The Absence of GH Signaling Affects the Susceptibility to High-Fat Diet-Induced Hypothalamic Inflammation in Male Mice

GH is important in metabolic control, and mice with disruption of the gene encoding the GH receptor (GHR) and GH binding protein (GHR−/− mice) are dwarf with low serum IGF-1 and insulin levels, high GH levels, and increased longevity, despite their obesity and altered lipid and metabolic profiles. Secondary complications of high-fat diet (HFD)-induced obesity are reported to be associated with hypothalamic inflammation and gliosis. Because GH and IGF-1 can modulate inflammatory processes, our objective was to evaluate the effect of HFD on hypothalamic inflammation/gliosis in the absence of GH signaling and determine how this correlates with changes in systemic metabolism. On normal chow, GHR−/− mice had a higher percentage of fat mass and increased circulating nonesterified free fatty acids levels compared with wild type (WT), and this was associated with increased hypothalamic TNF-α and phospho-JNK levels. After 7 weeks on a HFD, both WT and GHR−/− mice had increased weight gain, with GHR−/− mice having a greater rise in their percentage of body fat. In WT mice, HFD-induced weight gain was associated with increased hypothalamic levels of phospho-JNK and the microglial marker Iba-1 (ionized calcium-binding adapter molecule 1) but decreased cytokine production. Moreover, in GHR−/− mice, the HFD decreased hypothalamic inflammatory markers to WT levels with no indication of gliosis. Thus, the GH/IGF-1 axis is important in determining not only adipose tissue accrual but also the inflammatory response to HFD. However, how hypothalamic inflammation/gliosis is defined will determine whether it can be considered a common feature of HFD-induced obesity.

GH regulates metabolism and stimulates mitosis, proliferation, and cell differentiation (1), in addition to promoting neuroprotection (2). The GH receptor (GHR) belongs to the family of cytokine receptors with kinase activity (3, 4) and is abundantly expressed in the central nervous system, indicating the importance of its actions there (5). GHR/binding protein knockout (KO) mice (GHR−/−; Ghr/bp−/−) have a reduced body length and are long lived despite being obese (6). They have high serum GH concentrations, low or normal glycemia, greatly reduced plasma IGF-1 and insulin levels (7), and enhanced insulin sensitivity (8). Because GHR−/− mice are shorter than wild type (WT), most organs are proportionally smaller (allometrically scaled) but with some exceptions, such as the brain, pituitary, and specific adipose depots that are proportionally larger (9).

WT mice fed a high-fat diet (HFD) become obese and develop hyperinsulinemia and hyperglycemia (10–13). Although HFD also increases adiposity in GHR−/− mice (9), no significant alteration in glucose homeostasis is observed, which is most likely related to the effects of GH and IGF-1 on lipid and glucose metabolism (10–13). In addition to inducing inflammatory signaling in peripheral tissues resulting in insulin resistance, HFD is also associated with inflammation of the hypothalamus that is likewise linked to increased resistance to insulin and leptin. The absence of GH signaling has been related to obesity-associated inflammation and insulin resistance (14).

The hypothalamus is fundamental for the coordination of food intake and energy expenditure (15, 16), with specialized neurons relaying information provided by leptin and insulin regarding fat depot size to central mechanisms that regulate hunger and energy expenditure. Two neuron populations of the hypothalamic arcuate nucleus, one producing neuropeptide Y (NPY) and Agouti-related peptide (AgRP) and the other pro-opiomelanocortin (POMC) and cocaine and amphetamine regulated transcript (CART), are intricately involved in this process. Hypothalamic inflammation due to HFD-induced obesity has been associated with the development of secondary complications (17), and GH and IGF-1 have antiinflammatory actions (18, 19). Thus, the objective of the present study was to evaluate the effect of HFD intake on hypothalamic inflammation and gliosis in the absence of GH signaling and to determine how this correlates with changes in central and systemic metabolic parameters.

Materials and Methods

Electrophoresis reagents were from Bio-Rad Laboratories. The remaining chemicals and reagents were from Sigma or Merck unless otherwise indicated.

Animals

All procedures were approved by the Ohio University Institutional Animal Care and Use Committee and fully complied with all federal, state, and local policies.

Male GHR KO mice on a C57BL/6J background and homozygous WT littermates were used. Genotype was determined by PCR analysis of genomic DNA obtained from a tail sample collected at weaning. Mice were weaned at 25 days of age onto a standard rodent diet (14% fat, 26% protein, and 60% carbohydrates, ProLab RMH 3000; LabDiet). One week later, 2 male mice of the same genotype were housed per cage and fed with standard chow (normal diet [ND]) or with a HFD (20% carbohydrates, 20% protein, and 60% fat, D12492; Research Diets). This resulted in the next groups: WT + ND (n = 14), WT + HFD (n = 11), GHR−/− + ND (n = 12), and GHR−/− + HFD (n = 12).

After 50 days of HFD, mice were killed by cervical dislocation after 12 hours of fasting and hypothalami and trunk blood collected. Hypothalami were isolated on ice using the next boundaries: an anterior cut at the level of the optic chiasm, a posterior coronal cut anterior to the mammilary bodies, 2 sagittal cuts parallel to the lateral ventricles, and a dorsal horizontal cut at the level of the anterior commissure and then stored at −70°. Blood was allowed to clot, centrifuged (4°C, 10 min at 7000g), and the serum separated and stored at −70°C. Food intake and body weight were measured weekly from weaning until killing. Body length was measured at killing.

Body composition was assessed by nuclear magnetic resonance using a Bruker Minispec after 2 and 6 weeks of diet (1 wk before killing).

Blood glucose was measured 2 days before killing after 6 hours of fasting. Blood was collected from the tail and analyzed using a Lifescan One Touch glucometer (Johnson and Johnson).

