SIRT6 Promotes Hepatic Beta-Oxidation via Activation of PPARa
SUMMARY
The pro-longevity enzyme SIRT6 regulates various metabolic pathways. Gene expression analyses in SIRT6 heterozygotic mice identify significant decreases in PPARa signaling, known to regulate mul- tiple metabolic pathways. SIRT6 binds PPARa and its response element within promoter regions and acti- vates gene transcription. Sirt6+/— results in signifi- cantly reduced PPARa-induced b-oxidation and its metabolites and reduced alanine and lactate levels, while inducing pyruvate oxidation. Reciprocally, starved SIRT6 transgenic mice show increased pyruvate, acetylcarnitine, and glycerol levels and significantly induce b-oxidation genes in a PPARa- dependent manner. Furthermore, SIRT6 mediates PPARa inhibition of SREBP-dependent cholesterol and triglyceride synthesis. Mechanistically, SIRT6 binds PPARa coactivator NCOA2 and decreases liver NCOA2 K780 acetylation, which stimulates its activa- tion of PPARa in a SIRT6-dependent manner. These coordinated SIRT6 activities lead to regulation of whole-body respiratory exchange ratio and liver fat content, revealing the interactions whereby SIRT6 syn- chronizes various metabolic pathways, and suggest a mechanism by which SIRT6 maintains healthy liver.
INTRODUCTION
Metabolic diseases such as obesity, type 2 diabetes, and fatty liver are becoming the epidemic of the twenty-first century. Their influence on society, in terms of public health and their economic cost, is enormous. The worldwide increase in lifespan is associated with a corresponding rise in obesity-related pathologies. Thus, increasing our understanding of key regulators that play a role in healthy lifespan is crucial. Sirtuins are a family of NAD+-dependent deacylases homologous to yeast SIR2 deace- tylase (Imai et al., 2000), which were shown to regulate lifespan and age-related metabolic diseases (Tissenbaum and Guarente, 2001; Whitaker et al., 2013).Of the seven mammalian sirtuins, SIRT6 was shown to regu- late longevity (Kanfi et al., 2012) and various physiological pathways (Zhong et al., 2010). These include embryonic devel- opment, DNA repair, transposon stability, metabolism of carbo- hydrates, cholesterol and fat, inflammation, circadian rhythms, cancer, and aging (Kuang et al., 2018). SIRT6 deficiency in pri- mates or its inactivating mutation in human causes peri- and post-embryonic lethality, respectively (Ferrer et al., 2018; Zhang et al., 2018), while transgenic (TG) mice overexpressing SIRT6 have an increased male lifespan (Kanfi et al., 2012). Furthermore, in old age these mice are less susceptible to a spectrum of aging-related dysfunctions (Roichman et al., 2017). Similarly, calorie restriction (CR), known to extend healthy lifespan in both murine and primate models, increases SIRT6 levels (Kanfi et al., 2008a). This suggests that SIRT6 might mediate the bene- ficial effects of CR, resulting in increased lifespan. Indeed, male SIRT6 TG mice have a liver transcription profile highly similar to male mice under CR (Kanfi et al., 2012). Thus, SIRT6 is a key regulator of healthy aging, via its effects on metabolism.
SIRT6 is involved in many facets of metabolism, regulating key metabolic pathways such as insulin-like growth factor (IGF-1) (Sundaresan et al., 2012) and AMP-activated protein kinase (AMPK) (Cui et al., 2017). Glucose metabolism is also controlled by SIRT6 via inhibition of HIF1a-dependent transcription including glucose transporter genes, subsequently reducing glycolysis (Zhong et al., 2010). Interestingly, SIRT6 was also found to negatively regulate gluconeogenesis in the liver by regulating PGC1a and FOXO1 activities (Dominy et al., 2012; Zhang et al., 2014). SIRT6 is also an important regulator of lipidhomeostasis. SIRT6 inhibits cholesterol and triglyceride biosyn- thesis by inhibiting SREBP1/2, and controls hepatic fat meta- bolism by repressing miR122 (Elhanati et al., 2013, 2016). SIRT6 TG mice are protected against the physiological damages of a high-fat diet (HFD) (Kanfi et al., 2010), whereas fat-specific SIRT6 knockout (KO) increases obesity-related phenotypes (Kuang et al., 2017; Xiong et al., 2017). Liver-specific KO of SIRT6 results in increased hepatic steatosis and decreased b-oxidation (Kim et al., 2010). In contrast, SIRT6 TG mice fed an HFD have decreased micro-vesicular lipidosis and reduced body fat content compared to their wild-type (WT) littermates (Kanfi et al., 2010). Despite such extensive studies of SIRT6, our knowledge of its involvement in specific metabolic path- ways, particularly b-oxidation, is still limited. Moreover, how SIRT6 coordinates between these various metabolic pathways is unknown.PPARa, one of the three peroxisome proliferator-activated receptor (PPAR) isoforms, is a key transcription factor in hepatic b-oxidation. PPARa binds to a diverse group of compounds, such as WY 14,643 (WY) (Kersten, 2014), and to DNA as a mostly obligatory heterodimer with retinoic acid receptor RXRa, recog- nizing the PPAR response elements (PPREs).
