S6K1基因是糖尿病的终结基因？

Figure 1 Reduced adiposity in S6K1-/- mice.   Full legend   High resolution image and legend (59k)

Analysis of adipocytes in epididymal fat pads by either scanning electron microscopy or by haematoxylin and eosin staining showed a sharp reduction in size, with some adipocytes exhibiting a multi-locular-like phenotype (Fig. 1c and see below). Morphometric analysis revealed that S6K1^-/- adipocytes were consistently smaller compared with adipocytes from wild-type mice (Fig. 1c), with an average 71% decrease in size (Supplementary Fig. 1c). The decrease in fat accumulation in S6K1^-/- mice was not due to less food intake, which was increased 17% when food consumption was adjusted for body weight (Fig. 1d). Moreover, on the basis of normal fasting and feeding glucose levels and no increase in ketone body formation³ (Table 1), S6K1^-/- mice did not seem to be starving, nor was there an alteration in adaptive thermogenesis (data not shown). This raised the possibility that triglycerides were being broken down rather than stored in WAT. Consistent with this, basal rates of lipolysis were fivefold higher in S6K1^-/- versus wild-type adipocytes, although norepinephrine-induced fatty acid and glycerol release increased in both genotypes in a dose-dependent manner to the same final extent (Fig. 1e; see also Supplementary Fig. 1d). Moreover the metabolic rate was greatly enhanced in S6K1^-/- mice, as indicated by the 27% increase in oxygen consumption versus wild-type mice (Fig. 1f). The respiratory exchange ratio (RER) of 0.713 0.004 for wild-type mice and 0.709 0.003 for S6K1^-/- mice (P < 0.01) showed that both animals were largely using fatty acids as an energy source. Thus the failure to accumulate fat with age in S6K1^-/- mice seems to stem from a sharp increase in lipolytic and metabolic rates.

These increased responses, combined with the finding that levels of circulating triglycerides and free fatty acids (FFAs) were similar in both genotypes (Table 1), suggested that in S6K1^-/- mice triglycerides were being rapidly oxidized in WAT and/or muscle. WAT is not an energy-consuming tissue; however, electron micrographs revealed multi-locular adipocytes, with mitochondria of increased size and number—phenotypes that are absent in wild-type adipocytes (Fig. 2a). Consistent with this, analysis of the messenger RNA levels of genes involved in energy combustion and oxidative phosphorylation were found to be strongly increased in S6K1^-/- adipocytes compared with wild-type adipocytes, including uncoupling protein 1 (UCP1), UCP3, carnitine palmitoyltransferase 1 (CPT1) and PPAR- co-activator 1- (PGC1-) (Fig. 2a; see also Supplementary Fig. 2a). Mitochondrial content was also affected in S6K1^-/- skeletal muscle (Fig. 2b), consistent with increased expression of peroxisome proliferator-activated receptor / (PPAR-/), PGC1-, UCP3 and CPT1 (Fig. 2b; see also Supplementary Fig. 2b). As S6K1^-/- mice have reduced WAT and increased oxidative phosphorylation, this raised the possibility that S6K1^-/- mice are protected against diet-induced obesity, which is linked to the oxidative phosphorylation pathway^6-10. Indeed, when S6K1^-/- mice were challenged with a high fat diet (HFD) (60% total calories derived from fat, 4,057 kcal kg^-1) weight accumulation was significantly reduced compared with wild-type mice during the 4-month feeding period (Figs 2c, d). Although food intake in S6K1^-/- mice is similar to that of wild-type mice, when normalized to body weight they consume 44% more food (Supplementary Fig. 2c). Metabolic rate measured by indirect calorimetry increased for both genotypes on the HFD, with the effect more pronounced for S6K1^-/- mice (compare Figs 2e and 1f). Moreover RER remained unchanged in S6K1^-/- mice on a HFD (0.708 0.002), whereas in wild-type mice it increased from 0.713 0.004 to 0.729 0.002 (n = 6, P < 0.01), indicating an increase in carbohydrate relative to fatty acid oxidation. To determine the weight gain corresponding to fat, mice were subjected to magnetic resonance imaging (MRI) analysis at 2-month intervals while on the HFD. A transverse section through the abdomen showed a marked reduction in fat depots, depicted by the less intense signal (Supplementary Fig. 2d). MRI assessment of total fat, corrected for body weight, revealed that the body fat index for S6K1^-/- mice increased by 20% over this period, whereas it doubled for wild-type mice (Fig. 2f). Thus, consistent with the increase in oxidative phosphorylation, S6K1^-/- mice are protected against diet-induced obesity.