Determination of IGF-1

Serum IGF-1 was measured using an OCTEIA immunoenzymometric assay from Immunodiagnostic Systems Ltd according to the manufacturer’s instructions. This assay has a sensitivity limit of 63 ng/mL. The intra- and interassay coefficients of variation were 6.8% and 7.3%, respectively.

Determination of hypothalamic cytokines levels

Hypothalamic levels of TNF-α, IL-1β, IL-6, IL-2, IL-10, interferon-γ, monocyte chemoattractant protein (MCP)-1, leptin, and adiponectin were measured by a bead array assay (Procarta immunoassays; Affymetrix). Beads were analyzed in a Bio-Plex suspension array system 200 (Bio-Rad Laboratories). Raw data (mean fluorescence intensity) were analyzed using the Bio-Plex Manager software. Results are reported as pg/mg of protein.

Determination of serum leptin, insulin, and cytokines levels

Serum levels of TNF-α, IL-1β, IL-6, IL-2, IL-10, interferon-γ, MCP-1, and adiponectin were measured with a bead array assay (Procarta immunoassays; Affymetrix). Leptin and insulin levels were measured by using a mouse metabolic bead array assay (MMHMAG-44K; Millipore) in accordance with manufacturer’s instructions.

Homeostasis model assessment (HOMA) index

The HOMA index was calculated by using the formula: HOMA-IR = [insulin (μU/mL) × glucose (mg/dL)]/405.

Determination of serum cholesterol, triglycerides, free fatty acids, total lipids, and glycerol

Total lipids, triglycerides, and cholesterol were measured by using enzymatic colorimetric assay kits from SpinReact, nonesterified free fatty acids (NEFA) with a kit from Wako Chemicals and free glycerol levels with a kit from Sigma. All assays were performed according to the manufacturer’s instructions and absorbance read in a microtiter plate reader (Tecan Infinite M200).

Real-time PCR

Total RNA was purified after the instructions of RNeasy Plus Mini kit (QIAGEN). cDNA was synthesized from 1.5 μg of total RNA by using a high-capacity cDNA reverse transcription kit (Applied Biosystems). Quantitative RT-PCR was performed using assay-on-demand kits (Applied Biosystems) for NPY (Mn03048253), AgRP (Mn01431703), POMC (Mn00435874), CART (Mn04210469), IGF-1 (Mm00439561), and IGF-1 receptor (Mm00802831) and TaqMan Universal PCR Master Mix (Applied Biosystems) according to the manufacturer’s protocol in an ABI PRISM 7000 Sequence Detection System (Applied Biosystems). Relative TNF-α, IL-1β, and IL-6 mRNA levels were measured with SYBR Green (Roche). Primers (Sigma) were designed with Primer Express software. The forward and reverse sequences were the following: 5′-CATCTTCTCAAAATTCGAGTGA-3′ and 5′- GGGAGTAGACAAGGTACAAC-3′, 5′-GCAACTGTTCCTGAACTCAACT-3′ and 5′-ATCTTTTGGGGTCCGTCAACT-3′, and 5′-TAGTCCTTCCTACCCCAATTTCC-3′ and 5′-TTGGTCCTTAGCCACTCCTTC-3′, respectively. The reaction contained 300nM each primer. Values were normalized to glyceraldehyde 3-phosphate dehydrogenase (4352339E). The ΔΔC_T method was used to determine relative expression levels.

Protein purification and quantification

The supernatant collected in the RNA extraction process was diluted in acetone, frozen, and centrifuged. Proteins were resuspended in 100 μL of 3-[(3-Cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS) buffer, containing 7M urea, 2M thiourea, 4% CHAPS, and 1M Tris (pH 8.8). Protein concentration was measured using the Bio-Rad Protein Assay.

Immunoblotting

Proteins were resolved by SDS-PAGE and transferred onto polyvinylidene fluoride membranes. Filters were blocked with Tris-buffered saline with 0.1% (vol/vol) Tween 20 and 5% (wt/vol) BSA or nonfat dried milk and incubated overnight at 4°C with primary antibody (see Supplemental Table 1). Filters were washed and incubated with the corresponding secondary antibody conjugated with peroxidase (1:2000; Pierce). Peroxidase activity was visualized by chemiluminiscence (Bio-Rad) and quantified by densitometry using a Kodak Gel Logic 1500 Image Analysis system and Molecular Imaging Software, version 4.0. Data are normalized to control values on each membrane.

Immunofluorescence

Immunofluorescence for glial fibrillary acidic protein (GFAP) and vimentin was carried out on cryostat sections (20 μm) fixed with 4% paraformaldehyde (15 min at room temperature [RT]) and blocked in phosphate buffer (0.1M [pH 7.4] containing 3% BSA and 1% Triton X-100) for 24 hours at 4°C. Sections were incubated with anti-GFAP (1:500) or antivimentin (1:100) in blocking buffer for 48 hours at 4°C, washed with phosphate buffer with 0.1% Triton X-100, and incubated for 90 minutes at RT with antirabbit IgG-biotin (1:1000; Thermo Scientific). Slides were washed and incubated with streptavidin Alexa Fluor 488 (1:1000; Molecular Probes) for 90 minutes at RT. Immunofluorescence was visualized with a DM IRB confocal microscope (Leica).

Statistical analysis

Statistics were performed using GraphPad Prism 4.0 (GraphPad Software). Data are presented as mean ± SEM. Two-way ANOVA was used to analyze the effects of genotype and diet. Weight gain was analyzed by ANOVA with repeated measures. When significant effects were found by two-way ANOVA, to determine the differences among groups one-way ANOVA followed by the Newman-Keuls test or a Student’s t test was used. Results were considered statistically significant when P < .05. Only physiologically relevant differences are represented in the figures.