PPARa can be found in many large complexes and its ligand binding induces conformational changes, allowing coactivators to directly bind to PPARa, mainly via the LXXLL motif (Bugge and Mandrup, 2010). The known PPAR coactivator SRC2/NCOA2 binds PPARs and its main function is the recruitment of histone acetyltrans- ferases (HATs), such as P300 and CBP (Powell et al., 2007). These HATs acetylate the chromatin surrounding the PPRE and activate PPARa-dependent transcription.PPARa is expressed mainly in tissues with a high capacity for fatty acid oxidation, primarily the liver and heart (Moreno et al., 2010). Under conditions that demand fatty acids oxidation such as fasting, PPARa ensures energy availability by upregulat- ing the expression of enzymes necessary for fat oxidation, such as CPT1a (Leone et al., 1999; Mandard et al., 2004). CR was shown to increase hepatic PPARa-mediated fatty acid oxidation and whole-body fat oxidation rates (Takemori et al., 2011). Once activated, PPARa promotes b-oxidation of fatty acids, releasing free fatty acids, thereby increasing acetyl-coA production from fats. Aside from its direct role in b-oxidation, PPARa also indirectly drives the liver toward lipid oxidation by inhibiting glycolysis-derived pyruvate oxidation via PDK4 activation, and by increasing gluconeogenic precursors such as lactate and alanine (Baes and Peeters, 2010). Additionally, PPARa was shown to inhibit SREBP-mediated cholesterol and triglyceride synthesis (Ko¨ nig et al., 2007, 2009). Thus, PPARa also regulatesb-oxidation indirectly via several modes of regulation, including activation of gluconeogenesis and inhibition of pyruvate oxida- tion and cholesterol/triglyceride synthesis.PPARa shares several properties with SIRT6. Both were impli- cated in hepatic b-oxidation, inflammation, and circadian clock regulation (Chen and Yang, 2014; Vachharajani et al., 2016), and their activities increase under starvation and CR. SIRT6 and PPARa KO mice have shortened lifespans, and their levels decrease in aged mice (Howroyd et al., 2004; Kaluski et al., 2017; Mostoslavsky et al., 2006). Thus, these findings suggest that a coregulatory interaction between SIRT6 and PPARa may exist. Here, we underscore the interplay between SIRT6 and PPARa and demonstrate how SIRT6 coordinates different meta- bolic pathways.
RESULTS
To identify the physiological pathways regulated by SIRT6, we performed quantitative transcriptome analysis based on RNA sequencing (RNA-seq) from livers of WT and SIRT6 heterozygotic (HZ) mice (Table S1A). The top highly differentially expressed (DE) genes (Figure 1A) were validated by quantitative real-time PCR (Figure 1B; Table S1B). Interestingly, the top three pathways differ- entially regulated in SIRT6 HZ mice are all directly related to b-oxidation. Ingenuity pathway analysis (IPA) (Kra¨ mer et al., 2014) identified Acyl-CoA hydrolysis, an intermediate step in b-oxidation, as the most significantly regulated pathway. The sec- ond-ranked pathway was triacylglycerol degradation, which hydrolyzes triacylglycerol into fatty acids, used in b-oxidation. The third pathway, stearate biosynthesis, involves saturated fatty acid synthesis from palmitate and carbohydrates. Other pathways include Ehprin and VEGF signaling, involved in differentiation and angiogenesis, respectively (Figure 1C). Consistent with these find- ings, PPARa, a key regulator of liver fatty acid b-oxidation was identified as the most significantly inhibited upstream regulator of the DE genes. Likewise, PPARg was identified as a significantly inhibited regulator. Other regulators include TNFa, a known SIRT6 target of inflammation, and ACOX1 and EHHADH, which, similar to PPARa, regulate b-oxidation (Figure 1D). These findings strongly suggest that SIRT6 activates PPARa. Indeed, many genes activated by PPARa were inhibited in HZ mice, whereas genes inhibited by PPARa were activated in HZ mice (Figure 1E). Likewise, lipid metabolism, the primary pathway of PPARa, was a top network found in the DE genes. Both SIRT6 and PPARa are centrally placed in the lipid
metabolism network, indicating possible coordination between both proteins in regulating fat metabolism (Figure 1F). Other known SIRT6-regulated networks such as inflammation and car- bohydrate metabolism were identified as well (Figure S1). Furthermore, comparison of HZ liver DE genes to all microarrays currently in the GEO database using the ExpressionBlast tool (Zinman et al., 2013) identified the top three highly significant matches to be PPARa KO microarrays (Figure 1G). Moreover, 7 of the top 20 matches were from PPARa KO experiments (Table S2). This shows that SIRT6 HZ mice display a gene expression profile most similar to PPARa KO mice, compared to all other existing microarray expression profiles. Aside from PPARa, SIRT6 HZ expression data were similar to known SIRT6-regulated pathways such as infection and cancer (Lerrer et al., 2016; Table S2). Altogether, these findings identified PPARa as a candidate SIRT6-regulated pathway.
