Introduction

The epidemic of obesity is frequently attributed to the dual adverse effects of nutrient overabundance and diminished physical activity. While the vast majority of younger and middle-aged individuals are able to maintain normoglycaemia in the face of obesity-induced insulin resistance by a compensatory increase in glucose-stimulated insulin secretion, this ability diminishes with age so that dysglycaemia and diabetes ensue [1, 2]. Accordingly, obesity and the insulin resistance that it engenders are viewed as inexorably linked pathophysiological states of the modern human era that underlie the rapid increase in the incidence of type 2 diabetes. These same phenomena are paradoxically also observed in hibernating mammals as they ready themselves for the nutrient deprivation of winter. Indeed, during this preparatory phase, animals typically double their body weight [3] and increase their circulating insulin concentration more than fourfold [4]. However, despite these dramatic changes, plasma glucose levels in these animals do not rise [5].

Given the role of silent information regulator 2 (Sir2) and its mammalian orthologue, Sirt1, in allowing for survival and life extension in the setting of energy deprivation [69], we hypothesised that enhancing the activity of Sir2 would enable a transition from the obese, insulin-resistant diabetic phenotype to one that more closely resembles mammalian torpor or hibernation. With this in mind, we studied the db/db mouse, a rodent model that develops obesity, insulin resistance and ultimately beta cell failure to recapitulate many of the features of type 2 diabetes in humans. In addition, we took advantage of a new generation of SIRT1-activating compounds currently in clinical development that act by accelerating the SIRT1 catalysed deacetylation [10, 11].

Methods

Animal experiments

Thirty-six 6-week-old male db/db (BKS.Cg-Dock7m +/+ Leprdb/J) mouse models of diabetes and 12 age-matched db/m (Dock7m +/+ Leprdb, heterozygote from the same colony) mice were purchased from Harlan Laboratories (Indianapolis, IN, USA). At 8 weeks of age, the db/db and db/m mice were randomised to receive the SIRT1 activator, SRT3025 [12, 13] (gift of Sirtris/Glaxo Smith-Kline, Collegeville, Pennsylvania, USA) milled in chow (3.18 g/kg) or regular chow.

Mice were housed in a temperature-controlled room (22°C) with a 12 h:12 h light–dark cycle and free access to food and water at the St Michael’s Hospital Animal Research Vivarium (Toronto, ON, Canada). After 12 weeks of treatment, animals were killed at the end of a 12 h dark (feeding) period when urine and blood samples were collected and multiple organs, including liver, pancreas and brain were harvested and either fixed in formalin or snap frozen and stored at −80°C. The mediobasal hypothalamus was subsequently isolated as previously described [14].

All animal studies were approved by the St Michael’s Hospital Animal Ethics Committee in accordance with the Guide for the Care and Use of Laboratory Animals (National Institutes of Health Publication No. 85-23, revised 1996).

Metabolic studies and hormone measurements

Metabolic studies and hormone measurements were undertaken as previously reported [15] with measurements of blood glucose, food intake, HbA1c, insulin, C-peptide and free T3. The HOMA-IR was used to determine insulin resistance. Insulin tolerance tests were performed at the beginning of the 12th week on fasted animals [16]. See electronic supplementary material (ESM) Methods.

Indirect calorimetry, activity and temperature

The metabolic rate, motion and respiratory exchange ratio (RER) of mice were measured by indirect calorimetry as previously reported [17], as was body temperature [18]. See ESM Methods.

Histochemistry and immunohistochemistry

Pancreases and livers were fixed, processed, stained and quantified as previously reported [19]. See ESM Methods.

Real-time quantitative RT-PCR

Real-time quantitative RT-PCR was used to determine the relative expression levels of mRNAs. See ESM Methods.

Western blot

The in vivo lysine deacetylase activity of SRT3025 was assessed by examining kidney tissue where protein acetylation is constitutively present in diabetic animals [20]. See ESM Methods.