Results

Changes in weight, fat mass, and length

Weight gain was affected by genotype (F_(1,5) = 80.48; P < .001), diet (F_(1,5) = 20.48; P < .01), and time (F_(6,30) = 264.01; P < .001) with interactions between time and diet (F_(6,30) = 13.30; P < .01) and time and genotype (F_(6,30) = 19.02; P < .01). GHR−/− mice on ND gained less weight than all other groups. Mice of both genotypes fed a HFD gained more weight than ND-fed mice over time, with this gain being greater in GHR−/− mice compared with WT (Figure 1A).

A, Weight gain during HFD in WT and GHR KO (KO) mice fed with a ND or HFD. Statistical significance by repeated measures ANOVA. a, WT ND significantly different from WT HF; b, KO ND significantly different from KO HF; c, WT ND significantly different from KO ND; d, WT HF significantly different from KO HF. Food intake (B), change in fat mass (%) (C), change in lean mass (%) (D), and HOMA-IR index (E) during HFD in WT and GHR KO (KO) mice fed with a ND or HFD. Statistical significance by ANOVA. ***, P < .001 vs WT ND; ###, P < .001 vs KO ND; &&&, P < .001 vs WT HF; n = 4–14/group

Figure 1

Body length was modified by genotype (F_(1,45) = 1307; P < .0001) and diet (F_(1,45) = 31.36; P < .0001). GHR−/− mice were shorter than WT mice independently of diet. HFD intake increased body length (WT ND, 88.79 ± 0.60 mm; WT HF, 91.55 ± 0.72 mm; GHR−/− ND, 60.75 ± 0.55 mm; and GHR−/− HF, 66.25 ± 1.01 mm; P < .0001).

Energy intake was affected by genotype (F_(1,19) = 101.0; P < .0001) and diet (F_(1,19) = 28.55; P < .0001), with no interaction between factors. GHR−/− mice consumed less energy than WT. Energy intake was increased in WT and GHR−/− mice on a HFD (Figure 1B). However, when energy intake is expressed relative to body size, there was an effect of genotype (F_(1,19) = 30.19; P < .0001) but not diet with an interaction between diet and genotype (F_(1,19) = 5.95; P < .05). GHR−/− mice consumed more than WT ND and energy intake was increased in WT mice consuming a HFD but not in GHR−/− mice (WT ND, 3.17 ± 0.20 kcal/g body weight (bw); WT HF, 3.88 ± 0.19 kcal/g bw; GHR−/− ND, 4.41 ± 0.10 kcal/g bw; and GHR−/− HF, 4.36 ± 0.09 kcal/g bw; P < .0001).

The percentage of fat (Figure 1C) and lean mass (Figure 1D) after 7 weeks of HFD was affected by genotype (F_(1,45) = 13.20; P < .001 and F_(1,45) = 18.80; P < .0001, respectively) and diet (F_(1,45) = 39.65; P < .0001 and F_(1,45) = 34.81; P < .0001, respectively) with an interaction between factors on lean mass (F_(1,45) = 4.53; P < .05). WT ND mice had a decrease in fat mass and an increase in lean mass, whereas GHR−/− ND mice gained adipose tissue and lost lean mass. Mice fed a HFD gained more fat and decreased lean mass compared with ND fed mice, with this effect being greater in GHR−/− mice (Figure 1, C and D).

Glycemia and serum insulin and leptin levels

Glycemia was affected by genotype (F_(1,45) = 23.71; P < .0001) and diet (F_(1,45) = 21.85; P < .0001) with an interaction between factors (F_(1,45) = 7.45; P < .01). On a ND, glucose levels were decreased in GHR−/− mice compared with WT. HFD increased glucose levels in GHR−/− mice, with no effect in WT mice (Table 1). Insulin levels were affected by genotype (F_(1,34) = 12.99; P < .001) and diet (F_(1,34) = 4.15; P < .05), being decreased in GHR−/− ND compared with WT ND mice. HFD increased insulin levels, but this was only significant in GHR−/− mice (Table 1).

Table 1

Levels of Fasting Glucose, Insulin, Cytokines, and Lipids in Serum at Killing

	WT ND	WT HFD	KO ND	KO HFD
Glycemia	148.3 ± 6.3	163.6 ± 7.1	88.3 ± 5.0c	146.8 ± 11.7e
Insulin	0.95 ± 0.19	1.27 ± 0.25	0.22 ± 0.77b	0.66 ± 0.20d
TNF-α	1.84 ± 0.08	1.80 ± 0.11	2.83 ± 0.47a	2.31 ± 0.20f
Interferon-γ	0.16 ± 0.02	0.21 ± 0.02	0.31 ± 0.02c	0.33 ± 0.04h
IL-2	1.58 ± 0.23	2.23 ± 0.15b	1.33 ± 0.08	1.28 ± 0.14g
IL-10	1.31 ± 0.08	1.11 ± 0.04b	1.11 ± 0.12	1.75 ± 0.17dg
IL-6	0.88 ± 0.10	0.75 ± 0.06	1.04 ± 0.15	1.04 ± 0.09
MCP-1	3.39 ± 0.34	3.74 ± 0.39	4.19 ± 0.29	4.11 ± 0.29
IL-1β	4.52 ± 0.26	4.67 ± 0.45	5.23 ± 0.71	5.13 ± 0.61
Adiponectin	5837 ± 1557	5248 ± 2191	5808 ± 1317	9699 ± 2871
Triglycerides	112.5 ± 6.8	106.3 ± 3.2	94.9 ± 2.8	113.6 ± 7.7
Total cholesterol	169.7 ± 9.1	218.7 ± 14.4c	144.4 ± 9.2	230.2 ± 22.3e
NEFA	1.12 ± 0.08	0.86 ± 0.05c	1.76 ± 0.16c	1.26 ± 0.07eh
Total lipids	666.1 ± 17.2	663.9 ± 27.4	558.0 ± 20.1b	608.2 ± 18.9b
Glycerol	0.023 ± 0.003	0.025 ± 0.004	0.043 ± 0.006a	0.041 ± 0.009af