To investigate the mechanism underlying SIRT6 activation of PPARa signaling, PPARa levels and acetylation were examined in WT and HZ mice. PPARa mRNA and protein levels were similar in WT and HZ mice (Figures S2A and S2B). In addition, in line with previous studies (Oka et al., 2011), no acetylation was detected on PPARa (Figure S2C). Next, we examined whether SIRT6 and PPARa might physically interact. Recombinant PPARa-GST tagged protein co-immunoprecipitated with recombinant SIRT6-FLAG-tagged protein, but not with FLAG-tagged BIP used as a negative control. Additionally, GST protein alone did not co-immunoprecipitate with SIRT6 (Figure 2A). These findings show that SIRT6 specifically binds PPARa in vitro. The binding affinity of SIRT6 to PPARa in comparison to other known SIRT6 interactors was then measured. A microfluidics protein binding platform (Ben-Ari et al., 2013) was used to quantify SIRT6 binding affinity to selected proteins. Binding was measured as fluorescent signals of interactors normalized to SIRT6 levels. Interestingly, in comparison to two known SIRT6 interacting proteins, HIF1a and CTIP, the binding affinity of SIRT6 to PPARa was significantly higher than to HIF1a, suggest- ing strong interaction between PPARa and SIRT6 (Figure 2B). A representative binding signal is shown (Figure 2B, right panel). We then examined SIRT6 binding to PPARa in HEK293T cells. As seen in Figure 2C, PPARa-GFP specifically co-immunopre- cipitated with SIRT6-FLAG but not with GFP negative control. Reciprocally, endogenous SIRT6 specifically co-immunoprecip- itated with PPARa-FLAG (Figure 2D). In all the binding experi- ments using intact cells, the binding was not DNA dependent, as ethidium bromide did not abolish the interaction (Figures 2C and 2D). Endogenous SIRT6 and GFP-tagged PPARa or endog- enous PPARa and FLAG-SIRT6 interactions were shown in mouse Hepa1-6 hepatocytes as well (Figure S2D). These results show that SIRT6 and PPARa directly interact and are present in common protein complexes in cells.
We demonstrated that SIRT6 and PPARa associate with one another. Thus, the direct binding of SIRT6 to the PPARa-binding DNA element-PPRE and whether PPARa is required for this interaction were examined by using the quantitative protein interacting DNA (QPID) microfluidics method (Glick et al., 2016b). Recombinant SIRT6-FLAG was immobilized to the chip surface and incubated with Cy5-labeled PPREx3 or mutant DNA sequences (Oka et al., 2012) in the presence or absence of recombinant free Myc-tagged PPARa. The interaction ratio was calculated and is shown in Figure 2E. Importantly, significant and strong SIRT6 binding to PPRE was found specifically in the pres- ence of PPARa. In contrast, SIRT6 did not bind to a mutant motif regardless of the presence of PPARa. A representative binding experiment is shown (Figure 2E). SIRT6’s interaction with the PPRE was also examined in the presence or absence of PPARa using serial DNA dilutions (Glick et al., 2016b). In the presence of PPARa, SIRT6 displayed a logarithmic binding curve that reached saturation, indicating marked affinity to the PPRE but not to mutant PPRE sequence, whereas SIRT6 alone displays a non-specific linear binding curve similar to the PPRE mutant (Glick et al., 2016b; Figures S3A and S3B). These findings show that SIRT6 binds specifically to the PPRE in a PPARa- dependent manner. Next, we examined whether SIRT6 can activate the PPRE in vivo using a luciferase reporter assay. A construct containing the luciferase gene fused to three tandem repeats of the PPRE (Kim et al., 1998) was transfected into mouse Aml-12 hepato- cyte cells along with SIRT6 or control plasmids. SIRT6 overex- pression significantly induced the luciferase signal (Figure S3C). Importantly, SIRT6 does not activate negative control promoter sequences (Figure S3D). Thus, SIRT6 stimulates endogenous PPARa-dependent promoter activity in liver cells. To examine whether SIRT6 catalytic activity is required for PPARa transac- tivation, HEK293T cells were transfected with either SIRT6 or a catalytically inactive mutant, SIRT6 H133Y. Notably, SIRT6 but not the SIRT6 catalytic mutant activated PPRE transcriptional activity (Figure 2F). These findings suggest that SIRT6 enzy- matic activity is required to activate the PPRE. Moreover, induc- tion of the PPRE by PPARa overexpression was further increased in SIRT6 overexpressing cells (Figure 2G). Thus, the two proteins may work cooperatively to activate the PPRE.
