TCPTP Regulates Insulin Signaling in AgRP Neurons to Coordinate Glucose Metabolism With Feeding

Insulin is secreted by pancreatic β-cells in response to elevated blood glucose levels and signals via the insulin receptor (IR) in peripheral tissues, including skeletal muscle and fat, to promote glucose uptake and storage, and in the liver to repress hepatic glucose production (HGP) to prevent postprandial hyperglycemia. Insulin can also promote glucose uptake in brown adipose tissue (BAT) (1), where glucose can be stored as glycogen, used for fatty acid esterification and triglyceride synthesis, or converted to lactate by anaerobic glycolysis during nonshivering thermogenesis to produce ATP (2,3). The latter may compensate for the reduced ATP production that occurs during thermogenesis where fatty acid oxidation is uncoupled from ATP production to generate heat (2,3). The importance of BAT in glucose metabolism is highlighted by studies demonstrating that greater BAT abundance is accompanied by decreased glycemic variability (4,5). Beyond insulin’s roles in the periphery, a large body of evidence now also supports a role for insulin action in the central nervous system (CNS) in the regulation of systemic insulin sensitivity and glucose homeostasis (6,7).

Mice in which IR was deleted in neurons and astroglia using the Nestin Cre transgene become obese and develop systemic insulin resistance (8). Although the glucoregulatory effects of the IR can be ascribed to different regions of the brain, several nuclei within the hypothalamus are especially important. Neurons in the arcuate nucleus (ARC) of the hypothalamus, residing at the base of the third ventricle, are positioned to readily sense peripheral substances, such as insulin, that signal the nutritional and energy state of the organism (6,7,9). These include two molecularly defined neuronal populations, the anorexigenic POMC neurons that repress feeding and increase energy expenditure and the orexigenic AgRP/NPY neurons that can antagonize the actions of POMC neurons to promote feeding and decrease energy expenditure (6,7,9). Insulin signals via phosphatidylinositol 3-kinase (PI3K) to activate the Ser/Thr protein kinase AKT and other signaling cascades to elicit discordant effects in POMC and AgRP/NPY neurons (6,7,9). Insulin regulates POMC neuronal excitability and promotes POMC expression, which is processed into the neuropeptide α-melanocyte–stimulating hormone (MSH). α-MSH agonizes postsynaptic melanocortin-4 receptors on neurons in other regions of the brain to repress feeding and increase ambulatory activity and thermogenesis (6,7,9). By contrast, insulin hyperpolarizes AgRP neurons and inhibits their firing by opening K_ATP channels (10). Insulin also inhibits the expression of the neuropeptide AgRP that functions postsynaptically to antagonize α-MSH/melanocortin-4 receptors interactions (6,7,9). The inhibition of AgRP/NPY neurons by insulin can alleviate inhibitory constraints on POMC neurons and the melanocortin response (6,7,9). However, AgRP neurons can also elicit melanocortin-independent effects and signal through parallel and redundant neural circuits to affect feeding (11,12). Although the precise neuronal populations regulating glucose homeostasis remain to be defined, there is evidence that these can be distinct from those regulating feeding. Recent studies, for example, have shown that the acute pharmacogenetic activation of AgRP neurons represses BAT glucose uptake via projections to the bed nucleus of the stria terminalis that are not involved in the induction of feeding (13).

The infusion of insulin into the brain (lateral ventricle) results in the suppression of HGP and lowers blood glucose even in the context of diabetes (14–16). The CNS effects on HPG have been ascribed to AgRP neurons because the specific deletion of IR in AgRP but not POMC neurons results in the defective repression of HGP (10,17). The CNS-mediated repression of HGP is orchestrated by vagal efferents and α7-nicotinic acetylcholine receptors (18) that promote the expression and release interleukin-6 (IL-6) by Kupffer cells in the liver (15,19). IL-6 in turn acts on hepatocytes via STAT3 to repress the expression of gluconeogenic enzymes such as glucose-6-phosphatase (encoded by G6pc) and phosphoenolpyruvate-carboxykinase (encoded by Pck1) (20,21). Recent studies using pharmacogenetic approaches have shown that the acute activation of AgRP neurons promotes systemic insulin resistance by repressing the activity of sympathetic fibers supplying BAT and thereby BAT glucose uptake, without affecting HGP (13). These effects appear to be independent of the melanocortin system because the acute activation of POMC neurons has no effect on BAT glucose uptake (13). The extent to which hormones such as insulin influence BAT glucose uptake via AgRP neurons remains to be determined.