Glycemia (mg/dL), insulin (ng/mL), TNF-α (pg/mL), interferon-γ (pg/mL), IL-2 (pg/mL), IL-10 (pg/mL), IL-6 (pg/mL), MCP-1 (pg/mL), IL-1β (pg/mL), adiponectin (pg/mL), triglycerides (mg/dL), total cholesterol (mg/dL), NEFA (nM), total lipids (mg/dL), and glycerol (mg/mL) levels measured in serum at killing. One-way ANOVA; n = 5- 14/group.

^a

P < .05.

^b

P < .01.

^c

P < .001 vs WT ND.

^d

P < .01.

^e

P < .001 vs KO ND.

^f

P < .05.

^g

P < .01.

^h

P < .001 vs WT HFD.

HOMA, an index of insulin resistance, was affected by genotype (F_(1,29) = 23.66; P < .0001) and diet (F_(1,29) = 12.33; P < .01), with no interaction between factors. On a ND, HOMA was decreased in GHR−/− mice. HFD increased HOMA in all mice (Figure 1E).

Circulating leptin levels were affected by diet (F_(1,34) = 17.31; P < .001), being elevated after HFD in WT and GHR−/− mice, with no difference between KO ND and WT ND (Figure 2A). Positive correlations between leptin levels and fat gain (Pearson r = 0.77; P < .0001) (Figure 2B) and leptin and insulin levels (Pearson r = 0.46; P < .05) were observed.

A, Serum leptin concentration in WT and GHR KO (KO) mice fed with a ND or HFD. B, Linear correlation between serum leptin levels and gain of fat (%) during HFD. Relative mRNA levels of NPY (C), AgRP (D), POMC (E), and CART (F) in the hypothalamus of WT and GHR KO (KO) mice fed with a ND or HFD. Statistical significance by ANOVA. NS, no significant; **, P < .01 and ***, P < .001 vs WT ND; ##, P < .01 and ###, P < .001 vs KO ND; &&&, P < .001 vs WT HF; n = 3–10/group

Figure 2

Serum lipids levels

There was an interaction between diet and genotype (F_(1,44) = 4.47; P < .05) on triglyceride levels; although not statistically significant, there was a trend towards HFD decreasing triglycerides in WT and increasing them in GHR−/− mice. HFD was associated with elevated total cholesterol levels (F_(1,44) = 21.02; P < .0001) in both WT and GHR−/− mice. NEFA levels were affected by genotype (F_(1,43) = 28.26; P < .0001) and diet (F_(1,43) = 15.01; P < .001). GHR−/− ND had higher NEFA levels than WT ND mice. HFD decreased NEFA levels regardless of genotype, with this effect being greater in GHR−/− mice. Total lipids were modified by genotype (F_(1,45) = 15.48; P < .001) but not diet. GHR−/− mice had lower total lipids than WT ND mice. Circulating glycerol levels were affected by genotype (F_(1,41) = 10.16; P < .01), with GHR−/− mice having higher levels than WT (Table 1).

Serum cytokine levels

There was an effect of genotype on TNF-α (F_(1,20) = 8.10; P < .01) and interferon-γ (F_(1,20) = 25.82; P < .0001) (Table 1), with GHR−/− mice having higher levels than WT independently of diet. Serum IL-2 levels were affected by genotype (F_(1,20) = 13.85; P < .01) with an interaction between genotype and diet (F_(1,20) = 4.88; P < .05). HFD increased IL-2 levels in WT with no effect in GHR −/− mice. An interaction between genotype and diet was found on IL-10 (F_(1,20) = 13.71; P < .01), because HFD reduced serum IL-10 levels in WT and increased them in GHR−/− mice. There was no effect of genotype or diet on serum IL-6, MCP-1, IL-1β, or adiponectin levels.

Hypothalamic metabolic neuropeptides

The mRNA levels of NPY and AgRP were affected by genotype (F_(1,8) = 36.02; P < .001 and F_(1,8) = 76.64; P < .0001, respectively) and diet (F_(1,8) = 55.68; P < .0001 and F_(1,8) = 14.21; P < .01, respectively), with an interaction between factors (F_(1,8) = 17.74; P < .01 and F_(1,8) = 6.06; P < .05, respectively). The mRNA levels of these orexigenic peptides were lower in GHR−/− mice. HFD decreased NPY and AgRP mRNA levels in WT and GHR−/− mice (Figure 2, C and D).

POMC mRNA levels were affected by genotype (F_(1,8) = 15.13; P < .01) and diet (F_(1,8) = 55.96; P < .0001). GHR−/− ND mice had lower levels than WT ND. HFD increased POMC mRNA levels in WT and GHR−/− mice (Figure 2E). CART mRNA levels were not affected by genotype or diet (Figure 2F).

Positive correlations between leptin levels and POMC mRNA levels (Pearson r = 0.63; P < .05) and insulin levels and POMC mRNA levels (Pearson r = 0.63; P < .05) were observed.

