically in TORC2 animals whereas wild-type animals experience minor or unfavorable effects, such as reduced progeny production.

Our results show that dpy-21 and the DCC regulate development and metabolism downstream of TORC2 and SGK-1. RNAi of dpy-21 in rict-1 and sinh-1 single mutants and in the rict-1;sgk-1 double mutant resulted in suppression of slow developmental rate. This suggests that DPY-21 functions specifically in the TORC2 pathway downstream of sgk-1 by virtue of its suppression of developmental phenotypes associated with mutants in TORC2, rict-1, sinh-1 and sgk-1.

The finding that the DCC component DPY-21 physically associates with SGK-1 proteins suggests a mode of regulation of the DCC by the TORC2 pathway. We originally identified dpy-21 by virtue of its peptide sequence, which encodes two potential SGK phosphorylation sites. The DPY-21 and SGK-1 physical interaction supports the idea that SGK-1, and therefore TORC2, directly regulate the DCC, and may do so by phosphorylation of DPY-21. However, this conclusion requires further testing to determine the tissue and mechanism of action of the DCC in regulating TORC2 phenotypes.

Our data suggest that the DCC is negatively regulated in wild-type animals by normal activity of TORC2/SGK-1 signaling. We provide evidence that RNAi directed to dpy-21 decreased DCC activity post-embryonically, resulting in increased levels of H4K20me1 and H4K20me3. In rict-1 mutants, alternatively, we noted a decrease in mono- and tri-methylation of histone 4 lysine 20 (H4K20), epigenetic marks previously associated with DCC activity (Vielle et al., 2012; Wells et al., 2012). However, unlike previous studies, our data indicate that DCC activity is negatively associated with overall levels of H4K20me1 and H4K20me3. Thus, our data suggest that DCC activity is increased in TORC2 mutants, leading to lowered H4K20 methylation, growth delay, small body size, elevated fat mass and reduced brood size. It remains an open question as to how DCC activity is normally suppressed by TORC2, and what role this plays in development, although our data indicate that direct regulation by SGK-1 is a possibility. We speculate that TORC2 must modulate the DCC activity to allow for normal developmental rate while maintaining normal reproduction and metabolism (Fig. 7F). Thus, we build on the model of complex TORC2 biology in which this pathway serves as a novel descending input into the DCC linking metabolism and development to the chromatin state of C. elegans.

C. elegans TORC2 mutants have reduced life expectancy compared with wild-type animals (Fig. 2A) (Soukas et al., 2009). Lifespan analysis of wild-type and rict-1 animals when treated with control or dpy-21 RNAi indicates that loss of DCC functionality leads to a truncated lifespan in both backgrounds. Similarly, body size is also negatively impacted by dpy-21 RNAi in both wild-type and rict-1 animals. Thus, dpy-21 knockdown results in separable phenotypes with both beneficial and detrimental outcomes. The collective data indicate that dpy-21 acts downstream of TORC2 with regard to only certain pleiotropies (Fig. 7F). This suggests that signals from TORC2 are complex, and are not regulated by a single pathway process, but rather multiple signaling outputs. We conclude that the DCC is one of these novel outputs.

dpy-21 and the DCC act in a non-canonical manner to suppress developmental rate downstream of TORC2. Because dosage compensation, as described in hermaphrodites, does not occur in males (Csankovszki et al., 2004), we expected dpy-21 RNAi to have little effect on developmental timing in rict-1 males. Surprisingly, developmental delay of rict-1 male animals was suppressed, albeit to a lesser extent than in hermaphrodites. This indicates that at least part of the role of dpy-21 in the TORC2 pathway must be via a novel function outside of its canonical role in X chromosomal dosage compensation. How this is mediated and to what extent regulation of X chromosomal gene expression or autosomal gene expression is involved remains to be demonstrated.

