Insulin and IGF-I Inhibit GH Synthesis and Release in Vitro and in Vivo by Separate Mechanisms

Abstract

IGF-I is considered a primary inhibitor of GH secretion. Insulin may also play an important role in regulating GH levels because insulin, like IGF-I, can suppress GH synthesis and release in primary pituitary cell cultures and insulin is negatively correlated with GH levels in vivo. However, understanding the relative contribution insulin and IGF-I exert on controlling GH secretion has been hampered by the fact that circulating insulin and IGF-I are regulated in parallel and insulin (INSR) and IGF-I (IGFIR) receptors are structurally/functionally related and ubiquitously expressed. To evaluate the separate roles of insulin and IGF-I in directly regulating GH secretion, we used the Cre/loxP system to knock down the INSR and IGFIR in primary mouse pituitary cell cultures and found insulin-mediated suppression of GH is independent of the IGFIR. In addition, pharmacological blockade of intracellular signals in both mouse and baboon cultures revealed insulin requires different pathways from IGF-I to exert a maximal inhibitory effect on GH expression/release. In vivo, somatotrope-specific knockout of INSR (SIRKO) or IGFIR (SIGFRKO) increased GH levels. However, comparison of the pattern of GH release, GH expression, somatotrope morphometry, and pituitary explant sensitivity to acute GHRH challenge in lean SIRKO and SIGFRKO mice strongly suggests the primary role of insulin in vivo is to suppress GH release, whereas IGF-I serves to regulate GH synthesis. Finally, SIRKO and/or SIGFRKO could not prevent high-fat, diet-induced suppression of pituitary GH expression, indicating other factors/tissues are involved in the decline of GH observed with weight gain.

Circulating GH levels are negatively associated with body mass index (1, 2). Given the antilipogenic, prolipolytic, and anabolic actions of GH, it has been proposed that the low levels of GH observed in obesity contribute to the progression of metabolic disease. The mechanism(s) by which GH levels are reduced in obesity remains a subject of debate. However, several reports suggest insulin may play a key role in the suppression of GH secretion because circulating GH levels are negatively correlated with insulin levels when comparing obese subjects with lean controls (3, 4). Insulin may also contribute to lowering GH output in the absence of weight gain, as supported by a recent study showing healthy subjects who overate for just 3 days displayed a dramatic reduction in pulsatile GH release, which was associated with a rise in insulin (5). The inhibitory actions of insulin on GH production could be exerted through alterations in hypothalamic function (6, 7). In addition, insulin may also play an important role in directly suppressing pituitary somatotrope function in that insulin reduces GH synthesis and release in primary pituitary cell cultures from a variety of species, including primates (8–11). The fact that pituitaries from diet-induced obese, hyperinsulinemic mice displayed reduced expression of GH, but remained insulin sensitive (11) further supports the argument that elevated insulin could act directly on the pituitary somatotrope to suppress GH production in the obese state.

However, challenges arise when trying to determine the specific contributions of insulin in directly modulating GH production in response to changes in the metabolic environment because of the following: 1) insulin and bioavailable IGF-I are regulated in parallel in metabolic extremes (12), 2) IGF-I, as well as insulin, can suppress somatotrope function in vitro (10, 13), 3) the pituitary gland expresses both insulin receptor (INSR) and the IGF-I receptor (IGFIR) (10, 11), 4) INSR is structurally and functionally related to IGFIR, in which these receptors have been reported to form hybrids (14), 5) insulin and IGF-I, at high concentrations, can bind and activate each other's receptor (15), and 6) insulin/INSR and IGF-I/IGFIR can modulate cell function via similar intracellular signaling processes (15). Therefore, to circumvent these challenges and compare the relative contribution of insulin/INSR and IGF-I/IGFIR in modulating somatotrope function, the Cre-loxP system was used to inactive the Insr and IgfIr genes in the somatotrope in vitro and in vivo. In addition, pharmacological blockade of intracellular signals was used to identify the key pathways required for insulin and IGF-I actions.

