Diacylglycerol lipase regulates lifespan and oxidative stress response by inversely modulating TOR signaling in Drosophila and C. elegans

Target of rapamycin (TOR) signaling is a nutrient-sensing pathway controlling metabolism and lifespan. Although TOR signaling can be activated by a metabolite of diacylglycerol (DAG), phosphatidic acid (PA), the precise genetic mechanism through which DAG metabolism influences lifespan remains unknown. DAG is metabolized to either PA via the action of DAG kinase or 2-arachidonoyl-sn-glycerol by diacylglycerol lipase (DAGL). Here, we report that in Drosophila and Caenorhabditis elegans, overexpression of diacylglycerol lipase (DAGL/inaE/dagl-1) or knockdown of diacylglycerol kinase (DGK/rdgA/dgk-5) extends lifespan and enhances response to oxidative stress. Phosphorylated S6 kinase (p-S6K) levels are reduced following these manipulations, implying the involvement of TOR signaling. Conversely, DAGL/inaE/dagl-1 mutants exhibit shortened lifespan, reduced tolerance to oxidative stress, and elevated levels of p-S6K. Additional results from genetic interaction studies are consistent with the hypothesis that DAG metabolism interacts with TOR and S6K signaling to affect longevity and oxidative stress resistance. These findings highlight conserved metabolic and genetic pathways that regulate aging.

Previously, we have shown that a multi-stress screening strategy can be used to identify genes or mutants involved in the regulation of longevity (Wang et al., 2004Liu et al., 2009). Here, we report the characterization of one such gene, identified in a Drosophila multistress resistant strain DAGL/inaE EP1101 . This EP-element generated line is long-lived and resistant to oxidative stress. DAGL/inaE EP1101 shows upregulation of DAGL/inaE, a homolog of diacylglycerol lipase, and reduced levels of phosphorylated S6 kinase (p-S6K), consistent with the hypothesis that DAGL/inaE up-regulation causes a reduction in TOR signaling. Conversely, a second mutant with reduced DAGL/inaE expression, DAGL/inaE KG08585 , displays shortened lifespan, reduced tolerance to oxidative stress and elevated levels of p-S6K. Genetic manipulation of DAGL/inaE, rdgA, or S6K KQ (a dominant-negative form of S6 kinase) also suggest that reduced TOR signaling mediates the effects of DAGL/inaE overexpression on lifespan and stress resistance. Using Caenorhabditis elegans, we show that, as in flies, the nematode ortholog of DAGL/inaE, F42G9.6 (herein named dagl-1), also regulates lifespan and oxidative stress response via TOR. We propose that DAGL/ inaE and DGK regulate competing branches of pathways that metabolize DAG, ultimately resulting in altered PA levels, which in turn modulate TOR signaling. Collectively, our results show the modulation of longevity and oxidative stress response through conserved pathways that alter TOR signaling in Drosophila and C. elegans.

Results
Diacylglycerol lipase regulates longevity and oxidative stress response in Drosophila In a screen for long-lived mutants with enhanced resistance to simultaneous oxidative stress and starvation, we identified an EPelement insertion mutant DAGL/inaE EP1101 with a 66% increase (P < 0.001) in mean survival time compared to that of the control fly w 1118 (Fig. S1, Supporting information). The outcrossed DAGL/inaE EP1101 line was 72% longer lived than the control (Fig. 1A) and was similarly more resistant to oxidative stress induced by paraquat (Fig. 1B). To identify the target gene in DAGL/inaE EP1101 responsible for lifespan extension and stress resistance, we performed a plasmid rescue and verified that a single EP-element insertion was present in the 5 0 un-translated region of DAGL/inaE. The EP-element insertion in DAGL/ inaE EP1101 disrupts the binding site of a transcriptional repressor Tailless (Gui et al., 2011). Semi-quantitative RT-PCR analysis revealed a threefold increase of DAGL/inaE mRNA levels in DAGL/inaE EP1101 compared with the control (Fig. 1C). DAGL/inaE encodes diacylglycerol lipase (DAGL), which metabolizes DAG to 2-AG (Leung et al., 2008). Since increased DAGL/inaE expression extends lifespan and enhances resistance to oxidative stress ( Fig. 1A-C), we asked whether DAGL/inaE KG08585 , a mutant with reduced DAGL/inaE expression (Fig. 1G), would have the opposite phenotypes. As expected, DAGL/inaE KG08585 exhibited a 50% decrease (P < 0.001) in mean lifespan and a 34% reduction (P < 0.001) in mean survival time on oxidative stress compared to w 1118 (Fig. 1E,F). Together, the results suggest that DAGL/inaE regulates lifespan and oxidative stress resistance in Drosophila.