Subsequently, SIRT6 binding to the PPRE within promoters of PPARa target genes in vivo was measured using chromatin immunoprecipitation (ChIP) assay in primary hepatocytes. As shown in Figure 2H, in comparison to immunoglobulin G (IgG) control, endogenous SIRT6 significantly binds to the PPREs of several PPARa target genes. Strikingly, SIRT6 binds to the PPREs of Cpt1a, the rate-limiting enzyme of b-oxidation, as well as to the positive control Srebp2 promoter (Elhanati et al., 2013). This binding was specific, as SIRT6 does not bind to a negative control DNA sequence in the GAPDH gene promoter (Figure 2H). Srebp2 promoter was shown to be inhibited and de- acetylated on H3K9 by SIRT6 (Tao et al., 2013). Indeed, in com- parison to WT livers, Sirt6+/— livers showed significant increased Srebp2 promoter H3K9 acetylation levels. In contrast, specif- ically on the SIRT6-bound PPREs, Sirt6+/— livers showed signif- icantly decreased H3K9 acetylation levels. (Figure 2I). Notably, SIRT6 binds not only to PPREs within promoters but also to the PPRE localized 4 Kb distal of Angptl4 (Figures 2H and S3E). These findings further indicate that SIRT6 binding is PPREs specific and not due to its proximity to other transcription ele- ments near the promoter region. Moreover, these findings sug- gest that SIRT6 deacetylase activity promotes the activation of PPREs potentially via deacetylation of a PPARa cofactor and not via deacetylation of PPARa or the PPRE. SIRT6 was shown to bind to PPARa and PPREs under normal growth conditions (Figure 2). Next, we examined whether SIRT6 binding to PPRE depends on PPARa activity. Primary hepato- cytes were treated with the specific PPARa agonist, WY to induce PPARa activity. Interestingly, treatment with WY did not further increase SIRT6 binding to PPREs in comparison to untreated controls (Figure S3F). These findings imply that the in vivo association between SIRT6 and the PPRE is constant, irrespective of PPARa activation. Taken together, these results conclusively show that SIRT6 binds to and activates the PPRE in vivo in a PPARa-dependent manner.
As SIRT6 was found to bind to the PPRE (Figure 2), the effect of SIRT6 on transcription of various hepatic PPARa signaling path- ways was investigated. Gene expression of PPARa-regulated pathways was examined in WT or SIRT6 HZ primary hepatocytes treated with WY. As previously demonstrated, upon WY treat- ment a strong induction of PPARa transcription was observed both in mouse liver and primary hepatocytes relative to controls (Szalowska et al., 2014; Figure 3). These pathways include mitochondrial b-oxidation, peroxisomal b-oxidation, acyl-CoA binding/hydrolysis, lipid storage/transport, and ketogenesis. WY does not directly bind to SIRT6, nor does it affect SIRT6 deacylase activity or RNA levels (Figures S3A–S3C). Importantly, SIRT6 HZ primary hepatocytes consistently displayed reduced PPARa activation (Figure 3A). Conversely, in comparison to WT, SIRT6 TG mice had significantly increased induction of PPARa targets (Figure 3B) in primary hepatocytes. These find- ings show that SIRT6 uniformly promotes PPARa transcriptional activity in multiple pathways in primary liver cells.Next, in order to examine whether SIRT6 can activate PPARa in vivo, SIRT6 HZ and TG mice along with their appropriate WT controls were treated with WY. Consistent with the response of primary hepatocytes, SIRT6 HZ mice displayed significantly reduced PPARa activation (Figure 3C). Additionally, SIRT6 TG mice had significantly increased PPARa activation (Figure 3D). SIRT6 was previously shown to regulate b-oxidation via AMPK (Cui et al., 2017; Elhanati et al., 2016). Yet, WY is a specific PPARa activator, and no changes in phosphorylated AMPK were found following this treatment (Figure S4D). These results show that SIRT6 activates PPARa signaling in livers in vivo. Interestingly, as previously shown in mice but not in primary he- patocytes, the lipid transport gene Slc27a1 is induced by WY treatment (Rakhshandehroo et al., 2010), and SIRT6 further acti- vated this gene in mice (Figures 3C and 3D). In addition, SIRT6 also increases the expression of Fgf21 longevity hepatokine, a critical factor for PPARa activity (Figure 3; Goto et al., 2017). This indicates that in vivo, SIRT6 may have a broader effect in regulating PPARa signaling. Altogether, these results show that SIRT6 specifically activates PPARa-dependent transcription of several metabolic pathways.
