The Role of Hypothalamic mTORC1 Signaling in Insulin Regulation of Food Intake, Body Weight, and Sympathetic Nerve Activity in Male Mice

Abstract

Insulin action in the brain particularly the hypothalamus is critically involved in the regulation of several physiological processes, including energy homeostasis and sympathetic nerve activity, but the underlying mechanisms are poorly understood. The mechanistic target of rapamycin complex 1 (mTORC1) is implicated in the control of diverse cellular functions, including sensing nutrients and energy status. Here, we examined the role of hypothalamic mTORC1 in mediating the anorectic, weight-reducing, and sympathetic effects of central insulin action. In a mouse hypothalamic cell line (GT1–7), insulin treatment increased mTORC1 activity in a time-dependent manner. In addition, intracerebroventricular (ICV) administration of insulin to mice activated mTORC1 pathway in the hypothalamic arcuate nucleus, a key site of central action of insulin. Interestingly, inhibition of hypothalamic mTORC1 with rapamycin reversed the food intake- and body weight-lowering effects of ICV insulin. Rapamycin also abolished the ability of ICV insulin to cause lumbar sympathetic nerve activation. In GT1–7 cells, we found that insulin activation of mTORC1 pathway requires phosphatidylinositol 3-kinase (PI3K). Consistent with this, genetic disruption of PI3K in mice abolished insulin stimulation of hypothalamic mTORC1 signaling as well as the lumbar sympathetic nerve activation evoked by insulin. These results demonstrate the importance of mTORC1 pathway in the hypothalamus in mediating the action of insulin to regulate energy homeostasis and sympathetic nerve traffic. Our data also highlight the key role of PI3K as a link between insulin receptor and mTORC1 signaling in the hypothalamus.

Besides insulin's well-known effect in peripheral tissues (eg, liver, muscle, and adipocytes) on glucose metabolism, this hormone acts in the central nervous system to modulate a variety of physiological processes. For instance, insulin action in the brain is now recognized as an important regulatory element for energy homeostasis (1–3). Since the original observation by Woods and et al (4) showing that direct central administration of insulin decreases food intake and body weight in nonhuman primates, overwhelming evidence have been accumulated using various approaches supporting the notion that insulin is an important element of the adipostat, acting as an afferent signal in an endocrine feedback loop that controls the body's fat content (1–3). Consistent with its role in the regulation of body weight and energy homeostasis, central insulin stimulates thermogenesis (5) and increases the activity of the sympathetic nervous system supplying several beds, including skeletal muscle (6–8). In accordance with the ability of insulin to control regional sympathetic nerve traffic, brain action of insulin regulates a broad spectrum of physiological functions (9, 10).

The insulin receptor (IR) is expressed throughout the central nervous system, but its distribution is uneven with some regions such as the hypothalamus containing the highest density (11–14). Stimulation of hypothalamic IR increases the intrinsic tyrosine kinase activity of the receptor, leading to phosphorylation of IR substrate protein, which serves as a docking platform for the p85 regulatory subunit of phosphatidylinositol 3-kinase (PI3K) (15). The subsequent activation of the p110 catalytic subunit of PI3K converts phosphatidyl-inositol-4,5-bisphosphate to phosphatidyl-inositol-3,4,5-trisphosphate, which modulates the activity of various downstream signaling intermediates in a context-specific manner (15). The relevance of PI3K in the physiological effects of brain insulin is indicated by its requirement for the anorectic (16) as well as sympathetic nerve (7, 17) responses produced by central action of insulin.

The mechanistic target of rapamycin complex 1 (mTORC1) is composed of several proteins that include the highly conserved mTOR catalytic subunit. mTORC1 has been involved in the regulation of diverse cellular functions, including cell growth and protein synthesis. Activation of mTORC1 leads to phosphorylation of S6 kinase (S6K) and the downstream ribosomal protein S6 (S6), which promotes translation of mRNA transcripts into proteins (18). Within the hypothalamus, mTORC1 acts as a sensor of energy status with its activity being increased in fed and decreased in fasted states (19). Nutrients (eg, glucose and some branched-chain amino acids, particularly leucine) and hormonal signals can activate the negative feedback system regulating meal size and body mass by increasing mTORC1 kinase activity. For instance, direct activation of hypothalamic mTORC1 with leucine causes a significant decrease in food intake and body weight (19). In addition, leucine-mediated stimulation of hypothalamic mTORC1 signaling induces sympathetic nerve activation, suggesting the involvement of hypothalamic mTORC1 pathway in the regulation of sympathetic nerve traffic (20).