Materials and Methods

Animals

Somatotrope-specific inactivation of Insr and/or IgfIr was accomplished by crossbreeding rat GH promoter-driven Cre-recombinase (rGHpCre) mice (16) to mice homozygote for the loxP-modified Insr [Insr^fl/fl (17)], IgfIr [IgfIr^fl/fl (18)], or both alleles (Insr,IgfIr^fl/fl), in which all strains are in a C57BL/6J background. Insr^fl/wt,rGHpCre^+/− or IgfIr^fl/wt,rGHpCre^+/− female mice were crossbred with Insr^fl/fl or IgfIr^fl/fl male mice to obtain somatotrope-specific INSR knockout (Insr^rGHpCre, aka SIRKO) and somatotrope-specific IGFIR knockout (IgfIr^rGHpCre, aka SIGFRKO) mice, respectively. Experimental double-receptor knockout mice (Insr,IgfIr^rGHpCre) were generated as previously reported (19). It should be noted that all experimental mice within each floxed background were born within a 2-week time frame, and in vivo measurements and tissue collection were performed on the same day using littermate knockout and control genotypes. We would like to point out that mice with an Insr^fl/fl background were significantly fatter than mice with an IgfIr^fl/fl background (see nuclear magnetic resonance analysis in Supplemental Figure 1, published on The Endocrine Society's Journals Online web site at http://endo.endojournals.org). In fact, in preliminary studies, we observed Insr^fl/fl mice (2 months of age) had greater fat depot weight as compared with Insr^wt/wt and Insr^wt/fl mice (Supplemental Figure 2). These results emphasize the importance of using homozygous floxed controls to compare the impact of Cre-mediated gene recombination, in that insertion of the loxP sites within the gene of interest could modify basal gene function.

Mice were maintained on a standard rodent chow diet (CHOW; fat, 17% kcal; carbohydrate, 56% kcal; protein, 27% kcal; Formulab Diet, Purina Mills, Inc, Richmond, Indiana) or fed a low-fat (LF; fat, 10% kcal; carbohydrate, 70% kcal; protein, 20% kcal; Research Diets, New Brunswick, New Jersey) or a high-fat (HF; fat, 60% kcal; carbohydrate, 20% kcal; protein 20% kcal; Research Diets) diet, starting at 4 weeks of age. All experimental procedures were approved by the animal care and use committee of Jesse Brown Veterans Affairs Medical Center.

In vitro studies

To determine the requirement of phosphatidylinositol 3-kinase (PI3K), mammalian target of rapamycin complex 1 (mTORC1), and MAPK kinase (MEK) signaling in insulin- and IGF-I-mediated inhibition of GH expression and secretion, nonhuman primate pituitary cell cultures were prepared from random cycling female baboons (Papio anubis, 7–12 years of age) as previously reported (10), and mouse pituitary cell cultures were prepared from 9- to 11-week-old male wild-type mice as previously described (16). Cultures were preincubated for 48 hours in serum containing media, and then the medium was removed and cells were preincubated for 1 hour in fresh, serum-free medium. After the preincubation period, medium containing the inhibitors of key intracellular signaling pathways [LY294002 (1 μM) for PI3K, rapamycin (10 nM) for mTORC1, and PD098,059 (10 μM) for MEK] was added (medium alone was used in the vehicle treated controls). Ninety minutes later, the medium was replaced with medium alone (vehicle) or containing the inhibitor combined with 10 nM insulin or 10 nM IGF-I and incubated for an additional 24-hour period. Human insulin and IGF-I (Sigma, St Louis, Missouri) were used in baboon cultures, whereas mouse insulin and IGF-I were used in mouse cultures [Phoenix Pharmaceuticals, Inc., (Burlingame, California) and Sigma, respectively]. Although the effectiveness of these inhibitors on kinases phosphorylation was not directly tested herein, LY294002, rapamycin, and PD098,059 are widely used and validated inhibitors. In addition, the concentration of the inhibitors was based on our previous report (20) and was not cytotoxic based on the fact that the cell viability and end points measured did not differ between cultures treated with vehicle and those treated with the inhibitor alone (data not shown). The concentration of insulin and IGF-I used was selected based on dose response studies previously published (10, 11) and presented in Supplemental Figure 3, showing 10 nM of each hormone is the minimal concentration to achieve maximal inhibition of GH expression and release. Data represent the mean ± SEM of 2–4 independent experiments (with n = 3–4 replicates per treatment within experiment).

To study the impact of specific inactivation of Insr or IgfIr in vitro, mouse pituitary cell cultures were prepared from 9- to 11-week-old male and female Insr^fl/fl, IgfIr^fl/fl, or Insr,IgfIr^fl/fl mice (n = 7–10 pituitaries pooled within sex per experiment, with 2–4 independent experiments and 3–4 replicates per treatment within experiment) as outlined above. Eight hours after plating onto 24-well tissue culture plates in α-MEM containing 10% serum, cultures were treated overnight with cytomegalovirus (CMV)-Cre (Cre-recombinase enzyme gene under the control of CMV promoter) or CMV-Null (control) adenoviral (Ad) vectors (Vector BioLabs, Philadelphia, Pennsylvania) at a final concentration of 10 plaque-forming units per cell. The following morning the medium was removed and replaced with fresh α-MEM containing serum and incubated for an additional 48 hours. Cultures were then preincubated in serum-free medium for 2 hours and replaced with serum-free medium containing 0 (control group), 10 nM mouse insulin, or 10 nM mouse IGF-I. Cultures were incubated for an additional 24 hours, medium was collected for GH determination, and cells were recovered for total RNA extraction.

