Evidence that S6K1, but not 4E-BP1, mediates skeletal muscle pathology associated with loss of A-type lamins

The mechanistic target of rapamycin (mTOR) signaling pathway plays a central role in aging and a number of different disease states. Rapamycin, which suppresses activity of the mTOR complex 1 (mTORC1), shows preclinical (and sometimes clinical) efficacy in a number of disease models. Among these are Lmna−/− mice, which serve as a mouse model for dystrophy-associated laminopathies. To confirm that elevated mTORC1 signaling is responsible for the pathology manifested in Lmna−/− mice and to decipher downstream genetic mechanisms underlying the benefits of rapamycin, we tested in Lmna−/− mice whether survival could be extended and disease pathology suppressed either by reduced levels of S6K1 or enhanced levels of 4E-BP1, two canonical mTORC1 substrates. Global heterozygosity for S6K1 ubiquitously extended lifespan of Lmna−/− mice (Lmna−/− S6K1+/− mice). This life extension is due to improving muscle, but not heart or adipose, function, consistent with the observation that genetic ablation of S6K1 specifically in muscle tissue also extended survival of Lmna−/− mice. In contrast, whole-body overexpression of 4E-BP1 shortened the survival of Lmna−/− mice, likely by accelerating lipolysis. Thus, rapamycin-mediated lifespan extension in Lmna−/− mice is in part due to the improvement of skeletal muscle function and can be phenocopied by reduced S6K1 activity, but not 4E-BP1 activation.


Introduction
The mechanistic target of rapamycin (mTOR) kinase is a central regulator of growth factor signaling and metabolism [1][2][3] and is closely linked to aging and a wide range of diseases [4]. The mTOR protein is a component of two complexes, the mTOR complex 1 (mTORC1) and complex 2 (mTORC2). The bestdefined substrates of the mTORC1 complex are S6K1 (ribosomal protein S6 protein kinase 1) and 4E-BP1 (eIF4E (eukaryotic translation initiation factor 4E)binding protein 1) [5], both of which are important in the control of translation initiation [6]. Activation of the mTORC1 signaling cascade results in the phosphorylation of downstream substrates such as S6K1 and 4E-BP1, which in turn affect protein synthesis. Specifically, phosphorylation of S6K1 results in its activation and the subsequent phosphorylation of ribosomal protein S6 (rpS6), as well as other components of the translation machinery, whereas phosphorylation of 4E-BP1 disrupts its binding to eIF4E, freeing this initiation factor to promote cap-dependent translation [7].
Mutations in A-type lamins are associated with a range of dystrophic and progeroid syndromes in humans [8], including dilated cardiomyopathy with conduction-system disease (CMD1A) [9], Emery-Dreifuss muscular dystrophy (EDMD2/3) [10], familial partial lipodystrophy [11]. and Hutchinson-Gilford progeria syndrome [12]. Lmna −/− mice were generated nearly two decades ago to better understand the role of A-type lamins in nuclear organization and disease [13]. Despite a more recent report that these mice express a modified LMNA transcript and are actually hypomorphs [14], these mice have served as a workhorse disease model for the dystrophic syndromes. Lmna −/− mice rapidly develop dilated cardiomyopathy and muscular dystrophy, resulting in death by 6-8 weeks [13]. Recently, we demonstrated that rapamycin reverses elevated mTORC1 signaling in multiple tissues and rescues pathogenesis of dilated cardiomyopathy, Figure 1 Whole-body genetic knockdown of S6K1 extends survival of Lmna −/− mice. (a) Western blots of S6K1 protein expression in muscle (gastrocnemius), heart, liver, subcutaneous (Sub.) fat and brown adipose tissue (BAT). Representative blot derived from two mice for each genotype. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as loading control. Relative S6K1 protein levels (normalized to GAPDH) were quantified. Each value is mean ± s.e.m. for the number of mice indicated in parentheses. P-values were derived from unpaired two-tailed Student's t-test. skeletal muscle dystrophy and lipodystrophy, and as a result doubles the survival of Lmna −/− mice [15,16]. Given the multiple downstream substrates of mTORC1, it is critical to identify those that have a role in Lmna-mediated pathogenesis. Furthermore, understanding about how rapamycin suppresses mTORC1 and rescues pathologies in Lmna −/− mice may lead to better strategies to target the mTORC1 pathway in other disease states, as well as aging itself.
