Systemic SIRT1 insufficiency results in disruption of energy homeostasis and steroid hormone metabolism upon high-fat-diet feeding

Abstract

SIRT1 is a highly-conserved NAD⁺-dependent protein deacetylase that plays essential roles in the regulation of energy metabolism, genomic stability, and stress response. Although the functions of SIRT1 in many organs have been extensively studied in tissue-specific knockout mouse models, the systemic role of SIRT1 is still largely unknown as a result of severe developmental defects that result from whole-body knockout in mice. Here, we investigated the systemic functions of SIRT1 in metabolic homeostasis by utilizing a whole-body SIRT1 heterozygous mouse model. These mice are phenotypically normal under standard feeding conditions. However, when chronically challenged with a 40% fat diet, they become obese and insulin resistant, display increased serum cytokine levels, and develop hepatomegaly. Hepatic metabolomic analyses revealed that SIRT1 heterozygous mice have elevated gluconeogenesis and oxidative stress. Surprisingly, they are depleted of glycerolipid metabolites and free fatty acids, yet accumulate lysolipids. Moreover, high-fat feeding induces elevation of serum testosterone levels and enlargement of seminal vesicles in SIRT1 heterozygous males. Microarray analysis of liver mRNA indicates that they have altered expression of genes involved in steroid metabolism and glycerolipid metabolism. Taken together, our findings indicate that SIRT1 plays a vital role in the regulation of systemic energy and steroid hormone homeostasis.—Purushotham, A., Xu, Q., Li, X. Systemic SIRT1 insufficiency results in disruption of energy homeostasis and steroid hormone metabolism upon high-fat-diet feeding.

Yeast silent information regulator 2 (Sir2) protein and its homologues in other organisms, also known as sirtuins, are highly conserved NAD⁺-dependent protein deacetylases and/or ADP ribosyltransferases (1–3). SIRT1, the most conserved mammalian sirtuin (4), directly couples NAD⁺ hydrolysis to the deacetylation of a number of transcription factors and cofactors, and thereby is widely involved in a variety of cellular processes, including energy metabolism, genomic stability, stress response, development, and aging (5, 6). Recent studies in a number of transgenic and tissue-specific knockout mouse models have revealed essential roles of this metabolic sensor in the modulation of gene expression and metabolic activities in response to changes in cellular energy states (7). For example, in the liver, SIRT1 has been shown to regulate both glucose and lipid metabolism—such as gluconeogenesis, fatty acid oxidation, and cholesterol and bile acid metabolism—through deacetylation of PGC-1α and TORC2 coactivators, LXR and FXR nuclear receptors, and SREBP transcription factors (8–15). In the pancreas, SIRT1 stimulates insulin secretion in β cells by repressing transcription of uncoupling protein 2 (UCP2; refs. 16, 17). In white adipose tissue (WAT), SIRT1 represses PPARγ signaling, thus decreasing fat storage and promoting fat mobilization on fasting (18). In the brain, SIRT1 controls energy homeostasis through the hypothalamus/pituitary axis in response to nutrient signals (19–23). Recently, SIRT1 has also been linked to the regulation of animal circadian rhythm in response to oscillation of cellular NAD⁺ levels, thereby directly coupling metabolic processes to the circadian clock (24–27).

Elucidating the molecular mechanisms underlying the role of SIRT1 in the functioning of various metabolic organs is an area of intense research. In recent years, tissue-specific SIRT1-knockout mouse models have provided essential information on the organ-specific functions of this sirtuin, while bypassing the pleiotropy of a whole-body knockout. The systemic role of SIRT1, however, remains largely unknown, primarily due to the severe developmental defects resulting from a whole-body knockout of SIRT1 in mice (28, 29). Germline deletion of SIRT1 in mice has been shown to cause severe developmental defects, leading to a high frequency of neonatal death, as well as developmental delays and sterility in surviving animals (28, 29). Because these developmental defects result in growth retardation, which likely affects the metabolic homeostasis indirectly through reduced circulating insulin/insulin-like growth factor levels, it is difficult to utilize these conventional whole-body-knockout models to analyze the systemic functions of SIRT1. However, given that SIRT1 regulates a variety of metabolic processes in a number of tissues, investigation of SIRT1 pleiotropic effects is paramount to appreciating its role in systemic biological processes, such as aging and the development of certain diseases. In addition, better understanding of the systemic function of SIRT1 is crucial to validate therapeutic interventions that are based on SIRT1 small molecule regulators. In this report, we assessed the systemic functions of SIRT1 in a whole-body SIRT1 heterozygous mouse model (28). These mice are phenotypically normal under standard feeding conditions, therefore allowing us to investigate the systemic role of SIRT1 without the interference of developmental defects.