To analyze the impact of somatotrope-specific inactivation of Insr or IgfIr in vitro on basal and GHRH-stimulated GH release, whole-mouse anterior pituitaries were collected from 5- to 13 month-old male and female SIRKO and SIGFRKO mice and their respective controls (n = 4–24 pituitaries/group). Pituitaries were placed (1 per well) in 12-well tissue culture plates containing α-MEM with 0.1% BSA and penicillin-streptomycin. After 1 hour at 37°C, medium was recovered for assessment of basal GH secretion, and pituitaries were rinsed 3 times in sterile PBS. Pituitaries were incubated for an additional hour with medium alone (vehicle) or containing 100 nM GHRH (Sigma).

In vivo evaluation of metabolic status

In LF- and HF-fed mice, glucose tolerance tests (GTTs) were performed at 16 weeks of age (12 weeks of diet) after an overnight fast (2 g/kg glucose, ip), and insulin tolerance tests (ITTs) were performed at 18 weeks of age (14 weeks of diet) under ad libitum-fed conditions (1 U/kg Novolin [Novo Nordisk, Novo Allé, Denmark], ip), beginning between 8:00 and 9:00 am. Blood was collected at time 0 (t0) for hormone and nutrient measurements. In addition, whole-body composition (lean, fat, and extracellular water content) was assessed every 2 weeks by nuclear magnetic resonance (MiniSpec LF50; Bruker Optics, Billerica, Massachusetts).

Circulating hormones and metabolites

GH (Millipore, Bedford, Massachusetts), IGF-I (Immunodiagnostic Systems, Fountain Hills, Arizona), insulin (Millipore), and prolactin (Calbiotech, Spring Valley, California) levels were assessed using commercial ELISA kits. Blood glucose was assessed by glucometer (Alpha TRACK blood glucose monitoring system; Abbott, Abbott Park, Illinois). To assess differences in GH secretion pattern, GH levels were determined from tail vein samples taken 3 times a day during 3 consecutive days under fed conditions in 3-month-old animals (n = 5 mice/genotype per gender).

mRNA analysis by quantitative real-time PCR (qrtPCR)

Absolute expression levels of Insr, IgfIr, GH, GHRH, somatostatin, GH receptor, GHRH receptor, and IGF-I, as well as hepatic genes shown to be GH and sex dependent, were screened by qrtPCR as previously described (19), in which the primer sequences are provided in Supplemental Table 1. To control for variations in the amount of RNA used, cDNA copy number for the transcript of interest was adjusted by a normalization factor calculated from the cDNA copy number of 3 separate housekeeping genes (glyceraldehyde-3-phosphate dehydrogenase, β-actin, and cyclophilin-A) using the GeNorm 3.3 program (21).

Protein and immunohistochemical analysis

Pituitary protein was extracted and Western blot analysis for IGFIR and INSR was performed as previously described (19). In addition, a subset of pituitaries was formalin fixed, paraffin embedded, sectioned (5 μm), and processed for double immunolabeling for GH and INSR or IGFIR, as previously described (19).

Electron microscopy

Anterior pituitary tissue was prepared for electron microscopy (EM) by standard methods, and somatotropes were identified by immune-gold labeling, as previously described (22). For analysis of cell morphology 10 micrographs of somatotrophs per animal were taken at a magnification of ×4000. Negatives were scanned into Adobe Photoshop (version 5.5; Adobe, San Jose, California) and analyzed using AxioVision (version 4.5; Carl Zeiss, Jena, Germany) image analysis software. The analyst was blind to the sample code. The following parameters were calculated: cytoplasmic, nuclear, and total cell areas; and granule area, granule density, and granule diameter. For measurement of the cell and nuclear areas, margins were drawn around the cell or nucleus respectively and the area was calculated. Cytoplasmic area was determined by subtracting nuclear area from total cell area. Granule density was calculated by dividing total granule area by cytoplasmic area.

Statistics

Student's t tests were used to evaluate the impact of SIRKO and SIGFRKO on GH-axis function in LF- and CHOW-fed mice within gender. One-way ANOVA, followed by Newman-Keuls post hoc test, was used to evaluate the impact of signaling pathway inhibitors on insulin- and IGF-I-induced GH mRNA expression and secretion in baboon and mouse primary cell cultures. Two-way ANOVA, followed by Newman-Keuls post hoc test for multiple comparisons, was used to evaluate GH mRNA expression and secretion in primary cell cultures, GH mRNA in HF-fed SIRKO, SIGFRKO and double-receptor knockout mice vs LF-fed mice (within floxed background), GH levels in LF- and HF-fed double-receptor knockout mice vs controls (within gender), body composition in LF- and HF-fed SIRKO and SIGFRKO mice (within gender), and response to GTT and ITT and fat depot weights in LF- and HF-fed SIRKO and SIGFRKO mice (within gender and diet). Ranked plot analysis of GH was done by arranging GH values in order of magnitude as a rank plot (assigning to each GH value in the group the fraction of the mice that had a lower GH level) and the differences were compared by Wilcoxon signed-rank test, as previously reported (23).