We previously identified the tyrosine phosphatase TCPTP (encoded by Ptpn2) as a key negative regulator of IR signaling in the periphery and in the ARC (22–26). TCPTP dephosphorylates the IR and attenuates insulin-induced PI3K/AKT signaling in AgRP neurons (24,26,27). TCPTP deletion in AgRP neurons exacerbated the insulin-mediated inhibition of AgRP neurons to increase the sympathetic output to BAT and inguinal white adipose tissue (26). This both increased BAT activity and promoted the conversion of white adipocytes to brown-like or beige adipocytes (browning) and increased their thermogenic activity and energy expenditure to render mice resistant to diet-induced obesity (26). Importantly, we reported that TCPTP abundance in the hypothalamus exhibited diurnal fluctuations linked to feeding (26). Fasting increased hypothalamic TCPTP to repress IR signaling to facilitate AgRP neuronal activation, and feeding resulted in TCPTP being degraded so that IR signaling was exacerbated and AgRP neurons inhibited to promote white adipose tissue browning and energy expenditure (26). In this study, we explored the role of TCPTP in AgRP neurons on glucose metabolism and the extent to which TCPTP fluctuations may integrate feeding with the CNS control of glucose homeostasis.

Ptpn2^fl/fl, Agrp-Ires-Cre;Ptpn2^fl/fl (AgRP-TC), Npy-humanized renilla green fluorescent protein (Npy-GFP), Agrp-Ires-Cre;Ptpn2^fl/fl;Npy-GFP (AgRP-TC;Npy-GFP), Pomc-eGFP, and Agrp-Ires-Cre;Ptpn2^fl/fl;Insr^fl/+ (AgRP-TC-IR) mice have been described previously (26,28–30). To generate Agrp-Ires-Cre;Ptpn2^fl/fl;Insr^fl/+;Npy-GFP (AgRP-TC-IR;Npy-GFP) or Agrp-Ires-Cre;Insr^fl/+(AgRP-IR^fl/+) mice, AgRP-TC;Npy-GFP or Agrp-Ires-Cre mice were mated with Insr^fl/fl mice (31), respectively. Mice were maintained on a 12-h light-dark cycle in a temperature-controlled high barrier facility with free access to food and water. Mice were fed a standard chow (8.5% fat; Barastoc; Ridley AgriProducts, Melbourne, Victoria, Australia). The Monash University School of Biomedical Sciences Animal Ethics Committee approved the experiments.

Mice were injected intraperitoneally with vehicle or human insulin (0.85, 2.5, 5 mU/g, Actrapid; Novo Nordisk, Bagsvaerd, Denmark), then transcardically perfused with 4% weight (w)/volume (v) paraformaldehyde, and the postfixed brains were then processed for phosphorylated (p)-AKT (Ser-473) or GFP immunohistochemistry, as described previously (26).

Quantitative real-time PCR was performed as described previously (25,26). The following TaqMan gene expression assays were used: Pomc (Mm00435874_m1), Npy (Mm03048253_m1), Agrp (Mm00475829_g1), Gapdh (Mm99999915_g1), Pck1 (Mm01247058_m1), G6pc (Mm00839363_m1), and Il6 (Mm00446190_m1).

