Loss of growth hormone–mediated signal transducer and activator of transcription 5 (STAT5) signaling in mice results in insulin sensitivity with obesity

Abstract

Growth hormone (GH) has an important function as an insulin antagonist with elevated insulin sensitivity evident in humans and mice lacking a functional GH receptor (GHR). We sought the molecular basis for this sensitivity by utilizing a panel of mice possessing specific deletions of GHR signaling pathways. Metabolic clamps and glucose homeostasis tests were undertaken in these obese adult C57BL/6 male mice, which indicated impaired hepatic gluconeogenesis. Insulin sensitivity and glucose disappearance rate were enhanced in muscle and adipose of mice lacking the ability to activate the signal transducer and activator of transcription (STAT)5 via the GHR (Ghr-391^−/−) as for GHR-null (GHR^−/−) mice. These changes were associated with a striking inhibition of hepatic glucose output associated with altered glycogen metabolism and elevated hepatic glycogen content during unfed state. The enhanced hepatic insulin sensitivity was associated with increased insulin receptor β and insulin receptor substrate 1 activation along with activated downstream protein kinase B signaling cascades. Although phosphoenolpyruvate carboxykinase (Pck)-1 expression was unchanged, its inhibitory acetylation was elevated because of decreased sirtuin-2 expression, thereby promoting loss of PCK1. Loss of STAT5 signaling to defined chromatin immunoprecipitation targets would further increase lipogenesis, supporting hepatosteatosis while lowering glucose output. Finally, up-regulation of IL-15 expression in muscle, with increased secretion of adiponectin and fibroblast growth factor 1 from adipose tissue, is expected to promote insulin sensitivity.—Chhabra, Y., Nelson, C. N., Plescher, M., Barclay, J. L., Smith, A. G., Andrikopoulos, S., Mangiafico, S., Waxman, D. J., Brooks, A. J., Waters, M. J. Loss of growth hormone–mediated signal transducer and activator of transcription 5 (STAT5) signaling in mice results in insulin sensitivity with obesity.

It is well established that growth hormone (GH) decreases insulin sensitivity, as is evident in acromegaly (1, 2). Physiologic bursts of GH have clear insulin antagonistic effects with rapidly decreasing glucose uptake that is sustained for several hours (3). Insulin antagonism is relevant to metabolic control in the fed/unfed cycle and to type 2 diabetes, which is reported to have a 6-fold higher incidence in GH-treated children, although this incidence is very low (4). Insulin sensitivity is thought to relate to longevity because mice with Ghr knockout live longer and are unusually insulin sensitive despite displaying obesity and nonalcoholic fatty liver disease (5). We have previously shown the mechanistic basis for the signal transducer and activator of transcription (STAT)5 driven hepatic steatosis using a panel of Ghr mutant mouse models and hepatic Stat5a/b deleted mice (6). These mice showed marked increase in hepatic triglyceride uptake and synthesis by 4 mo of age as well as morphologic hallmarks of advanced steatosis, which was exacerbated with a high fat diet (6). Obesity evident in these Ghr mutant mouse models owing to lack of GH/STAT5 action is further supported by a reduction in their beige profile of white adipose tissue (WAT) (7).

Insulin insensitivity in Ghr knockout mice is not an IGF1-dependent phenomenon (8), and it initially does not result from elevation of free fatty acids by GH (9). In general, attempts to investigate insulin antagonism by GH have involved administration of supraphysiological doses of GH, or GH transgenic mice. These studies led to the hypothesis that an excess of p85α, the regulatory subunit of phosphatidylinositol-4,5-bisphosphate 3-kinase (PI3K), is induced by GH/STAT5 and acts as a competitive inhibitor for the catalytic p110 subunit of PI3K (10). p85α also activates phosphatase and tensin homolog (PTEN), which reduces PI3K activation through diminished PI(3,4,5)P3 levels (11). Although excess p85α in skeletal muscle is thought to be an important component of GH-induced insulin resistance, its role in liver, where GH is able to oppose insulin-mediated suppression of glucose production is unclear (12, 13). There is also evidence that free p85α is unstable and its induction by other agents actually increases PI3K activity (14). Glucose tracer studies in individuals with GH excess (acromegaly) before and after pituitary microsurgery have indicated that GH-dependent glucose production is largely derived from glycogen breakdown (15). However, direct GH/STAT5-dependent activation of hepatic gluconeogenesis by increasing phosphoenolpyruvate carboxykinase (PCK)-1 expression has also been reported (16), and up-regulation of pyruvate dehydrogenase kinase (PDK)4 with decreased pyruvate decarboxylase activity has been shown to be a GH/STAT5 action (17). A range of other mechanisms for the diabetogenic anti-insulin actions of GH have also been proposed in the last 2 decades, which include decreased insulin receptor and decreased tyrosine phosphorylation of insulin receptor substrate (IRS)1 (18). These proposals are confounded by the ability of GH-activated Janus kinase 1 (JAK2) to induce phosphorylation of insulin targets IRS1, IRS2, and PI3K, as well as glucose transporter (Glut)4 translocation (so-called insulin-like actions of GH), and the effects of hyperinsulinemia itself (19). Adiponectin (ADIPOQ) levels have also been suggested to play an important role in GH-induced insulin resistance as its levels are decreased in GH transgenic mice and increased in Ghr-null (Ghr^−/−) mice (20). Induction of suppressor of cytokine signaling (SOCS) proteins by GH has also been held accountable to explain the anti-insulin actions of GH because SOCS1 and SOCS3 are able to antagonize insulin action (21). Additionally, GH-induced c-Jun N-terminal kinase (JNK)-dependent phosphorylation of Ser307 of IRS1 has been invoked to explain GH-induced insulin resistance (22). Finally, up-regulation of PTEN can also contribute to insulin resistance (23).

Adults and children with GH deficiency exhibit fasting hypoglycemia with increased insulin sensitivity and reduced insulin secretion (24). Likewise, GH-deficient or Ghr^−/− mice are insulin sensitive and hypoglycemic, but the mechanisms involved are unclear and conflicting. In muscle-specific Ghr^−/− mice, down-regulation of insulin receptor protein in the muscle was reported, along with an increase in the inhibitory Ser1101 phosphorylation of IRS1 and in vivo insulin resistance (25). Differential expression of several hepatic genes related to insulin action in both young and old Ghr^−/− mice have also been reported (26). However, in insulin-sensitive young mice, these were restricted to changes in only the transcript level for Irs2, G6pc, and Foxo1.

Given these ranges of mechanisms, we have investigated the anti-insulin GH-mediated signaling pathways by utilizing mouse strains harboring GH receptor (GHR) signaling mutations to identify the molecular mechanisms involved. We have employed assessments of insulin secretion and action, transcript analyses, metabolic clamps, and immunoblotting in relevant tissues in mice comprising: 1) partial loss of GH-STAT5 activation (Ghr-569^−/−), 2) full loss of GH-STAT5 activation with remaining Src kinase activation (Ghr-391^−/−), and 3) total knockout of the receptor (Ghr^−/−) for this purpose.

MATERIALS AND METHODS

Animal studies

Animal studies were performed in accordance with The University of Queensland Animal Ethics Committee, Austin Health Animal Ethics Committee, as well as the Australian Office of the Gene Technology Regulator. Male mice were housed at the Institute for Molecular Biosciences animal facility under a 12-h light/dark cycle at 20 ± 2°C. Animals had ad libitum access to food (meat-free rat and mouse diet; Specialty Feeds, Glen Forrest, WA, Australia) and water, and food was withdrawn only if required for an experimental procedure. Animals were unfed overnight (16 h) unless indicated with animals having access to water ad libitum. All animals passed standard virus screens performed quarterly. The production of gene-targeted mice on a C57BL/6 background (Ghr-569^−/−, Ghr-391^−/−) used in this study has been previously described in detail (27). The Ghr^−/− mice previously generated (28) were kindly provided by Prof. John Kopchick (Department of Biomedical Sciences, Ohio University, Athens, OH, USA) and moved to a C57BL/6 background.