SIRT6 activates the PPARa-dependent transcription of genes involved in several metabolic pathways (Figure 3). Thus, we examined the downstream effects of SIRT6 on a range of PPAR- a-regulated pathways. Specifically, we investigated the ability of SIRT6 to activate PPARa pathways b-oxidation, gluconeogen- esis, and glycerol transport, and to inhibit glycogenolysis and py- ruvate metabolism, in WY treated mice. When activated, in order to increase b-oxidation, PPARa inhibits pyruvate oxidation by upregulating PDK4 and increases fatty acid and glycerol trans- port. PPARa also increases precursors of gluconeogenesis such as lactate and alanine (Baes and Peeters, 2010; Figure 4A). First, regulation of pyruvate oxidation by SIRT6 was examined. PKD4 kinase and its substrate PDC (P-PDC) are strongly induced in mouse livers following WY treatment (Figure 4B, left panel). Both PDK4 and P-PDC protein levels were significantly lower in HZ mice, while total PDC levels remained unchanged. This indicates that SIRT6 deficiency reduces PPARa-dependent inhibition of pyruvate oxidation (Figure 4B, middle and right panels). We next measured the regulation of fatty acid transport and b-oxidation. RNA levels of fatty acid transporter Cd36 were strongly induced following WY treatment and were significantly less activated in HZ mice (Figure 4C, left panel). Moreover, b-oxidation products were measured from labeled palmitate ex vivo in liver mitochondria. Acetylcarnitine metabolite levels, the product of long-chain fatty acid oxidation, were significantly reduced in SIRT6 HZ mice after WY treatment compared to con- trols (Figure 4C, middle panel). Likewise, significantly lower levels of the final product, palmitate-derived CO2, generated by the Krebs cycle from labeled palmitate, were found in isolated hepatic mitochondria from WY-treated HZ livers (Figure 4C, right panel). Thus, SIRT6 HZ mice have decreased b-oxidation compared to control mice upon WY treatment. These results support a role for SIRT6 in promoting PPARa-mediated
b-oxidation.
PPARa is also known to increase gluconeogenic precursors glycerol, lactate, and alanine. Thus, we first measured RNA levels of the glycerol transporter Aqp3 in the liver. As can be seen in Figure 4D, in comparison to WT littermates, upon WY treatment HZ mice showed significantly reduced induction of RNA levels of Aqp3. Likewise, the metabolites lactate and alanine were also significantly less induced in HZ mice in response to WY treatment (Figure 4E). We subsequently measured RNA levels of key enzymes of glycogenolysis, Gys2 and Pygl. Similar to previous studies, WY treatment inhibited the expression of the glycogenolysis gene Pygl and activated the glycogen synthesis gene Gys2. This effect was nearly completely abolished in HZ mice (Figure 4F). These data demon- strate that SIRT6 activates a broad spectrum of PPARa signaling pathways to regulate metabolism, ultimately increasing b-oxidation.Next, we aimed to examine whether the effects of SIRT6 on gene expression and metabolites are also found under normal physiological conditions in which PPARa is active. PPARa is known to be important for activating hepatic b-oxidation and in- hibiting pyruvate oxidation in response to fasting. Thus, we examined whether SIRT6 TG mice have any defects in these fasting-regulated pathways. Indeed, as can be seen in Figure 4G, under fasting conditions, SIRT6 overexpression results in a sig- nificant increase in pyruvate levels, a sign of inhibited pyruvate oxidation, and an increase in b-oxidation metabolite acetyl- carnitine (Figure 4G). In addition, a significant increase in glycerol levels was found in TG mice under starvation. These changes in metabolites were correlated with a significant increase in the transcription of the pyruvate oxidation inhibitor Pdk4 and the b-oxidation activators Cpt1a, Acot3, and Ehhadh genes and Gos2, a lipid storage gene, were found (Figure 4H). Altogether, these data show that SIRT6 activates PPARa, leading to decreased pyruvate oxidation, increased b-oxidation, and increased uptake of gluconeogenic precursors, as found under a normal fasting response.