Unless otherwise indicated, insulin, glucose, and pyruvate tolerance tests were performed on fasted (fasting from 9:00 a.m.) conscious mice by injecting human insulin (0.5 mU insulin/g body weight, 4-h fasted), d-glucose (2 mg/g body weight, 6-h fasted), or sodium pyruvate (1 mg/g body weight, 6-h fasted), respectively, into the peritoneal cavity. Glucose levels in tail blood were measured using an Accu-Chek glucometer (Roche, Mannheim, Germany). For the determination of fed and fasted blood glucose and corresponding plasma insulin levels, submandibular blood was collected at 9:00 a.m. after an overnight fast (food removed at 7:00 p.m. the previous day). Plasma insulin levels were determined using a Rat Insulin RIA kit (Linco Research, St. Charles, MO) or an in house ELISA (Monash Antibody Technologies Facility). Body composition was measured by EchoMRI (Echo Medical Systems, Houston, TX) or DEXA, as described previously (25). Lateral ventricle cannulations were performed as described previously (25,26).

For hyperinsulinemic-euglycemic clamps, Ptpn2^fl/fl, AgRP-TC, or AgRP-TC-IR mice (8–10 weeks old) were anesthetized under 2% (v/v) isoflurane in 250 mL/min oxygen, and the left common carotid artery and the right jugular vein were catheterized for sampling and infusions, respectively, as previously described (32). Catheters were kept patent by flushing daily with 10–40 μL saline containing 200 units/mL heparin and 5 μg/mL ampicillin. Animals were housed individually after surgery, and body weights were recorded daily. On the day of the experiment, food was removed at between 7:00 and 8:00 a.m. After 3.5 h fasting, a primed (2 min, 0.5 μCi/min) continuous infusion (0.05 μCi/min) of [3-³H]glucose was administered to measure whole-body glucose turnover, as described previously (32). After 5 h fasting, mice received a continuous insulin infusion (4 mU/kg/min), and blood glucose was maintained at basal levels by a variable infusion of a 50% (w/v) glucose solution. Arterial blood samples were collected during steady-state conditions (R_a = R_d) and at 80, 90, 100, 110, and 120 min for determination of R_d and R_a, as described above. At 120 min, a 13 μCi bolus of [¹⁴C]-2-deoxy-d-glucose was injected into the jugular vein, and arterial blood was sampled at 122, 125, 130, 135, and 145 min. Mice were then culled, and tissues were extracted and frozen for subsequent gene expression and glucose uptake determinations.

Statistical significance was determined by a one-way or two-way ANOVA with multiple comparisons or repeated measures, or by a two-tailed paired Student t test, as appropriate. P < 0.05 was considered significant.

To determine the extent to which TCPTP deficiency in AgRP/NPY neurons may influence whole-body glucose metabolism, we crossed Ptpn2^fl/fl mice with Agrp-Ires-Cre transgenic mice to excise Ptpn2 in AgRP-expressing neurons (Agrp-Ires-Cre;Ptpn2^fl/fl: AgRP-TC). TCPTP deletion in AgRP neurons (26) did not influence the overall number of AgRP/NPY neurons (Supplementary Fig. 1), as defined by the Npy-GFP reporter that marks >85% of AgRP neurons (33). To specifically assess the influence of TCPTP deficiency on insulin signaling in AgRP/NPY neurons (marked by Npy-GFP), we monitored for AKT Ser-473 phosphorylation (p-AKT) by immunofluorescence microscopy. In particular, we monitored for differences in p-AKT at intraperitoneal insulin doses where p-AKT predominated in AgRP/NPY neurons as opposed to higher insulin doses where p-AKT was also evident in POMC neurons (Fig. 1 and Supplementary Fig. 2). As noted previously (26), basal p-AKT in AgRP/NPY neurons (as reflected by the increased number of AgRP/NPY neurons positive for p-AKT) was enhanced in AgRP-TC mice (Fig. 1). Importantly, insulin-induced p-AKT in AgRP/NPY neurons was significantly increased in AgRP-TC mice (Fig. 1). Thus, TCPTP deficiency in AgRP/NPY neurons enhances insulin signaling.