MATERIALS AND METHODS

Animal experiments

Heterozygous mice carrying one null allele of the SIRT1 gene (28) were backcrossed 6 generations into the C57BL/6 background. Age-matched wild-type (WT; SIRT1^+/+) and heterozygous (het; SIRT1^+/−) littermate male mice, older than 6 wk of age, were fed ad libitum with either a control low-fat diet (D12329; Research Diets, New Brunswick, NJ, USA) or a high-fat diet providing 40% kcal from soybean and coconut oil (D12327; Research Diets) for 32 wk. Fat and lean body mass were determined by DEXA scanning in live mice at 26 wk of feeding on the low- or high-fat diet. At 32 wk, tissues were harvested after 4 h food withdrawal, starting at the beginning of the day-night cycle. All animal experiments were conducted in accordance with guidelines of U.S. National Institute of Environmental Health Sciences/National Institutes of Health Animal Care and Use Committee.

Histological and biochemical analysis

Paraffin-embedded liver sections were stained with hematoxylin and eosin for morphological examination. Serum lipids were measured using commercially available kits (Wako Chemicals USA, Richmond, VA, USA; and Sigma-Aldrich, St. Louis, MO, USA). Serum insulin, leptin, and inflammatory cytokine levels were measured using multiplexed ELISA plates (Meso Scale Discovery, Gaithersburg, MD, USA). Liver lipids were extracted as described previously (30), and liver triglycerides (TGs), phospholipids, and cholesterol were quantified using commercially available kits (Sigma and Wako). Serum testosterone levels were determined by ELISA (R&D Systems, Minneapolis, MN, USA).

Metabolomic analysis

To quantitatively analyze metabolic profiles in the liver, 50–100 mg of frozen liver tissues was submitted to Metabolon (Durham, NC, USA), where the relative amounts of small molecular metabolites were determined using 3 independent platforms: ultra-high-performance liquid chromatography/tandem mass spectrometry (UHPLC/MS/MS²) optimized for basic species, UHPLC/MS/MS² optimized for acidic species, and gas chromatography/mass spectrometry (GC/MS) (31, 32). Briefly, liver samples were cold methanol extracted and split into 3 aliquots. These aliquots were processed and characterized by one of the 3 analytical methods. Chromatographic timelines were standardized using a series of xenobiotics that elute at specified intervals throughout each chromatographic run. The technical variability of each analytical platform was assessed by repeated characterization of a pooled standard that contained an aliquot of each sample within the study. Metabolites were identified by automated comparison of the ion features in the experimental samples to a reference library of chemical standard entries, including retention time, molecular weight (m/z), preferred adducts, and in-source fragments, as well as associated MS spectra, and these were curated by visual inspection for quality control using software developed at Metabolon (33). Statistical analysis of log-transformed data was performed using R (http://cran.r-project.org/), and Welch's t tests were performed to compare data between experimental groups. Multiple comparisons were accounted for by estimating the false discovery rate (FDR) using q values (34).

RNA analysis

Total RNA was isolated from tissues using TriZOL (Invitrogen, Carlsbad, CA, USA) and Qiagen RNeasy minikit (Qiagen, Valencia, CA, USA). For real-time quantitative PCR (qPCR), cDNA was synthesized with the ABI reverse transcriptase kit, and analyzed using SYBR Green Supermix (Applied Biosystems, Carlsbad, CA, USA). All data were normalized to lamin A expression. To analyze the gene expression profiles of liver, total RNA was isolated, and RNA quality was determined using an Agilent bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). Gene expression analysis was conducted on 3 independent biological replicates, with 2 animals in each replicate, using Agilent Whole Genome Mouse 4 × 44 multiplex format oligo arrays (014868; Agilent Technologies), following the Agilent 1-color microarray-based gene expression analysis protocol. Starting with 500 ng of total RNA, Cy3-labeled cRNA was produced according to the manufacturer's protocol. For each sample, 1.65 μg of Cy3-labeled cRNAs was fragmented and hybridized for 17 h in a rotating hybridization oven. Slides were washed and then scanned with an Agilent scanner. Data were obtained using the Agilent Feature Extraction 9.5 software, using the 1-color defaults for all parameters. The Agilent Feature Extraction Software performed error modeling, adjusting for additive and multiplicative noise. The resulting data were processed using the Rosetta Resolver 7.1 system (Rosetta Biosoftware, Kirkland, WA, USA).

Western blot analysis

Tissue whole-cell homogenates were prepared with Nonidet P-40 buffer (50 mM Tris-HCl, pH 8.0; 150 mM NaCl; and 0.5% Nonidet P-40) containing Complete protease and phosphatase inhibitors (Roche Applied Science, Indianapolis, IN, USA), and then immunoblotted using antibodies against SIRT1 (Sigma-Aldrich).

Insulin tolerance test

To test the insulin sensitivity of WT and het mice fed control low-fat diet or high-fat diet, mice unfed for 6 h were intraperitoneally injected with 1 IU/kg body wt of insulin. Blood samples were then collected from the tail at 0, 15, 30, 45, 60, 90, and 120 min postinjection, and glucose levels were measured by OneTouch Ultra Glucose Monitor (LifeScan, Milpitas, CA, USA).