Overexpression of DAGL/inaE and knockdown of rdgA similarly extend lifespan To determine whether overexpression of DAGL/inaE is sufficient to extend lifespan and increase oxidative stress resistance, we generated transgenic flies expressing either the 2214-nt long isoform DAGL/inaE-PD cDNA (UAS-DAGL/inaE-PD) or the 1935-nt short isoform DAGL/inaE-PA cDNA (UAS-DAGL/inaE-PA). Since DAGL/inaE expresses mainly in adult fly brain, eye, and thoracic-abdominal ganglion according to the data from FlyAtlas (Chintapalli et al., 2007), thus we used GMR-Gal4 (eye and thoracicabdominal ganglion Gal4 driver), Appl-Gal4 (neuronal Gal4 driver), hs-Gal4, and da-Gal4 (ubiquitous Gal4 drivers) to express either UAS-DAGL/ inaE-PD or UAS-DAGL/inaE-PA to determine if overexpression of DAGL/ inaE would also enhance lifespan and oxidative stress response. In all cases, expression of either UAS-DAGL/inaE-PD or UAS-DAGL/inaE-PA by those drivers extended mean lifespan (Table S1, Supporting information) and enhanced oxidative stress resistance (Table S2, Supporting information). These results suggest that neurons are a target tissue for lifespan extension and oxidative stress resistance by DAGL/inaE overexpression. Since overexpression of both isoforms resulted in similar outcomes, we used only UAS-DAGL/inaE-PD in all subsequent experiments and hereafter refer to it as UAS-DAGL/inaE. Diacylglycerol can be converted to 2-AG by DAGL or metabolized to form phosphatidic acid (PA) by DAG kinase (encoded by retinal degeneration A (rdgA) in Drosophila (Hardie, 2003). In mammalian systems PA is reported to activate target of rapamycin (mTOR) kinase resulting in elevated levels of 4EBP and phosphorylated S6K (Fang et al., 2001). Thus, we hypothesize that the enhanced longevity of DAGL/ inaE EP1101 resulted from reduced TOR signaling, since DAGL/inaE overexpression shunts more DAG into 2-AG and it should also lower PA levels (Fig. S2, Supporting information). To examine this possibility, we measured the levels of phosphorylated S6 kinase (p-S6K), a downstream molecular marker of TOR signaling. Levels of p-S6K were reduced by 50% and 40% in young and old DAGL/inaE EP1101 flies, respectively, relative to levels in w 1118 (Fig. 1D). Conversely, in the shortlived DAGL/inaE KG08585 p-S6K levels were elevated by 1.5-and threefold in young and old flies, respectively (Fig. 1H).
If longevity resulting from DAGL/inaE overexpression is due to reduced PA formation and TOR signaling, then the knockdown of rdgA (DAG kinase) should produce similar phenotypes. Overexpression of DAGL/inaE resulted in a 41% increase (P < 0.001) in mean lifespan compared to Gal4 alone and 16% (P < 0.001) compared to UAS alone ( Fig. 2A). Knockdown of rdgA also significantly increases mean lifespan by 44% (P < 0.001) compared to Gal4 alone or 12% (P < 0.01) compared to UAS alone (Fig. 2B). Simultaneous DAGL/inaE overexpression and rdgA knockdown did not further extend lifespan of that achieved by either manipulation independently (Fig. 2C). Similar to DAGL/inaE overexpression, rdgA mutants rdgA BL33306 and rdgA BL20320 also displayed an increase of 53% (P < 0.001) and 48% (P < 0.001) in mean lifespan and 43% reduction for both in p-S6K levels compared to those in control w 1118 (Fig. S3A,B, Supporting information). Together, these results are consistent with the idea that DAGL/inaE and rdgA modulate lifespan via a common pathway.