To explore the effect of TCPTP deficiency and the promotion of IR signaling in AgRP/NPY neurons on glucose metabolism, we used 8- to 10-week-old chow-fed Ptpn2^fl/fl and AgRP-TC mice before any overt differences in body weight (Fig. 2A) and whole-body adiposity, as measured by DEXA (Fig. 2B). We found that TCPTP deficiency in AgRP/NPY neurons enhanced whole-body insulin sensitivity and glucose metabolism, as reflected by the improved glucose handling in response to boluses of insulin or glucose (Fig. 2C and D) and by the reduced fasted blood glucose and plasma insulin levels (Fig. 2E and F). To explore the extent to which the enhanced insulin signaling in AgRP/NPY neurons in AgRP-TC mice might be responsible for the improved whole-body glucose metabolism, we bred AgRP-TC mice onto an Insr^fl/+ background so that Insr gene expression in AgRP/NPY neurons would be reduced by 50%. We have shown previously that this largely corrects the otherwise enhanced hypothalamic insulin signaling, BAT activity, and white adipose tissue browning in AgRP-TC mice (26). We found that the enhanced insulin-induced p-AKT in AgRP/NPY ARC neurons in AgRP-TC mice was attenuated by 67% in AgRP-TC-IR (Fig. 3A). Importantly, glucose responses and fasted blood glucose and insulin levels in AgRP-TC mice were corrected so that AgRP-TC-IR mice resembled Ptpn2^fl/fl controls (Fig. 3C–F). By contrast, IR heterozygosity alone (Agrp-Ires-Cre;Insr^fl/+:AgRP-IR^fl/+) repressed insulin-induced p-AKT signaling in AgRP/NPY neurons compared with Insr^fl/+ controls (Supplementary Fig. 3A). This was not, however, sufficient to overtly affect glucose metabolism (Supplementary Fig. 3B–G), precluding any effects of IR heterozygosity in AgRP-TC mice being due to baseline effects on glucose homeostasis. TCPTP deficiency and the promotion of IR signaling in AgRP/NPY neurons therefore improves whole-body glucose metabolism independent of any effects on body weight.

To determine how TCPTP deficiency and the promotion of IR signaling in AgRP/NPY neurons might improve glucose metabolism, we used pyruvate tolerance tests as a means of assessing effects on HGP in the mice; administration of the gluconeogenic substrate pyruvate can increase blood glucose levels by promoting gluconeogenesis. We found that glucose excursions in response to pyruvate (mice fasted for 6 h from 9:00 a.m.) were significantly attenuated in AgRP-TC mice and reversed by Insr heterozygosity in AgRP-TC-IR mice (Fig. 4A). Moreover, genes encoding glucose-6 phosphatase (G6pc) and phosphoenolpyruvate-carboxykinase (Pck1), enzymes involved in the rate-limiting steps of gluconeogenesis, were decreased in AgRP-TC mice (Fig. 4B). The decreased glucose excursions in response to pyruvate in AgRP-TC mice were corrected by Insr heterozygosity, so that AgRP-TC-IR mice were indistinguishable from Ptpn2^fl/fl controls (Fig. 4A). These results are consistent with TCPTP deficiency and the promotion of IR signaling in AgRP/NPY neurons repressing hepatic gluconeogenesis.

Because the transformation of exogenous pyruvate into hepatic glucose is highly dependent on insulin sensitivity, we further explored the effects of TCPTP deficiency in AgRP/NPY neurons on whole-body insulin sensitivity and glucose metabolism using hyperinsulinemic-euglycemic clamps. The glucose infusion rate (GIR) needed to maintain euglycemia during the clamp was markedly increased in AgRP-TC versus Ptpn2^fl/fl mice, consistent with TCPTP deficiency in AgRP/NPY neurons improving whole-body insulin sensitivity (Fig. 4C and Supplementary Fig. 4). The improved insulin sensitivity was accompanied by an increased repression of endogenous glucose production (EGP), a measure of HGP (Fig. 4D and E), decreased hepatic expression of the gluconeogenic genes G6pc and Pck1 and increased hepatic expression of Il6 (Fig. 4F), and an increased R_d (a measure of glucose uptake) (Fig. 4G). Basal endogenous glucose production and glucose disposal were not altered in AgRP-TC versus Ptpn2^fl/fl mice in keeping with the improved glucose metabolism being a specific response to insulin (Fig. 4D and G). Furthermore, the increased GIR, decreased hepatic gluconeogenic gene expression, increased hepatic Il6 expression, decreased EGP, and increased R_d in clamped mice were completely corrected in AgRP-TC-IR mice (Fig. 4C–G), causally linking the improved systemic insulin sensitivity and glucose metabolism in AgRP-TC mice with the promotion of insulin signaling in AgRP/NPY neurons.