Statistical analysis

Values are expressed as means ± se. Significant differences between means were analyzed by 2-tailed, unpaired Student's t test, and differences were considered significant at values of P < 0.05. The microarray data were analyzed using a specific program (see above).

RESULTS

Single-copy loss of SIRT1 results in increased obesity, enhanced insulin resistance, and elevated circulating cytokine levels upon high-fat feeding

To examine the systemic function of SIRT1 in mice, we generated a whole-body heterozygous SIRT1-knockout mouse strain on the C57BL/6 background (SIRT1 het), as described in Materials and Methods. Our real-time qPCR and immunoblotting analyses confirmed that both SIRT1 mRNA and protein levels in SIRT1 het mice were decreased to 50–60% of those in WT mice in all tested tissues except the testis (Fig. 1A, B), indicating that this heterozygous strain is a true systemic SIRT1-insufficient line. The normal levels of SIRT1 in the testis probably reflect the exceptionally high expression levels of this sirtuin in germ cells (28), which can compensate for the loss of half-gene dosage of SIRT1 in this organ.

Figure 1.

Single-copy loss of SIRT1 increases obesity in mice after chronic high-fat feeding. A) Relative SIRT1 mRNA levels in indicated tissues. Expression of SIRT1 in WT and SIRT1 het mice was analyzed by qPCR (n=5 or 6). WAT, white adipose tissue; BAT, brown adipose tissue. B) SIRT1 protein in indicated tissues, analyzed by Western blot. C) SIRT1 het mice gained significantly more weight than WT control mice when fed the high-fat diet. WT and SIRT1 het mice were fed a low-fat (LF) or high-fat (HF) diet for 32 wk; body weight was measured weekly (n=15–20). D, E) SIRT1 het mice are more obese than WT mice after HF-diet feeding. Fat and lean body mass (D) and percentage of body weight (Bwt; E) were measured by DEXA scanning after 26 wk of the LF or HF diet (n=15–20). *P < 0.05.

Since SIRT1 het mice fed a chow diet were phenotypically normal, we challenged them with a high-fat diet, providing 40% kcal from soybean and coconut oil, or a low-fat control diet with 10.5% kcal from the same fat sources. Consistent with previous reports that loss of function of SIRT1 in various tissues disrupts energy homeostasis (5, 6), SIRT1 het mice started to gain significantly more weight than WT control mice after 8 wk of the high-fat diet (Fig. 1C), despite there being no detectable difference in food intake (data not shown). DEXA scanning after 26 wk of high-fat-diet feeding revealed that this increase in body weight gain in SIRT1 het mice was associated with significant increases in both fat mass and lean mass (Fig. 1D), as well as modest, but significant, elevations of relative whole-body fat percentage (Fig. 1E); thus, overall, these mice were more obese than WT mice on a high-fat challenge. Interestingly, SIRT1 het mice fed the low-fat control diet also displayed reduced relative lean body mass and a trend of increased fat mass percentage (Fig. 1E), although their body weights were normal (Fig. 1C).

Obesity has been associated with the activation of cellular stress signaling and inflammatory pathways, contributing to the development of insulin resistance, fatty liver, and proinflammatory and prothrombotic states—a collection of metabolic abnormalities also known as metabolic syndrome (35–37). The increase in body weight, particularly as diet-induced obesity, suggests that SIRT1 het mice may display metabolic defects related to obesity. While SIRT1 het mice showed normal levels of serum TGs, their serum cholesterol and free fatty acid levels were modestly elevated on high-fat feeding (Fig. 2A, C). Their serum concentrations of insulin, as shown in Fig. 2D, were comparable to those of WT controls. But unexpectedly, their leptin levels were decreased on high-fat feeding (Fig. 2E). Furthermore, their blood glucose levels were significantly higher than those of control mice (Fig. 2F), and in line with this observation, high-fat-fed SIRT1 het mice were more insulin resistant than WT controls, as revealed by an insulin tolerance test (Fig. 2G, H). The circulating levels of a number of serum proinflammatory cytokines, particularly IL-12 and TNF-α, were considerably elevated in high-fat-fed SIRT1 het mice (Fig. 3), indicating the development of an inflammatory state. Taken together, these observations demonstrate that systemic loss of one copy of the SIRT1 gene in mice is sufficient to drive impairments in energy homeostasis in response to nutrient challenges, leading to the development of metabolic syndrome.

Figure 2.

Single-copy loss of SIRT1 results in increased resistance to insulin. A–E) Serum TG (A), cholesterol (B), free fatty acid (C), insulin (D), and leptin levels (E) in WT and SIRT1 het mice. WT control and SIRT1 het mice were fed the low-fat (LF) or high-fat (HF) diet for 32 wk; serum lipid and hormone levels were measured after 4 h of food withdrawal. F) SIRT1 het mice have increased fasting glucose levels. WT control and SIRT1 het mice fed the LF or HF diet were fasted overnight (16 h). G) SIRT1 het mice are more insulin-resistant after high-fat diet feeding. Insulin tolerance test was carried out as described in Materials and Methods after 22 wk of LF- and HF-diet treatment (n=15–20). H) Area under curve (AUC) for insulin tolerance test. *P < 0.05.