Hypothalamic inflammatory signaling

Because hypothalamic inflammatory processes in response to HFD are reported to be associated with secondary complications of excess weight (see reference 40 below), inflammatory cytokine levels were measured in hypothalamic tissue. There was an effect of diet and an interaction between genotype and diet on TNF-α (diet: F_(1,19) = 10.32, P < .01; interaction: F_(1,19) = 11.66, P < .01), IL-1β (diet: F_(1,18) = 10.43, P < .01; interaction: F_(1,18) = 12.87, P < .01), IL-6 (diet: F_(1,18) = 16.61, P < .001; interaction: F_(1,18) = 8.69, P < .01), IL-2 (diet: F_(1,20) = 4.50, P < .05; interaction: F_(1,20) = 8.70, P < .01), MCP-1 (diet: F_(1,19) = 10.34, P < .01; interaction: F_(1,19) = 8.25, P < .01), and IL-10 (diet: F_(1,18) = 4.77, P < .05; interaction: F_(1,18) = 9.21, P < .01) levels. There was an interaction between genotype and diet on interferon-γ (F_(1,19) = 12.73; P < .01) and adiponectin (F_(1,19) = 10.29; P < .01). Hypothalamic leptin levels were affected by genotype (F_(1,19) = 5.39; P < .05) and diet (F_(1,19) = 7.77; P < .05), with an interaction between factors (F_(1,19) = 6.24; P < .05). GHR−/− ND mice had higher TNF-α, IL-1β, IL-2, MCP-1, IL-10, interferon-γ, adiponectin, and leptin levels than WT ND. HFD in GHR−/− mice reduced all cytokines to levels found in WT mice. However, in WT mice HFD had no effect (Table 2).

Table 2

Hypothalamic Levels of Cytokines

	WT ND	WT HF	KO ND	KO HF
TNF-α	7.76 ± 0.21	7.84 ± 0.39	9.75 ± 0.45c	7.16 ± 0.41f
IL-1β	31.97 ± 1.82	32.52 ± 1.51	36.91 ± 1.29b	26.43 ± 1.47eh
IL-6	3.47 ± 0.23	3.24 ± 0.19	4.01 ± 0.22	2.62 ± 0.06eh
IL-2	1.47 ± 0.09	1.55 ± 0,12	1.79 ± 0.10a	1.30 ± 0.07d
MCP-1	6.86 ± 0.25	6.75 ± 0.33	8.07 ± 0.36b	6.05 ± 0.34e
IL-10	7.22 ± 0.29	7.61 ± 0.32	9.07 ± 0.70b	6.70 ± 0.21eh
Interferon-γ	0.21 ± 0.01	0.24 ± 0.02	0.27 ± 0.01b	0.19 ± 0.01eh
Adiponectin	74.79 ± 4.14	105.40 ± 15.56a	94.61 ± 6.53a	65.40 ± 4.31dg
Leptin	5.46 ± 0.27	5.32 ± 0.45	7.74 ± 0.68b	5.24 ± 0.30e

Abbreviation: prot, protein. Levels of TNF-α (pg/mg prot), IL-1β (pg/mg prot), IL-6 (pg/mg prot), IL-2 (pg/mg prot), MCP- 1 (pg/mg prot), IL-10 (pg/mg prot), interferon-γ (pg/mg prot), adiponectin (pg/mg prot), and leptin (pg/mg prot) measured by bead array assay in hypothalamus. One-way ANOVA; n = 5–6/group.

^a

P < .05.

^b

P < .01.

^c

P < .001 vs WT ND.

^d

P < .05.

^e

P < .01.

^f

P < .001 vs KO ND.

^g

P < .05.

^h

P < .01 vs WT HFD.

To corroborate results obtained by Bio-Plex, Western blotting was used to measure IL-6, IL-1β, and adiponectin levels. There was an interaction between genotype and diet on adiponectin (F_(1,18) = 37.6; P < .001), IL-6 (F_(1,18) = 5.24; P < .05), and IL-1β (F_(1,18) = 7.05; P < .05) levels. GHR−/− ND mice had higher adiponectin and IL-1β levels than WT ND, with no change in IL-6. HFD increased adiponectin levels in WT mice and reduced adiponectin, IL-6 and IL-1β levels in GHR−/− mice (Supplemental Figure 1 .

Hypothalamic mRNA levels of TNF-α, IL-6 and IL-1β were measured to analyze local synthesis. There was an effect of diet (F_(1,19) = 12.73; P < .01) and interaction between genotype and diet (F_(1,19) = 12.73; P < .01) on TNF-α mRNA levels. IL-6 mRNA levels were affected by diet (F_(1,19) = 4.63; P < .05) and genotype (F_(1,19) = 17.83; P < .001). There was an interaction between diet and genotype on IL-1β mRNA levels (F_(1,16) = 9.68; P < .01). GHR−/− ND mice had lower levels of TNF-α, IL-6 and IL-1β mRNA than WT ND. HFD decreased the levels of the 3 cytokines in WT with no effect in GHR−/− mice (Supplemental Figure 2).

To study the possible existence of gliosis, GFAP, vimentin, and Iba-1 were measured. GFAP levels were affected by genotype (F_(1,20) = 72.03; P < .0001), with GHR−/− mice having lower levels than WT independently of diet (Figure 3A). Vimentin levels were affected by genotype (F_(1,8) = 36.05; P < .001) and diet (F_(1,8) = 17.83; P < .01), with an interaction between factors (F_(1,8) = 6.16; P < .05). GHR−/− ND mice had higher vimentin levels than WT ND. HFD reduced vimentin levels only in GHR−/− mice (Figure 3B). The neuronal marker neuron-specific class III beta-tubulin (Tuj-1) was not affected by diet or genotype (Figure 3C). In contrast, Iba-1 levels were affected by genotype (F_(1,20) = 16.61; P < .01) with an interaction between genotype and diet (F_(1,20) = 26.55; P < .001). HFD increased Iba-1 in WT with no effect in GHR−/− mice (Figure 3D).