Here, we tested the hypothesis that mTORC1 is critically involved in IR signaling in the hypothalamus. For this, we examined the ability of insulin to activate hypothalamic mTORC1 signaling and assessed the underlying mechanisms. We also investigated the role of mTORC1 in mediating the metabolic and sympathetic effects elicited by insulin action in the brain.

Materials and Methods

Reagents

Chemicals were purchased from Research Products International Corp except where indicated. All drugs were freshly prepared from concentrated stock solutions for in vitro and in vivo use. Human recombinant insulin (100 U/mL; Eli Lilly) was diluted in sterile PBS (pH 7.4). Rapamycin, LY294002, and Wortmannin (from Sigma-Aldrich) were dissolved in dimethyl sulfoxide (DMSO) (Fisher Scientific) and diluted in PBS. Antibodies used for this study are listed in Table 1.

Table 1.

List of Antibodies Used for Western Blotting and Immunohistochemistry

Peptide/Protein Target	Manufacturer, Catalog Number	Dilution Used (Application)
Akt	Cell Signaling, 9272	1:10 000 (WB)
Phospho-Akt (S473)	Cell Signaling, 406	1:10 000 (WB)
mTOR	Cell Signaling, 2972	1:1000 (WB)
Phospho-mTOR (S2448)	Cell Signaling, 2976	1:1000 (WB)
S6K	Cell Signaling, 2708	1:1000 (WB)
Phospho-S6K (T389)	Millipore, 04-392	1:1000 (WB)
S6	Cell Signaling, 2217	1:10 000 (WB)
Phospho-S6 (S240/244)	Cell Signaling, 5364	1:20 000 (WB), 1:1000 (IHC)
GAPDH	Santa Cruz, sc-32 233	1:20 000 (WB)
IRα subunit	Santa Cruz, sc-710	1:500 (IHC), 1:1000 (WB)
IRβ subunit	Santa Cruz, sc-711	1:1000 (WB)
Rabbit IgG, HRP conjugated	Cell Signaling, 7074	1:5000 (WB)
Rabbit IgG, Biotinylated	Vector, BA-1000	1:200 (IHC)
Mouse IgG, HRP conjugated	Thermo Scientific, 31 432	1:20 000 (WB)

WB, Western blotting; IHC, immunohistochemistry; GAPDH: Glyceraldehyde 3-phosphate dehydrogenase.

Cell culture

GT1–7 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) containing 4.5-g/L glucose (Gibco) supplemented with 10% (vol/vol) fetal bovine serum (Atlanta Biologicals) and 1mM sodium pyruvate (Gibco), at 37°C with 5% CO₂. For passaging the cells, 0.125% Trypsin-EDTA (Gibco) was used as a dissociation reagent. GT1–7 cells were split into 60-mm dishes and grown until they reached 70%–80% confluence. Cells were starved for 3 hours by replacing the complete growth medium with nonsupplemented DMEM before they were treated with insulin (100nM) or vehicle (PBS). To block the various signaling pathways, GT1–7 cells were pretreated with rapamycin (10nM), Wortmannin (100nM), LY294002 (10μM), or vehicle (DMSO) for 20 minutes before incubation with insulin (100nM) or vehicle for 20 minutes. The cells were washed by ice-cold PBS and immediately lysed in 50mM Tris-Cl buffer (pH 7.5) containing 0.1mM EDTA, 0.1mM EGTA, 1% sodium deoxycholic acid (wt/vol), 1% nonylphenoxypolyethoxyethanol-40 (vol/vol), 0.1% sodium dodecyl sulfate (vol/vol), 1mM phenylmethanesulfonylfluoride, protease inhibitor cocktail tablets (Complete Mini, EDTA-free; Roche Diagnostics), 1mM sodium orthovanadate, 10mM sodium fluoride, and 1mM sodium pyrophosphate for 30 minutes on ice. The cell extracts were centrifuged at 14 000g for 30 minutes at 4°C. Protein concentration of the cell lysate supernatant was measured using the Lowry et al (21) method with a spectrophotometer (Beckman Coulter, Inc).