Figure 2 Increased mitochondria and resistance to diet-induced obesity.   Full legend   High resolution image and legend (79k)

Although S6K1^-/- mice have a high metabolic rate when on a HFD, they exhibit a threefold increase in circulating FFAs (Table 1), consistent with an increase in fat on their skin and hair (data not shown). As increased circulating FFAs are implicated in insulin resistance^11-13 and S6K1^-/- mice are hypoinsulinaemic³, it seemed likely that they would become insulin resistant on the HFD. On a NCD both S6K1^-/- and wild-type mice exhibited similar fasting levels of glucose (Fig. 3a), although S6K1^-/- mice were more insulin sensitive, as indicated by faster glucose clearance in the insulin tolerance test (Fig. 3a). In contrast, both genotypes displayed increased hyperglycaemia on a HFD, although the effect was more pronounced in wild-type mice (Fig. 3b). However, despite the increases in glucose and in FFAs, S6K1^-/- mice remain as insulin sensitive on a HFD as on a NCD (Fig. 3b versus a). In the case of wild-type mice, insulin resistance on a HFD (Fig. 3b) can be explained by persistently elevated insulin levels (Table 1), inducing insulin receptor desensitization, as measured by reduced receptor auto-phosphorylation in response to insulin (Fig. 3c). In S6K1^-/- mice insulin levels fail to rise on a HFD (Table 1), consistent with higher insulin receptor auto-phosphorylation (Fig. 3c). However, insulin receptors still desensitize in S6K1^-/- mice on a HFD (Fig. 3c), probably due to the large increase in FFA levels (Table 1). That insulin receptors desensitize in S6K1^-/- mice but remain insulin sensitive suggests that absence of S6K1 facilitates insulin signalling downstream of the insulin receptor. To test this we monitored protein kinase B (PKB) phosphorylation, as insulin sensitivity is tightly linked to the phosphatidylinositol-3-OH kinase (PI(3)K)–PKB signalling pathway¹⁴. As with insulin receptor auto-phosphorylation in wild-type mice on a HFD, insulin-induced PKB phosphorylation was suppressed in fat, liver and muscle compared with wild-type mice maintained on a NCD (Fig. 3d). However, there was no significant effect on PKB phosphorylation in S6K1^-/- mice, regardless of diet or tissue (Fig. 3d). This suggested that S6K1 elicited a selective inhibitory effect on PKB activation at a point downstream of the insulin receptor, consistent with little effect on mitogen activated protein kinase and S6K1 activation in wild-type animals on a HFD (see below and data not shown).

Figure 3 Enhanced insulin sensitivity and insulin signalling in the absence of S6K1.   Full legend   High resolution image and legend (64k)

Recently, we demonstrated in Drosophila that dS6K negatively regulates dPKB activity in a cell-autonomous manner¹⁵. To test whether this was the mechanism responsible for the observed responses, S6K1 levels were reduced with small interfering RNAs. The results show that lowering S6K1 levels potentiates insulin-induced PKB phosphorylation, with no effect on insulin receptor auto-phosphorylation (Fig. 3e), suggesting that the target of inhibition resided downstream of the insulin receptor. Indeed, reduction of S6K1 levels is paralleled by a decrease in phosphorylation of S307 and S636/S639 of insulin receptor substrate 1 (IRS1) (Fig. 3e), sites shown to inhibit PI(3)K binding to IRS1 (ref. 4) and to be involved in insulin resistance in skeletal muscle cells from patients with type 2 diabetes⁵. Thus, removal of S6K1 can facilitate insulin signalling in a cell-autonomous manner.