Immunoblots probed with antibodies towards GFAP (A), vimentin (B), Tuj-1 (C), or Iba-1 (D) in the hypothalamus of WT and GHR KO (KO) mice fed with a ND or HFD

Figure 3

Statistical significance by ANOVA. **, P < .01 and ***, P < .001 vs WT ND; ###, P <.001 vs KO ND; &&, P < .01 and &&&, P < .001 vs WT HF; n = 3–6/group.

Immunofluorescent labeling for GFAP was less intense in the arcuate nucleus of GHR−/− mice (Figure 4). Labeling for vimentin was found in cells lining the third ventricle, which are most likely tanycytes, and was more intense in GHR−/− mice regardless of food intake. No vimentin labeled astrocytes were observed in any group (Figure 4). Projections labeled with GFAP and vimentin appeared to be longer in mice fed a HFD independently of genotype.

Immunostaining for GFAP and vimentin in the hypothalamus of WT and GHR KO (KO) mice fed with a ND or HFD

Figure 4

Scale bar, 50 μm. Arrows indicate longer projections.

Intracellular signaling

Intracellular signaling events involved in hypothalamic inflammation were analyzed. Levels of phospho (p)-Jun N terminal kinase (JNK) were affected by an interaction between genotype and diet (F_(1,20) = 9.85; P < .01). Increased p-JNK was observed in GHR−/− mice fed a ND compared with WT and HFD increased p-JNK levels in WT but not GHR−/− mice (Figure 5A). Nuclear factor-kappa B activation, assessed by p-inhibitor of kappa B levels, was not different between groups (data not shown).

Immunoblots probed with antibodies towards p-JNK (A), p-Akt (B) p-ERK1/2 (C), or SOCS-3 (D) in the hypothalamus of WT and GHR KO (KO) mice fed with ND or HFD

Figure 5

Statistical significance by ANOVA. *, P < .05 and **, P < .01 vs WT ND; #, P < .05 and ##, P < .01 vs KO ND; &&, P < .01 vs WT HF; n = 3–6/group.

Because hypothalamic inflammation can affect insulin and leptin sensitivity, their signaling pathways were analyzed. Serine phosphorylation of Akt (mammalian homologue of retroviral oncogene v-akt encoding a serine/threonine protein kinase), a marker of phosphatidylinositol-4,5-bisphosphate 3-kinase (PI3K) activation, was affected by genotype (F_(1,16) = 11.21; P < .01) and diet (F_(1,16) = 8.18; P < .05), with an interaction between factors (F_(1,16) = 5.82; P < .05). GHR−/− mice had lower p-Akt levels than WT. HFD reduced p-Akt levels in WT but had no effect in GHR−/− mice (Figure 5B). There was an effect of diet (F_(1,20) = 6.64; P < .05) but not genotype on pERK levels. HFD decreased pERK levels in both WT and GHR−/−, although it was only significant in GHR−/− mice (Figure 5C). No differences in phospho-signal transducer and activator of transcription 3 (Tyr705) were found (data not shown). There was an effect of genotype (F_(1,8) = 13.42; P < .01) and an interaction between genotype and diet (F_(1,8) = 5.26; P < .05) on suppressor of cytokine signaling 3 (SOCS-3) levels. GHR−/− mice fed a ND had higher SOCS-3 levels than WT ND. HFD reduced SOCS-3 levels only in GHR−/− mice, with no effect on WT (Figure 5E).

IGF-1 levels

Circulating IGF-1 was undetectable in GHR−/− mice. HFD increased IGF-1 levels in WT mice (WT ND, 303 ± 15 ng/mL and WT HF, 363 ± 15 ng/mL; P < .05).

There was an interaction between genotype and diet on hypothalamic IGF-1 mRNA levels (F_(1,8) = 22.35; P < .001). GHR−/− ND mice had higher IGF-1 mRNA levels than WT ND, with HFD increasing them in WT and decreasing them in GHR−/− mice (Figure 6A). IGF-1 receptor levels were affected by genotype (F_(1,8) = 16.84; P < .01) and diet (F_(1,8) = 19.24; P < .01) with an interaction between factors (F_(1,8) = 59.79; P < .0001) with levels being increased in GHR−/− ND mice and HFD increasing them in WT and reducing them in GHR−/− mice (Figure 6B).

Relative mRNA levels of IGF-1 (A) and IGF-1 receptor (B) in the hypothalamus of WT and GHR KO (KO) mice fed with a ND or HFD

Figure 6

Statistical significance by ANOVA. **, P < .01 vs WT ND; ##, P < .01 vs KO ND; &&, P < .01 vs WT HF; n = 3–6/group.

Discussion

Mice with disruption of the ghr gene are GH resistant with low circulating IGF-1 and high GH levels resulting in dwarfism (6). They are relatively obese (20–22), store a large fraction of their excess fat mass in the sc region (20), and are hyperphagic relative to their body size (21, 23). The response of male GHR−/− mice to dietary modifications has been reported (9), but there are no data in the literature regarding how the lack of GH signaling affects hypothalamic inflammatory processes and gliosis in response to HFD. Here, we found that although GH resistant mice gained relatively more fat mass than WT in response to a HFD, markers of gliosis and inflammation were actually reduced.

WT and GHR−/− mice increase energy intake on a HFD and gain substantial amounts of weight. However, when analyzed relative to body weight, energy intake and weight gain are greater in GHR−/− mice compared with WT regardless of the type of diet, as previously reported (24–26). On a ND, GHR−/− mice appear to have increased insulin sensitivity, because they have lower glucose and insulin levels and HOMA index compared with WT, even though they are more obese. Insulin levels are reported to increase in both genotypes in response to HFD with no significant rise in glycemia (26), but we found HFD to increase glucose and insulin levels only in GHR−/−. Here, the effects of HFD were analyzed after 7 weeks, whereas previous studies were performed after 12 or 17 weeks (26, 27). Hence, some of these differences may be due to the duration of the diet. On a ND, GHR−/− mice have a lower HOMA index than WT, and this possible increase in insulin sensitivity may be due to the lack of the antiinsulin effects of GH (28) and the low levels of circulating IGF-1 (29). HFD increased the HOMA index in both genotypes, although glycemia and insulin only increased in GHR−/− mice.