Results

In vitro, insulin and IGF-I inhibit GH expression and secretion through their own receptors

Because insulin and IGF-I can bind and activate each other's receptor, primary cell cultures from pituitary glands of Insr^fl/fl (17) or IgfIr^fl/fl (18) male mice were used to determine what receptors (INSR and/or IGFIR) are required for insulin- and IGF-I-induced suppression of somatotrope function. In control cells infected with CMV-Null Ad, insulin and IGF-I suppressed GH mRNA levels and release. Infection of Insr^fl/fl pituitary cultures with CMV-Cre Ad reduced the expression of Insr but did not alter IgfIr (Figure 1A), whereas infection of IgfIr^fl/fl pituitary cultures with CMV-Cre Ad reduced the expression of IgfIr but did not alter Insr (Figure 1A). Knockdown of the Insr gene completely blocked the inhibitory actions of insulin on GH expression and release but did not alter the inhibitory effect of IGF-I (Figure 1, B and C). Likewise, knockdown of the IgfIr gene completely blocked the inhibitory actions of IGF-I on GH expression and release but did not alter the inhibitory effect of insulin (Figure 1, B and C). As would be anticipated, CMV-Cre Ad-mediated knockdown of both receptors completely abolished the inhibitory actions of both insulin and IGF-I on somatotrope function (Figure 1, B and C). Similar results were observed in primary pituitary cultures obtained from female mice (Supplemental Figure 4).

Figure 1.

Testing the requirement of INSR and IGFIR to mediate insulin and IGF-I suppression of GH expression and release in vitro. Pituitary cell cultures were prepared from male Insr^fl/fl (left panels), IgfIr^fl/fl (middle panels), and Insr,IgfIr^fl/fl (right panels) mice and treated with CMV-Null (open bars) or CMV-Cre (solid bars) adenoviral vectors (Ad: 10 plaque-forming units/cell). A, Insr and IgfIr mRNA levels after treatment with CMV-Cre adenoviral vectors represented as a percentage of expression compared with CMV-Null vector-infected controls, set at 100% (dashed line). The impact of mouse insulin (10 nM) or IGF-I (10 nM) on GH mRNA (B) and 24-hour GH release (C) in cultures pretreated with adenoviral vectors represented as a percentage of expression compared with CMV-Null vector-infected vehicle-treated controls, set at 100% (dashed line). Values are shown as mean ± SEM (n = 2–4 independent experiments with 3–4 replicates per treatment within experiment). ***, P < .001; **, P < .01; *, P < .05 and indicate values that significantly differ from controls.

In vitro, insulin and IGF-I require distinct intracellular signals to suppress GH expression and release

To determine whether insulin and IGF-I require the same signaling pathways to suppress somatotrope function, we tested the impact of insulin and IGF-I on GH expression and release in primary baboon pituitary cell cultures in the absence or presence of widely used pharmacological blockers of PI3K (LY294002), mTORC1 (rapamycin), and MEK (PD098,059). The presence of the inhibitors did not alter basal levels of GH mRNA or release when compared with vehicle-treated control cultures (data not shown). Inhibition of PI3K completely blocked the ability of both insulin and IGF-I to suppress GH mRNA levels and release (Figure 2A). Surprisingly, inhibition of mTORC1 (Figure 2B) or MEK (Figure 2C) blocked IGF-I-induced suppression of GH production (mRNA and release) but did not affect the actions of insulin. These results demonstrate that insulin and IGF-I require distinct intracellular signaling pathways to suppress somatotrope function, and these pathways may be common across mammalian species in that we observed similar results using primary pituitary cell cultures obtained from mice (Figure 2, D–F).

Figure 2.

Effect of in vitro blockade of PI3K, mTORC1, and MEK on the regulation of GH expression and release by insulin and IGF-I. Pituitary cell cultures were prepared from female baboons (left panels) and male mice (right panels) and the impact of human (A–C) or mouse (D–F) insulin (10 nM) or IGF-I (10 nM) on GH mRNA expression, and 24-hour GH secretion was analyzed in cultures treated with inhibitor to PI3K [LY294002 (LY)] (A and D), mTORC1 [rapamycin (Rap)] (B and E), or MEK [PD098,059 (PD)] (C and F). The impact of insulin (INS) and IGF-I in the absence or presence of the inhibitors is shown as a percentage of the respective controls, set at 100% (dashed lines). Values are shown as mean ± SEM (n = 2–4 independent experiments with n = 2–4 replicates per treatment within experiment). ***, P < .001; **, P < .01; *, P < .05 and indicate values that significantly differ from vehicle-treated controls.