To show a PPARa dependency in the effect of SIRT6 on PPARa signaling, the following experiment was performed: PPARa was knocked down (KD) using two different small interfering RNAs (siRNAs) in both primary and Hepa1-6 hepatocytes overexpress- ing SIRT6. PPARa KD was validated in both RNA and protein levels (Figures 5A and S5A–S5C). First, we examined to what extent SIRT6 can stimulate PPRE luciferase reporter when PPARa is depleted. As seen in Figure 5B, in comparison to con- trol scrambled siRNA, SIRT6 activation of PPRE was abolished in PPARa KD Hepa1-6 cells. The importance of PPARa in medi- ating SIRT6’s effect on b-oxidation genes was investigated as well. Indeed, SIRT6 activation of b-oxidation genes was signifi- cantly reduced in PPARa KD primary hepatocytes (Figure 5C). Only two genes were found to be not wholly PPARa dependent (Figures S5D and S5E). Lastly, in comparison to control cells, the effect of SIRT6 on pyruvate signaling was also significantly inhibited in PPARa KD primary hepatocytes (Figures 5D and S5F). Altogether, these data conclusively show that PPARa me- diates the effect of SIRT6 on multiple PPARa signaling pathways.Finally, we further explored the broader effects of activation of PPARa by SIRT6. RNA-seq was performed from livers of WT and SIRT6 HZ mice treated with WY (Figure 6A; Table S1, tabs C and D). In line with the abovementioned results, PPARa was found to be one of the most strongly inhibited regulators in HZ mice (Figure 6B). Previous studies showed that induction of PPARa strongly represses SREBP1/2-mediated transcription (Zhang et al., 2015). Indeed, SREBP1/2 was among the most highly activated regulators found in HZ mice under WY treatment (Figure 6B). Importantly, this effect was not observed under normal conditions, in the absence of WY (Figure 1), suggesting that the effect of SIRT6 haploinsufficiency on the SREBP pathway is PPARa dependent. Moreover, under normal condi- tions, in contrast to SIRT6 KO (Tao et al., 2013), SIRT6 HZ mice have similar SREBP levels as WT mice (Figure S4D). Further pathway analysis showed that the top-five pathways differen- tially expressed in SIRT6 HZ mice were activation of SREBP- regulated pathways, cholesterol and sterol biosynthesis, and activation of inflammation-related pathways, B cell develop- ment, and antigen presentation (Figure 6C). Network analysis showed that the top significant network in the WY-treated hepat- ic transcriptome placed PPARa, SREBP, and TNFa/NFKb as central nodes (Figure S6). This indicates a broad and integrative role for SIRT6 in inflammation, cholesterol synthesis, and b-oxidation in a PPARa-dependent manner.
Next, we examined the overlap between PPARa-dependent genes and known SREBP targets in HZ DE genes. Notably, SIRT6 DE genes activated by SREBP1/2 are inhibited by PPARa, showing an opposite regulation (Figure 6D). These genes are involved in triglyceride and cholesterol synthesis. Importantly, as seen in Figure 6F, in the absence of PPARa activation, SIRT6 did not affect these genes in HZ mice. Yet, upon WY treatment, SIRT6 haploinsufficiency significantly blunted PPARa inhibition of SREBP-dependent genes (Figure 6E). Thus, these findings show that SIRT6 regulates PPARa inhibition of SREBP-dependent cholesterol synthesis.We then examined the physiological effect of the SIRT6 regu- lation of PPARa-mediated inhibition of SREBP. Hepatic lipid content is composed of triglycerides and cholesterol, regulated by SREBP-1 and -2, respectively (Xu et al., 2013). Therefore, we treated mice with WY and analyzed whole-liver fat content using NMR. WY treatment significantly decreased fat content in liver by almost 50%. Strikingly, in SIRT6 TG mice, this fat reduction was significantly enhanced (Figure 6F). Likewise, in HZ mice this fat reduction was significantly inhibited (Figure 6G). In addition, we examined the cumulative physiological outcome of these metabolic changes. Respiratory exchange ratio (RER) was quantified in WY treatment in WT and HZ mice. The RER ra- tio under a normal diet was similar between WT and HZ mice. Daily WY administration caused a decrease in RER shortly after WY supplementation in WT mice (Figure 6H). Strikingly, the HZ mice had a significantly weaker response to WY treatment, indi- cating reduced b-oxidation. In addition, once the WY treatment was stopped, the RER returned to normal, and the changes pre- viously seen between WT and HZ disappeared (Figure 6H). These findings suggest that SIRT6 directs the energy source uti- lization toward fat under broader conditions requiring PPARa activation.