Animals

Male mice (9–12 wk old) maintained on a mixed genetic background (FVB/N;B6;129) and p110α^D933A/WT mice originally created by Foukas et al (22) were obtained from our own colonies. PCR assay was used to genotype p110α^D933A/WT mice as described before (22). Mice were housed in air-conditioned conventional mouse cages on a 12-hour light, 12-hour dark cycle (light on from 6 am to 6 pm) with ad libitum access to water and food (7013 NIH-31 Modified Open Formula Mouse/Rat Sterilizable Diet; Harlan Laboratories). The Institutional Animal Care and Use Committee at the University of Iowa reviewed and approved all animal protocols used in the current studies.

Intracerebroventricular (ICV) cannulation and injection

Under ketamine/xylazine anesthesia (91 and 9.1 mg/kg, ip, respectively), 9-week-old mice were placed in a stereotactic device (David Kopf Instruments), and a custom-built guide cannula (25 G, 9 mm) was implanted into the left lateral brain ventricle (0.3 mm posterior and 1.0 mm lateral relative to bregma, and 2.0 mm below the dorsal surface of the skull). At least a 1-week period was given to mice for recovery. A slow ICV injection (2 μL) was delivered over a period of 20–30 seconds to each mouse gently restrained using the dorsal scruff method. After completion of ICV injection, inserted hypodermic needle (32 G) was cut off (2–3 mm left on the guide cannula) and sealed with instant glue to avoid leakage of injectate. Whenever leakage or removal of the hypodermic needle cap occurred during the experimental period, the animals were removed from the study.

Immunohistochemical studies

Mice were fasted for 3 hours by removing the feeders at 8 am. One hour after ICV injection of vehicle or insulin (100 μU) (17), mice were anesthetized using ketamine/xylazine, and transcardially perfused with PBS followed by ice-cold 4% paraformaldehyde (Fisher Scientific). Isolated brains were further fixed in 4% paraformaldehyde at 4°C overnight. The next day, the entire hypothalamic part of the brain was cut into 10-μm-thick sections on a vibratome (Leica Microsystems, Inc). Free-floating sections were treated with 0.3% hydrogen peroxide (Sigma-Aldrich) for 15 minutes to deactivate endogenous peroxidase and then blocked in Tris-buffered saline (pH 7.4) with 5% goat serum (Atlanta Biologicals) and 0.005% Triton X-100 (Sigma-Aldrich) for 1 hour. After overnight incubation with a primary antibody (against the IRα subunit or phospho (p)-S6) at 4°C, the sections were incubated with biotinylated antirabbit IgG for 1 hour and then transferred to a horseradish peroxidase (HRP)-conjugated avidin solution (VECTASTAIN Elite ABC kit; Vector) for 30 minutes. In a subset of brain sections, specificity of the antibody targeting the IRα was assessed by preabsorbing the antibody with the blocking peptide used for its generation (catalog number sc-710P; Santa Cruz Biotechnology, Inc) according to the manufacturer's peptide neutralization protocol. The signal was developed using diaminobenzidine substrate (DAB HRP Substrate kit; Vector), and sections were mounted on glass slides with Permount Medium (Fisher Scientific). Images collected from the brain sections were used for quantitative analysis of the immunostaining. Using ImageJ (http://rsb.info.nih.gov/ij/index.html), background and nonspecific signal were excluded by thresholding in 8 bit grayscale images. With the assistance of ImageJ's particle analyzer, the number of phospho-S6 positive cells in a side of the hypothalamic arcuate nucleus (ARC) was counted.

Western blotting

Mice fasted for 3 hours (beginning at 8 am) and treated with ICV insulin (100 μU) or vehicle were euthanized with CO₂ asphyxiation at the indicated time points. Extracted mouse brains were quickly semifrozen at −80°C for 5 minutes. Precooled razor blade was used to dissect a triangular prism-shaped mediobasal hypothalamus (MBH) wedge with equilateral sides from approximately −1.22 to −2.18 mm posterior to the bregma ensuring that the ARC was included. Microdissected MBH was snap frozen in liquid nitrogen and stored in −80°C ultralow freezer. MBH was homogenized in the lysis buffer using a plastic pellet pestle and processed as described above.