Leptin levels are reported to be elevated in GHR−/− mice on a ND, which is consistent with their increased adiposity, but here, they were not. This is possibly due to the younger age at killing, which was 11.5 weeks here in contrast to 6 months in previous studies (24). HFD increased leptin levels in both genotypes, and this correlated positively with fat mass gain. GHR−/− mice have a high percentage of body fat throughout their lifespan (30), with a disproportionate amount of fat deposition in the sc white adipose depot (30), probably due to loss of the antilipogenic effects of GH (28).

Hypothalamic mRNA levels of NPY, AgRP, and POMC are decreased in GHR−/− mice. The absence of GH signaling could underlie the decline in NPY and AgRP levels, because these neurons express GHR and are stimulated by GH (31). Leptin and insulin are important anorexigenic signals to the hypothalamus (32), and both NPY/AgRP and POMC/CART neurons express receptors for these 2 hormones (32, 33). The mRNA levels of POMC in GHR−/− mice are reduced probably due to the decline in circulating insulin and leptin. HFD decreased NPY and AgRP mRNA levels and increased POMC mRNA levels in WT and GHR−/− mice, as would be expected in attempt to reduce energy intake. It is of note that circulating leptin and insulin levels paralleled POMC mRNA levels in the 4 experimental groups.

Individuals with mutations in the GHR gene usually display high free fatty acid levels (34). Similarly, NEFA levels were increased in GHR−/− mice. In older GHR−/− mice, NEFA levels are reported to be normal (35), and here, HFD intake reversed the elevation of NEFA, resembling what occurs in older GHR−/− mice. Triglyceride levels are reported to be low to normal in young GHR−/− mice (8, 21, 35, 36), similar to that reported here, and were not affected by diet. Reduced total lipid and glycerol levels in GHR−/− mice are probably related to their increased lipid deposition (37). Cholesterol levels were increased in response to HFD in both genotypes as a normal consequence of high-fat intake (38).

HFD is reported to activate an inflammatory response by microglia and astrocytes in the hypothalamus (39, 40). This could be due, at least in part, to the elevation in NEFA levels as fatty acids can directly induce hypothalamic inflammation (41, 42). Signaling through JNK and NFκB induces cytokine gene transcription and a rise in local TNF-α, IL-1β, IL-6, and interferon-γ levels that promotes hypothalamic inflammation (39). Both TNF-α and the consumption of fat-rich diets activate the serine-kinase JNK in hypothalamic cells (43). Activated JNK targets insulin receptor substrate 1, catalyzing its serine phosphorylation and hampering its capacity to appropriately act as a docking protein for PI3K. This reduces insulin-induced activation of Akt and coincides with functional resistance to insulin (32). Inhibition of JNK or TNF-α reverses the effects of HFD or increased cytokines and reestablishes normal insulin activity in the hypothalamus (44). Hypothalamic phosphoserine Akt is strongly inhibited in GHR−/− mice probably as a consequence of increased JNK activation, as well as low insulin levels. HFD also inhibited Akt activation in WT mice. These data are in agreement with previous studies, whereby diet induced obesity results in central insulin resistance when measured either in terms of its ability to reduce food intake or to induce IRS-PI3K signal transduction (45). In GHR−/− mice, no further decrease in p-Akt levels was found in response to HFD, and how this relates to changes in central insulin sensitivity remains to be evaluated.

Both systemic and central inflammations have been implicated in development of secondary complications of obesity (46–48). On a ND, GHR−/− mice had increased circulating TNF-α and interferon-γ levels, with this possibly resulting from their increased adiposity. HFD increased serum levels of proinflammatory IL-2 and decreased antiinflammatory IL-10 in WT mice. In contrast, HFD did not modify the already altered pattern of circulating cytokines in GHR−/− mice.

The increase in serum inflammatory markers in GHR−/− mice on a ND was coincident with an inflammatory pattern in the hypothalamus. Adiponectin, leptin, TNF-α, IL-1β, IL-2, interferon-γ, and MCP-1 were increased in the hypothalamus of GHR−/−mice. Hypothalamic inflammation is reported to lead to the anomalous control of caloric intake and energy expenditure (49) and could be involved in the increased energy intake in GHR−/− mice. However, although TNF-α and IL-1β protein levels were increased and IL-6 unchanged, their mRNA levels were decreased in the hypothalamus, suggesting that there may be increased uptake of these cytokines from the circulation. In WT mice, HFD did not modify hypothalamic protein levels of cytokines but decreased mRNA levels of TNF-α, IL-6, and IL-1β. In GHR−/− mice, HFD did not change mRNA levels of cytokines in the hypothalamus, although the protein levels were normalized. These results suggest that the loss of GH signaling increases proinflammatory cytokines in serum, which is associated with a rise in the hypothalamus possibly due to increased cytokine uptake, because synthesis is decreased regardless of diet. How this relates to changes in systemic lipid metabolism, because NEFA levels are increased in GHR−/− mice and are implicated in obesity associated inflammation (41, 42), deserves further investigation. Differential maturation of hypothalamic metabolic circuits in GHR−/− mice could modify cellular metabolism, because these mice had low levels of NPY, AgRP, and POMC mRNA, and their response to HFD.