Our studies point toward the improved glucose homeostasis in AgRP-TC mice resulting from both the enhanced repression of HGP and increased R_d. The effects on the liver may therefore be secondary to the overall improvement in systemic insulin sensitivity or attributable to exacerbated hypothalamic-liver axis responses. To test the influence on the hypothalamic-liver axis, overnight fasted Ptpn2^fl/fl versus AgRP-TC mice were intracerebroventricularly administered vehicle or insulin and then processed for pyruvate tolerance tests, or the hypothalamus and liver were extracted and processed for quantitative real-time PCR (Fig. 5A–D). Although glucose excursions in response to pyruvate were not altered in Ptpn2^fl/fl mice at the insulin doses chosen, pyruvate responses were significantly repressed in AgRP-TC mice (Fig. 5B). Moreover, intracerebroventricularly administered insulin significantly repressed the hypothalamic expression of Agrp and Npy in AgRP-TC mice without altering Pomc expression (Fig. 5C) and increased hepatic Il6 expression while repressing Pck1 and G6pc expression (Fig. 5D). Although basal hepatic Pck1 and G6pc expression in AgRP-TC mice was increased (Fig. 5D), this was likely a compensatory response to prevent overt hypoglycemia after the overnight fast. Taken together, these results are consistent with the enhanced insulin signaling in AgRP/NPY neurons acting through the hypothalamic-liver axis to directly repress HGP and improve glucose metabolism.

Acute changes in AgRP neuronal activation may elicit effects on glucose metabolism by specifically influencing glucose uptake in BAT (13). Our studies indicate that the enhanced IR signaling in AgRP/NPY neurons in AgRP-TC mice increases systemic insulin sensitivity, at least in part by improving R_d. To determine the extent to which this may involve BAT glucose uptake, we administered mice [¹⁴C]-2-deoxy-d-glucose at the end of hyperinsulinemic-euglycemic clamps and assessed uptake in varied tissues (Fig. 4H). Although glucose uptake was not altered in the brain (hypothalamus), where glucose uptake is not insulin responsive, or in skeletal muscle or epididymal white adipose tissue, where insulin promotes glucose uptake, we found that glucose uptake was overtly increased in the BAT of AgRP-TC mice (Fig. 4H). Glucose uptake was also increased in the inguinal fat pads of AgRP-TC mice (Fig. 4H), where we have shown previously there is an increased abundance of thermogenically active beige adipocytes (26). By contrast, we did not note any increase in glucose uptake in epididymal white adipose tissue in AgRP-TC mice (Fig. 4H). Because the epididymal fat pad does not undergo browning in AgRP-TC mice (26), these results are consistent with TCPTP deficiency in AgRP/NPY neurons specifically increasing glucose uptake in brown and beige adipocytes. Moreover, the selective increase in BAT and inguinal white adipose tissue glucose uptake points toward this being a direct response to TCPTP deficiency in AgRP/NPY neurons rather than being an outcome of the systemic increase in insulin sensitivity. Importantly, the increased glucose uptake in BAT and inguinal white adipose tissue glucose were reduced to normal levels by Insr heterozygosity, so that AgRP-TC-IR mice were indistinguishable from Ptpn2^fl/fl controls (Fig. 4H). Our studies point toward TCPTP deficiency in AgRP/NPY neurons promoting IR signaling to improve whole-body glucose metabolism via the repression of HGP and the promotion of glucose uptake in brown/beige adipocytes.