Here, we assess the role of two key mTORC1 substrates, S6K1 and 4E-BP1, in Lmna −/− mice, finding that S6K1 is the more important mediator of pathogenesis and early mortality. Moreover, our findings indicate that a reduction in S6K1 activity in muscle underlies the benefits to these mice, indicating that altered skeletal muscle function is a contributor to mortality in the Lmna −/− mouse model of laminopathies.
Previously, we observed that rapamycin treatment led to improved maintenance of adiposity in Lmna −/− mice, underscoring its critical role in their survival [16]. Specifically, enhanced lipolysis in WAT and deficient thermogenesis in BAT are at least partially rescued by rapamycin in Lmna −/− mice. Unlike rapamycin, however, the body weight (BW) and fat content of longlived Lmna −/− S6K1 +/ − mice are indistinguishable from Lmna −/− littermate controls (Figure 1c and d, Supplementary Figure S2B, C, E and F). These results are further supported by the observation that the lipid metabolism of long-lived Lmna −/− S6K1 +/ − mice is not changed. For instance, we reported that levels of adipose triglyceride lipase (ATGL) are elevated in WAT of Lmna −/− mice, and here show that they are unaltered by S6K1 heterozygosity (Supplementary Figure S3A). Similarly, the low levels of uncoupling protein 1 (UCP1) in BAT are not rescued (Supplementary Figure S3B). This is in spite of the fact that phosphorylation of ribosomal protein S6 (rpS6) protein (p-S6 S240/244), a well-documented readout of mTORC1 and S6K1 activity, is suppressed in both WAT and BAT (although not statistically significant) in Lmna −/− S6K1 +/ − mice (Supplementary Figure S3A, B). Thus, reduced S6K1 signaling in adipose tissues does not likely underlie the enhanced survival of Lmna −/− S6K1 +/ − mice.
Analysis of BW and composition revealed that double knockout mice (Lmna −/− S6K1 −/− ) were consistently smaller than age-matched Lmna −/− littermates (Supplementary Figure S2H and I). This smaller body size, which is analogous to the phenotypes of S6K1 −/− mice compared to littermate controls [18], may exacerbate dystrophic phenotypes of Lmna −/− mice and thus override the potentially beneficial effects derived from reduced S6K1 signaling (see Discussion). Given a lower-than-expected number of double knockout mice obtained from crosses (Supplementary Figure S1A Rapamycin induces glucose intolerance as indicated by glucose tolerance test (GTT) in wild-type mice (WT; Lmna +/+ S6K1 +/+ ) [19] (Supplementary Figure S4A). In contrast, Lmna −/− mice are hypoglycemic [20] and have increased sensitivity upon glucose infusion (Supplementary Figure S4A). After 1 week of rapamycin treatment, Lmna −/− mice had reduced glucose tolerance compared with untreated controls; however, they were comparable to untreated Lmna +/+ mice (Supplementary Figure S4A). After 3 weeks of treatment, rapamycin induces glucose intolerance in Lmna −/− mice (Supplementary Figure S4A). Given the lack of response to glucose challenge in Lmna −/− mice, it is possible that the glucose intolerance and hyperglycemia induced by rapamycin may play a positive role in life extension in Lmna −/− mice. S6K1 −/− mice are long-lived, and in these mice glucose tolerance is improved in 600-day-old females but impaired in 8week-old mice [18]. In contrast to rapamycin-treated Lmna −/− mice (Supplementary Figure S4A), however, long-lived Lmna −/− S6K1 +/ − mice have comparable glucose profiles in response to GTT to Lmna −/− mice (Supplementary Figure S4B). These results suggest life extension in Lmna −/− S6K1 +/ − mice is not owing to altered glucose sensitivity and support the hypothesis that, whereas rapamycin targets multiple metabolic tissues to extend lifespan in Lmna −/− mice [15,16], a reduction in S6K1 may only provide a subset of those benefits.