Insulin sensitivity was assessed in liver tissue by immunoblotting for phosphorylated and total insulin receptor substrate with antibodies directed against phosphorylated IRS-1/2 and total IRS-1 (sc-17195 and sc-8038, respectively; Santa Cruz Biotechnology, Dallas, TX, USA). SIRT1 abundance was similarly assessed in liver by immunoblot (antibody no. 2028, Cell Signaling, Boston, MA, USA) with equivalent protein loading determined by beta actin. See ESM Methods.

Statistics

Data are expressed as means ± SEM. Between-group differences were analysed by one way ANOVA with Fisher’s Protected Least Significant Difference test post hoc. All statistics were performed using GraphPad Prism 5 for Mac OS X (GraphPad Software Inc, San Diego, CA, USA). A p value of <0.05 was considered statistically significant.

Results

Animal characteristics

Diabetic db/db mice gained substantially more weight than db/m mice during the study period with db/db mice that had received the SIRT1 activator, SRT3025, gaining more weight than mice not receiving the compound (Table 1). Despite their greater weight gain, food intake among SRT3025-treated db/db mice was lower than that of their untreated counterparts.

Table 1 Animal characteristics

Immunoblot

Consistent with its mechanism of action as a lysine deacetylase, SRT3025-treated animals displayed substantially reduced lysine acetylation in the in vivo setting compared with their untreated counterparts (ESM Fig. 1).

SIRT1 protein was detected in the livers of all four animal groups but in greater abundance in those db/db mice that had received SRT3025 (ESM Fig. 2).

Energy expenditure, body temperature and activity

SIRT1 activation lowered energy expenditure in both db/m and db/db mice, and this was most evident during the dark phase (Fig. 1). Reduction in body temperature was also evident but was lower in db/db mice receiving SRT3025 where a prominent circadian pattern was also noted. Physical activity was reduced in db/db mice but lowered further in animals that received SRT3025 compared with their non-treated counterparts. While this may have been in part due to the greater body mass of mice receiving the compound, SIRT1 activation similarly lowered activity in db/m mice whose weight was unchanged.

Fig. 1
figure 1

(a) Heat production was reduced in db/db compared with non-diabetic db/m mice and reduced in both animal groups by the administration of SRT3025 (SRT). (b) Body temperature was lower in db/db than in db/m mice and fell further in db/db mice that had received SRT3025. (c, d) A circadian pattern in body temperature was noted in db/m and to a greater extent in db/db mice that had received SRT3025, whereby body temperature began to dip during the latter part of the dark phase, returning to its baseline in the early hours of the light phase. (e) Activity was reduced in db/db compared with non-diabetic db/m mice, and reduced in both animal groups by the administration of SRT3025. (f, g) RER was increased in db/db mice that had received SRT3025 compared with untreated diabetic mice. db/m control, blue; db/m + SRT3025, orange; db/db control, red; db/db + SRT3025, green. *p < 0.05 vs db/m control, p < 0.05 vs db/db control, p < 0.05 vs db/m + SRT3025

When compared with untreated db/db mice, those that had received SRT3025 showed a higher RER throughout a 24 h period (Fig. 1).

Plasma free T3 concentrations differed between db/m and db/db mice but were unaffected by SRT3025 treatment (ESM Fig. 3).

Sirt1 activation ameliorates hyperglycaemia in db/db mice

Within 1 week of receiving SRT3025, non-fasting blood glucose levels were lower in treated db/db mice in comparison with db/db mice that had received regular chow (Fig. 2). Fasting glucose levels were also lower in SRT3025-treated db/db mice, although this only became manifest after 4 weeks. Both SRT3025 and regular chow-treated db/m mice remained normoglycaemic throughout the study, with no difference in blood glucose between the two groups in either the fasting or non-fasting states (Fig. 2).

Fig. 2
figure 2

(a, b) Non-fasting and (c) fasting blood glucose concentrations. Non-fasting blood glucose was lower in SRT3025-treated db/db mice after 1 week, while fasting glucose became lower by the second month of the study. Non-diabetic db/m mice were unaffected by drug treatment. db/m control, blue; db/m + SRT3025 orange; db/db control, red; db/db + SRT3025, green. *p < 0.01 vs db/db control mice

In alignment with the blood glucose findings, HbA1c measured after 4 weeks and at the end of the study was approximately twofold higher in untreated db/db mice but approached levels of normoglycaemic db/m animals in db/db mice that had received SRT3025. HbA1c levels were unaffected by SRT3025 in non-diabetic db/m mice (Table 1).

