Context-Dependent Roles for SIRT2 and SIRT3 in Tumor Development Upon Calorie Restriction or High Fat Diet

Calorie restriction (CR) is considered one of the most robust ways to extend life span and reduce the risk of age-related diseases, including cancer, as shown in many different organisms, whereas opposite effects have been associated with high fat diets (HFDs). Despite the proven contribution of sirtuins in mediating the effects of CR in longevity, the involvement of these nutrient sensors, specifically, in the diet-induced effects on tumorigenesis has yet to be elucidated. Previous studies focusing on SIRT1, do not support a critical role for this sirtuin family member in CR-mediated cancer prevention. However, the contribution of other family members which exhibit strong deacetylase activity is unexplored. To fill this gap, we aimed at investigating the role of SIRT2 and SIRT3 in mediating the anti and pro-tumorigenic effect of CR and HFD, respectively. Our results provide strong evidence supporting distinct, context-dependent roles played by these two family members. SIRT2 is indispensable for the protective effect of CR against tumorigenesis. On the contrary, SIRT3 exhibited oncogenic properties in the context of HFD-induced tumorigenesis, suggesting that SIRT3 inhibition may mitigate the cancer-promoting effects of HFD. Given the different functions regulated by SIRT2 and SIRT3, unraveling downstream targets/pathways involved may provide opportunities to develop new strategies for cancer prevention.


INTRODUCTION
Calorie restriction (CR); 20-40% reduction of daily calorie intake without malnutrition, is the most studied and reproducible non-genetic intervention known to extend lifespan in different species (1)(2)(3)(4)(5)(6)(7)(8). In addition to positively affecting longevity, CR improves also healthspan as evidenced by the reduced risk for a variety of age-associated diseases, including diabetes, sarcopenia, cardiovascular diseases, and hearing loss in both rodents and humans (9). The opposite effect is exerted by obesity and high-fat diets. In this regard, the high-fat diet (HFD)-fed mouse has been well-established as a model for impaired glucose tolerance (IGT) and type 2 diabetes (10). In a similar way, obesity in humans is linked with increased mortality due to several pathologies and major reductions in life expectancy compared with normal weight (11).
Given that cancer is considered an age-related disease (12), any interventions shown to improve healthspan could be exploited to either prevent or slow down tumorigenesis. Indeed, CR delays tumorigenesis in a p53-knockout mouse model (13) and recent studies show a significant reduction in the incidence of cancers in rhesus monkeys fed a CR diet compared to a control diet (14). Accordingly, a recent meta-analysis showed that 40 of the 44 pre-clenical studies (90.9%) support the anticancer role of CR despite the different measurements (15). On the contrary, HFD-induced obesity was found to shorten life expectancy through mediating various diseases, including cancer (16)(17)(18). While the anti and pro-tumorigenic effects of CR and HFD, respectively, are well-established, the underlying mechanisms remain obscure.
Considering that sirtuins mediate the CR-induced anti-aging effects (19,20) through their function as NAD-dependent protein deacetylases directly activated by CR, it could be proposed that they may contribute to the tumor suppressive effects of CR. Despite extensive research regarding the broader role of sirtuins in cancer (21), this is a relatively unexplored area. Therefore, it is rather surprising that there is a lack of experimental data to assess the involvement of sirtuins, specifically, in the CRinduced effect on tumorigenesis. Among the different family members, SIRT1, is the only sirtuin studied so far in this context. More specifically, its overexpression failed to influence the anticancer effects of every-other-day fasting-a variation of CR- (22). Although these result suggest that SIRT1 may play a limited role in the effects of CR on cancer, the possible involvement of other sirtuins in the diet-induced effects on tumorigenesis has yet to be elucidated.

Genotyping
Mouse genomic DNA was isolated from tail biopsies using the REDExtract-N-Amp TM Tissue PCR Kit following the standard protocol. PCR was performed using the primer pairs and conditions as described below: For p53 Forward wild-type 5 ′ -AGGCTTAGAGGTGCAAGCTG-3 ′ ,

Diets
Each of the 6 genotypes (n = 9-12) was subjected to 3 dietary regimes; ad libitum control diet (3.85 kcal/g, D12450Ji), 30% calorie restriction diet (D15032801Bi), and ad libitum high-fat diet (5.24 kcal/g with 60% kcal from fat, D12492Ri), starting at about 7-8 weeks after birth. Diets were purchased from Research Diet Inc. (Brunswick, NJ, USA), and were designed to provide variations in calories while maintaining similar micronutrient composition. Heterozygous p53 knockout mice were subjected to two cycles of either CR or HFD-14 weeks each, intercepted by 21 weeks of exposure to control diet pellets. Homozygous p53 knockout mice displayed shorter lifespan (∼60% shorter than p53 heterozygous knockout mice), and were subjected to one 14 weeks long cycle of either CR or HFD.