dagl-1 modulates lifespan and oxidative stress response through reduced TOR signaling in C. elegans To confirm that dagl-1 also modulates TOR signaling in C. elegans, we first inspected the levels of p-S6K in the dagl-1 mutant worms compared to N2. In the mutants, p-S6K levels were 41-46% higher than in control worms (Fig. 3G), a result similar to those seen in Drosophila (Fig. 1H). The transgenic worms overexpressing dagl-1 showed significantly lower levels of p-S6K relative to that of the control (Fig. 3H). Moreover, the elevated levels of p-S6K in both dagl-1 mutant worms were reduced after introducing transgenic dagl-1 (Fig. 3H). Thus as in Drosophila, a clear correlation exists between the effects on TOR signaling and the resulting longevity and stress resistance phenotypes.
Knockdown of dgk-5 rescues the shortened lifespan and reduced oxidative stress tolerance in dagl-1 mutants Since dagl-1 mutation may enhance TOR signaling by converting more DAG to PA by diacylglycerol kinase (DGK) (model in Fig. S2), we predicted that the enhancement of TOR signaling in dagl-1 mutants could be blocked by using RNAi knockdown of the other branch of the pathway, DGK/dgk-5. In both dagl-1 mutants p-S6K levels were significantly reduced by RNAi knockdown of DGK/dgk-5 (Fig. 4A). In addition, DGK/dgk-5 RNAi knockdown fully rescued the shortened lifespan (Fig. 4B, and Table S5, Supporting information) and improved the response to oxidative stress in both the dagl-1 mutant worms (Fig.  S6C, D, and Table S4, Supporting information). These results showed that knockdown of DGK/dgk-5 can rescue the shortened lifespan and the reduced oxidative stress tolerance in both dagl-1 mutants.
RNAi knockdown of Tor/let-363 or raptor/daf-15 reduces the elevated p-S6K levels, rescues the shortened lifespan and improves the oxidative stress response in dagl-1 mutants To further verify that TOR signaling plays a role in dagl-1-mediated lifespan and oxidative stress response in C. elegans, we examined whether RNAi knockdown of the Tor kinase, Tor/let-363, blocks the increase in p-S6K levels in the dagl-1 mutants. As expected, p-S6K levels were dramatically reduced in dagl-1 mutant worms treated with Tor/let-363 RNAi-containing bacteria (Fig. 4C). Raptor binds to TOR to form TOR complex 1 and regulates TOR downstream signaling (Wullschleger et al., 2006). Therefore, we checked whether RNAi knockdown of raptor/daf-15 expression could also diminish the elevated levels of p-S6K in the dagl-1 mutants and found that the enhanced p-S6K levels were also significantly reduced in both dagl-1(tm2908) and dagl-1(tm3026) mutants treated with raptor/daf-15 RNAi (Fig. 4C). Moreover, treatment of either Tor/let-363 or raptor/daf-15 RNAi to dagl-1(tm2908) and dagl-1(tm3026) mutant worms also rescued their shortened lifespan ( Fig. 4D-G, and Table S5, Supporting information).
knockdown of Tor/let-363 or raptor/daf-15 not only lowers the elevated p-S6K levels but also rescues the shortened lifespan and partially improves the oxidative stress response in the dagl-1 mutant worms.
To exclude the possibility that 2-AG itself may also reduce TOR signaling, we exogenously supplemented 2-AG into NIH3T3 and Hep3B cell lines and examined the levels of p-S6K. 2-AG did not cause any reduction in the levels of p-S6K in both NIH3T3 and Hep3B cell lines, while rapamycin treatment dramatically reduced the levels of p-S6K (Fig.  S7, Supporting information).
In summary, our parallel analysis using Drosophila and C. elegans demonstrate that DAGL/inaE/dagl-1 regulates lifespan and modulates oxidative stress response through inversely modulating TOR signaling (Fig. 4H).

