Hypothalamic Neurogenesis Is Not Required for the Improved Insulin Sensitivity Following Exercise Training

Obesity dramatically increases the likelihood of the development of metabolic complications, including dyslipidemia, fatty liver, glucose intolerance, and cardiovascular disease, which increase morbidity and mortality (1,2). Insulin resistance is a prominent metabolic defect in many obese individuals and contributes to the development of type 2 diabetes. “Lifestyle” interventions involving reduced caloric intake and increased physical activity remain the cornerstone for the treatment of obesity-related diabetes. Laboratory-based studies demonstrate improved insulin sensitivity in the muscle and liver after a period of exercise training (3,4), and exercise reduces the progression from glucose intolerance to type 2 diabetes in at-risk individuals (5–7). The exercise-induced improvements in insulin action are mediated by a host of molecular adaptations that improve metabolic, vascular, and endocrine functions in the major peripheral glucoregulatory tissues, which include skeletal muscle, liver, and adipose tissue (for review, see Strasser [8]).

The regulation of plasma glucose was traditionally viewed as a process controlled by the coordinated actions of peripheral tissues. The pancreas responds to postprandial rises in glucose levels by secreting insulin, which in turn increases glucose disposal into skeletal muscle and other insulin-sensitive tissues, and reduces glucose output by the liver to restore euglycemia. However, studies in the last decade have shown that neuronal populations within the hypothalamus and hindbrain influence autonomic systems controlling hepatic glucose and lipid fluxes, pancreatic insulin secretion, and peripheral glucose uptake. For example, central injections of neuropeptide Y (NPY) inhibits hepatic glucose production (9) and stimulates hepatic VLDL-triglyceride secretion (10) via the sympathetic nervous system. Moreover, insulin signaling in the hypothalamus suppresses hepatic glucose production (11) via phosphoinositide 3-kinase–dependent activation of ATP-dependent potassium channels in agouti-related peptide–expressing neurons within the arcuate nucleus (ARC), which in turn regulate efferent vagal innervation of the liver (12). Insulin action in the central nervous system (CNS) increases muscle glucose uptake, independent of insulin signaling in muscle (13,14), and this involves the activation of ATP-dependent potassium channels. Although somewhat controversial (11,15), it is suggested that the insulin-dependent CNS pathway may account for more than half of the insulin-mediated glucose uptake in lean mice (14), making this an important regulatory point for systemic glucose homeostasis. The deletion of insulin receptors in the CNS causes insulin resistance (16), and insulin resistance of the CNS occurs in obese rodents (17–19) and humans (20). These results suggest that the management of impaired insulin action in obesity may require restoration of CNS insulin sensitivity. Indeed, successful treatment of uncontrolled diabetes depends on intact brain insulin action (21), and selective modification of the CNS in several animal models of insulin resistance and type 2 diabetes improves peripheral insulin action and reverses hyperglycemia (22).

Emerging evidence shows that new neurons are generated within the adult hypothalamus (2,23–25). In sedentary mice, the rate of neuronal turnover is relatively slow, with new neurons accounting for ∼6% of the total neuronal population (26). Markers of hypothalamic subtypes such as proopiomelanocortin, agouti-related peptide, and NPY are expressed in the newly formed neurons, suggesting involvement in metabolic regulation (2,23,25). Pioneering studies (2) have demonstrated pronounced neurogenesis in the energy balance circuitry of the ARC in response to central ciliary neurotropic factor (CNTF) administration. That neurogenesis was required to elicit the sustained and powerful anorexigenic effects of CNTF further indicates that postnatal neurogenesis may be important in regulating hypothalamic functions. Consistent with this notion are the findings that obesity is associated with impaired neurogenesis in the ARC (24,26), which leads to a relative aging of hypothalamic neuronal populations. Further, reducing neurogenesis by expressing constitutively active inhibitory κB kinase in hypothalamic neural stem cells (NSCs) causes overeating, weight gain, glucose intolerance, and hyperinsulinemia (26). Together, these data suggest that neurogenesis within the hypothalamic circuitry and/or turnover of energy-balance neurons is important for the maintenance of normal energy balance and glucose metabolism in mice. On the basis of these observations, strategies aimed at enhancing hypothalamic neurogenesis may be a viable strategy to combat obesity and diabetes.

Regular exercise training is commonly used to treat obesity and diabetes and is also known to induce neurogenesis in several brain regions (27), although exercise-induced neurogenesis in the hypothalamus is poorly described. In the current study, we tested the hypothesis that neurogenesis is an adaptive response to exercise that is important for the weight loss and insulin-sensitizing effects of regular exercise training in obesity.