Protein lysate extracted from GT1–7 cells or microdissected MBH was resolved by SDS-PAGE and transferred to polyvinylidene fluoride membranes (Millipore). The membranes were blocked in 5% nonfat dry milk or bovine serum albumin in Tris-buffered saline with 0.1% Tween 20 for 1 hour at room temperature and incubated with an appropriately diluted primary antibody at 4°C overnight. The membranes were incubated with antirabbit IgG conjugated with HRP for 1 hour at room temperature. Visualization was done with enhanced chemiluminescence (ECL plus or prime; General Electric). The Western blotting films were scanned at 600 dots per inch in grayscale mode. Background signal was eliminated by selecting bands of interest using wand (tracing) tool. The band's intensity was calculated using ImageJ's gel analyzer.

Food intake and body weight studies

After ICV cannulation and postsurgical recovery period, mice were single housed. After 5–7 days of acclimation to new environment and handling stress, basal food consumption and body weight were monitored daily for 5 days. On the day of treatment, mice received a combined ICV injection (2 μL), 30 minutes before the beginning of dark cycle, as follows: DMSO/PBS (vehicles), DMSO/insulin (100 μU), or rapamycin (1 ng)/insulin (100 μU). Food intake and body weight were measured 4 and 24 hours later.

Sympathetic nerve activity (SNA) recording

Under ketamine/xylazine anesthesia, a nerve fascicle to the hindlimb was carefully isolated. A bipolar platinum-iridium electrode (Cooner Wire) was suspended under the nerve and secured with silicone gel (Kwik-Cast, WPI). The electrode was attached to a high-impedance probe (HIP-511, Grass Instruments), and the nerve signal was amplified 10⁵ times with a Grass P5 AC preamplifier (Natus Neurology, Inc). After amplification, the nerve signal was filtered at a 100- and 1000-Hz cutoff with a nerve traffic analysis system (model 706C, University of Iowa Bioengineering). Subsequently, the amplified and filtered nerve signal was routed to an oscilloscope (model 54501A, Hewlett-Packard) to monitor the quality of the SNA recording and its quantification. Sympathetic activity obtained after death (background noise) was subtracted from the total integrated voltage to calculate real SNA. Catheterization of the carotid artery and the jugular vein was also performed in the same animal for hemodynamic monitoring and maintenance of anesthesia with α-chloralose (25 mg/kg·h), respectively. Recording of baseline parameters was performed during a 10-minute control period, then mice received ICV rapamycin (1 ng/μL) or vehicle (DMSO, 1 μL), followed 10 minutes later by ICV insulin (100 μU/μL) or PBS (1 μL). SNA was recorded every 15 minutes and monitored for 4 hours after last ICV injection.

Statistical analysis

Data are presented as means ± SEM. All data were analyzed by Student's t test, one- or two-way ANOVA with or without repeated measures. A Tukey test was employed post hoc when appropriate. P < .05 was considered statistically significant. Prism 6 for Macintosh (GraphPad Software, Inc) was used for statistical analysis.

Results

Insulin activates hypothalamic mTORC1 signaling

We first examined whether insulin is capable of increasing mTORC1 activity in a mouse hypothalamic neuroblastoma GT1–7 cell line, which expresses the endogenous IR (23). The dose of insulin (100nM) was selected based on previous studies in cells (24, 25). Western blot analysis revealed that insulin treatment significantly increased mTORC1 activity in GT1–7 cells, indicated by elevated phosphorylation levels of S6K and S6 in a time-dependent manner (Figure 1, A and B). In addition, insulin treatment led to a significant increase in Akt phosphorylation reflecting PI3K pathway activation (Figure 1C). In contrast, vehicle treatment (PBS) had no effect on basal phosphorylation levels of the mTORC1 effectors or Akt in GT1–7 cells (data not shown).

Figure 1. Activation of hypothalamic mTORC1 signaling by insulin.

A–C, Time-dependent increase in the phosphorylation of S6K (A), S6 (B), and Akt (C) in GT1–7 cells in response to treatment with insulin (100nM), n = 3–4 experiments per time point. D, Time-dependent effect of ICV insulin (100 μU) on S6 activity in the MBH explants of mice, n = 4–6 experiments per time point. E, Immunohistochemical staining for p-S6 in the ARC of mice treated with ICV vehicle (PBS) or insulin (100 μU). Scale bar, 100 μm. 3V, third ventricle. Values are mean ± SEM. *, P < .05 vs time 0.