Tissue and plasma collection

Unless specified, all mice were 16–18 wk of age at collection. Male mice were either in the fed state or subjected to withdrawal from food 16 h prior to tissue collection. Mice were anesthetized by placing them into a chamber flushed with CO₂ until unconscious. Plasma samples were obtained by blood collection via cardiac puncture using 27 gauge needles and syringes coated in EDTA. Blood was kept on ice until centrifugation at 1000 g at 4°C for 10 min. Following cardiac puncture, mice were euthanized by cervical dislocation. Tissue samples were dissected and snap frozen in liquid nitrogen. Snap frozen tissues and plasma aliquots were stored at −80°C until used for analysis.

Insulin tolerance test

Insulin tolerance tests were performed on 14–16-wk-old male mice (n = 6/genotype) and 18-mo-old mice (n = 5/genotype) with age-matched wild type (WT). Insulin (MilliporeSigma, Burlington, MA, USA) was freshly diluted in sterile saline and administered by intraperitoneal injection at the doses of 0.25 or 0.5 IU/kg following a 6-h starvation. Blood glucose was measured using a glucometer (Accu-Chek Active; Roche, Basel, Switzerland) by drawing a drop blood from the tail tip before insulin injection and then 10, 20, 30, 45, and 60 min after insulin injection.

Glucose tolerance test

Glucose tolerance tests were performed on 14–16-wk-old male mice (n = 5/genotype) with age-matched WT. A glucose solution (50% wt/vol in sterile water) was administered by intraperitoneal injection following withholding food overnight at a dose of 2 g/kg. Blood glucose was measured prior and then 15, 30, 60, 45, 90, and 120 min after glucose injection.

Pyruvate tolerance test

Pyruvate tolerance tests (PTTs) were performed on 15–17-wk-old male mice (n = 6/genotype) with age-matched WT. Pyruvate solution (25% w/v in sterile saline; MilliporeSigma) was administered by intraperitoneal injection at a dose of 2 g/kg following overnight food withdrawal. Blood glucose was measured before and then 15, 30, 60, 45, 90, and 120 min after pyruvate injection.

Glucose clamp studies

On the day of the experiments, mice were unfed for 5-h and surgery was performed followed by a 20-min recovery period. Hyperinsulinemic and euglycemic clamps were performed as previously described (29). In the current study, an initial priming dose of insulin (200 mU/kg) was followed by constant infusion at a rate of 10 mU/kg per min. Euglycemia was maintained by variable infusion of 6.5–12.5% glucose solution. Basal blood glucose for WT and Ghr-391^−/− was 8.2 ± 0.7 and 5.7 ± 0.2 mM [mean ± sem (n = 4)], respectively. During the clamp, values were 7.7 ± 0.3 and 8.1 ± 0.3 mM for WT and Ghr-391^−/−, respectively. Glucose turnover was calculated using Steele’s steady-state equation. Basal = 0 min (20 min after recovery from surgery); clamp = average of 90, 100, and 110-min plasma samples of the clamp in which the mice were at steady state. At the conclusion of the clamp experiment, peripheral glucose uptake was assessed following intravascular bolus of 2-[1-¹⁴C] deoxyglucose (2-DG) as previously described (29). Briefly, a 2-DG bolus was administered, and plasma was collected at 2, 5, 10, 15, and 30 min thereafter. At the conclusion of the 30-min experiment, mice were overdosed with sodium pentobarbitone (200 mg/kg), and white quadriceps muscle and epididymal adipose tissue were immediately excised, snap frozen in liquid nitrogen, and stored at −80°C for subsequent analysis.

Serum cytokine measurement

Rat and mouse Milliplex Kits (Merck, Darmstadt, Germany) were used for measurement of cytokines, insulin, thyroxine, glucagon, IL, testosterone, and corticosterone. Circulating fibroblast growth factor 1 (FGF1) levels was measured using an ELISA kit (Abcam, Cambridge, MA, USA) as per the manufacturer’s guidelines.

Glycogen assay

Hepatic glycogen was measured in mice unfed for 16 h using the Colorimetric Kit (Abcam) as per the manufacturer’s guidelines, and total protein content was measured by the Bicinchoninic Acid Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA) as per the manufacturer’s guidelines.

Protein extraction and Western blot analysis

All tissues (∼100 mg) were homogenized in 500 μl of cold lysis buffer using handheld Kinematica Polytron Homogenizer (Thermo Fisher Scientific) at full speed. Tissues were homogenized in either lysis buffer (50 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid pH 7.4, 150 mM NaCl, 1% Triton X-100, 1 mM Na₃VO₄, 30 mM NaF, 10 mM Na₄P₂O₇, 10 mM EDTA, or radioimmunoprecipitation assay buffer (150 mM NaCl, 50 mM Tris pH 7.5; 0.5% SDS, 1% nonidet-P (NP)-40 sodium deoxycholate, 30 mM NaF, 10 mM Na₄O₇P₂, 2 mM Na₃VO₄) both supplemented with cOmplete EDTA-free protease inhibitor cocktail (Roche) prior to use. Immunoblotting was performed as described in Barclay et al. and Chhabra et al. (6, 30). Blots were probed by incubating with specific primary antibodies (Supplemental Table S1) overnight shaking in cold room and developed as previously described (30). The PDK4 antibody was a generous gift from Professor Robert Harris at the Veterans Affairs Medical Center, Indianapolis, IN, USA.

RNA extraction and quantitative PCR

Liver and muscle tissue were homogenized in Trizol Reagent (Thermo Fisher Scientific), and RNA was extracted according to the manufacturer’s instructions for both fed and unfed states. RNA was further purified, and DNase was treated using RNeasy Mini Kit clean up protocol (Qiagen, Germantown, MD, USA). RNA was reverse transcribed using Superscript III (Thermo Fisher Scientific) with random hexamers and used for quantitative PCR (qPCR) with SYBR Green mix (Thermo Fisher Scientific) in the 7900 Real Time Cycler (Thermo Fisher Scientific) or ViiA7 Real Time Cycler (Thermo Fisher Scientific). Analysis was performed by calculating the change in C_t values between the gene of interest normalized against the housekeeping gene, β2-microglobulin (β2m), and expressed as relative mRNA expression or as a fold change relative to WT control. For glucose metabolism and adipogenesis RT² Profiler PCR Arrays (Qiagen), RT² First Strand Kit (Qiagen) was used for cDNA synthesis. The ViiA7 Real Real Time Cycler was used to perform the qPCR, and data were analyzed by calculating the change in C_t between the gene of interest normalized against multiple housekeeping genes and expressed as a fold change relative to WT control. Primers for qPCR were designed using Primer Express 2.0 software (Thermo Fisher Scientific).

STAT5 chromatin immunoprecipitation

Chromatin immunoprecipitation (ChIP)-seq data for GH pulse-activated STAT5 in male mouse liver (31) were downloaded from Gene Expression Omnibus (Accession number GSE31578). Raw sequence reads were mapped to the reference mouse genome (mm9) and uploaded to the University of California–Santa Cruz (UCSC) Genome Browser (https://genome.ucsc.edu/cgi-bin/hgGateway) as files for visualization. Also visualized were DNase-I hypersensitive sites (DHSs) based on a set of 72,862 mouse liver DHSs (32).