Our work shows that TORC2 mutants have reduced H4K20me1, an epigenetic mark associated with the DCC and with increased gene silencing (Vielle et al., 2012). Chromatin modification through epigenetic marks, such as histone acetylation and methylation, is a conserved, dynamic process, which alters the physical structure of the DNA and can affect other cellular processes such as transcription (Grant, 2001). In the fission yeast S. pombe, TORC2 signaling mutations mimic chromatin structural mutants by increasing expression from regions typically existing as heterochromatic and through underpacking of the chromatin, resulting in increased sensitivity to DNA damage and other stressors (Schonbrun et al., 2009). The epigenetic mark most associated with C. elegans dosage compensation complex is the transcriptional inhibiting monomethylation of lysine 20 of histone H4 of the X chromosome in the hermaphroditic worm (Vielle et al., 2012; Wells et al., 2012). This modification is controlled through H4K20 methyltransferases SET-1 (monomethyltransferase) and SET-4 (di- and tri-methyltransferase). The DCC also mediates H4K16 deacetylation on the X chromosome where SET-1, SET-4 and the Sir-2.1 histone deacetylase play regulatory roles (Wells et al., 2012).

RNAi directed toward set-1 and set-4 histone methyltransferases and sir-2.1 histone deacetylase revealed that RNAi of each suppressed the slow developmental rate of rict-1 mutants. This indicates that rict-1 pleiotropies might in part be due to altered methylation at H4K20. This concept is further supported by western blot analysis revealing that rict-1 animals show reduced levels of H4K20me1 and H4K20me3. However, it is clearly not simple decreases in H4K20me1 and H4K20me3 that are solely responsible for TORC2 mutant phenotypes. First, suppression of most TORC2 phenotypes studied was partial with knockdown of DCC components, along with set-1 and set-4. Second, we found that rict-1 mutants have a lower level of H4K20me1 and H4K20me3, and that this was reversed by knockdown of dpy-21. This suggests that post-embryonically the overactive DCC in TORC2 mutants suppresses SET-1 and SET-4, not activates them. Why then would knockdown of set-1 and set-4, if their activities are already reduced in TORC2 mutants, suppress TORC2 phenotypes? It is possible that TORC2 phenotypes are due predominantly to global reduction of H4K20me1, as the levels of this mark are decreased in rict-1 mutants and increased by dpy-21 knockdown and set-4 knockdown. However, because knockdown of set-1 can also suppress TORC2 mutant phenotypes and this is associated with decreased H4K20me1, it is likely that it is not an overall level of H4K20 methylation that regulates metabolism, but rather the balance of H4K20me1 and H4K20me3 at specific promoters. Future studies of the localization of H4K20me1 and H4K20me3 in TORC2 mutants will be necessary to illuminate the specific mechanisms by which these marks act to mediate metabolic defects in TORC2 mutants.

DPY-21 and the DCC are emerging as powerful, conserved regulators of developmental and metabolic regulatory pathways. dpy-21 also modulates development through negative regulation of the insulin signaling pathway (Dumas et al., 2013). Unlike in TORC2 signaling, the findings of K. Dumas and P. Hu show that dpy-21 acts in its canonical role in dosage compensation as male dauer diapause entry was not affected. These observations, although probably disparate mechanistically, suggest that, either via its canonical role in dosage compensation or in a non-canonical role in TORC2 signaling, the DCC and DPY-21 are conserved regulators of growth, development and metabolism.

Further studies into the genetic mechanisms of chromatin modification in TORC2 mutants will undoubtedly offer exciting new avenues of research with the ultimate aim of understanding the role of TORC2 in the pathogenesis of disease states such as type 2 diabetes and cancer. Cellular control of transcription requires a complex series of post-translational events in order to maintain homeostatic levels of transcripts. Identification of DPY-21 and the DCC as novel regulators and signaling components of the TORC2 pathway helps to shed a new light not only on its genetic function but also the role of TORC2 in epigenetic regulation.

We are grateful to K. Dumas and P. Hu for sharing their results on dpy-21. We thank J. Avruch, E. Pino and M. Runcie for discussions and for reading the manuscript. Thanks to M. Kacergis for technical support in all aspects of the project.