Body Weight, Tumor Incidence, and Survival
Diets were administered by members of the Center for Comparative Medicine (CCM) at Northwestern University. Body weights and health condition of the mice were closely monitored and reports were provided to the lab members. All data were analyzed by lab members. Tumors and tissues were collected at the end of the experiments.

Statistical Analysis
All graphs, statistical analysis, and calculation of p-values were carried out using GraphPad Prism software (GraphPad Software Inc., San Diego, CA, USA). Overall survival and tumor-free survival were plotted by Kaplan-Meier curves and analyzed using the log-rank (Mantel-Cox) test. The hazard ratios, indicating changes in tumor incidence rate, were calculated from the slopes of tumor-free survival curves and estimated based on log-rank test.

Effect of SIRT2 and SIRT3 Loss on Body Weight Upon CR and HFD
No change in body weight of Sirt2 −/− and Sirt3 −/− mice fed on control diet was observed, as compared to control mice with the exception of Sirt3 −/− ; p53 −/− mice which exhibited increased body weight (Figures 1A,D). Upon CR, all mice responded to the diet as evidenced by the decrease in body weight and the weight gain following withdrawal from calorie restriction (Figures 1C,F). Of note, Sirt2 −/− mice appeared to show the greatest effect, based on the significant decreased body weight compared to other mice (Figures 1C,F). On the other hand, Sirt3 −/− mice were indistinguishable compared to wild-type ( Figures 1C,F) with regards to body weight following CR. This finding is consistent with previous reports showing no difference in the body weight of Sirt3 deleted mice compared to Sirt3 wildtype ones under CR conditions (29,30). Upon HFD, although a similar gain weight was observed in all mice in the p53 +/− background ( Figure 1B), both Sirt2 −/− and Sirt3 −/− mice in the p53 −/− background displayed decreased body weights as compared to control mice ( Figure 1E).   (Figures 3A,B). Accordingly, survival and tumor incidence curves upon CR were identical in  Table 1). It is worth mentioning that p53 +/− control diet-fed mice deleted for Sirt3 exhibited accelerated tumor incidence when compared to Sirt3 wild-type mice, (Figure 3D) which could possibly mask any benefits from CR. All together, our data suggest that among the two family members studied here, SIRT2 exhibits strong tumor suppressive properties upon CR regardless of the p53 genetic background, consistent with a role for SIRT2 in mediating the cancer preventive effect of CR.

Tumor Spectrum Analysis in Sirt2 and Sirt3 Null Mice Upon CR and HFD
We noticed that p53-related tumors were the major tumors developed in mice of our study, with sarcomas and lymphomas representing more than 50% of all tumors ( Figure 6A). Interestingly, we observed strong association between the loss of Sirt2/Sirt3 and certain tumors (Figure 6B and Supplementary Table 2). Under different dietary regimens, Sirt3 deficient mice experienced the highest incidence rate of liver tumors and the lowest incidence of sarcomas, either in p53 +/− or p53 −/− background, compared to the other groups. On the other hand, colon/intestinal and renal tumors were increased in Sirt2 knockout mice as compared to other genotypes. Moreover, we observed an increased incidence of splenomegaly in mice with Sirt3 deletion, when compared to the other genotypes, regardless of p53 status (Figures 6C-E).