IR is widely expressed in the brain, which is consistent with the involvement of insulin in the regulation of various neurobiological processes. Nonetheless, hypothalamic nuclei, particularly the ARC, contain the highest density of IR, as previously determined by immunohistochemistry against IRβ subunit (10), a binding assay with isotope-labeled insulin (11, 12), or in situ hybridization for IR mRNA (13). Using immunostaining, we detected extensive DAB labeling of IRα subunit in the ARC with lesser staining intensity in other hypothalamic nuclei, including the dorsomedial and ventromedial hypothalamic nuclei (Supplemental Figure 1, left). Specificity of the antibody used for immunostaining was demonstrated using the blocking peptide (Supplemental Figure 1, right, and Supplemental Figure 2). This evidence confirmed the presence of insulin-responsive neurons in the hypothalamus with the highest density in the ARC.

Next, we tested the effect of a single ICV administration of insulin (100 μU) on S6 phosphorylation by Western blotting in mice. Interestingly, relative to the previous report (26) that used 900 μU of ICV insulin (for 1 h) to induce hypothalamic mTORC1 stimulation, we found that 100 μU of ICV insulin is sufficient to elicit mTORC1 activation in the MBH as indicated by the increased phospho-S6 levels. The increase in the phospho-S6 in response to ICV insulin occurred in a time-dependent manner without modifications in total S6 protein (Figure 1D). The maximal response to insulin was detected at 30 minutes after treatment, and the response was sustained at 60 minutes with a trend toward a decrease (Figure 1D). A marked increase in phospho-S6 immunoreactivity in the ARC of mice treated with ICV insulin for 60 minutes confirmed further the ability of insulin to active mTORC1 signaling in the hypothalamus (Figure 1E).

Rapamycin blocks hypothalamic mTORC1 activation by insulin

In order to verify the mTORC1-dependent activation of S6K and S6 in the hypothalamic IR signaling, GT1–7 cells were pretreated with rapamycin, at a dose (10nM) that was previously shown to specifically inhibit mTORC1 in cells (27), for 20 minutes before insulin treatment. Rapamycin significantly blunted the increase in phosphorylated levels of mTOR, S6K, and S6 in response to insulin (Figure 2, A–C). In contrast, rapamycin treatment did not affect insulin-induced Akt activation (Figure 2D). These results demonstrate that in the hypothalamic neuronal cell line, insulin activates S6K and S6 directly through mTORC1.

Figure 2. Effect of rapamycin on insulin-induced activation of hypothalamic mTORC1 pathway.

A–D, Rapamycin (10nM, 20 min) inhibits insulin-induced phosphorylation of mTOR (A), S6K (B), and S6 (C) but not Akt (D) in GT1–7 cells, n = 3–4 experiments per treatment. E–G, Rapamycin (1 ng, ICV) inhibits phospho-S6 immunoreactivity induced by ICV insulin (100 μU) in the ARC by immunohistochemistry (n = 5 mice per treatment) and MBH explants by Western blotting. Scale bar, 100 μm. 3V, third ventricle. Values are mean ± SEM. *, P < .05 vs other groups; #, P < .05 vs Veh/Veh.

Next, we tested whether ICV rapamycin blocks hypothalamic mTORC1 activation induced by ICV insulin in mice. As shown in Figure 2, E and F, the number of phospho (p)-S6 positive cells in the ARC was increased by ICV insulin (45 ± 2) when compared with vehicle-treated group (22 ± 3). ICV rapamycin significantly diminished insulin-evoked mTORC1 activation (15 ± 4). This finding was confirmed by Western blotting demonstrating that ICV rapamycin suppressed insulin-induced increase in phospho-S6 in MBH explants (Figure 2G). Thus, rapamycin abolishes insulin-induced activation of hypothalamic mTORC1 signaling.