Statistical analysis

Densitometric quantitation was performed on immunoblots using ImageJ software (National Institutes of Health, Bethesda, MD, USA). Values were normalized to β-Tubulin or glyceraldehyde 3-phosphate dehydrogenase (Gapdh) and represented as arbitrary units. Statistical analyses were performed using Prism software (GraphPad, La Jolla, CA, USA) based on a minimum of 3 independent experiments using Student’s t test for comparison between 2 groups or 1-way ANOVA test followed by a Tukey’s posttest for analyzing means at 1- or 2-way ANOVA followed by a Bonferroni posttest for analyzing means at 2 independent variables. Values of P < 0.05 were considered significant.

RESULTS

Mice lacking ability to generate active STAT5 from GH are insulin sensitive but glucose intolerant. Their ability to secrete insulin is impaired

All Ghr mutant mice (Ghr-391^−/−, Ghr-569^−/−, and Ghr^−/−) had significantly lower overnight fasting blood glucose levels, but mice lacking GH-dependent STAT5 activation, Ghr-391^−/−, and Ghr^−/− displayed the lowest levels. Decreased blood glucose was evident in these obese mice even at 18 mo (or 52 wk) of age (Fig. 1A), but fasting plasma insulin levels were not significantly different from WT mice (Fig. 1B). When challenged with glucose (2 g/kg, i.p.) after an overnight withdrawal from food, all Ghr mutant mice showed an impaired ability to clear glucose (Supplemental Fig. S1A–C). Plasma insulin levels revealed defective insulin secretion in Ghr-391^−/− and Ghr^−/− mice, both lacking GHR-STAT5 activation (Supplemental Fig. S1D).

Figure 1.

Mouse models exhibiting complete abrogation of GH-mediated STAT5 activation are hypoglycemic and insulin sensitive. A) Fasting blood glucose levels in young (14–16 wk) (n = 6/genotype) and old (52 wk) (n = 5/genotype) Ghr mutant male mice with age-matched WT controls. B) Fasting plasma insulin levels in Ghr mutant male mice at 14–16 wk of age (n = 5/genotype). C–E) Insulin tolerance tests in 14–16 wk of age Ghr mutant mice following a 6-h food withdrawal by insulin injection at 0.25 IU/kg, i.p. in Ghr-569^−/− (n = 6/genotype) (C), Ghr-391^−/− (n = 6/genotype) (D), and Ghr^−/− (n = 6/genotype) (E) indicates increased insulin sensitivity in Ghr-391^−/− and Ghr^−/− mice but not in Ghr-569^−/− that retains 30% of GH-mediated STAT5 activation in comparison to age-matched WT mice. IP, immunoprecipitation. Data represented as means ± sem. Analysis by 1-way ANOVA, followed by Tukey’s posttest for analyzing means at 1- or 2-way ANOVA, followed by a Bonferroni posttest for analyzing means at 2 independent variables. *P < 0.05, **P < 0.01 (Ghr mutants vs. WT).

When 0.25 IU/kg of insulin was administered to each mouse line, Ghr^−/− and Ghr-391^−/− mice both showed similar increases in insulin sensitivity, whereas mice with 30% residual GHR-STAT5 signaling (Ghr-569^−/−) were insulin resistant compared to WT (Fig. 1C–E). Insulin insensitivity was strikingly evident in Ghr-569^−/− when 0.5 IU/kg of insulin was administered (Supplemental Fig. S2A), a dose that was lethal for the Ghr-391^−/− and Ghr^−/−, but not WT. This insulin sensitivity was still evident at 18 mo of age despite obesity (Supplemental Fig. S2B–D). Given the insulin sensitivity in Ghr-391^−/− and Ghr^−/−, continued insulin action in the unfed state would be unexpected, where insulin levels do not differ from WT (Fig. 1B).

Complete loss of GH-activated STAT5 results in increased insulin-mediated suppression of gluconeogenesis and insulin sensitivity in peripheral tissues

Reduced gluconeogenic capacity in Ghr-391^−/− and Ghr^−/− relative to age-matched WT was supported by PTT, with no change evident in Ghr-569^−/− mice despite their insulin resistance (Fig. 2A–C). Because Ghr-391^−/− and Ghr^−/− had similar insulin sensitivity, we focused our glucose clamp studies on the Ghr-391^−/−. To assess whole body glucose turnover, mice were subjected to a hyperinsulinemic clamp with tracer kinetics. The GH-mediated STAT5 signaling deficient Ghr-391^−/− were profoundly insulin sensitive as indicated by a 3-fold increase in whole body glucose infusion rate (GIR) required to maintain euglycemia at the same insulin level compared to WT mice (Fig. 2D, E). The elevated GIR was associated with enhanced whole body rate of glucose disappearance and a reduced rate of endogenous glucose production (Fig. 2F, G). This suggests the Ghr-391^−/− have enhanced peripheral sensitivity (i.e., more glucose clearance into muscle and fat) as well as enhanced hepatic insulin sensitivity compared to WT, indicating that liver is able to reduce hepatic glucose production in response to insulin. Clamp administration was also able to show that Ghr-391^−/− have reduced basal insulin levels (Fig. 2H). The rate of glucose uptake under insulin-stimulated conditions in skeletal muscle (white quadriceps) and in epididymal adipose tissue (Fig. 2I, J) was enhanced in Ghr-391^−/− compared to WT, which is consistent with the observed increase in the whole body rate of glucose disappearance.

Figure 2.

A–C) PTTs by intraperitoneal pyruvate injections at 2 g/kg in 15–17-wk-old Ghr mutant male mice following an overnight fast; Ghr-569^−/− (n = 6/genotype) (A), Ghr-391^−/− (n = 6/genotype) (B), and Ghr^−/− (n = 6/genotype) (C) indicates impaired gluconeogenesis in Ghr-391^−/− and Ghr^−/− mice but not in Ghr-569^−/− relative to WT. Hyperinsulinemic and euglycemic clamps with tracer kinetics in 14-wk-old Ghr-391^−/− (n = 4) mice in comparison to age-matched WT male mice indicates increased insulin sensitivity and decreased gluconeogenesis in Ghr-391^−/− mice. D–G) GIR (D), insulin levels during clamp studies (E), endogenous glucose production (F), and rate of glucose disappearance (G) following clamp analysis. H) Basal insulin levels during the clamp studies are indicated. I, J) Peripheral glucose uptake was assessed following intravascular bolus 2-DG with increased glucose uptake in skeletal muscle (white quadriceps) (I) and adipose tissue (epididymal fat) (J). Data represented as means ± sem. Analysis by Student’s t test or 1-way ANOVA, followed by Tukey’s posttest for analyzing means at 1 variable. *P < 0.05, **P < 0.01, ***P < 0.001 (Ghr mutants vs. WT).

Circulating hormone and cytokine profile favors insulin sensitivity in Ghr mutant mice

In order to assess the effect of the lack of GH-mediated STAT5 activation resulting in insulin sensitivity in Ghr mutant mice, we measured plasma levels of relevant cytokines and hormones across the 2 genotypes (Ghr-391^−/− and Ghr^−/−) in comparison to WT mice. No significant difference in fasting plasma glucagon, IFN-γ, IL-2, IL-6, and IL-10 (Supplemental Fig. S3A–E) was observed between Ghr mutant mice and WT, although TNF was elevated in both fasting Ghr-391^−/− and Ghr^−/− (Fig. 3A). Plasma testosterone and corticosterone concentrations remain unchanged (Supplemental Fig. S3F, G), whereas thyroxine (T4) was decreased significantly in Ghr^−/− compared to WT, but not in Ghr-391^−/− (Fig. 3B). Fasting plasma ADIPOQ was elevated in both Ghr-391^−/− and Ghr^−/− (Fig. 3C), which is consistent with a previous report (20), as well as in Laron dwarfs that are obese (33). Finally, the insulin sensitizing agent FGF1 was significantly elevated in both Ghr-391^−/− and Ghr^−/− compared to WT (Fig. 3D).