Discussion
Genetic studies in model organisms have led to the discovery of many genes that can modulate aging. In addition many of these studies  suggest that the pathways that control aging have been evolutionarily conserved. TOR signaling is one of the conserved nutrient sensor pathways involved in metabolism, growth, and nutrient sensing, and plays an important role in the regulation of aging from yeast to mammals including humans (Kapahi et al., 2010). TOR is proposed to be a lipid sensor that modulates cell growth and proliferation (Foster, 2013). However, we did not observe any changes in the eye and wing sizes (Fig.  S9, Supporting information) neither the body size (data not shown) upon DAGL/inaE overexpression, suggesting that DAGL/inaE overexpression does not affect developmental growth. Accumulated evidence has shown that lipid metabolism is linked to lifespan regulation (Oldham, 2011;Ackerman & Gems, 2012). In this study, we demonstrated that diacylglycerol lipase (DAGL/inaE/dagl-1) regulates lifespan and oxidative stress response through TOR signaling in both Drosophila and C. elegans. Overexpression of DAGL/inaE/dagl-1 may shunt more DAG toward the production of 2-AG, thereby leaving less DAG available to produce PA, and consequently resulting in reduced TOR signaling. Both in flies and worms, DAGL/inaE/dagl-1-mediated lifespan is negatively correlated with levels of p-S6K. Both the shortened lifespan and the elevated levels of p-S6K can be rescued and reverted by the RNAi knockdown of dgk-5, daf-15, or let-363 in the dagl-1 mutants, suggesting that TOR signaling plays a role in DAGL/inaE/dagl-1 mediated lifespan. In addition, we also showed that both RNAi knockdown of DGK/rdgA/dgk-5 and their mutants extend lifespan and exhibit reduced level of pS6K in Drosophila and C. elegans. This is the first demonstration showing that reduced rdgA and dgk-5 expression extend lifespan in Drosophila and C. elegans.
Together, it suggests that genetically altered DAG metabolism may influence PA levels to affect TOR signaling mediated lifespan and stress response. Lipid homeostasis is critical to aging. Several genes involved in lipid metabolism control lifespan (Ackerman & Gems, 2012;McCormick et al., 2012). DAG is a lipid metabolic intermediate as a second messenger involved in complex signaling (Carrasco & Merida, 2007). DAG can activate protein kinase D (PKD). It has been suggested that DGK functions upstream of PKD in the regulation of oxidative-induced intestinal cell injury (Song et al., 2008). Thus, genetic manipulation of DGK and PKD should produce similar phenotypes. Indeed, it was reported that PKD/DFK-2 deficiency increases adult lifespan by 40% in C. elegans (Feng et al., 2007), implying that a lower level of DAG may extend lifespan in C. elegans. This is in agreement with our idea that lower DAG levels results in less PA formation, reduced TOR signaling, and thus to an extension of lifespanan effect mimicked by knockdown of DGK/rdgA/dgk-5 both in Drosophila and C. elegans. It was reported that Drosophila microRNA mir-14 inhibits reaper-dependent cell death and is required for lipid metabolism (Xu et al., 2003). Depletion of mir-14 results in reduced lifespan and lowered stress tolerance and is accompanied with increased levels of triacylglycerol and diacylglycerol and the above phenotypes are reverted upon increasing mir-14 copy number in Drosophila. This suggests that lifespan negatively correlates with DAG level. DAG activation of protein kinase C (PKC) is linked to hepatic insulin resistance, a risk for type 2 diabetes (Jornayvaz & Shulman, 2012). In addition, PKC activity is associated with prefrontal cortical decline in aging and pharmacological inhibition of PKC rescues working memory malfunction in aged rat and increased working memory in aged rhesus monkeys (Brennan et al., 2009), indicating accumulated DAG is deleterious to lifespan and health. DAG is a second messenger triggering activation of PKC to enhance calcium influx for the activation of mTORC1. Overexpression of DAGL/inaE in neurons may result in less DAG levels for lowered PKC activity leading to reduced calcium influx and hence diminished mTORC1 activity to account for the extended lifespan and oxidative stress resistance. Thus, altered lipid metabolism achieved by lowering DAG levels is beneficial to lifespan and stress response.