Figure 3.

SIRT1 het mice display increased serum cytokine levels after high-fat-diet feeding. Serum IL-12 (A), TNF-α (B), IL-6 (C), and IL-10 (D) levels were measured as described in Materials and Methods (n=8). HF, high fat; LF, low fat, *P < 0.05; ^#0.05 < P < 0.10.

Single-copy loss of SIRT1 impairs glycerolipid/free fatty acid (GL/FFA) cycle and induces gluconeogenesis on high-fat diet

To further evaluate the pathophysiological effects of SIRT1 insufficiency on energy homeostasis, we examined the nutrient metabolism of SIRT1 het mice in liver, the central metabolic organ that controls key aspects of glucose and lipid metabolism (38). As shown in Fig. 4A, chronic high-fat feeding induced hepatomegaly in SIRT1 het mice. The liver was significantly enlarged, even after normalization to body weight. Surprisingly, this enlargement of liver was not accompanied with significantly greater lipid accumulation, as indicated by hematoxylin and eosin staining of liver sections (Fig. 4B), and by enzymatic colorimetric quantification of extracted hepatic TGs, cholesterol, and phospholipids (Fig. 4C–E). In fact, the hepatic TGs were significantly decreased compared to those of WT control mice (Fig. 4C). Further metabolomic analysis of liver tissues, with 3 platforms developed by Metabolon (Fig. 5A and Supplemental Table S1), confirmed that the liver tissues of SIRT1 het mice had normal levels of cholesterol, glucose, and many other metabolic intermediates. However, the GL/FFA cycle—a set of processes that occurs in liver, WAT, and other metabolic tissues to recycle FFAs released by lipolysis (39, 40)—was dramatically altered in the liver tissue of SIRT1 mice. As shown in Fig. 5, in addition to the decreased levels of hepatic TGs, SIRT1 het liver was severely depleted of free fatty acids, monoacylglycerol, glycerol, and many other intermediates in glycerolipid metabolism (Fig. 5A–C). On the other hand, there was accumulation of lysolipids—products of hydrolysis of phospholipids—in the liver of SIRT1 het mice (Fig. 5A, D). These observations, together with the normal metabolomic profile of metabolites in hepatic Krebs cycle (Fig. 5A and Supplemental Fig. S1), the normal expression of hepatic genes involved in fatty acid oxidation (not shown), and the elevated serum free fatty acid levels (Fig. 2C), all suggest that the liver of SIRT1 het mice has enhanced lipolysis and release of free fatty acids. Consistent with this observation, WAT of SIRT1 het mice, a fat storage tissue that is primarily responsible for the release of free fatty acids into the bloodstream in response to nutrient and hormonal signals, accumulated glycerol and free fatty acids, two primary lipolysis products (Fig. 5E, F). Since the SIRT1 het mice are more resistant to insulin, which is coupled with increased incomplete fatty acid oxidation in peripheral tissues, such as muscle (41), our data suggest that the liver and WAT of SIRT1 het mice may augment lipolysis to meet the increased demand for fatty acids in response to insulin resistance in other peripheral tissues after chronic high-fat diet feeding.

Figure 4.

SIRT1 het mice display hepatomegaly but not hepatic steatosis after high-fat-diet feeding. A) Hepatomegaly in SIRT1 het mice. B–E) SIRT1 het mice do not accumulate more lipids in the liver, as indicated by hematoxylin and eosin staining of liver sections from control and SIRT1 het mice (B), and total liver TG (C), cholesterol (D), and phospholipid levels (E); n = 15–20. *P < 0.05.

Figure 5.