mTORC1 inhibition reverses the anorectic and weight-reducing effects of insulin

To investigate the putative physiological importance of mTORC1 for the hypothalamic IR signaling, we assessed the role of mTORC1 in mediating the anorectic and weight-reducing actions of this hormone in mice. Relative to ICV vehicle, ICV treatment with 100 μU of insulin at the onset of the dark phase significantly reduced food intake at both 4 and 24 hours (Figure 3, A and B). This effect was accompanied with decreased body weight at 4 and 24 hours (Figure 3, C and D) after treatment with ICV insulin. Notably, inhibition of mTORC1 signaling with rapamycin (1 ng, ICV) completely reversed the anorectic (Figure 3, A and B) and body weight-lowering (Figure 3, C and D) effects of insulin. Of note, this dose of rapamycin alone caused no significant change in 4- and 24-hour food intake and body weight in mice (Supplemental Figure 3). These findings highlight the crucial role of mTORC1 signaling for insulin regulation of food intake and body weight.

Figure 3. Effect of mTORC1 blockade (rapamycin) on the anorectic and body weight-reducing effects of insulin.

A–D, ICV rapamycin (Rap) (1 ng) reversed the reduction in food intake (A and B) and body weight (C and D) induced by ICV insulin (Ins) (100 μU) 4 hours (A and C) and 24 hours (B and D) after treatment in mice, n = 6–7 per group. Values are mean ± SEM. *, P < .05 vs other groups.

PI3K ties the IR to mTORC1 signaling

We next investigated the mechanism underlying insulin-induced activation of hypothalamic mTORC1. An upstream effector suspected to underlie mTORC1 stimulation by the IR is PI3K/Akt pathway. This is based on their relationship in the peripheral IR signaling (28, 29). To test whether PI3K/Akt pathway couples mTORC1 signaling to the IR in the hypothalamus, we performed several experiments. In hypothalamic GT1–7 cells, we found that PI3K inhibition with Wortmannin (100nM, 20 min) (30–32) substantially inhibited insulin-mediated increase in the phosphorylation levels of mTOR, S6K, and S6 (Figure 4, A–C). As predicted, Wortmannin also blocked insulin-induced activation of Akt (Figure 4D). Similar findings were obtained when PI3K inhibition was achieved with LY294002 (Supplemental Figure 4). It should be noted that Wortmannin alone reduced levels of p-S6K and p-Akt (Figure 4, B and D), whereas LY294002 alone significantly decreased levels of p-S6K and p-S6 (Supplemental Figure 4, B and C). These results demonstrate that insulin activates mTORC1 pathway in a PI3K-dependent manner in hypothalamic GT1–7 cells.

Figure 4. Effect of PI3K inhibition (Wortmannin) on insulin-induced activation of hypothalamic mTORC1 pathway in vitro.

A–D, Pretreatment with Wortmannin (100nM, 20 min) blocked insulin-activation of mTOR (A), S6K (B), S6 (C), and Akt (D) in GT1–7, n = 4–7 experiments per treatment. Values are mean ± SEM. *, P < .05 vs other groups; #, P < .05 vs Veh/Veh.

To investigate the role of PI3K pathway in linking hypothalamic mTORC1 signaling to the IR in vivo, we used p110α^D933A/WT mice, which carry a heterozygous mutation in p110α, a catalytic subunit of PI3K. This missense mutation in p110α inhibits PI3K activity in the hypothalamus (22). In wild-type mice, ICV insulin caused a significant increase in phospho-S6 immunoreactivity in the ARC. In contrast, in p110α^D933A/WT mice, ICV insulin failed to increase phospho-S6 immunoreactivity in the ARC (Figure 5, A and B). These data demonstrate the PI3K-dependent mTORC1 activation by insulin in the hypothalamic ARC.

Figure 5. Effect of PI3K mutation on insulin-induced activation of hypothalamic mTORC1 pathway in vivo.

A and B, Insulin-induced increase in p-S6 immunoreactivity in the ARC is absent in mice carrying a missense mutation in p110α (p110 α^D933/WT) blocking PI3K signaling. Examples of p-S6 staining in the ARC are shown in A and quantified data in B, n = 4 per group. Scale bar, 100 μm. 3V, third ventricle; WT, wild-type. C and D, Insulin-evoked increase in lumbar SNA is blunted in p110 α^D933/WT mice, n = 7–8 per group. Values are mean ± SEM. *, P < .05 vs other groups.