Plasma insulin and insulin resistance are increased with Sirt1 activation in db/db mice

At 6 weeks of age, db/db mice are characteristically obese and insulin resistant, maintaining normoglycaemia by compensatory hyperinsulinaemia. By 10–14 weeks, however, the circulating insulin concentration falls and blood glucose rises. As expected, we observed these changes in untreated db/db mice. By contrast, animals that had received SRT3025 had markedly augmented plasma insulin concentrations in relation to their untreated counterparts (Fig. 3). However, the C-peptide:insulin ratio was reduced in these animals suggesting that diminished hepatic clearance of insulin may also contribute to the hyperinsulinaemia seen in these animals (ESM Fig. 4).

Fig. 3
figure 3

(a) Fasting plasma insulin, (b) insulin tolerance test and (c) HOMA-IR. When compared with untreated db/db mice, those that had received SRT3025 (SRT) had higher plasma insulin, an increase in plasma glucose in response to insulin administration and higher HOMA-IR. Non-diabetic db/m mice were unaffected by drug treatment. *p < 0.01 vs control db/m mice; p < 0.01 vs control db/db mice db/m control, blue; db/m + SRT3025 orange; db/db control, green; db/db + SRT3025, red

When compared with untreated db/db mice those that had received SRT3025 displayed increased insulin resistance as indicated by the results of the insulin tolerance test, HOMA-IR and measurement of hepatic phosphorylated IRS-1 (Fig. 3, ESM Fig. 5).

Gene expression

In the setting of insulin resistance, impaired insulin signalling fails to repress hepatic gluconeogenesis, thereby contributing to fasting hyperglycaemia in type 2 diabetes. Consistent with the insulin resistance of the db/db mouse, expression of the key gluconeogenic enzymes Pepck and G6pc (also known as G6Pase) mRNA was increased approximately fivefold in db/db mice when compared with db/m controls. By contrast, in mice that had received SRT3025 the overexpression of gluconeogenic enzymes was substantially reduced, consistent with the ability of high insulin concentrations to overcome insulin resistance (Fig. 4).

Fig. 4
figure 4

Liver gene expression of (ad) key gluconeogenic enzymes and insulin-suppressible transcription factors that initiate gluconeogenic gene transcription, G6pc, Pepck, Foxo1 and Foxo6, and (eg) factors associated with fat deposition, Fasn, the fatty acid transporter Cd36 and Srebp1c. mRNA abundance was expressed relative to that of RPLI3a. The ratio, so-derived, was then expressed as fold change relative to that in db/m control mice that was arbitrarily set at 1. G6pc and Pepck mRNA were increased approximately fivefold and Foxo1 and Foxo6 mRNA were increased approximately twofold in db/db mice compared with db/m controls. SRT3025 (SRT) administration led to a reduction in this overexpression in db/db mice to levels similar to those seen in non-diabetic db/m mice and that were unaffected by drug administration. *p < 0.01 vs control db/m mice; p < 0.01 vs control db/db mice

Expression of the lipogenic enzyme fatty acid synthase (FAS), but not the fatty acid transporter cluster differentiation 36 (CD36) was increased in SRT3025-treated db/db mice (Fig. 4).

Consistent with a shift from gluconeogenesis to lipogenesis, changes in mRNA levels of related transcription factors forkhead box O1 (Foxo1), forkhead box O6 (Foxo6) and sterol regulatory element-binding protein-1c (Srebp-1c) were also found. Here, the diabetes-associated overexpression of the gluconeogenic Foxo transcription factors was reduced with SRT3025 (Fig. 4).

SIRT1 activation had no effect on gene expression of any of the gluconeogenic or lipogenic enzymes, or on transcription factors in db/m mice (Fig. 4).