Improved skeletal muscle function in long-lived Lmna
Since partial knockdown of S6K1 improves the survival of Lmna −/− mice (Figure 1b), and neither cardiac function nor metabolic parameters are improved in long-lived Lmna −/− S6K1 +/ − mice (Supplementary Figure S4B), we further evaluated whether genetic reduction of S6K1 rescues skeletal muscle deficits in Lmna −/− mice. If elevated mTORC-S6K1 activity contributes to the muscle dystrophy and this phenotype also reduces survival, then reduced S6K1 signaling may be exerting its protective effect in skeletal muscle. Our findings are consistent with this hypothesis. For instance, muscle function was improved in long-lived Lmna −/− S6K1 +/ − mice evaluated by rotarod at 4 and 5 weeks of age with double-mutant mice displaying both enhanced latency to fall and increased maximum speed reached ( Figure 2a).
Consistent with a prior report, we find reduced rotarod performance in Lmna −/− mice [15]. (Figure 2a). Muscular dystrophy in Lmna −/− may also be driven by reduced levels of peroxisome proliferator-activated receptor gamma coactivator-1-alpha (PGC-1α), a master regulator of mitochondrial biogenesis [22]. (Figure 2b). We interrogated Lmna −/− S6K1 +/ − mice to see if PGC-1α levels were affected by reduced S6K1 activity. Interestingly, we found PGC-1α protein levels were restored in muscle tissue (gastrocnemius) of longlived Lmna −/− S6K1 +/ − mice (Figure 2b and c). Furthermore, mitochondrial protein subunit 4 of cytochrome c oxidase complex (Cox IV), a nuclear-encoded mitochondrial protein of the electron transport chain, is reduced in skeletal muscle of Lmna −/− mice and also restored in Lmna −/− S6K1 +/ − mice (Figure 2b and c). This improved mitochondrial function in skeletal muscle of long-lived Lmna −/− S6K1 +/ − mice is further supported by a trend toward increased nuclear respiratory factor 1 (NRF1) and increased mitochondrial transcription factor A (mtTFA) (Figure 2c). NRF1 is a PGC-1α-inducible transcription activator for the gene encoding cytochrome c [23]. PGC-1α also could activate the expression of mtTFA through the coactivation of NRF1-mediated transcription [24]. mtTFA is a nuclear-encoded gene product that is imported into the mitochondrial for mitochondrial biogenesis, including the replication and transcription of mitochondrial DNA [24]. However, the amount of Relative PGC-1α and Cox IV protein levels (normalized to GAPDH) were quantified. (c) Signaling through the mTORC1 pathway, indicated by p-S6, in muscle (gastrocnemius) tissue of long-lived Lmna −/− S6K1 +/ − mice. Relative p-S6 levels (normalized to S6) and relative desmin, PGC-1α, Cox IV, NRF1 and mtTFA protein levels (normalized to GAPDH) were quantified. Each value is mean ± s.e.m. for replicate numbers indicated in parentheses, and statistical significance was determined by unpaired two-tailed Student's t-test.
Chen-Yu Liao et al . 5 desmin was not reduced in skeletal muscle of long-lived Lmna −/− S6K1 +/ − mice (Figure 2c). Of note, we detect a modest trend toward reduced phosphorylation of the S6K1 substrate, ribosomal protein S6 (rpS6), although it was not statistically significant ( Figure 2c). We speculate that phosphorylation is maintained by S6K2 and/or other reported kinases [25,26]. and it is also possible that the rescue in skeletal muscle may occur through phosphorylation of other S6K1 substrates.