We recently showed that TCPTP levels in the ARC are coordinated by diurnal feeding rhythms (26). Accordingly, we asked whether diurnal fluctuations in TCPTP might also help regulate the hypothalamic control of HGP. To explore this, we determined whether TCPTP deficiency in AgRP neurons might differentially influence the hypothalamic-liver axis in response to feeding and fasting. To test this we intracerebroventricularly administered vehicle or insulin to mice where food was withheld (food restricted) at the start of the dark cycle (Fig. 6A–F) versus ad libitum fed mice 4 h after the start of the dark cycle (Fig. 6G–K), when we have shown previously mice are satiated (26); under these conditions, hypothalamic TCPTP levels are high and low, respectively (26). To explore the influence on hepatic glucose metabolism in the mice, we used pyruvate tolerance tests or extracted livers for gene expression analyses. In food-restricted Ptpn2^fl/fl control mice, when hypothalamic TCPTP levels were high, we found that intracerebroventricular insulin had no effect on glucose excursions in response to pyruvate (Fig. 6B and E). By contrast, when hypothalamic TCPTP levels were low in fed Ptpn2^fl/fl mice, intracerebroventricular insulin effectively repressed pyruvate responses (Fig. 6H and K). Strikingly, pyruvate responses in food-restricted AgRP-TC mice were not only lower than in Ptpn2^fl/fl controls but reduced further in response to intracerebroventricular insulin (Fig. 6C and E), whereas responses in fed mice were similar to those of controls (Fig. 6I and K). Moreover, the precocious intracerebroventricular insulin-mediated repression of pyruvate responses accompanying AgRP TCPTP deficiency in otherwise unresponsive food-restricted mice were corrected by Insr heterozygosity (Fig. 6D and E), whereas pyruvate responses in fed mice were not altered by Insr heterozygosity (Fig. 6I and K). Similarly, in food-restricted Ptpn2^fl/fl mice, hepatic Il6, Pck1, and G6pc expression was unaltered by intracerebroventricular insulin, whereas in AgRP-TC mice, intracerebroventricular insulin increased hepatic Il6 expression and repressed Pck1 and G6pc expression, and these responses were corrected in AgRP-TC-IR mice (Fig. 6F). These results point toward diurnal feeding-associated fluctuations in TCPTP in AgRP neurons serving to coordinate hepatic glucose metabolism.

Gene deletion studies in rodents have established that insulin action on AgRP neurons in the hypothalamus is important for insulin’s ability to suppress HGP (10). Moreover, recent human studies exploring the utility of intranasal insulin administration have substantiated insulin’s ability to act via the CNS to suppress HGP and promote glucose uptake in lean but not in overweight individuals (34,35). However, the mechanisms by which hypothalamic insulin action is coordinated to influence whole-body glucose metabolism remain unclear.

We have shown previously that the abundance of the IR phosphatase TCPTP in hypothalamic neurons, including POMC and AgRP neurons, is altered by diurnal feeding rhythms in mice (26). Fasting increases TCPTP expression, whereas feeding represses TCPTP expression and promotes its rapid degradation (26). Increases in TCPTP serve to attenuate IR signaling in AgRP neurons after a fast to facilitate AgRP neuronal activation by hormones such as ghrelin (26), whereas the postprandial elimination of TCPTP helps promote IR signaling in AgRP neurons to facilitate AgRP neuronal inhibition by circulating insulin (26). We have shown that the regulation of IR signaling by TCPTP in AgRP neurons coordinates the browning of white adipose tissue and the expenditure of energy with feeding, so that fasting and AgRP neuronal activation repress browning and feeding and AgRP neuronal inhibition promotes browning and the expenditure of energy (26). In this way, diurnal fluctuations in hypothalamic TCPTP associated with feeding and fasting help maintain energy balance.