Hyperglycaemia, hyperaminoacidaemia and hyperlipidaemia are associated with obesity and insulin resistance¹⁶; however, the role of increased nutrients in insulin action is not well understood^17-19. As S6K1 is activated by nutrients^20-22 and acts negatively on PI(3)K signalling, this raised the possibility that on a HFD S6K1 is involved in inducing insulin resistance. This hypothesis is supported by the reversal of amino acid inhibition of insulin-induced PI(3)K signalling by rapamycin, which inhibits mTOR²³, an immediate upstream S6K1 kinase². Consistent with this hypothesis, phosphorylation of S6K1 T389, S6 S240/S244, IRS1 S307 and IRS1 S636/S639 is highly elevated in wild-type mice maintained on a HFD compared with wild-type mice maintained on a NCD (Fig. 4a, b). Furthermore, the increase in IRS1 S307 and S636/S639 phosphorylation is absent in S6K1^-/- mice on a HFD (Fig. 4b). Under these conditions there were no apparent alterations in IRS1 levels (Fig. 4b). These findings suggest that nutrient-induced S6K1 activation acts to suppress insulin signalling through modulating IRS1 S307 and S636/S639 phosphorylation. To test this further, two genetic models of obesity were examined: K/K A^y and ob/ob mice^{24, 25}. The results show that the K/K A^y and ob/ob mice maintained on a NCD have elevated S6K1 T389, S6 S240/S244, IRS1 S307 and IRS1 S636/S639 phosphorylation as compared with wild-type mice on a NCD (Fig. 4c). Moreover, in contrast with PKB S473 phosphorylation in adipose and muscle (Fig. 3d), insulin stimulates S6K1 T389 phosphorylation to even higher levels in wild-type animals on a HFD compared with a NCD (Fig. 4d), potentially further suppressing PI(3)K signalling. Thus, either nutritionally or genetically driven obesity leads to the upregulation of S6K1, which may in turn act to suppress PI(3)K signalling, contributing to insulin resistance.

Figure 4 S6K1 activation in obesity.   Full legend   High resolution image and legend (42k)

The results presented here indicate that S6K1^-/- mice are protected against obesity and insulin resistance due to the upregulation of the oxidative phosphorylation pathway and increased insulin sensitivity. Enhanced oxidative metabolism is consistent with the increase in mitochondria number, as well as the induction of genes that control the oxidative phosphorylation pathway. That S6K1^-/- mice remain insulin sensitive despite high circulating FFAs may be explained by the strong protection against metabolic syndrome by overexpression or activation of PPAR-^6-8. Despite the increase in the oxidative phosphorylation pathway and reduced insulin levels, circulating glucose and FFA levels still rise in S6K1^-/- mice; however, these animals remain sensitive to insulin, and PI(3)K signalling is unaffected. This observation may be explained by the loss of S6K1, whose activation by either nutrients or insulin leads to increased IRS1 serine phosphorylation (Fig. 4e). In the case of insulin, this is mediated by a negative feedback loop triggered by PKB phosphorylation of the TSC1/2 tumour suppressor complex, leading to Rheb activation and stimulation of S6K1 (ref. 26). Thus, in a homeostatic setting, as nutrients and amino acids are consumed, mTOR/S6K1 activity would decrease, restoring PI(3)K signalling, whereas the incessant supply of nutrients associated with the obese state would lead to constitutive activation of mTOR/S6K1 and desensitization of insulin signalling (Fig. 4e). Taken together the results suggest that S6K1 may have a central function along with other signalling components in development of obesity and insulin resistance, and may be an important drug target in the treatment of patients suffering from these pathological disorders.

Methods
Mice S6K1^-/- mice were generated as previously described3. Male C57BL/6J, K/K A^y and ob/ob (C57BL/6J background) mice were obtained from E. Janvier, CLEA Japan Inc., and The Jackson Laboratory, respectively.

Metabolic studies At 10 weeks of age male mice were placed on either a NCD (diet number 3807, KLIBA-NAFAG) or HFD ad libitum (diet D12492, Researchdiets) and monitored for 24 weeks. Body weight was recorded weekly and food intake was measured every second day for 15 consecutive days. Insulin tolerance tests, oxygen consumption, RER measurements and quantification of blood metabolites were performed as previously described²⁷.

Histology and morphometric analysis of tissues Adipose tissue was analysed by haematoxylin and eosin staining as previously described²⁷. Morphometric analysis of epididymal WAT from 500 or more cells from three different animals per genotype was performed with ImageJ software (NIH). Adipose and plantaris muscle tissue were prepared as described for scanning and transmission electron microscopy²⁷.