The rescue of PGC-1α in skeletal muscle was not observed in cardiac tissue (Supplementary Figure S5B), further supporting our findings that cardiac function is not improved in long-lived Muscle-specific S6K1 knockdown improves survival in Lmna −/− mice Given our findings that Lmna −/− mice with reduced S6K1 activity have enhanced survival and improved skeletal muscle (but not cardiac and metabolic) function, we decided to target S6K1 specifically in skeletal muscle to provide a more direct test of the role of elevated mTORC1 signaling in this tissue. Thus, we generated a muscle-specific S6K1 knockout mice in the Lmna −/− context to test this hypothesis (Supplementary Figure S1B). A breeding strategy utilizing Lmna +/ − mice [13], mice bearing one allele floxed S6K1 gene (S6K1 flox/+ or S6K1 f/+ mice) [27], mice expressing Cre recombinase under the control of the muscle creatine kinase promoter (Ckmm-Cre) [28] was utilized to generate muscle-specific S6K1 knockdown in Figure S1B). As expected, western blot analysis of a broad spectrum of tissues showed that the S6K1 protein level is reduced in skeletal muscle and to some extent in heart, but not in liver, WAT or BAT of  Figure S1B) for the remainder of the study.
As with the Lmna −/− S6K1 +/ − mice and unlike rapamycin-treated Lmna −/− mice [16], BW and fat content of long-lived Lmna −/− S6K1 f/f Ckmm mice was comparable to Lmna −/− mice (Figure 3c and d). Previously, we showed that elevated lipolysis in WAT, indicated by higher levels of ATGL, may underlie the lipodystrophic phenotype in Lmna −/− mice [16]. Interestingly, ATGL is further elevated in long-lived S6K1 f/f Ckmm mice (Supplementary Figure S7A), whereas thermogenic protein UCP1 is indistinguishable in BAT (Supplementary Figure S7B). The reasons for this elevated lipolysis are unclear given that we did not observe a dramatic change in BW and adiposity (Figure 3c and d). Nonetheless, consistent with the data from long-lived Lmna −/− S6K1 +/ − mice (Figure 1c   Relative p-S6 levels (normalized to S6) and relative PGC-1α, Cox IV, NRF1 and mtTFA protein levels (normalized to GAPDH) were quantified. Each value is mean ± s.e.m. for replicate numbers indicated in parentheses, and statistical significance was determined by unpaired two-tailed Student's t-test.
Ablation of S6K1 improves survival of Lmna − / − mice 8 creatine kinase promoter is also partially expressed in cardiac tissue [28]. This raised a caveat that extended survival of Lmna −/− S6K1 f/f Ckmm mice may be attributable to improved cardiac function due to genetic ablation of S6K1. However, the rescue of PGC-1α and Cox IV protein levels in muscle tissue (Figure 4b) were not observed in heart tissue of long-lived Lmna −/− S6K1 f/f Ckmm mice (Supplementary Figure S8). In addition, even though long-lived Lmna −/− S6K1 +/ − mice had reduced levels of phospho-rpS6 in cardiac tissue (Supplementary Figure S5B), no improvement of cardiac function was observed (Supplementary Figure S5A) and PGC-1α protein levels were unaffected (Supplementary Figure S5B).  Figure S9B, C, E, F), a finding consistent with our previous study showing that wild-type mice overexpressing 4E-BP1 have smaller body size and less adiposity [29].
We evaluated lipolysis in WAT of Lmna −/− 4E-BP1 mice, as measured by levels of ATGL and monoacylglycerol lipase (MGL), as well as thermogenesis, as indicated by levels of UCP1 in BAT. Surprisingly, lipolysis was further elevated in WAT and thermogenesis was further suppressed in BAT of Lmna −/− 4E-BP1 mice (Figure 5d and e). Increased lipolysis and suppressed thermogenesis suggest that Lmna −/− 4E-BP1 mice may experience further increased energy expenditure. This Lmna −/− 4E-BP1 mouse model also further supports our hypothesis that elevated energy expenditure is one factor that shortens the survival of Lmna −/− mice [16].
In summary, overexpression of 4E-BP1 exaggerates the small phenotype in Lmna −/− mice and further enhances early mortality. These results also echo our previous study that rapamycin did not affect phosphorylation nor total levels of 4E-BP1 protein in Lmna −/− mice, especially in heart and muscle tissues [15]. Thus, all these results further suggest that rapamycin extends survival of Lmna −/− mice at least in part by mediating the mTORC1-S6K1 branch of the pathway, but not the mTORC1-4E-BP1 branch.