Here we report that the regulation of IR signaling by TCPTP in AgRP neurons is also important in coordinating whole-body glucose metabolism. We demonstrate that TCPTP deletion in AgRP neurons (emulating the fed state when hypothalamic TCPTP is eliminated) promotes IR signaling to enhance whole-body insulin sensitivity and glucose homeostasis. Mice lacking TCPTP in AgRP neurons show improved responses to glucose, pyruvate, and insulin, reduced fasted blood glucose and plasma insulin levels, and reduced GIRs during hyperinsulinemic-euglycemic clamps, independent of any differences in body weight/adiposity. Importantly, the enhanced glucose metabolism could be corrected by Insr heterozygosity in AgRP neurons, which largely corrected insulin-induced PI3K/AKT signaling in AgRP neurons. The improved glucose metabolism was partly attributable to the enhanced suppression of HGP. The enhanced suppression of HGP was a direct consequence of the hypothalamic-liver axis (7,10,15,16,18–21), because the CNS insulin-induced promotion of hepatic Il6 expression and STAT3 signaling, and consequent suppression of gluconeogenic genes and glucose excursions in response to pyruvate, were exacerbated by TCPTP deficiency in AgRP neurons. Although TCPTP deficiency in AgRP neurons attenuated hepatic gluconeogenic gene expression, we cannot rule out a contribution of glycogenolysis to the overall suppressed HGP in AgRP-TC mice, because previous studies have indicated that AgRP neurons may regulate glycogenolysis (36).

The extent to which feeding-/fasting-associated TCPTP fluctuations in AgRP neurons might help coordinate HGP was highlighted by the lack of any overt effect of intracerebroventricular insulin on pyruvate-induced glucose excursions and hepatic gluconeogenic gene expression in food-restricted control mice, where hypothalamic TCPTP levels would be high (26). This was contrasted by the striking ability of TCPTP deletion in AgRP neurons to reinstate such responses. By comparison, ad libitum fed and satiated mice readily responded to intracerebroventricular insulin by repressing hepatic gluconeogenic gene expression and pyruvate-induced glucose excursions, and these responses were unaltered by TCPTP deletion. Our results point toward the control of IR signaling by TCPTP in AgRP neurons serving to coordinate hepatic glucose metabolism so that fasting is accompanied by elevated HGP to prevent hypoglycemia. In obesity, where we have shown hypothalamic TCPTP levels are high and remain elevated even after feeding (26,28), the resulting sustained repression of IR signaling in AgRP neurons would be expected to contribute to the elevated HGP and hyperglycemia characteristic of the obese and insulin-resistant state. However, the decreased weight gain evident in high fat–fed AgRP-TC mice prohibited us from testing this directly.

Beyond influencing glucose production, our studies indicate that the regulation of IR signaling in AgRP neurons might also affect glucose clearance by specifically influencing glucose uptake in BAT and inguinal white adipose tissue, where browning in mice predominates (37–39). Activated brown and beige adipocytes contain high amounts of the uncoupling protein-1 (UCP-1), allowing for the uncoupling of fatty acid oxidation from ATP production to generate heat (2,3,39). Although largely ignored as a tissue involved in glucose homeostasis, early hyperinsulinemic-euglycemic clamp studies in rodents highlighted BAT as a major insulin-responsive depot for glucose uptake, exceeding on a per unit mass basis glucose uptake in muscle or white adipose tissue (1). Consistent with this, we found that when normalized for tissue mass, BAT was more effective than muscle or epididymal adipose tissue in taking up glucose under hyperinsulinemic-euglycemic conditions. In humans, brown and beige adipocytes are found interspersed in different white fat depots, including the supraclavicular depot as well as in the supraspinal, pericardial, and neck regions (2,39–45). Implantation of human brown/beige adipocytes into chow-fed, high fat–fed, or glucose-intolerant NOD-scid IL2^(null) mice dramatically enhances systemic glucose tolerance (46). Moreover, variables such as high BMI, increased age, or type 2 diabetes have been shown to correlate with attenuated brown/beige glucose uptake at least in some studies (41,47–49). Noteworthy, Lee et al. (5) recently highlighted the importance of brown/beige fat in humans in systemic glycemic control, demonstrating that higher brown/beige fat activity results in lesser glycemic variability. Moreover, the same study demonstrated that the thermogenic activity of brown/beige fat in humans was increased in response to a glucose challenge, which increases circulating insulin (5). Our recent studies have shown that enhanced leptin plus insulin signaling in POMC neurons or insulin signaling in AgRP neurons can function to promote BAT activity and the browning of inguinal white adipose tissue in rodents (25,26). In particular, we reported that in fed/satiated mice when TCPTP was degraded, the enhancement of insulin signaling and resultant inhibition of AgRP neurons increased the sympathetic innervation and browning of white adipose tissue to promote the expenditure of energy (26). Mice lacking TCPTP in AgRP neurons were remarkably resistant to weight gain due to the increased white adipose tissue browning as well as BAT activity (26). In this study we assessed the influence of TCPTP loss and the promotion of insulin signaling in AgRP neurons on glucose metabolism. For these studies, we used 8- to 10-week-old mice before any overt differences in adiposity/body weight resulting from the increased browning and BAT activity. Our studies indicate that the promotion of insulin signaling and inhibition AgRP neurons by TCPTP deletion at least partly improves glucose metabolism through the promotion of BAT and beige adipocyte glucose uptake. On the other hand, Steculorum et al. (13) reported that the pharmacogenetic activation of AgRP neurons promotes insulin resistance by repressing BAT glucose uptake.