DISCUSSION
SIRT2 is the least comprehensively studied member of mammalian sirtuins, especially for its role in the context of calorie restriction. The results from this study provide evidence, for the first time, to support a role for SIRT2 in mediating the protective effects of CR against tumorigenesis.
Although controversial roles for SIRT2 in cancer have been described (32), our data show that SIRT2 exhibits tumor suppressive properties upon calorie restriction. Interestingly, this happened despite the decreased overall weight of calorie restricted Sirt2 −/− mice as compared to Sirt2 wild-type mice suggesting that SIRT2-directed downstream signaling is impaired. The observed effect is consistent with previous studies showing that SIRT2 functions as a tumor suppressor (28,(33)(34)(35). Sirt2 loss negates the protective effect of CR on tumor development in a p53 mouse model and, therefore, unraveling downstream pathways regulated by SIRT2 in this context is warranted. It is worth mentioning that despite that several SRT2 deacetylation targets have been reported, a comprehensive analysis of SIRT2 targets using unbiased high throughput approaches is currently missing. We are in the process of completing such an analysis (data not show), which will enable to associate SIRT2 targets with molecular pathways related to CR and provide mechanistic insights. Very recently, Diaz-Ruiz et al. showed that upregulation of NQO1, an essential NADH-dehydrogenase that mediates redox control of metabolic homeostasis, mimicks aspects of CR including protection against carcinogenesis in mice (36). Of note, our latest study showed that NQO1 interacts with and activates SIRT2 in an NAD-dependent manner (37). Therefore, it is intriguing to further investigate whether SIRT2 activation mediates the CR-mimetic effects of NQO1 overexpression or whether SIRT2 overexpression alone can mimic the beneficial effects exerted by CR on cancer prevention, further emphasizing the potential for exploiting SIRT2 as a target for cancer prevention. SIRT3 has been previously described to mediate CR-related health benefits (25,30,38). However, the involvment of SIRT3 in regulating diet-dependent tumor development is obscure. By using a p53 mouse model, here we show that SIRT3 plays a prominent role in HFD-induced tumorigenesis. Notably, Sirt3 deficient mice were resistant to the tumor promoting effect of HFD. Although it has been previously shown that Sirt3 loss in mice on a high-fat diet accelerates obesity, insulin resistance, hyperlipidemia, and steatohepatitis (39), and SIRT3 preserves heart function and capillary density (40), our results show that Sirt3 loss increases survival and decreases tumor incidence upon HFD. Therefore, it can be proposed that SIRT3 exhibits oncogenic properties in the context of HFD-induced tumorigenesis and SIRT3 inhibition might be suggested as a strategy to mitigate the tumor promoting effects of HFD. The protective effect of Sirt3 loss, especially in the p53 +/− background, was observed while mice gained weight as a response to the HFD. Therefore, it is reasonable to suggest that SIRT3 functions previously shown to support its role as an oncogene might be involved. In this regard, regulation of ROS production and mitochondrial metabolism (41)(42)(43) might provide the mechanistic link to explain the protective effect of SIRT3 inhibition in HFDinduced tumorigenesis. Of note, our unbiased proteomics-based analysis to detect proteins with increased acetylated levels in Sirt3 −/− liver tissue as compared to Sirt3 +/+ tissue revealed that fatty acid metabolic process was the most significantly enriched pathway using (Supplementary Figure 1). This is consistent with previous studies showing that SIRT3 regulates fatty acid oxidation (25,44), while Perturbations in fatty acid metabolism have been associated with cancer (45). More importantly, regulation of fatty acid metabolism is critical especially in mice fed a HFD diet. Therefore, it is possible that dysregulation of this pathway in Sirt3 −/− mice might provide some mechanistic insights to explain the observed resistance to HFD-induced tumorigenesis.  Both p53 −/− and p53 +/− mice were included in our analysis to address whether p53-dependent functions are involved. Upon CR, Sirt2 deletion abolished the protective effect in both genotypes, supporting a critical role for SIRT2 upon CR. In the HFD-fed mice, Sirt3 deletion rendered mice more resistant to HFD-induced tumorigenesis in both p53 genotypes, suggesting a role for SIRT3 in this context. p53 has been previously shown to regulate metabolism and mitochondrial function (46)(47)(48). The observed phenotypes in mice lacking p53 favor a role for these sirtuin family members involving p53independent mechanisms. Although some responses seem to be stronger in the p53 heterozygous as compared to the p53 homozygous knockout mice, we believe that this might due to the different total time that the mice have been exposed to the two diets. In this regard, p53 +/− mice develop tumors later in life compared to the p53 −/− mice which allowed 2 cycles of feeding with the different diets as opposed to one cycle in the p53 −/− mice.
Collectively, our data support new context dependent roles for SIRT2 and SIRT3 in regulating the effects of CR and HFD on tumorigenesis. Given the lack of experimental data to address the contribution of situin family members other than SIRT1, our results reignite interest regarding the contribution of sirtuins under these specific metabolic conditions. Therefore, SIRT2 activation upon CR and SIRT3 inhibition upon HFD could be explored as new strategies to enhance CR-induced cancer prevention and decrease HFDinduced tumor development, respectively (Figure 7). Future studies focusing on unraveling downstream pathways and mechanisms hold the promise for identifying novel interventions to either mimic the cancer-preventive effect of CR or protect against HFD-induced tumorigenesis.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article/Supplementary Material.

ETHICS STATEMENT
The animal study was reviewed and approved by Institutional Animal Care and Use Committee (IACUC) Northwestern University.

AUTHOR CONTRIBUTIONS
MA and CO'C performed experiments and collected data. HJ maintained the mice colonies. EC helped with tissue collection and tissue sample analysis. MA performed statistical analyses. CO'C and AV designed the study. MA and AV prepared the manuscript.