Discussion
Rapamycin-mediated mTORC1 inhibition rescues cardiac, skeletal muscle and adipose function and robustly enhances survival in Lmna −/− mice [15,16], a model for the cardiomyopathy and muscular dystrophy associated with human mutations in LMNA [8].
Here, we identified a key molecular mechanism underlying lifespan extension by rapamycin in Lmna  Figure S10).
We found that reducing S6K1 activity improved the survival of Lmna −/− mice, likely by improving muscle function and possibly by rescuing PGC-1α protein levels in skeletal muscle. Given that PGC-1α is a master co-transcriptional factor regulator of mitochondrial biogenesis and mitochondrial function, [22,30]. which declines with age [31], this is likely a critical factor underlying muscle-specific defects in Lmna −/− mice. Increased PGC-1α expression in longlived Lmna −/− mice (Lmna −/− S6K1 +/ − and Lmna −/− S6K1 f/f Ckmm mice) presumably induces mitochondrial subunits Cox IV and drives more ATP generation. Consistently, PGC-1α expression and ATP production are reduced in fibroblasts derived from Hutchinson-Gilford progeria syndrome, a lethal genetic disease caused by point mutation in LMNA, and both can be rescued by methylene blue, a mitochondrial-targeting antioxidant [32]. Of note, long-lived S6K1 −/− mice Lmna −/− 4E-BP1 (n = 6) mice. Relative UCP1 levels (normalized to GAPDH) were quantified. Each value is mean ± s.e.m. for replicate numbers indicated in parentheses, and statistical significance was determined by unpaired two-tailed Student's t-test. ns, no significance. display improved rotarod performance [18] as well as elevated PGC-1α expression in both muscle and adipose tissues [33]. Life extension by dietary restriction, which also suppresses S6K1 signaling, improves rotarod performance in mice [34] and increases both PGC-1α and Cox IV expression in skeletal muscle [35]. Furthermore, PGC-1α is also required for the dietary restriction-induced increases in mitochondrial gene expression and mitochondrial density in skeletal muscle [36]. These independent studies, coupled with our findings, further demonstrate the conserved roles of mTORC1-S6K1 signaling in normal aging and Lmna gene-derived laminopathies, particularly with respect to muscle function.
Although reducing mTORC1-S6K1 activity improves muscle function and extends lifespan in both normal and Lmna −/− mice, many studies show that intact mTORC1-S6K1 signaling is required for muscle function. For instance, muscle-specific inactivation of mTOR leads to severe myopathy, resulting in premature death [37]. Skeletal muscle-specific ablation of raptor (mTORC1) can cause metabolic changes, reduces mitochondrial biogenesis and results in muscle dystrophy [38]. During muscle hypertrophy, S6K1 is required for skeletal muscle force production [39]. Furthermore, mice deficient in ribosomal protein S6 phosphorylation (rpS6 P − / − mice), a downstream target of S6K1, suffer from muscle weakness that reflects a growth defect and energy deficit [40]. These studies based on genetic mouse models indicate that mTORC1 pathway is a crucial regulator of skeletal muscle growth and function. Notably, mTORC1 signaling is aberrantly elevated in skeletal muscle in Lmna −/− mice [15], and the reduced mTORC1 activity in Lmna −/− S6K1 +/ − mice may compensate for this aberrant elevation. Double knockout (Lmna −/− S6K1 −/− ) mice may not be long-lived because of effects in other tissues. For instance, these mice exhibit an even greater reduction in BW (Supplementary Figure S2H, I), which is likely deleterious to survival in an already cachexic background (Supplementary Figure S2G). Together, these findings suggest that the levels of mTORC1 activity must be well-balanced-too much contributes to pathology in disease states (and possibly normal aging), whereas too little impairs the ability of tissues to respond to stress and/or regenerate.