Phophatidic acid is implicated in the activation of mammalian target of rapamycin (mTOR) and the control of cell growth and differentiation (Fang et al., 2001;Merida et al., 2008). Overexpression of DAGL/inaE/ dagl-1 may result in lower level of PA for reduced TOR signaling in extending lifespan. It was reported that the expression of a specific isoform DGKf, which modulates PA levels, regulates the levels of seruminduced phosphorylation of S6K for mTOR signaling in HEK293 cells (Avila-Flores et al., 2005). Interestingly, the closest homologs of DGKf in Drosophila and C. elegans are rdgA and dgk-5, respectively. Our data showed that not only the mutants of rdgA and dgk-5 but also both knockdown of rdgA in fly and knockdown of dgk-5 in worm extend lifespan and reduce the levels of p-S6K. This provides the first in vivo evidence that reducing DGK extends lifespan via its effect on TOR signaling in both Drosophila and C. elegans. As we hypothesized that DAGL overexpression may result in more 2-AG formation, and 2-AG can be further metabolized to become arachidonic acid, also known as omega-6 polyunsaturated fatty acids, and glycerol. Omega-6 polyunsaturated fatty acids recently have been reported to extend C. elegans lifespan via activation of autophagy (O'Rourke et al., 2013). Therefore, it is also possible that DAGL/inaE/dagl-1 overexpression may result in more 2-AG for increased levels of omega-6 polyunsaturated fatty acids to activate autophagy for lifespan extension.
Insulin signaling is a well-studied and conserved pathway that also regulates lifespan (Kenyon, 2010). The interplay between insulin and TOR signaling pathways is well characterized (Hay, 2011). Interestingly, we found that the longevity and oxidative stress resistance of daf-2 can be partially inhibited by knockdown of dagl-1, and increased dagl-1 expression was detected in a daf-2 mutant. Two putative daf-16 binding sites were identified in the regulatory region of dagl-1 (Liu et al., 2012). In addition, we also found increased levels of phosphorylated Akt (p-Akt) in the two dagl-1 mutants compared to N2 (Lin and Wang, unpublished data), suggesting that dagl-1 plays a role in the lifespan and oxidative stress response of the daf-2 mutant and insulin signaling may also modulate dagl-1 expression in C. elegans. However, we did not detect any changes in the levels of p-Akt in DAGL/inaE EP1101 and DAGL/ inaE KG08585 compared to w 1118 (data not shown). It suggests that there is a discrepancy between Drosophila and C. elegans in insulin signaling for the DAGL/inaE/dagl-1-mediated lifespan regulation.
In summary, our study shows that DAGL/inaE/dagl-1 regulates lifespan and oxidative stress response via negatively modulating TOR signaling in both Drosophila and C. elegans. Since TOR signaling is a conserved pathway among different species regulating nutrient sensing, cell growth, and aging, our discovery may be relevant in mammals. Our results provide new insights on how the altered genetic regulation of DAG metabolism affects lifespan and stress response and may help in developing therapies to DAG imbalance-related diseases.

Lifespan and oxidative stress assays in Drosophila and C. elegans
For Drosophila, the lifespan assay and paraquat-induced oxidative stress assay for the progeny from specific crosses were carried out as described previously (Liao et al., 2008;Liu et al., 2009;Wang et al., 2012). We found female flies in DAGL/inaE EP1101 and DAGL/inaE KG08585 showed similar results to males in the lifespan and stress assays and thus only results from male flies were used in this paper. Most experiments were carried out at 25°C unless otherwise stated. For C. elegans, lifespan assays were performed at 20°C as described previously (Liu et al., 2009) but without adding 5 0 flourodeoxyuridine (FUdR). N2, dagl-1(tm2908), dagl-1(tm3026), N2; Ex[Pdpy-30::GFP], N2; Ex[Pdpy-30::dagl-1::GFP](3) and N2; Ex[Pdpy-30::dagl-1::GFP](4) worms were grown on NGM plates seeded with E. coli OP50 bacteria. For RNAi treatment, worms were placed on NGM plates with E. coli HT115 containing the control L4440 plasmid or L4440 expressing dsRNA targeting the specific gene. All the worms were initially transferred daily for the first seven days and later every 2 or 3 days. Dead worms not responding to gentle prodding were scored until all were dead. The oxidative stress assay for worms was conducted at 20°C. Young adult hermaphrodites were immersed in S-media containing either 10 or 40 mM of paraquat (1,1-dimethyl-4,4bipyridinium dichloride, Sigma-Aldrich, St. Louis, MO, USA). The number of dead worms was scored every hour until all worms were dead. All experiments were repeated at least three times. Gene expression changes were monitored by RT-PCR and real-time PCR. Statistical differences in survival were calculated by the log-rank test. Differences in oxidative stress resistance were determined by Student's t-test.