SIRT6 deacetylase activates PPARa, yet no acetylation was de- tected on PPARa (Figure S2C). In addition, previously SIRT6 was reported as a histone deacetylase, which usually results in reduced levels of transcription. Intriguingly, here, SIRT6 induces PPRE H3 K9 acetylation and activity (Figure 2). Therefore, we examined whether the mechanism underlying SIRT6’s activation of PPARa is potentially via the deacetylation of one of its coacti- vators or corepressors. The acetylation levels of 12 PPARa coac- tivators or corepressors were compared between WT and SIRT6 TG livers using stable isotope labeling by amino acids in cell cul- ture (SILAC) mass spectrometry (MS) analyses. Within these, only one protein, NCOA2, a known PPARa coactivator, showed significantly lower differentially acetylated levels on K780 in TG mice (Figure 7A). Therefore, SIRT6 and NCOA2 interaction was examined. As seen in Figure 7B, endogenous SIRT6 interacts with FLAG-tagged NCOA2. Conversely, endogenous NCOA2 in- teracts with FLAG-tagged SIRT6. This suggests that NCOA2 might be the PPARa coactivator through which SIRT6 mediates the induction of PPARa activity.To explore the role of NCOA2 acetylation in PPARa activation, NCOA2 K780 was mutated into arginine (K780R) or glutamine (K780Q), mimicking constitutive deacetylated or acetylated lysine, respectively. PPRE-luciferase activity was measured in HEK293T cells overexpressing WT, K780R, or K780Q with or without overexpression of SIRT6, PPARa, or both. As previ- ously published, NCOA2 significantly induced PPRE activity (Røst et al., 2009). Notably, in comparison to WT or K780Q, the K780R mutant significantly further increases PPRE activation (Figures 7C and S7A). WT and K780Q showed similar levels of PPRE activation, suggesting that NCOA is highly acetylated and its deacetylation is required for further PPRE activation. Strikingly, indeed, SIRT6 overexpression significantly induced WT NCOA2-dependent PPRE activation to the same levels as K780R, the deacetylated NCOA2 mimicker. Moreover, SIRT6 overexpression had no effect on either K780Q- or K780R- induced PPRE activity or on NCOA2-PPARa association (Figures 7C and S7B). These findings suggest a model in which SIRT6 promotes PPARa activity via NCOA2 K780 deacetylation, which leads to increased HAT acetylation of PPRE chromatin and subsequent activation of PPARa-dependent signaling (Figure 7D).
DISCUSSION
Here, we show that SIRT6 regulates hepatic b-oxidation by activating PPARa via NCOA2 deacetylation. This regulation culminates in a PPARa-dependent maintenance of normal whole-body RER and improved liver fat content. These findings significantly contribute to our understanding of how SIRT6 coor- dinates central metabolic pathways in response to energy fluctuations. Previously, we showed that SIRT6 overexpression mimics key aspects of the rodent CR response (Kanfi et al., 2012). b-oxida- tion is a known hallmark of CR in both rodents and monkeys (Bruss et al., 2010; Rhoads et al., 2018). Remarkably, SIRT6 acti- vation of PPARa also increases fatty acid degradation and peroxisomal b-oxidation, revealing an additional similarity to the CR response. Moreover, both SIRT6 and PPARa levels increase under starvation and CR (Masternak and Bartke, 2007). Altogether, in parallel to increased PPARa levels, we suggest here a pathway of PPARa regulation under CR by induction of its activity via SIRT6. What are the physiological outcomes of this PPARa induction by SIRT6? A significant amount of data show a role for SIRT6 as a guardian of healthy liver (Kim et al., 2010). HFD or liver-specific deletion of SIRT6 in mice causes fatty liver formation while SIRT6 overexpression significantly reduced fatty liver incidence (Kanfi et al., 2010). Thus, potentially SIRT6 protects against he- patic steatosis through PPARa activation and the increase in fatty acid b-oxidation. In addition, liver-specific SIRT6 KO
impairs hepatic ketogenesis via induced Fsp27b. When fed a ketogenic diet, these mice also showed exacerbated hepatic steatosis and inflammation (Chen et al., 2019). This adds another interesting layer for SIRT6’s protective effect on liver pathol- ogies. Yet, it is possible that SIRT6 expression in other tissues also contributes to this and other liver phenotypes observed in the current study due to the use of a whole-body HZ model. However, samples from human patients with fatty liver or liver fibrosis exhibited significantly lower levels of SIRT6 than did normal controls (Ka et al., 2017; Kim et al., 2010). Thus, SIRT6 activation has great therapeutic potential in treating fatty liver diseases.