SIRT1 het mice have increased hepatic gluconeogenesis and defective GL/FFA cycle. A) Altered levels of metabolites in the gluconeogenesis and GL/FFA cycle pathways in the liver of SIRT1 het mice. Hepatic metabolites of high-fat-fed WT and SIRT1 het mice were analyzed by metabolomics as described in Materials and Methods. Amounts of metabolites in WT livers were set as 1 (dashed line). Metabolites increased in SIRT1 het mice are labeled red (P<0.05) or light red (0.05<P<0.10); metabolites that decreased in SIRT1 het mice are labeled green (P<0.05) or light green (0.05<P<0.10); metabolites that were not measured are indicated within parentheses. B) Fatty acid depletion in the liver of SIRT1 het mice after high-fat feeding. C) SIRT1 het mice were depleted of monoacylglyerol and other metabolites in the glycerolipid metabolism pathway. D) SIRT1 het mice accumulate lysolipids in the liver. E) SIRT1 het mice accumulate phospholipids, free fatty acids, and glycerol in the WAT. F) SIRT1 het mice display an abnormal GL/FFA cycle in the WAT. (n=7). Acsl, acyl-CoA synthetase long-chain family; Acsm, acyl-CoA synthetase medium-chain family; Acss, acyl-CoA synthetase short-chain family; Agpat, 1-acylglycerol-3-phosphate-O-acyltransferase; Aldo, fructose-bisphosphate aldolase; Atgl, adipose triglyceride lipase; DAG, diacylglycerol; Dagl, diacylglycerol lipase; Dgat, diacylglycerol O-acyltransferase; Dgkg, diacylglycerol kinase θ; Eno, enolase; Fbp, fructose-1,6-bisphosphatase; FFA, free fatty acid; G6Pase, glucose-6-phosphatase; Gapdh, glyceraldehyde-3-phosphate dehydrogenase; Gk, glucokinase; GL, glycerolipid; Gpat, glycerol-3-phosphate acyltransferase; GPC, glycerolphosphocholine; GPE, glycerolphosphoethanolamine; GPI, glycerolphosphoinositol; Gpdh, glycerol-3-phosphate dehydrogenase; Hsl, hormone-sensitive lipase; LPA, lysophosphatidic acid; MAG, monoacylglycerol; Mgat, monoacylglycerol O-acyltransferase; Mgl, monoglyceride lipase; PA, phosphatidic acid; Pgm, phosphoglycerate mutase; Pc, pyruvate carboxylase; Pdh, pyruvate dehydrogenase; Pepck, phosphoenolpyruvate carboxykinase; Pfk, phosphofructokinase; Pgi, glucose-6-phosphate isomerase; Pgk, phosphoglycerate kinase; Pk1, pyruvate kinase; PL, phospholipid; Ppap, phosphatidic acid phosphatase. *P < 0.05; ^#0.05 < P < 0.10.

Hepatic metabolomic studies further revealed that SIRT1 heterozygous mice had elevated levels of metabolites, such as phosphoenolpyruvate (PEP) and 3-phosphoglycerate, as well as increased levels of oxidized glutathione (GSSG) (Fig. 5A and Supplemental Table S1), indicating that they display a higher trend of gluconeogenesis and suffer from increased oxidative stress. This increased trend of gluconeogenesis may contribute to the elevated blood glucose levels in these animals (Fig. 2F). Collectively, our findings demonstrate that systemic SIRT1 insufficiency impairs not only hepatic nutrient metabolism, but also systemic energy expenditure.

To further elucidate the molecular mechanisms underlying the impaired hepatic energy metabolism in SIRT1 het mice fed the high-fat diet, we assessed their hepatic gene expression profiles by microarray analysis. Consistent with the fact that SIRT1 het mice fed the low-fat diet were relatively normal, only 101 of the 41,175 tested gene probes were differentially expressed between SIRT1 WT and het mice (Fig. 6A). Thirty-two weeks of high-fat-diet feeding induced significant changes in the expression of 4719 gene probes in WT mice, whereas 12,783 gene probes were altered in SIRT1 het mice. Of these genes, 1352 and 2330 gene probes were significantly changed by >1.5-fold, respectively, and only 737 were common to both WT and het mice (Fig. 6B), suggesting that WT and het mice respond to the chronic high-fat diet differently. Further, ingenuity pathway analysis (IPA) showed that SIRT1 WT and het mice shared similar responses in pathways involved in metabolism of xenobiotics by cytochrome P-450 and tryptophan metabolism (Table 1). However, their responses in pathways that mediate fatty acid metabolism and glycerolipid metabolism were distinct (Table 1). These genes included Agpat6, a lysophosphatidic acid acyltransferase involved in TG and phospholipid synthesis, and Mgat1 and Mgat2, two monoacylglycerol O-acyltransferases that catalyze the synthesis of diacylglycerols (Figs. 5A and 6C). Interestingly, the expression of Pnpla3 (adiponutrin), a patatin-like phospholipase domain-containing lipase that has been strongly associated with fatty liver in humans and mice (42, 43), was lower in the SIRT1 het mice (Fig. 6C). These data were consistent with our observations that the SIRT1 het mice had a defective GL/FFA cycle (Fig. 5).

Figure 6.

SIRT1 het mice fed the high-fat (HF) diet display altered gene expression profiles. A) Numbers of differentially expressed gene probes between WT and SIRT1 het mice fed the low-fat (LF) or HF diet. Liver mRNAs were analyzed by mouse whole-genome microarray as described in Materials and Methods. B) Venn diagram representation of the subset of hepatic genes significantly altered by 1.5-fold with HF diet feeding in WT and SIRT1 het mice (n=3, P<0.05). C) Relative mRNA levels of genes in the GL/FFA cycle. D) Relative mRNA levels of genes in glycolysis/gluconeogenesis pathways (n=6). E) Relative mRNA levels of genes in energy metabolism pathways in BAT. Expression of the indicated genes was determined by qPCR (n=6). Lpgat, lysophosphatidylglycerol acyltransferase; Glut, glucose transporter. *P < 0.05; ^#0.05 < P < 0.10.