Our results demonstrate that the deletion of TCPTP and inhibition of AgRP neurons (26) promotes systemic insulin sensitivity accompanied by increased glucose uptake in BAT and beige adipocytes in inguinal fat depots. Our findings are consistent with the TCPTP control of AgRP neuronal activation being instrumental in coordinating BAT/beige adipocyte activity with glucose uptake. Consistent with this assertion, Lee et al. (5) reported that BAT glucose uptake in humans correlates with thermogenesis (as assessed by measuring heat production with infrared thermography). Because fatty acid oxidation is essential for BAT thermogenesis (50–53), glucose likely indirectly contributes to BAT/beige adipocyte activity by promoting lipogenesis and/or supporting ATP production during thermogenic responses through anaerobic glycolysis (2,3). By contrast, other studies argue that BAT activity and glucose uptake can be dissociated. Blondin et al. (49), for example, reported recently that despite glucose uptake in BAT being diminished in older men with type 2 diabetes, cold-induced BAT oxidative metabolism and thermogenesis were not altered. However, this does not preclude feeding-induced beige adipocyte thermogenesis/WAT browning (26) normally being coordinated with glucose metabolism. This would provide an effective mechanism for coordinating both the removal and utilization of glucose by beige adipocytes to prevent of postprandial hyperglycemia.

Our results underscore the critical role of CNS insulin signaling in coordinating peripheral glucose metabolism through effects on both HGP and BAT/beige adipocyte glucose uptake. Taken together with our previous findings (26), our results point toward feeding-/fasting-associated alterations in hypothalamic TCPTP integrating the systemic control of glucose metabolism and energy expenditure with the nutritional state of the organism to maintain glucose and energy homeostasis. Thus, the promotion of CNS insulin sensitivity is likely to provide an important means by which to concomitantly promote weight loss and improve whole-body glucose metabolism and glycemic control in obesity and type 2 diabetes.

Acknowledgments. The authors thank Sunena Bhandari for technical support (Department of Biochemistry and Molecular Biology, Monash University, Clayton, Melbourne, Victoria, Australia) and Herbert Herzog (Neuroscience Division, Garvan Institute of Medical Research, St. Vincent’s Hospital, Sydney, New South Wales, Australia) for access to Insr^fl/fl mice.

Funding. This work was supported by the Diabetes Australia Research Trust (Y16G-DODG to G.T.D.) and by the National Health and Medical Research Council of Australia (APP1100240 to T.T.). T.T. is a National Health and Medical Research Council Principal Research Fellow.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. G.T.D. and R.S.L.-Y. performed investigations. G.T.D., R.S.L.-Y., J.C.B., and T.T. contributed to the methodology and wrote, reviewed, and edited the manuscript. G.T.D. and T.T. conceptualized the study and wrote the original draft of the manuscript. T.T. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.