Confirmation of the somatotrope-specific knockout of Insr or IgfIr, in vivo

To investigate the separate roles insulin and IGF-I play in the regulation of somatotrope function in vivo, Insr^fl/fl (17) and IgfIr^fl/fl (18) mice were cross-bred to rat GH promoter driven Cre-recombinase (rGHpCre) mice (16), resulting in the generation of SIRKO and SIGFRKO mice. Specificity of the knockout was confirmed by qrtPCR (Figure 3A), Western blot analysis (Figure 3B), and double immunolabeling of receptors and GH (Figure 3C). Importantly, somatotrope-specific inactivation of the Insr or IgfIr genes did not alter the gross appearance or overall cell morphology of the anterior pituitary gland, including the appearance and function of lactotropes (Supplemental Figure 5, A and B), in which a subpopulation of lactotropes is thought to be derived from somatotropes and require IGF-I for differentiation (16, 24). Also, there was no major alteration in the expression levels of key genes of other pituitary cell types (Supplemental Figure 5C).

Figure 3.

Validation of Insr and IgfIr SIRKO and SIGFRKO. A, Pituitary mRNA levels of Insr and IgfIr in 4-month-old male and female SIRKO and SIGFRKO compared with controls. B, INSR and IGFIR protein levels in pituitaries from 4-month-old SIRKO and SIGFRKO, as assessed by Western blot, using β-actin as a control. ***, P < .001 and indicated values that significantly differ from controls. The partial reduction in receptor mRNA and protein levels is consistent with the fact that, although 99% of somatotropes express Cre recombinase (16), somatotropes represent only 30%–50% of all pituitary cells. Therefore, non-GH-producing cells, which account for half of the pituitary cell population, would retain the ability to synthesize INSR or IGFIR in the SIRKO and SIGFRKO mice, respectively. C, Double immunohistochemistry for GH and INSR or IGFIR confirmed somatotrope-specific receptor knockout. Signals were visualized with fluorescence-labeled secondary antibodies [anti-GH (green), anti-IGFIR (red), anti-INSR (red), and nuclei stained with 4′,6′-diamino-2-phenylindole (DAPI) (blue)]. In the merged images of control pituitaries, an asterisk marks the nucleus of a receptor-positive somatotrope (yellow). In SIRKO pituitary images, #, GH-immunopositive cell that lacks the INSR signal; @, an example of a GH-immunonegative cell that retains receptor signal. In SIGFRKO pituitary images, #, GH-immunopositive cell that lacks the IGFIR signal; @, an example of a GH-immunonegative cell that retains receptor signal. Data represent mean ± SEM of 3–6 mice/group.

Somatotrope-specific knockout of Insr or IgfIr differentially alters the GH secretion pattern

After confirming the somatotrope-specific knockout of INSR or IGFIR, we analyzed the impact on GH production in CHOW-fed mice. The average GH levels in blood samples taken over 3 consecutive days were elevated in KO mice compared with controls (Figure 4, A and B, insets), consistent with that previously reported in SIGFRKO mice (25). However, because average GH levels do not accurately reflect the pulsatile nature of GH secretion, GH data from these samples were arranged as ranked plots because it has been previously shown that an increase in intermediate ranked GH values corresponds to more frequent small amplitude peaks or broadening of high amplitude peaks, whereas differences in high GH values correspond to high peak amplitude (23). This analysis suggests SIRKO mice have wider or more frequent GH pulse release, whereas SIGFRKO mice display higher GH peak amplitude (Figure 4, A and B).

Figure 4.

Effects of in vivo somatotrope-specific knockout of IgfIr or Insr on GH/IGF-I axis. Plasma GH (A and B) and IGF-I (C and D) levels in CHOW-fed SIRKO (A and C) and SIGFRKO (B and D) males (left panels) and females (right panels) compared with appropriate controls are shown. GH levels were determined from tail vein samples taken 3 times a day during 3 consecutive days under fed conditions in 3-month-old animals (n = 5 mice/genotype per gender), whereas IGF-I levels were determined in trunk blood plasma obtained from the same animals at the time they were killed (5 months old). In A and B, upper graphs show mean GH values and lower graphs show ranked plot data. E and F, GH mRNA level in the pituitary gland of 8-month-old CHOW-fed SIRKO (E) and SIGFRKO (F) males (left panels) and females (right panels) compared with appropriate controls. G and H, GHRH-R mRNA level in the pituitary gland of 8-month-old CHOW-fed SIRKO (E) and SIGFRKO (F) males (left panels) and females (right panels) compared with appropriate controls. Data represent mean ± SEM of 5–9 mice/group. * P < .05; ** P < .01; ***, P < .001 and indicate values that significantly differ from controls.