Figure 3.

Circulating cytokine profile in Ghr mutant mice indicates factors involved with increased insulin sensitivity and decreased glycogenolysis. A–D) Plasma from a 16–18-wk-old male Ghr-391^−/− and Ghr^−/− (n = 6–8/genotype) unfed overnight was analyzed for TNF (A), thyroxine (T4) (B), adiponectin (C), and FGF1 (D) and compared with age-matched WT. No significant differences in fasting plasma glucagon, testosterone, corticosterone, IFN-γ, IL-2, IL-6, and IL-10 were observed between Ghr mutant mice and WT. E–H) Hepatic transcripts for Glut2 (E) and Gbe1 (F) normalized to β2m were measured and immunoblots with quantification performed for Agl and GBE1 (G) and G6pc (H) using β-Tubulin as loading control in overnight fasted Ghr mutants (n = 3–4/genotype) in comparison to age-matched WT. I) Hepatic glycogen levels in 16-h unfed 14-wk-old Ghr mutants (n = 5/genotype) normalized to total protein content. Data represented as means ± sem and analyzed by Student’s t test or 1-way ANOVA, followed by Tukey’s posttest. *P < 0.05, **P < 0.01, ***P < 0.001 (Ghr mutants vs. WT).

Glucose metabolism transcripts are altered in liver of fed and unfed Ghr-391^−/− mice

We have previously analyzed changes in hepatic transcripts from Ghr-391^−/− and Ghr^−/− across both the fed and unfed states (6). However, this analysis did not reveal a basis for the phenotypic differences (particularly suppressed gluconeogenesis) previously reported. Accordingly, we used PCR arrays targeted specifically to determine glucose metabolism attributes (Supplemental Table S2A, B). Unfed Ghr-391^−/− liver transcripts, hexokinase 2, enolase 1, and pyruvate dehydrogenase (PDH) phosphatase catalytic and regulatory subunits (Pdp2 and Pdpr) that are involved in promoting the glycolytic pathway were significantly increased compared to WT (Supplemental Table S2A). Mitochondrial hexokinase-2 expression is induced by peroxisome proliferator-activated receptor γ (Pparg) (34), and Pparg transcripts are increased in the fatty liver of Ghr-391^−/− and Ghr^−/− mice (6), providing glucose carbons for triglyceride synthesis. Similarly, the increase in Glut2 transcript in Ghr-391^−/− may also favor increased uptake of glucose for triglyceride and glycogen synthesis (Fig. 3E).

Transcripts for genes involved in glycogen metabolism were also altered in Ghr-391^−/− with increased 1,4-α-glucan branching enzyme 1 (Gbe1) responsible for glycogen synthesis. Conversely, a decrease in the glycogen debranching enzyme amylo-1,6-glucosidase, 4-α-glucanotransferase (Agl), involved in glycogen degradation was detected. In addition, a modest decrease in glycogen phosphorylase kinase β (Phkb) was identified, although glycogen phosphorylase, liver form transcripts did not change (Supplemental Table S2A). In agreement with the elevated transcript for Gbe1 (Fig. 3F), glycogen branching enzyme protein was increased in Ghr-391^−/− compared to WT (Fig. 3G). However, despite a decrease in the transcript for the glycogen debranching enzyme Agl (Supplemental Table S2A), we could detect no significant change in Agl protein levels (Fig. 3G). Although expression for phosphoglucomutase (Pgm)2 increased modestly, the gluconeogenic transcript for the catalytic subunit of glucose-6-phosphatase (G6pc) was substantially decreased (Supplemental Table S2A). However, we could not identify a significant decrease in G6pc protein expression using antibodies from 2 different vendors (Fig. 3H and Supplemental Fig. S3H), indicating that release of glucose from either glycogen or gluconeogenesis is not significantly affected by loss of GH-activated STAT5. The elevated glycogen level seen in unfed liver Ghr-391^−/− and Ghr^−/− mice following a 16-h food withdrawal (Fig. 3I) is consistent with an increase in glycogen synthesis, resulting in less glucose being released into the periphery.

In unfed Ghr-391^−/− compared to WT mice, there was a small but significant increase in PDH transcripts (Pdhb and Pdha1), which is responsible for conversion of pyruvate to acetyl CoA (Supplemental Table S2A). However, the protein level for PDHA1 and the inhibitory phosphorylated form of pyruvate dehydrogenase E1 component α subunit (PDHE1a) PDHE1a (S300) remained unaltered (Supplemental Fig. S3I). This correlates with unchanged protein expression of PDK4 (Supplemental Fig. S3I) despite transcript for Pdk4 being elevated in unfed Ghr-391^−/− mice. Sirtuin (SIRT)3, which is known to increase PDH activity by promoting deacetylation (35), was elevated at the protein and transcript level in Ghr-391^−/− mice (Supplemental Fig. S4A, B). This would be consistent with the increase in transcripts for several enzymes in the tricarboxylic acid (TCA) cycle (Aco2, Dist, Dld, Idh2, Idh3b, Sdha, Sdhc, Sdhd, Suclg2) (Supplemental Table S2A) in accord with our previously reported metabonomics data in Ghr-391^−/− mice (36).

In relation to gluconeogenic enzymes, transcripts for both cytosolic (Pck1) and mitochondrial PCK (Pck2) were unchanged in Ghr mutant mice as well as PCK2 protein (Supplemental Table S2A and Supplemental Fig. S4C), although only the protein expression of PCK1 was decreased in liver from unfed Ghr-391^−/− mice (Fig. 4A, B). This is congruent with the decreased expression of the cytosolic SIRT2 deacetylase protein (Fig. 4C) and higher extent of acetylation of PCK1, which is known to promote its degradation (Fig. 4A, B). Pyruvate carboxylase (Pcx) transcript level was unchanged in fed liver and marginally increased in liver of unfed Ghr-391^−/− mice (Supplemental Table S2A, B).

Figure 4.

Altered activation and expression of insulin signaling cascade in hepatic tissue of overnight fasted 14-wk-old male Ghr-391^−/− (n = 3–5/genotype) mice with age-matched WT. A–C) Acetylated (Ac) PCK1 levels following immunoprecipitation (IP) with acetylated-lysine antibody and immunoblotted (IB) for PCK1 and Ac-Lysine (A) and total level of PCK1 (B) and SIRT2 (C) with densitometric quantifications. D–F) Immunoblots indicating expression levels of total and active insulin receptor β (D), total and active IRS1 (E), and negative regulator of insulin signaling, PTP1B (F), with densitometric quantification. β–Tubulin expression was used as loading control for all the immunoblots. Ns, not significant. Data represented as means ± sem and analyzed by Student’s t test. *P < 0.05, **P < 0.01 (Ghr-391^−/− vs. WT).