The role of mTORC1-S6K1 in cardiovascular aging is well documented [41,42]. Specifically, suppression of mTORC1-S6K1 signaling, either by rapamycin or dietary restriction, improves cardiac functions in normal mice [43][44][45]. Previously, we also found that suppression of S6K1 signaling by rapamycin improves cardiac function in Lmna −/− mice [15]. Intriguingly, improved cardiac function was not observed in long-lived Lmna −/− S6K1 +/ − mice (Supplementary Figure S5). Although it may be possible that complete inhibition of S6 kinase activity is required for restoration of cardiac function, it is also possible that the pathology evoked by aberrant mTORC1 signaling in this tissue occurs through phosphorylation of other mTORC1 substrates. We have previously shown that autophagy is impaired in heart tissue of Lmna −/− mice and speculate that this may be mediated through phosphorylation of ULK1, or other autophagy-related substrates [15].
It has been suggested that the regulation of mTORC1-mediated fat metabolism involves signaling through S6K1 and 4E-BP1 [46][47][48]. More direct evidence of mTORC1-S6K1 pathway's role in fat metabolism has been obtained from S6K1 −/− mice [18,33], where a lean phenotype and resistance to obesity in part owing to increased lipolysis [33]. Furthermore, mice with adipose-specific deficiency of raptor, an mTORC1 component, show a lean phenotype, which is related to increased energy expenditure and increased mitochondrial uncoupling [49]. However, we did not observe a significant difference in the adiposity of longlived Lmna −/− S6K1 +/ − and Lmna −/− S6K1 f/f Ckmm mice compared with controls. Thus, altered adiposity may not be linked to enhanced survival in Lmna −/− S6K1 +/ − and Lmna −/− S6K1 f/f Ckmm mice [16].
To our surprise, overexpression of 4E-BP1 shortened the survival of Lmna −/− mice, especially in females ( Figure 5, Supplementary Figure S9). The results are unexpected, considering the beneficial effects of 4E-BP1 in metabolism of wild-type mice. For instance, wild-type mice with increased 4E-BP1 expression are resistant to high-fat diet-induced obesity [29]. Conversely, the amount of WAT is significantly decreased in male 4E-BP1 −/− mice [50]. Combined genetic ablation of 4E-BP1 and 4E-BP2, that is, hyperactivation of mTORC1 signaling throughout body (4E-BP1 and 2 −/− mice), also led to decreased lipolysis, increased TGA accumulation, and increased insulin resistance [51]. A follow-up study showed that mouse embryonic fibroblasts derived from mice lacking 4E-BPs accumulate more fat by suppressing ATGL, the enzyme involved in the first step of triglyceride hydrolase activity in lipolysis [52]. In line with this regulatory mechanism, short-lived Lmna −/− 4E-BP1 mice also experienced elevated ATGL in WAT (although only a trend) as well as decreased UCP1 in BAT ( Figure 5). Thus, we speculate that the deleterious effect of 4E-BP1 overexpression in Lmna −/− mice might be a result of further increased energy expenditure, causing a reduction in adiposity and a shorter lifespan [16]. In sum, these findings indicate that altered mTORC1-4E-BP1 signaling cannot explain rapamycin-mediated life extension in Lmna −/− mice.
Based on results from our genetic interventions, the life extension by rapamycin in Lmna −/− mice is likely mediated by suppressing mTORC1-S6K1 signaling, which improves mitochondrial activity by rescuing the suppressed PGC-1α expression in skeletal muscle. However, mTORC1-4E-BP1 signaling may not be involved in lifespan extension. It appears that 4E-BP1 overexpression may reduce the adiposity in already lipoatrophic Lmna −/− mice. Although a reduction in S6K1 activity leads to a 20~30% lifespan extension in Lmna −/− mice, modulation of S6K1 cannot recapitulate all the beneficial effects by rapamycin, which doubles their survival [15,16]. This suggests that mTORC1 phosphorylation of multiple substrates may account for different aspects of toxicity. As stated the cardiac toxicity may reflect mTORC1-mediated control of autophagy [15] and the adipose phenotypes may be mediated through inhibition of UCPs in BAT [46]. Therefore, to recapitulate the effects of rapamycin, it may be necessary to target multiple downstream substrates. We posit that from a therapeutic perspective, with respect to laminopathies, other disease and possibly aging itself, it may be necessary for full efficacy to target mTORC1 directly, or perhaps upstream components of the signaling pathway. Furthermore, human diseases mimicking Lmna −/− mice-muscular dystrophy and progeroid syndromes-targeted inhibition of S6K1 might be an effective therapeutic approach, whereas not as powerful as rapamycin but might limit associated immunosuppressive side effects.