The lumbar sympathetic activation evoked by insulin requires PI3K and mTORC1

Hypothalamic action of insulin is known to elevate regional sympathetic nerve traffic. We previously used pharmacological approaches to demonstrate that insulin-induced increase in lumbar SNA is PI3K mediated (7, 17). Thus, we hypothesized that the lumbar sympathetic nerve response to insulin was altered in p110α^D933A/WT mice. In wild-type mice, ICV insulin caused a slowly developing, but robust, increase in lumbar SNA (Figure 5, C and D). In contrast, the p110α^D933A/WT mice displayed absent lumbar sympathetic nerve response to ICV insulin (Figure 5, C and D). These findings demonstrate the key role of PI3K signaling in mediating insulin-induced lumbar sympathetic excitation.

Our data above showing PI3K-dependent mTORC1 activation by insulin (Figures 4, A–C, and 5, A and B) led us to test whether inhibition of mTORC1 signaling would also interfere with the ability of insulin to increase lumbar SNA. Interestingly, mTORC1 inhibition with ICV rapamycin (1 ng) abolished the lumbar sympathetic nerve response evoked by ICV insulin (Figure 6, A and B) in a manner comparable with the effect of PI3K blockade (Figure 5, C and D). Vehicle or rapamycin alone had no effect on lumbar SNA (Figure 6B). These results suggest that mTORC1 activation is a key signaling transduction event underlying insulin-induced lumbar sympathetic nerve activation. These data also support the idea that central insulin regulates lumbar sympathetic traffic through hypothalamic PI3K-mTORC1 signaling axis.

Figure 6. Effect of mTORC1 inhibition (rapamycin) on the lumbar sympathetic nerve activation evoked by insulin.

A and B, Rapamycin (1 ng) blocks insulin-induced increase in lumbar SNA in mice. Time course of the lumbar SNA responses over 4 hours (A) and averages of the fourth hour of SNA recording (B) are displayed, n = 6–8 per group. C and D, Lack of effect of rapamycin on the lumbar sympathetic response induced by MTII (melanocortin 3/4 receptor agonist, 0.1 μg) in mice. Time course of lumbar SNA response over 4 hours (C) and average of the fourth hour of SNA recording (D) are shown, n = 4 per group. Values are mean ± SEM. *, P < .05 vs other groups.

Finally, we tested the specificity of rapamycin inhibition of insulin-evoked sympathetic activation. For this, we assessed the effect of rapamycin on the sympathetic nerve activation induced by another stimulus, melanotan-II (MTII), an agonist of the melanocortin-3 and melanocortin-4 receptors. ICV administration of MTII caused a robust elevation in lumbar sympathetic outflow. However, rapamycin failed to alter the MTII-induced lumbar SNA response (Figure 6, C and D), suggesting that the blockade of insulin-induced sympathetic nerve response by rapamycin is specific. Of note, it was previously shown that rapamycin does not interfere with the anorectic and body weight-reducing actions of MTII (19). Interestingly, the lumbar SNA response to MTII was also unaltered in p110α^D933A/WT mice (338 ± 135%) relative to littermate controls (384 ± 64%, P = .35).

Discussion

Central insulin signaling has an important role in the regulation of energy homeostasis and SNA (1, 6). The current study demonstrates that insulin activates mTORC1 signaling in hypothalamic GT1–7 cells and mouse hypothalamus. We also found that insulin regulation of physiological processes required mTORC1 activation. Indeed, hypothalamic mTORC1 activation appears essential for the control of food intake and body weight by insulin. Our data also reveal the functional requirement for intact PI3K signaling for insulin to activate hypothalamic mTORC1 and to cause lumbar sympathetic excitation. In addition, we observed that blockade of mTORC1 signaling abolished the lumbar sympathetic response to insulin. Together, these findings demonstrate the physiological significance of the PI3K-mTORC1 axis for the central actions of insulin on energy balance and sympathetic outflow.