Fed Ghr-391^−/− liver transcripts involved in the glycolytic pathway, aldolase b and enolase 1 were increased compared to WT (Supplemental Table S2B). G6pc transcript was unchanged, whereas fructose bisphosphatase 1 transcript was moderately increased in Ghr-391^−/− liver. Elevated Glut2 transcript in Ghr-391^−/− would favor increased uptake of glucose for triglyceride and glycogen synthesis, whereas expression of Phkβ, involved in the activation of glycogen phosphorylase, the first step in glycogenolysis, was decreased (Supplemental Table S2B). Thus, despite an increase in the β subunit of the PDC (Pdhb) transcript, the increase of the regulatory subunit of PDH phosphatase (Pdpr) has an inhibitory effect on PDH phosphatase (37), suggesting that PDH activity is reduced in the fed and unfed state (Supplemental Table S2B). Key enzymes indicative of pyruvate flux into the TCA cycle were reduced, whereas those supporting β-oxidation were increased in Ghr-391^−/− in comparison to WT (Supplemental Table S2B) in accord with increased hepatic triglyceride synthesis and lipid oxidation (6).

Ghr-391^−/− mice have increased insulin sensitivity and elevated insulin signaling

To investigate the basis for decreased hepatic glucose output and increased muscle and adipose glucose uptake (insulin sensitivity), we analyzed Ghr-391^−/− hepatic tissues for known regulatory steps in glucose metabolism by immunoblotting and qPCR. Insulin receptor-β protein (Insrβ) expression was elevated in Ghr-391^−/− in comparison to WT mice (Fig. 4D). This was supported by increased tyrosine phosphorylation (Y1146) of Insrβ when normalized to total insulin receptor in Ghr-391^−/−. IRS1 protein remained unchanged (Fig. 4E). However, tyrosine phosphorylation of IRS1 on Y608 was increased in Ghr-391^−/− (Fig. 4E), indicating activation, whereas phosphorylation of IRS1 at S318 that potentiates down-regulation of insulin signaling (38) was unchanged (Supplemental Fig. S4D). Protein expression of protein tyrosine phosphatase 1B (PTP1B), a negative regulator of insulin receptor activation was decreased in Ghr-391^−/− in the unfed state (Fig. 4F).

To further assess if elevated insulin receptor contributes to an increase in downstream signaling, even in the unfed state, protein kinase B (AKT) activation was assessed. Although the p85 subunit of PI3K displayed no significant change in level (Fig. 5A), phosphorylated AKT (Thr308 and Ser473) downstream was increased in Ghr-391^−/−, indicating activation even at low fasting insulin levels (Fig. 5B). 3-Phosphoinositide-dependent protein kinase 1PDPK1 is known to activate AKT (39, 40) and the increased level of SIRT1 protein and transcript (Fig. 5C, D) would potentially increase the PDPK1 activity, adding to the effect of elevated protein itself (Fig. 5A) and further promoting AKT activation. In accord with increased AKT action, phosphorylated forkhead box protein O1 (S256) and especially inactivating phosphorylated glycogen synthase kinases α and β (GSK3α/β) levels are increased in Ghr-391^−/− (Fig. 5E, F), the former inhibiting gluconeogenesis, whereas the latter facilitating glycogen synthesis. Likewise, phosphorylated mammalian target of rapamycin (mTOR) (S2448) and phosphorylated p70S6K (T389) were both increased in Ghr-391^−/− mice (Fig. 5E). Further evidence of AKT activation is the 3-fold decrease in Insig2 in Ghr mutants reported in our previous study (6). AKT has been shown to repress Insig2 (41), allowing activation of sterol regulatory element-binding protein 1 and lipid synthesis.

Figure 5.

Altered activation and expression of insulin signaling pathway and intermediates in hepatic tissue of overnight unfed 14-wk-old male Ghr-391^−/− and Ghr^−/− (n = 3–5/genotype) mice with age-matched WT. A) Total protein levels of p85 and PDPK1 with densitometric quantifications. B, C, E–G) Protein levels of phosphorylated AKT (B), SIRT1 (C), phosphorylated GSK3α/β, phosphorylated p70S6K, and phosphorylated mTOR (E) with densitometric quantification. Levels of total and inhibitory phosphorylated FOXO1 (F) and total and inhibitory phosphorylated PTEN (G) with densitometric quanitification are indicated. β–Tubulin and Gapdh expression were used as loading controls for the immunoblots where indicated. D) Relative hepatic mRNA levels of Sirt1 normalized to β2m is indicated. Data represented as means ± sem and analyzed by Student’s t test or 1-way ANOVA, followed by Tukey’s posttest. *P < 0.05, **P < 0.01, ***P < 0.001 (Ghr mutants vs. WT).

PTEN regulates PI3K activity, and knockout of Pten results in fatty liver with increased insulin sensitivity (42), phenocopying the Ghr-391^−/− mice. Although total PTEN protein remained unchanged, phosphorylated PTEN (S300/T382/T383) was significantly increased in Ghr-391^−/−, and these phosphorylation events are inhibitory (Fig. 5G) and driven in part by increased GSK3α/β action (Fig. 5E). Additionally, the Pfkfb1 transcript levels were increased in Ghr-391^−/− and Ghr^−/−, although it remained unchanged at the protein level (Supplemental Fig. S4E, F). Similarly, AKT-dependent phosphorylated PFKFB1 (S483) remained unchanged (Supplemental Fig. S4E).

Gluconeogenesis in Ghr^−/− and Ghr-391^−/− mice is reduced

Foxo1 transcripts were modestly increased in Ghr-391^−/− and Ghr^−/−, but at the protein level, we observed increased phosphorylation of FOXO1 (Fig. 5F), which results in its exclusion from the nucleus, degradation, and inhibition of gluconeogenesis. We also observed an increase in the transcription factor small heterodimer partner (SHP) (Fig. 6A), which inhibits gluconeogenesis through blocking activated cyclic AMP response element–binding (CREB) protein (43). Hepatic Shp transcription is increased by SIRT1 (44), which is also elevated in Ghr-391^−/− and Ghr^−/− mice (Fig. 5C). SHP is itself activated by AMPK (43), which showed increased phosphorylation (T172) in both Ghr-391^−/− and Ghr^−/− fed liver (Fig. 6B), at least partially because of the increased plasma ADIPOQ. We also found a marked decrease in hepatocyte nuclear factor 4α (HNF4α) (Fig. 6C), which synergizes with FOXO1 in promoting G6pc transcription (45). Further, phosphorylated STAT3 levels are increased in Ghr-391^−/− mice (Fig. 6D and (6)), and this is known to repress G6pc expression (46). We also note that histone acetyltransferase (GCN5) protein is increased (Fig. 6E) and GCN5 acts as an acetyltransferase to suppress hepatic gluconeogenesis by acetylating and modulating peroxisome proliferator-activated receptor γ coactivator 1α (PGC1α) levels in Ghr-391^−/− mice, thereby decreasing G6pc expression (46). We have previously shown elevated Pgc1α levels in unfed Ghr mutants (6) but see no differences in the PGC1α protein level in unfed Ghr-391^−/− compared to WT mice (Fig. 6F). These actions are similar to those of metformin, which suppresses gluconeogenesis via induction of SIRT1 and GCN5 (47).

Figure 6.

Altered activation and expression of gluconeogenesis messengers in hepatic tissue of 14-wk-old male Ghr-391^−/− and Ghr^−/− (n = 3–5/genotype) mice unfed overnight with age-matched WT. A–C) Immunoblots indicating levels of SHP (A), phosphorylated AMPK (B), and HNF4α (C) in Ghr-391^−/−. FL, full length isoform. D–F) Immunoblots indicating expression levels of total and active STAT3 (D), GCN5 (E), and PGC1α (F) in Ghr-391^−/−. β–Tubulin and Gapdh expression were used as loading controls for all the immunoblots as indicated. Data are represented as means ± sem and analyzed by Student’s t test. *P < 0.05, **P < 0.01 (Ghr mutants vs. WT).