Table 1.

Top 10 canonical metabolic pathways that are differentially changed in WT and SIRT1 het livers in response to HF diet feeding

Canonical pathway	P		Ratio
	WT	Het	WT	Het
Arachidonic acid metabolism	2.19E-07	1.35E-02	22/191 (0.115)	18/191 (0.094)
Biosynthesis of steroids	9.10E-07	5.99E-01	9/120 (0.075)	2/120 (0.017)
Metabolism of xenobiotics by cytochrome P-450	3.32E-06	2.48E-05	18/175 (0.103)	22/175 (0.126)
Linoleic acid metabolism	5.27E-06	9.68E-03	15/99 (0.152)	12/99 (0.131)
Fatty acid metabolism	6.44E-05	7.81E-03	18/171 (0.105)	19/171 (0.111)
Nitrogen metabolism	5.99E-04	2.20E-01	8/120 (0.067)	5/120 (0.042)
Androgen and estrogen metabolism	6.85E-04	2.71E-02	11/119 (0.092)	11/119 (0.092)
Tryptophan metabolism	1.85E-03	1.03E-03	16/227 (0.07)	23/227 (0.101)
Glycerolipid metabolism	9.80E-03	1.79E-01	11/138 (0.08)	11/138 (0.08)
Glycosphingolipid biosynthesis	1.27E-02	8.01E-02	5/55 (0.091)	5/55 (0.091)

WT and SIRT1 het mice responded similarly in metabolism of xenobiotics by cytochrome P-450 and tryptophan metabolism pathways, but differently in other pathways involved in lipid metabolism and steroid hormone metabolism.

In agreement with our metabolomic data, SIRT1 het mice displayed a modest increase in the expression of genes involved in the gluconeogenesis pathway. For example, glycerol kinase 2, an enzyme that converts free glycerol to glycerol-3-P, thereby enabling free glycerol released from lipolysis to enter the glycolysis/gluconeogenesis pathway (Fig. 5A), was significantly increased in SIRT1 het mice (Fig. 6D). Moreover, glucose-6-phosphatase (G6Pase) and pyruvate carboxylase (Pc), two important gluconeogenesis enzymes, were also induced in these mice. Furthermore, in line with our lipid profile data, it appears that the metabolic rates were significantly induced in brown adipose tissue (BAT). As shown in Fig. 6E, SIRT1 het mice had elevated expression of Dio2, the type 2 iodothyronine deiodinase that functions to activate thyroid hormone by converting the prohormone thyroxine (T4) to bioactive 3,3′,5-triiodothyronine (T3), and elevated expression of UCP1, a brown fat-specific uncoupling protein that functions to produce heat. Together, our results demonstrate that systemic loss of a single copy of the SIRT1 gene in mice increased hepatic gluconeogenesis, enhanced lipolysis in liver and WAT, and induced metabolic activity in BAT. These metabolic alterations eventually result in an inflammatory state and the development of insulin resistance with a high-fat diet. These observations further suggest that SIRT1 het mice may have an increased risk of other related metabolic diseases, such as cardiovascular diseases.

Systemic SIRT1 insufficiency alters steroid hormone metabolism and leads to development of hyperandrogenism with the high-fat diet

In addition to fatty acid metabolism, microarray analyses unexpectedly revealed that SIRT1 het mice had altered expression profiles of genes involved in biosynthesis of steroids and androgen and estrogen metabolism (Table 1). In fact, two highly related pathways of the top 5 canonical pathways, among 766 final differentially expressed gene probes, were C21-steroid hormone metabolism and androgen and estrogen metabolism (not shown). Among the genes in these two pathways, two of the most significantly reduced genes in SIRT1 het mice were 3β-hydroxy-Δ⁵-steroid dehydrogenase-4 (Hsd3b4) and 3β-hydroxy-Δ⁵-steroid dehydrogenase-5 (Hsd3b5). Real-time qPCR confirmed that mRNA levels of Hsd3b5 were significantly decreased in the SIRT1 het mice after 32 wk of high-fat feeding (Fig. 7A). Hsd3b5 is a male-prevalent, testosterone-sensitive gene with hepatic expression that has been negatively associated with steatosis (44, 45). Although the exact in vivo function of this enzyme in steroid hormone metabolism is still unclear, previous studies have reported that it is a male-specific hepatic 3-ketosteroid reductase that can inactivate dihydrotestosterone (DHT), an active androgen (46, 47). Therefore, the dramatic decrease of Hsd3b5 gene expression in SIRT1 het mice suggests that these mice may have androgen inactivation defects. Indeed, one of most intriguing phenotypes of SIRT1 het mice after chronic high-fat-diet feeding was the enlargement of the seminal vesicles, a pair of testosterone-sensitive male reproductive glands that are near the prostate (Fig. 7B, C). In line with this observation, SIRT1 het mice displayed significantly elevated levels of serum testosterone (Fig. 7D). To explore the possibility that the rise in serum testosterone levels in SIRT1 het mice, relative to WT mice, is due to their defective hepatic androgen inactivation, but not due to increased androgen production, we analyzed the expression of a number of genes involved in androgen metabolism in the testis and the adrenal gland, two primary steroidogenic tissues in male mice. Consistent with the observation that the SIRT1 het mice have normal levels of SIRT1 in testis (Fig. 1A, B), the expression levels of 3 genes involved in testosterone synthesis—including Cyp11a1, a cholesterol desmolase that catalyzes the first and rate-limiting step in the synthesis of the steroid hormones; Cyp17, a key enzyme in the steroidogenic pathway that produces androgens; and Hsd17b3, an enzyme that catalyzes the conversion of androstenedione to testosterone—were normal in the testis of SIRT1 het mice (Fig. 7E, left panel). The expression of Hsd17b1, an enzyme that reduces testosterone, was also normal. Moreover, it appears that the elevated circulating testosterone concentrations in SIRT1 het mice even induced the expression of testicular levels of Cyp19, the aromatase that converts testosterone to estradiol, thereby inactivating it (Fig. 7E). Similar results were observed in the adrenal gland (Fig. 7E, right panel). Together, these observations suggest that the increased serum testosterone levels in the SIRT1 het mice might be induced by defective hepatic androgen inactivation and clearance.