Western blot
Fly heads of specific age for each strain were collected and homogenized in lysis buffer containing protease inhibitor (Cat#: 04693159001, Roche, Indianapolis, IN, USA) and phosphatase inhibitor (Cat#: 04906837001, Roche). Synchronized four-day-old adult worms of each strain, with or without RNAi treatment, were collected in 15ml centrifuge tubes, washed three times by M9 buffer, transferred to new microfuge tubes, and homogenized by lysis buffer containing protease inhibitor and phosphatase inhibitor. In the cell lines, NIH3T3 and Hep3B cells were treated with DMSO as a mock, 10 or 20 lM of 2-AG (Cat#: 1298, TOCRIS, Bristol, UK), or 10 nM Rapamycin as a positive control (Cat#: 553210, Millipore, Billerica, MA, USA). After 24-h incubation, the treated cells were collected in 15-mL centrifuge tubes and the cell pellets were lysed in lysis buffer containing protease inhibitor and phosphatase inhibitor as mentioned above. After homogenization, 2% SDS was added to each sample again and then the sample was vortexed and incubated 5 min at 70°C. The sample was centrifuged at 13 000 rpm for 10 min at room temperature and the supernatant was transferred into new tube to measure protein concentration. Equal amounts of protein for each sample were loaded and separated in a 12% SDS-PAGE gel, and transferred to a nitrocellulose membrane. The membrane was blocked with 5% BSA in TBST for 1 h, and later incubated with anti-pS6K (Cell Signaling, Billerica, MA, USA, #9209, 1:500 dilution in 5% BSA /1XTBST or Abbomax Inc., #601-030, 1:1000 dilution in 5% BSA /1XTBST), anti-pERK (Epitomics, #1481-1, 1:1000 in 5% BSA/1XTBST), anti-aÀactin or b-actin or tubulin (aÀactin, Santa Cruz, Dallas, Texas, USA, #SC-1616; b-actin, Spring, #E4554, 1:10 000; tubulin, Epitomics, #1871-1, 1:1000 dilution, in 5% BSA/1XTBST), or anti-GAPDH (Epitomics, #S0011, 1:2000 in 5% BSA/1XTBST) at 4°C overnight. The membrane then was washed three times with TBST, and incubated with the secondary antibody (goat anti-rabbit IgG, 1:10 000 in 5% BSA/ 1XTBST) for 2 h at 4°C, again washed three times with TBST, incubated with ECL reagent (Cat#: RPN 2132, Amersham, GE Healthcare, Fairfield, CT, USA) and exposed to the X-ray film (Kodax, Rochester, NY, USA). The protein image was quantified by ImageJ â to calculate the fold of changes by normalizing each measurement to its control.

Semi-quantitative RT-PCR and quantitative real-time PCR assays
Drosophila total RNA extraction and reverse-transcription following by semi-quantitative polymerase chain reaction (RT-PCR) were described in Wang et al. (2004). For C. elegans, worms with or without RNAi treatment were collected into 1.5-mL microfuge tube, washed three times by M9 buffer, and lysed by using 1 ml TRIzol â reagent (Life Technologies, Grand Island, NY, USA) to extract RNA. Subsequent procedures were similar to those used for Drosophila. Each gene was amplified by gene specific primers (sequences available upon request). The genes rp49 and actin were used as internal controls in the PCR reactions for Drosophila and C. elegans, respectively. The fold changes for gene expression were calculated, normalized to the internal control, by quantification of the image of the DNA in agarose gel by ImageJ â software. Alternatively, the cDNAs were used as templates in quantitative real-time PCR utilizing SYBR Green PCR Master Mix in the Applied Biosystems 7900HT Fast Real-Time PCR System (7900HT Fast System, Life Technologies). Each gene was amplified with the specific real-time PCR primer set, and was normalized to the control (rp49 for Drosophila and actin for C. elegans). The relative transcriptional levels of the genes were presented as fold of 2 ÀDDCt ¼ 2 ÀðDCt tested gene ÀDCt controlÞ . C t is the threshold cycle value clarified as the fractional cycle number at the time of target fluorescent signal passed a threshold above baseline.