ChIP analysis determines GH-induced STAT5 targets in hepatic tissue

We sought to determine key genes with known function in glucose metabolism implicated in this study that may harbor GH-induced STAT5 binding sites, based on the previously reported hepatic ChIP data in WT mice (31). STAT5 binding was generally found at open chromatin regions of DNase hypersensitivity (Fig. 7) and positively associated with H3K4me1 and H3K4me3 (activating) marks, but negatively with H3K27me3 (repressive) marks in male mice. We identified direct STAT5 target genes (Insr, Irs1, Glut2, Gbe1, Agl, G6pc, Ptpn1, Shp, and Hnf4α) based on multiple strong STAT5 binding sites typically located at the promoter on several nearby sites (Fig. 7). No STAT5 binding was evident for G6pd, Gcn5, Prkcd, Pdpk1, Pdk2, Pdpr, Phkb, and Sirt1, 2, and 6.

Figure 7.

Schematic of ChIP-seq data indicating active STAT5 binding sites on genomic DNA from hepatic tissue of male mice following a GH-induced STAT5 activity. Gbe1 (A), G6pc (B), Irs1 (C), Agl (D), Ptpn1 (E), Insr (F), and Fgf1 (G). UCSC Genome Browser tracks are shown indicating DHSs, which mostly coincide with the active STAT5 binding sites (75, 76). Also shown are the UCSC Genome Browser tracks indicating locations of liver topologically associating domains (TADs) and subtopologically associating domains, which segment the genome into functional units. Chr, chromosome.

Ghr-391^−/− mice have altered glucose metabolism transcripts and protein expression in fed skeletal muscle

In muscle tissue, transcripts encoding enzymes involved in the glycolytic pathway aldolase A (AldoA), enolase 3 (Eno3) Gapdh, and triosephosphate isomerase (Tpi1) were modestly up-regulated in Ghr-391^−/− mice in the fed state compared to WT (Supplemental Table S2C). Transcripts of uridine diphosphate glucose phosphorylase (Ugp2) and the branching enzyme Gbe1 and both glycogen synthesis enzymes were modestly increased, as was the transcript for Gsk3α/β. Although expression of the Pdha1 subunit was unchanged, expression of the dihydrolipoamide S-acetyltransferase (Dlat: E2 subunit of the PDC) was increased, as was Dld, favoring glucose utilization. Although Pdk2 expression was increased in fed Ghr-391^−/− muscle, Pdk4 transcript was unchanged, but protein decreased in fasted muscle (Fig. 8A and Supplemental Table S2C). In addition, several enzymes involved in the TCA cycle were up-regulated (Supplemental Table S2C). Importantly, expression of the Glut4 transcript was increased in both Ghr-391^−/− and Ghr^−/− (Fig. 8B), as was the transcript for Akt1 (Fig. 8C). IL-15, a known insulin sensitizer (48), showed significantly increased levels of transcript and protein in Ghr-391^−/− (Fig. 8D, H). Muscle insulin resistance regulator, Foxo1 was considerably decreased in fed skeletal muscle of Ghr-391^−/− mice, whereas Pparα transcript was significantly increased, favoring β-oxidation (Fig. 8E, F). PDHA protein was markedly decreased in Ghr-391^−/− unfed skeletal muscle (Fig. 8H) despite the transcript expression being unchanged (Fig. 8G). However, in accord with the decreased active STAT5 levels (49), Pdk4 transcript and protein were decreased so the extent of inhibitory Ser293 and Ser300 phosphorylation of PDHE1a was decreased (Fig. 8H). This suggests that the residual PDHA is more active, although diminished in amount. The outcome of these changes would favor glycolysis and fatty acid oxidation, aligning with the significantly lower respiratory quotient observed in the Ghr^−/− mice (48).

Figure 8.

Altered activation and expression of glucose metabolism related genes in quadriceps femoris skeletal muscle tissue of overnight fasted 14-wk-old male Ghr mutant mice (n = 3–5/genotype) with age-matched WT mice. A–G) Relative RNA levels of Pdk4 (A), Glut4 (B), Akt1 (C), Il15 (D), Foxo1 (E), Pparα (F), and Pdha1 (G) normalized to β2m in skeletal muscle tissue of Ghr mutants. H) Immunoblots indicating total PDHA1 and inhibitory phosphorylated PDHA1 levels, total levels of PDK4 and IL-15 with Gapdh as loading control, and densitometric quantification (H). Data represented as means ± sem and analyzed by Student’s t test or 1-way ANOVA, followed by Tukey’s posttest. *P < 0.05, **P < 0.01, ***P < 0.001 (Ghr mutants vs. WT).

Ghr-391^−/− fed inguinal WAT show altered glucose metabolism transcripts

Importantly, transcript for the insulin target Pdk4 was substantially decreased in inguinal WAT (iWAT) in fed Ghr-391^−/− compared to WT mice (Supplemental Table S2D). This would promote glucose oxidation and ATP synthesis in conjunction with increased PDH phosphatase (Pdpr) levels (Supplemental Table S2D). Expression of enzymes involved in the pentose phosphate pathway providing NADPH for fatty acid synthesis (G6pdx, Rpia, Prps2) were also increased (Supplemental Table S2D). Transcript for Phkγ2 was markedly decreased, which would reduce glycogen breakdown. Transcripts involved in the glycogenolytic/glycogenic pathways, Pgm1 and Pgm3, were up-regulated, whereas transcript for mitochondrial pyruvate carboxylase (Pck2) was increased, which could provide glycerol for triglyceride synthesis.

Analysis of an adipogenesis array in Ghr-391^−/− on fed iWAT revealed a 5-fold increase in leptin (Lep) transcript but no change in Adipoq transcript in comparison to WT mice (Supplemental Table S2E). The increased mass of WAT in Ghr mutant mice compared to WT (7) presumably accounts for the 50% increase in the circulating ADIPOQ level. The 10-fold decrease in acetyl CoA carboxylase β transcript (Supplemental Table S2E) would favor increased mitochondrial fatty acid oxidation over fatty acid synthesis. The modest increase in Foxo1 and the decrease in Irs1 and Srebf1 would not favor increased insulin action. However, the 3.3-fold increase in Fgf1 transcript (Supplemental Table S2E) may counter this because FGF1 is a potent insulin sensitizer (50). This is supported by a substantial increase in circulating FGF1 (Fig. 3D) with ChIP data, indicating the presence of 3 active STAT5 binding sites in the Fgf1 promoter (Fig. 7F). The schematic (Fig. 9) indicates that the effect of the GH-STAT5 deficiency in regulating hepatic glucose metabolism and insulin sensitivity is a collective effect of our findings in liver and complementary effects mediated in part by endocrine sensitizers (IL-15, FGF1, and ADIPOQ) from distinct tissues. Several, possibly cumulative, mechanisms that cause impaired gluconeogenesis from reduced glycogen breakdown and reduced PCK1 stability to persistent insulin action, albeit at low levels, promote elevated insulin sensitivity in Ghr mutant mice that is maintained with age despite obesity.

Figure 9.

Schematic showing actions of lack of GH-STAT5 on hepatic gluconeogenesis with direct STAT5-binding genes based on ChIP promoter analysis shown as red dots and altered regulatory phosphorylation or acetylation sites shown as green dots. G6-P, glucose-6-phosphate; G1-P, glucose-1-phosphate; OAA, oxaloacetic acid; PEP, phosphoenol pyruvate; DHAP, dihydroxyacetone phosphate; GAP, glyceraldehyde-3-phosphate; 6-Pgl, 6-phosphogluconolactonase.