In the hypothalamus, gene expression of Npy and Trh was reduced in both untreated and treated diabetic mice when compared with their non-diabetic db/m counterparts. Pomc and Agrp mRNA, on the other hand, were similar in all four groups (ESM Fig. 6).

Sirt1 activation exacerbates hepatic steatosis

Non-alcoholic liver disease is closely associated with obesity, insulin resistance and type 2 diabetes in population studies and while excessive fat deposition (steatosis) is found in the livers of db/db mice it does not progress to steatohepatitis or cirrhosis [21]. When compared with lean db/m mice, the livers of db/db mice showed marked lipid deposition when stained with Oil Red O (Fig. 5). A further, substantial increase in the extent of hepatic steatosis was seen in animals that had received the SIRT1 activator, SRT3025. In line with these findings, db/db mice that had received the compound showed increased hepatic expression of the transcription factor SREBP-1c and FAS but not the fatty acid transporter CD36 compared with their non-treated counterparts (Fig. 4).

Fig. 5
figure 5

Liver from non-diabetic db/m mice treated with (a) vehicle or (b) SRT3025 (SRT) and diabetic db/db mice treated with (c) vehicle or (d) SRT3025 stained for neutral lipid with Oil Red O, and (e) their quantification. db/db mice show abundant lipid deposition that is further increased in those that had received SRT3025. Non-diabetic db/m mice had minimal lipid that was unaffected by drug. Magnification ×160. *p < 0.01 vs control db/m mice; p < 0.01 vs control db/db mice

Beta cell mass is increased with Sirt1 activation in db/db mice

Early in their development, db/db mice display islet enlargement allowing them to maintain normoglycaemia in the face of insulin resistance. Subsequently, however, the expanded islets involute leading to relative insulin deficiency and hyperglycaemia. When compared with untreated, age-matched db/db mice, SRT3025-treated db/db mice displayed increased beta cell mass, predominantly as a consequence of islet enlargement (Fig. 6). No effect of SRT3025 on beta cell mass was noted in non-diabetic db/m mice.

Fig. 6
figure 6

Pancreases showing immunolabelled insulin in non-diabetic db/m mice treated with (a) vehicle or (b) SRT3025 (SRT) and diabetic db/db mice treated with (c) vehicle or (d) SRT3025, and (e) their quantification. While the islets of db/db mice show reduction in immunostainable insulin (brown) the reverse is seen in those that had received SRT3025. Non-diabetic db/m mice were unaffected by drug treatment. Magnification ×160. *p < 0.01 vs control db/db mice

Betatrophin

Betatrophin is a recently identified peptide hormone, primarily expressed in liver and fat, which promotes pancreatic beta cell proliferation and expansion of beta cell mass to improve glucose tolerance [22]. While no difference in hepatic betatrophin expression was evident between untreated db/db and db/m mice, expression was elevated in db/db mice that had received SRT3025 (Fig. 7).

Fig. 7
figure 7

Betatrophin expression in the livers of experimental animals. mRNA abundance was expressed relative to that of RplI3a. The ratio, so-derived, was then expressed as fold change relative to that in db/m control mice that was arbitrarily set at 1. Betatrophin expression was increased in the livers of db/db mice that had received SRT3025 (SRT) compared with untreated db/db and db/m mice. *p < 0.01 vs control db/m mice; p < 0.01 vs control db/db mice

Discussion

Obesity accompanied by insulin resistance and diabetes is commonly viewed as a consequence of excessive food consumption and reduced physical activity, often erroneously attributed to a lack of will power and discipline. Notably, individuals with type 2 diabetes who lose relatively modest amounts of weight experience substantially improved insulin sensitivity and reductions in blood glucose concentrations. Here we show, however, that obesity, insulin resistance and hyperglycaemia are not inextricably intertwined. While pharmacological activation of SIRT1 lowered blood glucose to near-normal levels, these changes were not accompanied by weight loss and improved insulin sensitivity in db/db mice. Indeed, the reverse occurred with augmented insulin resistance and further weight gain during the study. Rather, blood glucose lowering appeared to be a consequence of substantially increased circulating insulin concentrations and augmented pancreatic islet beta cell mass in association with elevated hepatic expression of the beta cell growth factor, betatrophin [22].