As shown in Figure 4, C and D, irrespective of the GH pattern obtained, both SIRKO and SIGFRKO mice exhibited elevated IGF-I levels, which was not associated with changes in hepatic IGF-I or GH receptor expression (Supplemental Figure 6A). Although the rise in IGF-I levels in male SIRKO did not reach significance, significant elevations were observed in 2 other cohorts (Supplemental Figure 6B).

To better understand the impact of somatotrope-specific INSR or IGFIR knockout, we analyzed pituitary expression of GH and GHRH receptor (GHRH-R). Specifically, SIRKO mice exhibited lower GH mRNA levels (Figure 4E), suggesting the existence of a compensatory mechanism, which may include elevated IGF-I, acting through the intact IGFIR to suppress GH gene expression. Consistent with a role of IGF-I in suppressing pituitary GH expression, the rise in GH observed in SIGFRKO mice was associated with an elevation in pituitary expression of GH (Figure 4F) as well as GHRH-R (Figure 4H). These changes in pituitary expression were independent of changes in the expression of hypothalamic GH regulators (somatostatin or GHRH) in either model (Supplemental Figure 7), although we cannot discount the possibility of nucleus-specific expression changes or altered neuronal activity.

Taken together, these observations suggest insulin plays a dominant role in suppressing GH release, whereas IGF-I is more effective in suppressing GH synthesis. This conclusion is further supported by EM analysis showing that SIRKO, but not SIGFRKO, somatotropes have a greater proportion of GH secretory granules closely associated with the plasma membrane (Figure 5, A and B), and these changes occur without any alteration in total cell, nuclear or cytoplasmic area, granule density, or granule diameter (Supplemental Figure 8). Basal GH release from pituitary explants of SIRKO and SIGFRKO mice did not differ from explants of respective controls (Supplemental Figure 9). However, GHRH-stimulated GH release was enhanced in pituitary explants from SIRKO mice but attenuated in SIGFRKO explants (Figure 5, C and D). These results suggest enhanced GH output observed in SIRKO mice in vivo is dependent on GHRH.

Figure 5.

Impact of in vivo somatotrope-specific knockout of IgfIr and/or Insr on GH-producing cell morphology and on in vitro basal and GHRH-stimulated GH secretion. A and B, Morphological analysis of pituitary GH-producing cells by EM. Representative micrographs showing the localization of GH-containing immunogold-labeled granules in SIRKO and SIGFRKO pituitaries and quantification of percentage of granules within 300 nm margin of the plasma membrane are shown. Data represent mean ± SEM. **, P < .01 and indicate values that significantly differ from controls. C and D, GHRH-stimulated GH secretion from freshly isolated whole-mouse anterior pituitaries in vitro, obtained from 5- to 13-month-old male and female SIRKO or SIGFRKO mice and their respective controls (n = 4–24 pituitaries/group). Data represent mean ± SEM. *, P < .05; **, P < .01; ***, P < .001 and indicate values that significantly differ from vehicle-treated. a, Values that significantly differ from respective controls.

Somatotrope-specific INSR and/or IGFIR loss modifies, but does not prevent, high-fat diet-induced suppression of somatotrope function

Because insulin has been shown to be negatively correlated with GH levels in obese subjects (1–4), we examined whether the direct inhibitory action of insulin or IGF-I on somatotrope function contributes to reduced GH secretion. HF feeding induced a suppression of pituitary GH mRNA in male and female control mice (Insr^fl/fl, IgfIr^fl/fl, and Insr, IgfIr^fl/fl; Figure 6A, open bars), consistent with previous findings (11). Elimination of insulin or IGF-I negative feedback on somatotrope cells was not able to prevent the decline in GH expression (Figure 6A). To avoid the potential confounding effects of INSR compensating for loss of IGFIR or vice versa, as well as to avoid the confounding effects of elevated insulin and IGF-I observed after long-term HF feeding (Supplemental Figure 10B), we investigated whether loss of both the INSR and IGFIR would prevent the obesity-induced decline in GH expression (Insr,IgfIr^rGHpCre). Like single-receptor knockout, double-receptor knockout could not prevent the obesity-induced reduction in GH expression (Figure 6A), and this reduction translated into a reduction in circulating GH levels. As shown in Figure 6, B and C, and as previously reported (19), GH levels were elevated in double-receptor knockout mice under both LF and HF diets, compared with diet-matched controls. However, HF feeding still suppressed GH output in double-receptor knockout male and female mice (Figure 6, B–D), although it did not reach the low levels observed in HF-fed receptor intact controls.