Mice husbandry
Mice were bred and maintained under specific pathogen-free conditions. A tail biopsy was performed in weaning mice at 3 weeks of age for genotyping by polymerase chain reaction with specific primers. All the animals had food and water ad libitum and were kept in standard temperature conditions (22°C) and 12:12-h light-dark cycles. All animal care and experimental procedures were approved by the Institutional Animal Care and Use Committee at the Buck Institute for Research on Aging.

Lifespan study
All mice on the lifespan studies were monitored everyday from 4-week of age until the end of life. BW was measured every other day in all mice. No mice used for the lifespan study were used for any other biochemical or metabolic tests.

Body composition
Whole-body composition (fat mass, lean mass and free water) analysis was conducted weekly using quantitative nuclear magnetic resonance machine (EchoMRI-2012; Echo Medical Systems, Houston, TX, USA) starting at 4 weeks of age.

Rapamycin injection
Mice were injected intraperitoneally with 8 mg kg − 1 BW rapamycin (LC Laboratories, Woburn, MA, USA) or vehicle every other day according to our previous study [16]. A stock solution of 50 mg ml − 1 rapamycin was prepared in 100% ethanol and stored at − 20°C. Rapamycin was then diluted in vehicle (5% polyethylene glycol and 5% Tween 80) before injection. The vehicle control consisted of the same volume ethanol.

GTT
GTT performed on non-anesthetized animals. Mice were fasted with access to water for 16 h (overnight) before being given a single injection intraperitoneally with 20% glucose at a dose of 2 g kg − 1 of BW. The tail prick was used for blood glucose measurement at time points 0, 30, 60, 90, 120 and 180 min with an ACCU-CHEK Aviva glucometer (Roche Diagnostics, Dallas, TX, USA) and the test strips. The mice for rapamycin study were on 129Sv-C57BL/6J genetic background [16].

Transthoracic echocardiography
Mouse cardiac structure and contractility were imaged using VisualSonics Vevo2100 at 5-6 weeks of age. These experimental approaches were adapted from our previous study [15,43,53]. In brief, mice were initially anesthetized with~2.25% isofluorane, placed on a heating pad (37°C) and maintained under light anesthesia so as to maintain the highest possible heart rate during data collection. LV parameters were obtained from M-mode recordings. LV end diastolic diameter and systolic diameter were calculated from the mean of at least three separate cardiac cycles. To calculate myocardial performance index and the E/A ratio, pulsed wave Doppler measurements were taken in the four-chamber view of the heart. All standard cardiac parameters were calculated off-line using VisualSonics software (v1.3.0).

Rotarod
In forced motor activity on a rotating rod (rotarod) assays to assess motor and neurological function was adapted from our previous study [15]. In brief, the day before the actual rotarod testing, the mice were placed on the rotarod set to a beginning speed of 5 r.p.m., with an acceleration rate of 0.1 r.p.m. s − 1 . The max speed was set at 80 r.p.m.. They were allowed to practice the rotarod five times, one repetition every 5 min. If the animal did not fall off, each repetition would end at 5 min. The animals were tested 24 h after the practice day, and the procedure was the same except scores were recorded. The score for each repetition was the time in second and the speed in r.p.m. until the animal fell off the rotarod. The average of five repetitions was used to score the sessions.

Statistical analysis
All statistical analyses were conducted using GraphPad Prism 6 (GraphPad, La Jolla, CA, USA). The survival curves were completed using a Kaplan-Meier curve. We used a logrank (Mantel-Cox) test to perform the statistical analyses of the survival curves. All the other data are shown as mean ± s.e.m. The statistical significance of differences between two groups was determined using unpaired, two-tailed Student's t-test.