Although imperfect, the db/db mouse recapitulates many of the features of type 2 diabetes in humans. As a consequence of a point mutation in the receptor for the anorexogenic adipokine, leptin, homozygous db/db mice are hyperphagic, obese and insulin resistant, developing hyperinsulinemia in an attempt to maintain glucose homeostasis. As it ages, however, islet beta cell mass becomes reduced so that relative insulin insufficiency and hyperglycaemia ensue. As such, this mouse model resembles obesity with type 2 diabetes in humans where insulin and leptin resistance, rather than their deficiencies, are commonly found [23]. In the present study, the otherwise natural history of beta cell failure in the db/db mouse was averted by SIRT1 activation. Instead of the age-dependent diminution in insulin secretion, mice treated with SIRT1 activator maintained high levels of insulin that were near-commensurate with their requirements, reducing hepatic gluconeogenesis, lowering plasma glucose and achieving an HbA1c that was approximately half that of untreated db/db mice and only slightly higher than non-diabetic db/m mice.

Pancreatic islet beta cells constitutively express SIRT1 that, when upregulated, increase glucose-stimulated insulin secretion in both the cell culture and in vivo settings [24, 25]. The present study suggests that another SIRT1-dependent mechanism may also serve to augment insulin secretion in the face of increased demands. Rather than the age expected decline in beta cell mass, pancreases from db/db mice that had received SRT3025 displayed enlarged insulin-immunopositive islets. To examine possible mechanisms that may account for these salutary findings, we explored changes in the expression of betatrophin, a recently identified hormone that increases beta cell proliferation in response to insulin resistance and hyperglycaemia [22]. Here, we show that hepatic betatrophin expression was increased a further three- to four-fold in SRT3025-treated db/db animals. By comparison, SIRT1 activation did not affect betatrophin expression in non-diabetic db/m mice.

The ability to adapt energy expenditure to nutrient availability is a key survival mechanism. Although there is marked variation among ‘warm-blooded’ animals, considerable energy is used in the maintenance of body temperature and physical activity [26]. Reduction in both, however, can be achieved by bouts of torpor that may manifest as either hibernation or daily torpor. Hibernation tends to be prolonged, often lasting the entire winter, relying critically on the stored fat that has been accumulated in its preliminary phase. Daily torpor, on the other hand, typically lasts only a few hours, is characterised by lesser reductions in body temperature, relies on energy from ingested food rather than stored fat and can be inhibited by leptin [27]. As expected, db/db mice deficient in leptin receptor underwent shallow torpor that extended from the latter part of dark phase to the early light phase, as is typical for nocturnal animals [28]. Consistent with its role in survival during energy deprivation, SIRT1 activation in the present study not only increased the magnitude of temperature reduction but also diminished energy expenditure and activity. While the short duration of these effects is typical of torpor, the fat accumulation that was also noted is more typical of the early phase of hibernation. In db/m mice, a similar, albeit incomplete response to SIRT1 activation was noted whereby mice that received SRT3025 showed reduced energy expenditure and activity but not body temperature. Although, blood glucose was reduced in diabetic mice, other features that accompanied the administration of SRT3025 may be less desirable in diabetic patients. However, given the importance of genetic background to the development of diabetes in both the animal and human setting, how these findings might translate to human type 2 diabetes, where neither hibernation nor torpor occur, remains to be determined.

Heat production and body temperature are reduced during periods of lower activity and food consumption, ostensibly to conserve energy. In the present study, we observed dramatic changes in the amplitude in the circadian rhythm of body temperature in db/db mice treated with SRT3025 when compared with untreated db/db mice or db/m mice that had also received the SIRT1 activator. Together, these findings suggest that while both SIRT1 and leptin are known to modulate circadian rhythms [2932], when combined their effects may be more than additive.