DISCUSSION

This study sought to define the metabolic state of male mice lacking the ability to activate STAT5 via GH (Ghr-391^−/−) and focused on their increased insulin sensitivity and decreased hepatic glucose output. Although these mice are glucose intolerant because they have insufficient insulin secretion when challenged, they are very insulin sensitive, both in relation to suppression of hepatic glucose output and to enhanced glucose uptake into skeletal muscle and WAT. Therefore, a normal fasting insulin level, such as we observed, may still allow the postprandial actions of insulin even in the face of a chronically low blood glucose level. However, GH itself has important metabolic actions, and the absence of these has additional actions to those resulting from the enhanced insulin sensitivity.

We sought in particular to identify key elements responsible for the suppressed gluconeogenesis that was evident in Ghr^−/− mice and Ghr-391^−/− mice based on the PTT and the insulin clamp study. These elements should be viewed in the context of the hepatosteatosis in these mice and their elevated hepatic glycogen during a period of food withdrawal, both of which would lessen fasting glucose production because they compete for triose sugars. Our data indicate that these mechanisms are regulated in part directly by GH-mediated STAT5 transcriptional activity as well as the phosphorylation and acetylation status of signaling molecules, controlling its activity or stability, with a defined role in gluconeogenic and insulin signaling processes (Fig. 9).

GH is known to promote glycogenolysis in humans (15), and in GH-STAT5–deficient mice we found a significant decrease in the transcript for Phkb, the debranching enzyme (Agl), and G6pc transcript in unfed liver. However, the latter 2 transcript changes did not manifest as a significant decrease in protein level. Alternatively, glycogen synthesis would be enhanced through the inhibitory phosphorylation evident on GSK3α/β and by up-regulation of the glycogen branching enzyme (Gbe1), which occurs at both transcript and protein level. ChIP analysis showed that both Gbe1 and Agl are direct STAT5 targets. Additionally, Gbe1 increased at both the transcript and protein level and is known to be up-regulated by nuclear factor erythroid 2–related factor 2 (51), which is induced by STAT1 (52). We have previously reported elevated hepatic STAT1 levels in the Ghr-391^−/− and Ghr^−/− mice (6).

We observed strong activation of the canonical AKT pathway in the livers of GH-STAT5–deficient mice, even at low fasting insulin levels. This is partly due to an increase in the levels of insulin receptor as well as Irs1 and Glut2. Both insulin receptor and IRS1 protein showed increased tyrosine phosphorylation, indicating increased activation in the unfed state. This is in agreement with previous data showing elevated Insr, Irs1, and Glut2 transcripts in young Ghr^−/− male mice (26). As we show here, these changes are a result of deficient GH-STAT5 signaling, and this is supported by ChIP data showing all 3 to be direct STAT5 targets. It is likely that the decrease in PTP1B protein phosphatase expression we see would also amplify insulin action because this is a negative regulator (and a ChIP STAT5 target) (53). Our ChIP analysis also indicated that Ptp1b is a STAT5 target gene. We found no change in expression of the p85 subunit of PI3K, arguing against a role for GHR signaling in hepatic insulin sensitivity through control of free p110. However, the negative regulator of PI3K, PTEN, was phosphorylated on regulatory serine and threonine residues S300/T382/T383, and these are inhibitory (54). The phosphorylation on PTEN are likely to be a consequence of PDPK1 action because this regulatory enzyme is activated and has increased expression in Ghr-391^−/−.

In addition to residual insulin activation in the liver of Ghr-391^−/− mice, increased PDPK1 levels have been shown to promote AKT activation resulting from binding to PIP3 (39) and S473 phosphorylation by mTORC (40), which we also find activated in Ghr-391^−/−. One of the key roles of increased AKT activity is phosphorylation of gluconeogenic FOXO1, which results in its inactivation because of exclusion from the nucleus (55), and such phosphorylation is evident in Ghr-391^−/−. This, together with reduced levels of HNF4α, a direct STAT5 target and elevated levels of active STAT3 shown previously (6), would contribute to the decrease in G6pc. Indeed, phosphorylated STAT3 not only represses G6pc directly, but also activates microRNA-23a, which targets G6pc (56). Loss of HNF4α would compromise the ability of PGC-1α in activating gluconeogenic genes such as G6pc (57). Further evidence for activated AKT is evident from increased inhibitory phosphorylation of GSK3α/β in Ghr-391^−/−, promoting glycogen synthesis, along with increased phosphorylated p70S6K and its target, phosphorylated mTOR (S2448). The latter can also be activated by PDPK1, in an insulin-independent activation process via SIRT1 (58), which itself can be regulated by STAT5 phosphorylation status (59). SIRT1 has been shown to activate PDPK1 as well as to promote loss of gluconeogenic CREB regulated transcription factor 2 (60).

SIRT1 is a master metabolic regulator activated by nicotinamide adenine dinucleotide (NAD⁺), which increases during energy deficit, such as during unfed and caloric restriction, and as a result of AMPK action (61). We find AMPK to be activated, potentially as a result of elevated ADIPOQ and through the energy demands of hepatosteatosis. Another action of increased AMPK is induction of the transcription factor SHP, which inhibits gluconeogenesis by blocking cAMP activated CREB (43). Hepatic SHP transcription is also increased directly by SIRT1 (44). It is notable that SHP has also been reported to directly bind to GH-activated STAT5, inhibiting its ability to activate the Pck1 and G6pc promoters (16). For this reason and because the contributing elements outlined above clearly demonstrate a decrease in hepatic glucose production in Ghr mutant mice, we were surprised that a 2-fold increase in G6pc was reported in unfed male Ghr^−/− (26). The molecular basis for this difference is unclear, although the Ghr^−/− used in this study (26) were in a mixed genetic background of 129Ola and BALB/c as opposed to C57BL/6 used in our study.

We expected that Pck1 and Pck2 transcripts would also be decreased in Ghr mutant mice because of suppressed glucose production and increased phosphorylation of FOXO1, particularly in the light of the functional STAT5 binding site on SHP promoter (16), and decreased HNF4α expression. However, we could find no change in transcript expression for Pck1 and Pck2, as also reported in Ghr^−/− mice (26). Further investigation revealed that the inhibitory acetylation of PCK1 was elevated, which results in increased protein degradation evident in the immunoblot. This acetylation is regulated by SIRT2 (62), which is in accord with the decrease in its transcript and protein expression we observed in Ghr-391^−/− mice. Hence, we propose that 1 reason for suppressed gluconeogenesis in Ghr-391^−/− mice is inhibition of PCK1 because of its degradation via acetylation owing to reduced SIRT2. Note that the lack of change in Pck1 transcript parallels the lack of a difference from unfed WT mice in insulin-induced hepatic glucokinase (Gck) transcript, which is similar in both Ghr-391^−/− and WT mice.

PDKs are a major control point for directing triose sugars to the TCA cycle or to gluconeogenesis. In particular, PDK4 phosphorylates PDHA to inhibit triose sugar flow into the TCA cycle. In unfed Ghr-391^−/− liver, we found increased transcripts for Pdk4, but protein expression of PDK4 remained unchanged. However, we did observe an increase in SIRT3, which is known to increase PDH activity by promoting deacylation (35). The lack of alteration in PDK4 protein expression in our chronic GH-STAT5–deficient model contrasts with the previous report demonstrating STAT5-dependent induction of hepatic PDK4 expression in response to acute GH administration at supraphysiological concentrations in mice (16).