Magnetic resonance imaging analysis MRI experiments were carried out on a Biospec 47/30 spectrometer (Bruker Medical) at 4.7 T equipped with a self-shielded 12-cm bore gradient system²⁸. Animals were anaesthetized with 1.5% isoflurane (Abbott). Adipose tissue was measured with an optimized turbo-RARE2 imaging sequence. Acquisition parameters were: repetition delay (TR) = 250 ms; echo delay (TE) = 8.6 ms; RARE factor = 32 (effective echo time 73.1 ms); number of averages = 8; slice orientation transverse, image matrix = 128 128 pixels; field-of-view = 3.5 3.5 cm; slice thickness = 1.2 mm (contiguous). Fat pad volumes were assessed with an in-house software algorithm based on IDL software package (Research Systems Inc.). Body fat indices were calculated by dividing adipose tissue weight by body weight.

Lipolysis in isolated adipocytes Primary adipocytes were prepared from epididymal fat pads as described previously²⁹. Cells were incubated for 30 min at 37 °C with or without norepinephrine (Sigma-Aldrich SARL) at the indicated concentrations.

Real-time quantitative RT–PCR Total RNA was extracted from frozen tissue samples or cells using the RNeasy kit (Qiagen). Complementary DNA was synthesized from total RNA with the SuperScript First-Strand Synthesis System (Invitrogen) and random hexamer primers. The real-time polymerase chain reaction (PCR) measurement of individual cDNAs was performed using SYBR green dye to measure duplex DNA formation with the Roche Lightcycler system and normalized to the expression of either -actin or 18S ribosomal RNA. The primers and probes used in the real time RT–PCR were the following: UCP1 sense 5'-GGCCCTTGTAAACAACAAAATAC-3', antisense 5'-GGCAACAAGAGCTGACAGTAAAT-3'; UCP3 sense 5'-ACTCCAGCGTCGCCATCAGGATTCT-3', antisense 5'-TAAACAGGTGAGACTCCAGCAACTT-3'; mCPT1 sense 5'-TTGCCCTACAGCTCTGGCATTTCC-3', antisense 5'-GCACCCAGATGATTGGGATACTGT-3'; mPPAR- sense 5'-CTCTTCATCGCGGCCATCATTCT-3', antisense 5'-TCTGCCATCTTCTGCAGCAGCTT-3'; PGC-1 sense 5'-AAGTGTGGAACTCTCTGGAACTG-3', antisense 5'-GGGTTATCTTGGTTGGCTTTATG-3'.

Measurement of insulin receptor and IRS1 phosphorylation in vivo After a 6-h fast mice were injected intravenously with 0.75 U kg^-1 insulin (Eli Lilly) or equal volume of vehicle. All indicted tissues were collected in liquid nitrogen 5 min after injection. Protein extracts from tissue samples were analysed as described³⁰. Antibodies were from Santa Cruz (anti-insulin receptor and anti-S6K1 antibodies), Upstate Biotechnology (anti-phosphotyrosine and anti-IRS1 antibodies) and Cell Signaling (anti-PKB, anti-phospho-PKB Ser 473, anti-phospho-IRS1 Ser 636/639, anti-phospho-S6K Thr 389 and anti-phospho-S6 240/244 antibodies). Antibodies to S6 and phospho-IRS1 Ser 307 were from J. Mestan and Y. Le Marchand-Brustel, respectively.

RNA interference RNA interference (RNAi) duplexes corresponding to human S6K1 (5'- AAGGGGGCTATGGAAAGGTTT-3') were purified, annealed and transfected into HeLa cells using oligofectamine (Invitrogen). After 60 h cells were deprived of serum overnight and either lysed directly or stimulated with 200 nM insulin for 30 min. The effect of RNAi on S6K1 expression and PKB phosphorylation was measured by western blot analysis. Cell lysates were incubated for 4 h with anti-IRS1 or anti-insulin receptor antibody pre-absorbed on protein A Sepharose at 4 °C, and analysed by western blot analyses after gel electrophoresis.

Statistical analysis Data are presented as mean s.e.m. The main and interactive effects were analysed by analysis of variance (ANOVA) factorial, repeated measurements or by one-way ANOVA followed by Bonferroni t-test (MRI analysis). Differences between individual group means were analysed by Fisher's PLSD test. Analyses were performed using Statview Software (Brainpower).