Figure 6.

Impact of in vivo somatotrope-specific knockout of IgfIr and/or Insr on obesity-induced suppression of somatotrope function and circulating GH levels. A, Relative expression of GH in the pituitary glands of male and female HF-fed SIRKO, SIGFRKO and double receptor KO mice expressed as the percent of LF-fed controls within gender and genotype. B and C, Plasma GH levels in male (B) and female (C) LF- and HF-fed double-receptor knockout mice (Insr,IgfIr^rGHpCre) vs controls (Insr,IgfIr^fl/fl) from tail vein samples taken under fed conditions at 14 weeks of the diet (18 weeks old). Bar graphs show mean of GH values and line graphs show ranked plot analysis. Data represent mean ± SEM of n = 8–14 mice/group. *, P < .05; **, P < .01; ***, P < .001 and indicate values that significantly differ from LF-fed controls. a, Values that significantly differ from matched HF-diet receptor intact controls. The P values of ranked plot data analyzed by Wilcoxon signed-rank test are shown in panel D.

Somatotrope-specific knockout of Insr or IgfIr differentially alters hepatic gene expression, body composition, and glucose homeostasis in a sex-dependent fashion

It has been previously demonstrated that alterations in GH pulse pattern differentially alter metabolism (26, 27) and liver function (28). Therefore, we examined the impact of somatotrope-specific knockout of Insr or IgfIr on these end points and the data are summarized in Table 1, whereas full data are provided in Supplemental Figure 1 and Supplemental Figures 10–12. In general, enhanced GH levels observed in the SIRKO mice had a more dramatic impact on metabolic function, as compared with SIGFRKO mice. In addition, hepatic expression of genes previously reported to be GH dependent and female specific (28) were modestly but significantly increased in female SIRKO, but not SIGFRKO, mice across diets (Supplemental Figure 12). These results support the conclusion that the pattern of GH release differs between the SIRKO and SIGFRKO mice.

Table 1.

Impact of SIRKO and SIGFRKO on Body Composition, Glucose Homeostasis, Plasma Hormones/Metabolites Levels, and Liver Gene Expression vs Controls

	SIRKO				SIGFRKO
	Males		Females		Males		Females
	LF	HF	LF	HF	LF	HF	LF	HF
Body weight	↑	<	=	=	=	=	=	=
Lean	↑	<	=	=	=	=	=	=
Fat	=	=	↓	<	=	=	↓	=
Water	=	=	=	=	=	=	=	=
Fat depots	=	=	↓	↓	=	=	↓	=
GTT	=	=	=	=	=	=	=	=
ITT	=	↓	=	=	<	=	↓	=
Insulin	=	=	=	=	=	=	=	=
IGF-I	↑	=	↑	↑	=	=	↑	=
Fed glucose	=	=	=	=	=	=	=	=
Fast glucose	=	=	=	=	↑	=	=	=
Liver PRL-R	=	=	>	>	=	=	=	=
Liver Cyp3a16	=	=	↑	>	=	=	=	=
Liver Cyp3a41	=	=	↑	↑	=	=	=	=
Liver Cutl2 (Cux2)	=	=	>	↑	=	=	=	=
Liver C4a (Slp)	=	=	=	=	=	=	=	=

Abbreviation: PRL-R, prolactin receptor.

↑ and ↓ indicates parameters significantly (P < .05) increased or decreased, compared to diet- and gender-matched controls. > and < indicates parameters that trend (P < .1) to be increased or decreased, compared to diet- and gender-matched controls.

Discussion

More than 20 years ago, Melmed and colleagues (8, 9) presented a series of papers demonstrating that insulin can directly suppress GH synthesis and release from rat primary pituitary cell cultures and GH-producing pituitary cell lines, and this action could be blocked by immunoneutralization of the INSR. Since that time, the direct inhibitory actions of insulin on somatotrope function have been observed in a variety of other species including mice and primates (10, 11). In the present study, we confirmed and extended the early work of Melmed and colleagues and used Cre-mediated knockdown of INSR and IGFIR to show insulin and IGF-I work separately through their own receptors (INSR and IGFIR, respectively) to selectively suppress GH expression and release in vitro. Further evidence presented herein using pharmacological blockade in mouse and primate primary pituitary cultures indicated insulin and IGF-I require different pathways to mediate these actions because inhibition of mTORC1 or MEK blocked IGF-I- but not insulin-mediated suppression of GH expression and release in vitro. To our knowledge, this is the first clear-cut evidence showing that insulin and IGF-I require distinct receptors/intracellular signaling pathways to mediate their regulatory actions on the same end point in a primary culture system. However, it has been previously reported that the INSR and IGIFR require different signaling pathways (PI3K-Akt and MAPK, respectively) when heterologously transfected in NIH3T3 mouse fibroblasts to mediate regulation of VEGF gene expression (29).