Previous studies have established the relevance of hypothalamic mTORC1 for the regulation of food intake and body weight. For instance, opposing changes in the activity of mTORC1 signaling produced contrasting effects on food intake and body weight (33). Moreover, mTORC1 signaling in the hypothalamus was found to mediate the metabolic effects induced by various stimuli. Since the original study by Cota et al (19) implicating mTORC1 in the anorectic and weight-reducing actions of leucine and leptin, the requirement of this signaling pathway has been extended to other hormones, including thyroid hormone (34) and ghrelin (35). Our current study further demonstrates that mTORC1 is a key mediator of hypothalamic IR signaling to promote negative energy balance. Indeed, inhibition of mTORC1 signaling with rapamycin reversed the ability of ICV insulin to decrease food intake and body weight. Collectively, these findings point to the hypothalamic mTORC1 signaling as a critical regulator of feeding and energy homeostasis. However, the implication of hypothalamic mTORC1 in mediating both anorexigenic (eg, leptin and insulin) and orexigenic (eg, ghrelin) effects is intriguing. Additional studies are warranted to explain this seemingly paradoxical role of hypothalamic mTORC1.

To gain insight into the mechanisms mediating IR activation of mTORC1 signaling, we investigated the role of the PI3K, which is a critical upstream element in the intracellular signaling cascade regulating mTORC1 activity (36, 37). In GT1–7 cells, rapamycin treatment reversed insulin-induced phosphorylation of mTORC1, S6K, and S6 but not Akt. This is consistent with PI3K/Akt being upstream of the mTORC1 pathway. This is further supported by our finding that pharmacological inhibition of PI3K eliminated insulin-induced activation of mTOR, S6K, and S6 in GT1–7 cells. Moreover, we show that mice bearing a knock-in mutation in p110α (p110α^D933A/WT) that abrogates PI3K lipid kinase activity (22) have blunted insulin-induced hypothalamic mTORC1 activation. Collectively, these data identify PI3K as a mediator between IR and mTORC1 in the hypothalamus.

Hypothalamic PI3K was previously shown to mediate the anorectic effect of central action of insulin (15). Activation of PI3K by insulin was also implicated in the modulation of neuropeptide expression, neuronal firing, and sympathetic traffic (2, 7, 17, 38). Using a genetic approach, we confirmed our previous findings regarding the involvement of PI3K signaling in the lumbar sympathetic nerve response to insulin (7, 17). p110α^D933A/WT mice had absent lumbar sympathetic response when treated with ICV insulin. Likewise, the lumbar sympathetic nerve response was abolished by rapamycin treatment. These results support the importance of PI3K-mTORC1 axis in the regulation of lumbar sympathetic traffic by insulin. Of note, we previously found that obese/insulin-resistant mice displayed an intact lumbar sympathetic response to insulin, supporting the concept of selective insulin resistance in the brain as a mechanism linking hyperinsulinemia to sympathetic overdrive in obesity (17). We also showed the key role of PI3K in mediating the preserved lumbar sympathetic response to insulin in obesity (17). Our finding in the current study demonstrating that PI3K lies upstream of mTORC1 raises the possibility that mTORC1 signaling may contribute to obesity-induced sympathetic overdrive. However, additional studies are needed to test such possibility.

IR expression is widely distributed throughout the brain, but the density of IR is relatively higher in the hypothalamus (10–13), where it colocalizes with its essential signaling components such as IR substrate protein and PI3K (39–42). Our immunohistochemical analyses showing high density of IRα subunit and the strongest insulin-induced activation of mTORC1 signaling in the ARC are consistent with the notion that this nucleus has a predominant role in mediating the effects of insulin on energy homeostasis (43). This is further supported by the previous finding that down-regulation of ARC IR with antisense oligodeoxynucleotide promotes feeding and adiposity in rats (44, 45). Moreover, ARC-targeted microinjection of insulin causes lumbar sympathetic nerve activation (46). Conversely, ARC-specific injection of antiinsulin affibody inhibits the lumbar sympathetic activation by insulin (47). These studies demonstrate the importance of IR signaling in the ARC for the regulation of physiological processes. It should be noted that IR is present in other hypothalamic and extrahypothalamic nuclei (10–13) that are implicated in the regulation of feeding (48, 49), adiposity (50), and sympathetic nerve traffic (51). The contribution of these other brain sites to insulin regulation of energy homeostasis and sympathetic activity remains poorly understood.

In summary, the current study identified a novel role for mTORC1 as a key downstream pathway of IR signaling in the regulation of feeding, body weight, and sympathetic traffic. In addition, we showed the requirement of PI3K signaling for insulin activation of hypothalamic mTORC1. Collectively, these findings demonstrate the physiological significance of the PI3K-mTORC1 axis for the central actions of insulin to regulate energy balance and sympathetic nerve traffic.