TNF is reported to inhibit hepatic gluconeogenesis prior to pyruvate carboxylase enzymatic step in vivo (63). Circulating levels of TNF were found to be elevated in both Ghr-391^−/− and Ghr^−/− mice. In addition, elevated ADIPOQ would also be expected to contribute to the suppression of hepatic gluconeogenesis in Ghr-391^−/− and Ghr^−/− mice by mechanisms that can be AMPK-dependent (64) or independent (65). Although the primary purpose of our study was to determine the reasons for the suppression of hepatic glucose production in the absence of GH-STAT5 signaling, we also sought the reasons for the enhanced glucose uptake in skeletal muscle and WAT. In skeletal muscle of Ghr-391^−/− mice, several changes could contribute to the increased glucose uptake we observed. These include increased Glut4 and Akt1, decreased Foxo1, and an increase in glycogen synthesis transcripts for Gbe1 and Ugp2. The increased expression of Il15 transcript and protein we observed could contribute to the up-regulation of Glut4 and glucose uptake (66) and confirms the finding in muscle-specific Ghr^−/− mice (67). In addition, we found reduced Pdk4 transcript and protein expression in Ghr-391^−/− muscle, which is concordant with the report that GH-activated STAT5A up-regulates Pdk4 expression in 3T3-L1 adipocytes (49). This decrease in PDK4 correlates with a decrease in phosphorylation of PDE1a at serine 300. The converse of this was observed in healthy human males administered GH, which increased muscle PDK4 transcript and inhibitory phosphorylation of PDHE1a at Ser300 (68).

Given the lack of change in plasma corticosterone, the increase in Lep transcripts in iWAT could be a consequence of the loss of β-3 adrenergic receptor evident in WAT of Ghr-391^−/− mice (7). This is supported by the fact that administration of β-3 agonist CL316243 is able to induce a suppression of Lep transcript in iWAT (69). Alternatively, because knockout of Jak2 in adipocytes is reported to have no effect on Lep levels (70), whereas liver-specific Jak2 knockout increases them (71), it may be that the decreased circulating IGF1 in Ghr-391^−/− is derepressing Lep production. The lack of change in Adipoq transcript in obese Ghr-391^−/− mice is intriguing as there is a 48% increase in total plasma ADIPOQ in Ghr-391^−/− and a 52% increase in circulating ADIPOQ seen in liver-specific Ghr^−/− mice (71). Previously it has been proposed that a GH-regulated hepatic factor governs Adipoq production from adipose tissue because liver-specific knockout of GHR resulted in the increase of circulating ADIPOQ (71), whereas adipose tissue-specific knockout of Ghr or Jak2 results in a small decrease in circulating ADIPOQ (70, 72). Based on our finding of unchanged Adipoq transcript in Ghr-391^−/−, we suggest that such a factors acts at the translational level. Of greater importance in understanding the insulin sensitivity of Ghr-391^−/− and Ghr^−/− is the 3-fold increase in Fgf1 transcript in iWAT, which is matched by a 3–4 fold increase in circulating FGF1 levels in Ghr-391^−/− and Ghr^−/− mice. FGF1 has powerful insulin synergistic actions and is synthesized and secreted from adipose tissues (50, 73). Importantly, a recent study in adipose tissue-specific Jak2 knockout mice has revealed a marked increase in hepatic insulin sensitivity and postulated that this was a result of altered secretion of an adipose tissue factor (74). It is plausible that this factor is indeed FGF1 and that it acts in an endocrine manner to increase insulin sensitivity in the obese Ghr-391^−/− and Ghr^−/− mice. This is likely one of the ways that GH regulates insulin sensitivity throughout the body, through STAT5 activation to decrease circulating FGF1 from WAT. Elevated FGF1 would also promote adipogenesis and increase in fat mass, leading to an obese yet insulin-sensitive mouse model (50) as observed in Ghr mutants used in our study.

In summary, we have shown GH-mediated STAT5 activation acts on multiple sites in the major insulin responsive tissues to promote insulin sensitivity. These actions are regulated at both transcriptional and posttranscriptional levels, and although there is ChIP analysis indicating direct STAT5 action at the promoter level of key genes, it is apparent that many of the insulin-sensitizing actions of GH-STAT5 deficiency are indirect and likely involve altered expression of other transcription factors, such as elevated STAT1, which promotes hepatic Pparg expression (6), or reduction in the levels of STAT5-dependent repressors such as Bcl6 (75, 76). STAT5 binding is also accompanied by binding of other transcription factors to proximal promoters and enhancers of STAT5-regulated genes, evidenced by DNase hypersensitivity and active histone marks corresponding to the binding of active STAT5 in response to a GH pulse (76). Finally, in the light of the lack of the ability of tissue-specific Ghr^−/− to promote insulin sensitivity (except for muscle-specific Ghr^−/− with a high fat diet), there may be synergistic actions between tissues to promote whole body sensitivity through the combined actions of several important factors such as IL-15 from muscle and ADIPOQ and notably FGF1 from WAT. Together, these elements may account for the interesting phenomenon of an obese, steatotic, insulin-sensitive, and long-lived mouse model resistant to diabetes and reduced tumor burden (77).

Supplementary Material

This article includes supplemental data. Please visit http://www.fasebj.org to obtain this information.

Glossary

2-DG

2-[1-¹⁴C] deoxyglucose

β2m

β2-microglobulin

AKT

protein kinase B

Adipoq

adiponectin

Agl

amylo-1,6-glucosidase, 4-α-glucanotransferase

ChIP

chromatin immunoprecipitation

CREB

cyclic AMP response element –binding

DHS

DNase-I hypersensitive site

FGF1

fibroblast growth factor 1

G6pc

glucose-6-phosphatase

Gapdh

glyceraldehyde 3-phosphate dehydrogenase

Gbe1

1,4-α-glucan branching enzyme 1

GCN5

histone acetyltransferase

GH

growth hormone

GHR

GH receptor

Ghr^−/−

Ghr null

GIR

glucose infusion rate

Glut

glucose transporter

GSK3α/β

glycogen synthase kinases α and β

HNF4α

hepatocyte nuclear factor 4α

HNFIRS

insulin receptor substrate

iWAT

inguinal white adipose tissue

JAK2

Janus kinase 2

mTOR

mammalian target of rapamycin

PCK

phosphoenolpyruvate carboxykinase

PDC

pyruvate dehydrogenase complex

PDH

pyruvate dehydrogenase

PDK

pyruvate dehydrogenase kinase

PDPK1

3-phosphoinositide-dependent protein kinase 1

PGC1α

peroxisome proliferator-activated receptor γ coactivator 1α

Pgm

phosphoglucomutase

Phkb

phosphorylase kinase β

PI3K

phosphatidylinositol-4,5-bisphosphate 3-kinase

Pparg

peroxisome proliferator-activated receptor γ

PTEN

phosphatase and tensin homolog

PTP1B

protein tyrosine phosphatase 1B

PTT

pyruvate tolerance test

qPCR

quantitative PCR

SHP

small heterodimer partner

SIRT

sirtuin

SOCS

suppressor of cytokine signaling

STAT

signal transducer and activator of transcription

TCA

tricarboxylic acid

WT

wild type

AUTHOR CONTRIBUTIONS

Y. Chhabra, C. N. Nelson, and M. J. Waters designed the research; Y. Chhabra, C. N. Nelson, M. Plescher, J. L. Barclay, A. G. Smith, S. Andrikopoulos, S. Mangiafico, D. J. Waxman, and M. J. Waters analyzed data; Y. Chhabra, A. J. Brooks, and M. J. Waters wrote the manuscript; and all authors performed the research.