Although the in vitro results clearly demonstrate that separate systems are involved in insulin- and IGF-I-mediated GH suppression, in vitro administration of insulin and IGF-I does not accurately reflect the physiological condition. Indeed, insulin levels rise in response to a meal and subsequently decline to low baseline levels, whereas total IGF-I levels are more abundant and stable. Of note, IGF-I circulates in plasma bound to IGF-binding proteins, and less than 1% of the IGF-I is thought to circulate in a biologically active unbound form (30). In addition, IGF-I is also locally produced in the pituitary gland, mainly in corticotrope cells and, to a lesser extent, in GH-producing cells, in which it could exert an autocrine feedback on GH expression/release (31, 32). Given the in vivo complexity in the presentation of insulin (intermittent) and IGF-I (more constant) to the somatotrope, the Cre-loxP system was used to inactivate the Insr and/or IgfIr in the somatotropes, in vivo, to determine the impact of somatotrope-specific loss of INSR and/or IGFIR on GH secretion in the context of physiologically relevant levels and patterns of insulin and IGF-I. The data presented herein demonstrate somatotrope-specific loss of INSR or IGFIR in vivo result in an increase in circulating GH sufficient to raise IGF-I levels, in which the negative feedback of GH/IGF-I to the hypothalamus (33, 34) cannot compensate for the loss of feedback at the level of the somatotrope. However, the differential impact of the loss of each receptor on the pattern of GH release, GH expression, somatotrope morphometry, and in vitro sensitivity to GHRH-stimulated GH release strongly suggests that, in an intact system, the dominant role of insulin is to suppress GH secretory vesicle release, whereas IGF-I serves to regulate synthesis under basal metabolic conditions.

Because GH has been negatively associated with insulin sensitivity in mice (35) and humans (2, 36), it is tempting to speculate that short-term increases in insulin due to regular meals or short bouts of overeating may rapidly initiate suppression of GH release, which would in turn serve to augment insulin-mediated nutrient storage, thus preparing the body to weather short periods of fasting. This may be particularly important in females, wherein maintenance of body fat is critical to maintain reproductive function (37). Although it has been previously reported that pituitary GH expression and circulating GH levels are suppressed by prolonged high-fat feeding (11, 38), our data indicate that the somatotrope-specific actions of long-term chronic elevations in insulin (or IGF-I) cannot solely account for obesity-induced GH suppression. Therefore, in the context of long-term high-fat feeding and obesity, the fall in circulating GH levels may be driven by insulin- and/or IGF-I-induced changes in hypothalamic function because there is clinical and experimental evidence supporting a role for insulin- and IGF-I-mediated increases in somatostatin tone (39). Alternatively, other systemic factors may act directly on the somatotrope and/or centrally to reduce GH output in the obese state (1, 2).

In summary, although both insulin and IGF-I can directly suppress GH output, the results of the current study indicate that the mechanisms by which each one mediates this effect are distinct. This conclusion is firmly based on the following collective observations: 1) Cre-loxP-mediated inactivation of the receptors in vitro demonstrated insulin and IGF-I can induce maximal suppression of GH expression and release when only the cognate receptor is present, 2) pharmacological blockade of MEK and mTORC1 in vitro blocked the inhibitory actions of IGF-I but not insulin, and 3) the pattern of GH release and associated changes in body composition, metabolic, and hepatic function differed between SIRKO and SIGFRKO mice. Our observations are of specific relevance to the field of GH-axis regulation, in that it has been previously assumed that IGF-I is the primary systemic factor that is directly responsible for homeostatic control of GH synthesis and release. The fact that insulin works through a receptor/signaling pathway unique from IGF-I and plays a different role in regulating GH secretion pattern in vivo dramatically shifts this commonly accepted paradigm. These results suggest circadian GH secretion is more tightly controlled by nutrient-induced insulin release than previously envisioned. However, because somatotrope-specific loss of INSR (and/or IGFIR) could not prevent obesity-induced GH suppression, additional mechanisms come into play to control GH output as metabolic disease progresses.

Our results are also of significance to the broader scientific community in that they provide clear evidence that the mechanisms by which insulin and IGF-I regulate common cell functions can be distinct and thus provide the rationale to continue to compare and contrast the separate roles of insulin and IGF-I. Such studies could reveal unique targets critical in designing drugs to selectively modify insulin or IGF-I signaling, which may be useful in the treatment of a variety of diseases, including diabetes and cancer.