Figure 7.

Systemic SIRT1 insufficiency leads to hyperandrogenism after 32 wk of high-fat-diet feeding. A) SIRT1 het mice have decreased mRNA levels of Hsd3b5, a hepatic gene involved in inactivation of androgen. Expression of Hsd3b5 was determined by qPCR; n = 6. B, C) SIRT1 het mice display increased seminal vesicle size (B) and percentage of body weight (Bwt; D); n = 15–20. D) Serum testosterone levels are increased in SIRT1 het mice; n = 15–20. E) Relative mRNA levels of testicular and adrenal steroid hormone metabolism genes; n = 6. Cyp, cytochrome P-450. *P < 0.05.

DISCUSSION

As the most conserved mammalian sirtuin, SIRT1 has been under intense study in the past 2 decades. While it has been reported that SIRT1 is a vital regulator for many aspects of energy metabolism in a number of metabolic tissues (5, 6), the mechanism by which SIRT1 coordinates metabolic processes in these tissues and modulates whole-body energy homeostasis in response to nutrient signals remains to be elucidated. We show in the present study that mice lacking one copy of the SIRT1 gene are more susceptible to chronic high-fat feeding-induced metabolic abnormalities than WT controls, including obesity, insulin resistance, inflammation, and hepatomegaly. Our study also uncovers novel functions of SIRT1 in hepatic and adipose GL/FFA cycles and systemic steroid hormone homeostasis. These findings present additional evidence that solidify SIRT1's function as an essential systemic metabolic sensor, which will likely provide a molecular basis for the development of therapies for a number of human metabolic diseases. Our findings will also offer crucial information that could help to evaluate the efficacy of therapeutic interventions with SIRT1 small-molecule regulators.

Several previous reports from our group and others have demonstrated that SIRT1 plays important roles in the regulation of hepatic metabolism and systemic inflammation. For example, we have previously reported that on high-fat-diet feeding, mice lacking SIRT1 in hepatocytes develop hepatomegaly/steatosis and hepatic inflammation (10), and mice with SIRT1 deletion in myeloid-linage cells display elevated inflammation and insulin resistance (48). Other groups have shown that manipulation of SIRT1 levels in the liver affects the expression of a number of genes involved in glucose and lipid metabolism (11). In addition, systemic administration of SIRT1-activating molecules, including polyphenol resveratrol and several synthetic pharmacologic activators, has been shown to protect mice against high-fat-diet-induced obesity and metabolic abnormalities (49–51). Moreover, modest overexpression of SIRT1 in mice has been shown to result in a protective effect against high-fat-diet-induced hepatic steatosis, glucose intolerance, and inflammation (52, 53). Our observations from SIRT1 het mice confirm many of these findings, and suggest that SIRT1 mediates a fine balance between energy influx and energy expenditure.