SIRT6 possesses histone deacetylase activity that is usually associated with repressed transcription. Here, we show an example of direct activation of gene transcription by SIRT6. SIRT6 binds to PPARa and its PPREs in a PPARa-dependent manner, inducing PPARa-mediated transcription. Importantly, SIRT6 enzymatic activity is required for this induction. Therefore, here, we aimed to identify the precise target for SIRT6 within the PPARa complex on the PPRE. There are at least three possible options. First, SIRT6 might directly affect PPARa, either by deac(et)ylation, or mono-ADP-ribosylation (monoADPR). Yet, we and others were unable to detect acetylation on PPARa (Figure S2C; Oka et al., 2011). However, this possibility cannot be rigorously excluded at this stage. A second possibility is that SIRT6 binding to PPARa results in a conformational change, allowing increased binding or recruitment of PPARa coactiva- tors. In support of this model, SIRT6 contains the well-known LXXLL motif (Figure S7C) in its coactivator domain, a motif commonly found in PPARa-binding coactivators. Yet, given that SIRT6 enzymatic activity is required for PPARa activation, such a model does not fully explain the activity of SIRT6 on PPARa. Thus, a final, more comprehensive model suggests that SIRT6 activates PPARa via deacetylation and activation of a member of the PPARa complex, such as a coactivator acetyl- transferase. Indeed, we found that SIRT6 regulates PPARa activ- ity via NCOA2 K780 deacetylation. In analogy, SIRT1 was shown to activate PPARa indirectly via deacetylation and activation of its coactivator, PCG1a (Purushotham et al., 2009). Furthermore, in contrast, SIRT4 was shown to inhibit PPARa signaling and b-oxidation indirectly by inhibiting SIRT1 activity by NAD+ sub- strate competition (Laurent et al., 2013). While SIRT6 and SIRT1 increase in starvation, SIRT4 decreases in starvation, which may potentially explain their opposite roles in b-oxidation (Laurent et al., 2013; Kanfi et al., 2008a, 2008b).
Deacetylation of NCOA2 K780 promotes its activation of PPARa. Likewise, NCOA2 Ser736 phosphorylation is also known to augment PPAR activity via increased binding and recruitment of P300 acetyltransferase to NCOA2 and PPREs (Frigo et al., 2006). Thus, it would be interesting to further investigate the interaction between these posttranslational modifications on PPARa activation potentially by increased P300 recruitment. In support of this model, indeed, we found an inverse correlation between NCOA2 and H3K9 acetylation along with increased PPARa activation.
Regarding the involvement of PPARa in SIRT6-dependent phenotypes, an interesting question is the potential contribution of PPARa to the gender-specific effects found in SIRT6 TG mice.In mice, SIRT6 regulation of body and fat mass is more pronounced in males. In addition, the reduced lifespan in SIRT6-deficient mice is more severe in males (Peshti et al., 2017). Similarly, SIRT6 overexpression extends lifespan, mimics a DR-like transcription profile, and reduces IGF-1 signaling only in males (Kanfi et al., 2012). This gender-specific effect is main- tained over the course of evolution, in both monkeys and humans (Naiman and Cohen, 2018). In SIRT6 KO monkeys, only females survive to birth, and in humans, SIRT6 deficiency results in sex reversal in the male fetus (Ferrer et al., 2018; Zhang et al., 2018). Interestingly, the effects of PPARa are also gender- specific. PPARa activators decrease adiposity in male but not female animals fed HFD, and reduce total cholesterol and triglyc- erides only in males in normal diet (Yoon et al., 2002). Moreover, the response of b-oxidation gene expression to PPARa agonist is much stronger in male mice and ovariectomized females than in WT females, indicating the involvement of gonadal sex steroids (Yoon, 2010). Similarly, in humans, males but not females carrying a PPARa polymorphism demonstrated improved triglyceride levels in response to statin treatment (Khan et al., 2004). Therefore, PPARa and SIRT6 share several gender- specific effects, particularly in protecting against obesity and its physiological consequences. It would be of great interest in future research to explore whether PPARa is involved in other male-specific SIRT6-regulated pathways such as lifespan extension.
Interestingly, in comparison to their WT littermates, no signif- icant physiological effects were observed in SIRT6 HZ mice under normal growth conditions. For example, no changes in body weight and composition, daily activity, RER, or food intake were detected. This suggests that SIRT6 haploinsuffi- ciency retains sufficient activity under unstressed conditions, at least to adulthood. Yet, under conditions in which PPARa is activated, such as increased b-oxidation (i.e., under food limitation), both copies of active SIRT6 are required. Notably, gene transcription changes in PPARa genes were found in SIRT6 HZ even under normal conditions, but these were not phenotypically evident until PPARa was activated. This may indicate that SIRT6 primes transcription to respond quickly to energy fluctuations such as changes in nutrient availability. We suggest that a certain minimum level of SIRT6 is necessary for long-term survival under fluctuating or chronic mild meta- bolic stress, such as that found in obese or old animals. Thus, OSS_128167 SIRT6 may have therapeutic potential under chronic long-term stress conditions, for example, those that can be found in age-related diseases.