Though currently viewed as a pathophysiological state of disordered lipid and carbohydrate metabolism with considerable associated morbidity [33], the ability to accumulate fat is a normal physiological process that aids survival and averts starvation in the setting of impending food shortage [34]. Indeed, just prior to hibernation, animals gain substantial body and hepatic fat mass. The accumulation of fat in the livers of SRT3025-treated diabetic mice, as seen in the present study, is consistent with the known ability of SIRT1 to activate Liver X receptor alpha [35], the consequences of which would include increased expression of Srebp1c [36], Fasn [37] mRNAs and hepatic steatosis [37, 38], as observed in our study. In addition, changes in FOXO1 expression were also noted in the current study with increased Foxo1 mRNA in db/db mice that was reduced with the administration of SRT3025. These alterations would be expected to lead to changes in the liver’s major metabolic pathways with a reduction in gluconeogenesis and increased lipogenesis [39], findings consistent with the reduced Pepck and G6pc mRNA, along with the increase in Fasn mRNA and increased RER demonstrated in our study. Importantly, however, while increased lipogenesis and hepatic steatosis were features of db/db mice treated with SRT3025, other studies using a similar SIRT1 activating compound, SRT1720, showed reduced fatty liver in diet-induced obese, non-diabetic mice [40, 41]. These findings highlight the importance of genetic background and metabolic state in modulating the response to pharmacological intervention.

Notably, despite the associated insulin resistance, hibernating animals do not develop hyperglycaemia with its attendant glycosuria and insensible energy loss. Suggestive of transition to this more physiological, torpor-like state, obese insulin-resistant db/db mice that received the SIRT1 activator SRT3025 in the present study displayed lower plasma glucose concentrations and did not develop the polyuria seen in untreated animals.

Diminished insulin secretion and enhanced insulin resistance together contribute to the pathogenesis of type 2 diabetes in the db/db mouse and the vast majority of humans. While SIRT1 activation greatly augmented circulating insulin in the present study, insulin resistance, as indicated by both HOMA-IR and IRS-1 phosphorylation, was exacerbated in response to drug administration. These findings are consistent with some, but by no means all, previous studies that have examined the effects of SIRT1 modulation. For instance, Sirt1 overexpression by the systemic administration of adenovirus expressing Sirt1 transcript reduced insulin sensitivity in 6–8-week-old male BALB/c mice [42]. Consistent with this, reducing Sirt1 expression with either small hairpin RNA or antisense oligonucleotide improved insulin sensitivity [42, 43]. By contrast, hepatic Sirt1 deficiency was found to worsen insulin resistance in Sirt1 flox albumin-cre mice [44]. Previous studies have also addressed the effects of pharmacological Sirt1 activation in the diabetes setting. For instance, administration of SRT1720 reduced blood glucose in high-fat diet fed, obese Lep ob/ob mice and while they did not gain weight, the mice received the drug for only 1 week [45]. Moreover, in other studies, mice with high-fat feeding-induced metabolic syndrome (but not diabetes), preliminary data suggest that these animals are protected from weight gain and hepatic steatosis with SRT3025 [46, 13].

By responding to metabolic signals such as leptin, insulin and glucose, the orexigenic, neuropeptide Y (NPY)/agouti-related peptide and thyrotrophin releasing hormone (TRH) neurons of the hypothalamus coordinate changes in food intake and energy expenditure, respectively, that together aim to maintain homeostasis. While reductions in hypothalamic Npy and Trh mRNA and circulating free T3 were seen in db/db mice in the present study, levels in these mice were unaffected by Sirt1 activation.

In summary, SIRT1 activation in db/db mice led to substantial expansion of beta cell mass, increased circulating insulin and improved glycaemic control. These ostensibly beneficial changes were, in this mouse model of type 2 diabetes, also accompanied by weight gain, reduced physical activity, diminished energy expenditure and hepatic steatosis. While in some ways reminiscent of torpor and hibernation, the summative impact of these changes on the long-term progression of type 2 diabetes remain to be unravelled, particularly in the setting of the complex interaction between Sirt1 modulation and genetic factors, as seen in this study.