However, it is important to note that although SIRT1 het mice in our study display liver enlargement with high-fat feeding, they do not develop a higher degree of hepatic steatosis than control animals (Fig. 4) and have normal expression profiles of many lipogenesis genes (Fig. 6 and data not shown). In fact, a number of metabolites in the GL/FFA cycle are depleted in their liver tissues, whereas highly proinflammatory lysolipids, such as LPCs (54), are accumulated (Fig. 5); this suggests that hepatic lipid metabolism in these mice is influenced by metabolic processes occurring in other tissues, such as adipose tissues and muscle. These observations are in contrast to those recently reported by Xu et al. (55). Using a SIRT1 het line generated in a separate study (29), Xu et al. observed similar phenotypes, such as obesity, decreased energy expenditure, hepatomegaly, and systemic inflammation in these mice when chronically fed a high-fat diet. However, the researchers detected hepatic steatosis and liver damage, as well as increased expression of several key genes involved in lipogenesis. One possible factor contributing to the discrepancy between our observations and those of Xu et al. (55) may be the difference in the components of the high-fat diets. Our high-fat diet (D12327; Research Diets) contains 20% kcal from protein, 40% kcal from sucrose, and 40% kcal from coconut and soybean oil. In contrast, the high-fat diet in the Xu et al. (55) study (D12331; Research Diets) contains 16% kcal from protein, 25.5% kcal from sucrose, and 58% kcal from hydrogenated coconut oil, and their medium-fat diet (Labdiet 5015; PMI Nutrition, St. Louis, MO, USA) has completely different sources and percentages of protein, carbohydrate and fat. As a result, in our study, the SIRT1 het mice become defective in the hepatic GL/FFA cycle in response to insulin resistance in other peripheral tissues, while SIRT1 het mice suffered hepatic steatosis and increased liver damage in the study carried out by Xu et al. (55) Nevertheless, both studies confirm that single-copy deletion of the SIRT1 gene in mice is sufficient to drive increased obesity, inflammation, and other metabolic abnormalities in mice on high-fat diet challenge.

The GL/FFA cycle, also known as the TG/FFA cycle, has been increasingly recognized as an important signaling pathway for many metabolic expenditure-associated biological processes in response to various stresses, such as burns, cancer, and exercise (39, 40). Alterations in GL/FFA cycling have been implicated in the pathogenesis of a number of human diseases, including obesity, type 2 diabetes, and cancer (39, 40). Our observations that SIRT1 het mice have altered GL/FFA cycles in liver tissues and WAT suggest that SIRT1 may exert its protective effects against high-fat-diet feeding through modulation of this pathway. It would be of great interest to test whether SIRT1 also regulates thermogenesis in BAT, another process governed by the GL/FFA cycle.

Another intriguing observation in our study is that SIRT1 het mice develop hyperandrogenism after chronic high-fat feeding. Although our present data suggest that defects in hepatic androgen inactivation, but not increased androgen production, may be responsible for this phenomenon, additional experiments will be needed to analyze the rates of androgen production and clearance in these mice. It is worth noting that hyperandrogenism has been tightly linked to metabolic disorders in women with polycystic ovary syndrome (56). This syndrome is characterized by hyperandrogenism, disordered gonadotropin secretion, insulin resistance, and, frequently, obesity. It has been shown that antagonizing androgen action ameliorates some features of metabolic syndrome. Our observations that SIRT1 het mice have increased testosterone levels suggest that androgen excess may partially underlie some of the abnormalities observed in these mice. Additional experiments with castrated/adrenalectomized SIRT1 WT and het mice will be helpful to test this possibility.

How loss of single copy of SIRT1 gene in mice leads to alternations in energy homeostasis and steroid hormone metabolism remains to be determined. As a NAD⁺-dependent protein deacetylase, SIRT1 has been shown to play a central role in the regulation of metabolism through deacetylation of a number of transcriptional factors, cofactors, metabolic enzymes, and signaling molecules involved in the fuel metabolism (5, 6). However, we failed to observe notable differences of the acetylation levels of many of these factors (data not shown). SIRT1 has also been shown to work together with the AMP-activated protein kinase (AMPK), a fuel-sensing enzyme that is essential in cellular energy homeostasis, to modulate metabolism, inflammation, and mitochondrial function (reviewed in Ruderman et al.; ref. 57). A number of studies have demonstrated that SIRT1 and AMPK reciprocally activate each other and share many common target molecules (53, 59–63). Moreover, resveratrol-mediated protective effects against high-fat diet involve AMPK (64, 65). Furthermore, in muscle cells, the AMPK/SIRT1 axis appears to be under control of adiponectin, an antidiabetic adipokine, through Ca²⁺ signaling and changes of the NAD⁺/NADH ratio (58). Yet we again failed to detect any significant alternations in the phosphorylation levels of the T172 residue of AMPK, a hallmark of an activated AMPK enzyme, in both liver and muscle of SIRT1 het mice (data not shown). Additional experiments with more quantitative methodologies will be needed to pinpoint the final mechanisms in these mice. Nevertheless, our observations suggest that the phenotypical abnormalities in SIRT1 het mice may result from combinations of modest changes in a variety of signaling pathways and that the SIRT1 het mouse may be a more relevant animal model than the knockout mouse for human physiology and disease associated with SIRT1 gene polymorphisms.

In summary, we have shown that SIRT1 is an essential factor in the regulation of systemic metabolic homeostasis and steroid hormone metabolism. Our findings not only confirm previously reported roles of SIRT1 in obesity, insulin resistance, and inflammation, but also reveal novel functions of SIRT1 in systemic GL/FFA cycles and steroid hormone metabolism. These findings suggest that new therapeutic strategies designed to modulate SIRT1 activity may be beneficial for a number of human obesity-associated diseases.