The adiponectin detected in the hypothalamus, which is most likely the result of transport across the blood brain barrier (50), could exert antiinflammatory actions and promote insulin sensitivity (51). HFD increased hypothalamic adiponectin levels in WT but reduced them in GHR−/− mice. However, circulating levels of this cytokine were not different between experimental groups regardless of diet. Previous studies report that GHR KO mice have increased serum adiponectin levels (9). It is possible that we do not find these differences in these relatively young mice and that this difference would appear with maturation.

Hypothalamic GFAP levels were reduced in GHR−/− mice, which could be related to the role of GH in the differentiation of astrocytes and suggests that these mice may have fewer GFAP positive astrocytes (52). In contrast to other studies (40), no increase in GFAP levels or vimentin positive astrocytes was found in response to HFD, although an increase in the length of GFAP positive projections can be appreciated in the arcuate nucleus of HFD mice. The hypothalamic astroglial response to weight gain may depend on a number of factors, including developmental events, the type of diet, the length of time exposed to the diet, and the genetic/epigenetic background of the animal (53, 54). Moreover, whether HFD-induced astrocyte responses represent other functions of these glial cells, such as structural changes involved in adaptation of the animal to the new metabolic situation, in addition to inflammatory responses and protection of neurons, remains to be demonstrated.

The microglial marker Iba-1 was induced by HFD in WT mice, as previously reported (40). However, HFD decreased Iba-1 levels in GHR−/− mice, which is in line with the decrease in cytokine production. Thus, diet is not the only signal involved in activation of microglia, as previously reported (55). SOCS-3 levels were increased in GHR−/− mice fed normal chow and HFD reduced them. Expression of this gene is induced by various cytokines, including IL-6, IL-10, and IFN-γ (56). Hence, the increase in SOCS-3 may be related to the rise in hypothalamic cytokine levels in GHR−/− mice, and this increase in SOCS-3 could inhibit insulin and leptin signaling.

The absence of GH signaling impairs liver IGF-1 synthesis and disruption of IGF signaling pathways during early life causes growth retardation and defects in the development of metabolic organs that can alter energy homeostasis for life (57, 58). GHR−/− mice have increased expression of hypothalamic IGF-1 and IGF-1 receptor, most likely as a compensatory response to the decrease in circulating IGF-1 levels, but this increased local production of IGF-1 does not appear to protect this brain area from inflammation. Enhanced IGF-1 expression raises the possibility that compensatory increases in the local production and paracrine/autocrine actions of IGF-1 could account for some characteristics of long-lived GH-deficient mutants (59), because both GH and IGF-1 exert important neurostimulatory and neuroprotective effects (60, 61).

Tanycytes are the normal port of entry for IGF-1 and numerous other circulating factors (62). These specialized glial cells take-up IGF from the cerebrospinal fluid and transport it along their basal processes (63). Leptin is also transported by tanycytes by an ERK-dependent mechanism (64). This transport could play a critical role in the metabolic and inflammatory responses of the hypothalamus. In fact, vimentin, a marker of tanycytes, was increased in GHR−/− mice, suggesting a possible increase in their activity.

One caveat that should be taken into consideration in these studies is that we found HFD to actually decrease cytokine mRNA levels in WT mice in contrast to the expected rise. However, p-JNK and Iba-1 levels were increased, with this normally taken to indicate inflammatory processes. This raises the question as to how hypothalamic inflammation should be defined in order to clearly delineate the mechanisms involved. Moreover, this process most likely includes various stages that could differ depending on the metabolic/dietary alteration involved, as well as the basal inflammatory/glial conditions, for example, cytokine production by microglia depends on their underlying activational state.

In summary, the absence of GH signaling is associated with hypothalamic changes in inflammatory factors even on a ND, with these modifications possibly being involved in their response to HFD intake. The absence of GH signaling increases NEFA levels, which could underlie the increase in hypothalamic inflammatory markers. The intake of HFD by young hypoinsulinemic, hypoglycemic and obese GHR−/− mice increases fat mass and reduces the levels of NEFA and hypothalamic inflammatory markers. Collectively, these studies identify a potential link between GH signaling, hypothalamic inflammation, and obesity. However, further studies are necessary to analyze the outcome of a more extensive exposure to poor dietary habits and to determine the mechanisms involved in the observed hypothalamic modifications.

Acknowledgments

We thank Sandra Canelles and Francisca Díaz for excellent technical assistance.

This work was supported by the Ministerio de Ciencia e Innovación Grant BFU2011–27492, Fondo de Investigación Sanitaria Grants PI100747 and PI13/02195, Centro de Investigación Biomédica en Red Fisiopatología de Obesidad y Nutrición, Instituto de Salud Carlos III, and Fundación de Endocrinología y Nutrición.

Disclosure Summary: The authors have nothing to disclose.

Abbreviations

 
• AgRP

  Agouti-related peptide

 
• Akt

  mammalian homologue of retroviral oncogene v-akt encoding a serine/threonine protein kinase

 
• bw

  body weight

 
• CART

  cocaine and ampheramine regulated transcript

 
• GFAP

  glial fibrillary acidic protein

 
• GHR

  GH receptor

 
• HFD

  high-fat diet

 
• HOMA

  homeostasis model assessment

 
• JNK

  Jun N terminal kinase

 
• KO

  knockout

 
• MCP

  monocyte chemoattractant protein

 
• ND

  normal diet

 
• NEFA

  nonesterified free fatty acids

 
• NPY

  neuropeptide Y

 
• p

  phospho

 
• PI3K

  phosphatidylinositol-4,5-bisphosphate 3-kinase

 
• POMC

  pro-opiomelanocortin

 
• RT

  room temperature

 
• SOCS-3

  suppressor of cytokine signaling 3

 
• WT

  wild type.

Author notes

^*

L.M.F., J.J.K., and J.A. contributed equally to this work.