Experimental procedures were approved by the School of Biomedical Sciences Animal Ethics Committee (Monash University). Male C57BL/6J mice were purchased from Monash Animal Services. Mice were fed a standard low-fat laboratory diet (10% of total energy from fat) or a micronutrient matched high-fat diet (HFD; 59% of total energy from fat; Specialty Feeds, Glen Forrest, WA, Australia).

C57BL/6J mice were familiarized with the treadmill (Columbus Instruments, Columbus, OH) for 3 days before the experiment, where mice ran for 30 min at 15 m/min on a 5% grade. Sedentary “control” mice were placed on the stationary treadmill for the same amount of time. Mice were culled 6 h later, and the hypothalamus was dissected (defined caudally by the mamillary bodies, rostrally by the optic chiasm, laterally by the optic tract, and dorsally by the apex of the hypothalamic third ventricle).

RNA from the hypothalamus was extracted in Qiazol extraction reagent and was isolated using an RNeasy Tissue Kit (Qiagen). RNA quality and quantity was determined (NanoDrop p2000 Spectrometer; Thermo Scientific) and reverse transcribed (Invitrogen). Gene products were determined by real-time quantitative RT-PCR (ep realplex Mastercycler; Eppendorf) using an RT² profiler PCR Array (SA Biosciences) with targeted expression of genes related to neurogenesis and NSC activation. Hspcb was used as a reference gene and did not vary between groups. The mRNA levels were determined by a comparative C_T method.

An intracerebroventricular (ICV) cannula and osmotic mini-pump were implanted in 20-week-old C57BL/6J mice (12 weeks on their respective diet). One day later, mice commenced training, which consisted of 30 min of treadmill running daily for 7 days at ∼12 m/min and a 5% slope (see protocol in Supplementary Table 1). Mice were perfused transcardially with 4% paraformaldehyde (PFA) 24 h or 28 days after the final exercise bout, and brains were removed for immunohistochemical analysis (Fig. 1A).

Mice were fed an HFD for 4 weeks, then were randomized into a vehicle or arabinosylcytosine (AraC) treatment group. Mice were implanted with an ICV cannula attached to an osmotic mini-pump (see below) and recovered for 2 days. Thereafter, mice remained sedentary or commenced exercise training, which consisted of daily treadmill running, five times a week for 4 weeks (Supplementary Table 2). Body weight was monitored throughout the training period, and exercise capacity was assessed after the exercising training period. Insulin sensitivity was assessed 72 h after the final exercise bout (Fig. 2A).

Mice were anesthetized by isoflurane inhalation and stereotactically implanted with steel guide cannulae (Plastics One, Roanoke, VA) targeted to the right lateral ventricles (−0.3 mm anteroposterior, 1.0 mm laterally to bregma, and −2.5 mm below the skull).

Cannulae were connected to subcutaneously implanted osmotic mini-pumps (flow rate 0.5 μL/h, 7 days; model 1007D; Alzet, Cupertino, CA) via 65-mm-long vinyl tubing filled with artificial cerebrospinal fluid (aCSF). Pumps were filled with vehicle solution, vehicle solution containing the known neurogenic agent Axokine (100 ng/μL, 1.2 µg/day; a modified CNTF, Regeneron Pharmaceuticals), AraC (3.3 μg/μL, 40 µg/day; Sigma-Aldrich, St. Louis, MO; selectively inhibits DNA synthesis), or Axokine plus AraC. The vehicle solution was aCSF containing 1 μg/μL BrdU (12 µg/day; Sigma-Aldrich). BrdU is incorporated into the DNA of dividing cells, and is commonly used for the birth dating and monitoring of cell proliferation (2,24,26).

Cannulae were connected to subcutaneously implanted osmotic mini-pumps (flow rate 0.11 µL/h, 28 days; model 1004; Alzet) via 33-mm-long vinyl tubing. Pumps and their tubing extensions were filled with vehicle solution or vehicle solution containing AraC (15.2 μg/μL, 40 µg/day; Sigma-Aldrich). The vehicle solution was aCSF containing 4.5 μg/μL BrdU (12 µg/day). The different concentrations of compounds between experiments 2 and 3 ensured that mice received the same total amount of chemical per day, irrespective of the flow rate. Mice were housed singly and monitored daily for body weight and food intake for all experiments.

Mice were anesthetized under isoflurane inhalation and perfused transcardially with 0.9% NaCl with 10 mg/L heparin followed by 4% PFA (Sigma-Aldrich). Brains were removed, postfixed by standard methods, and sectioned on a cryostat in the coronal plane. Sections (30 µm thick) were collected in four series. For BrdU immunostaining, after mounting (Superfrost Ultra Plus Slides; Thermo Scientific) and drying overnight, sections were fixed with 4% PFA for 10 min at room temperature, rinsed in PBS, then rinsed with 100% methanol for 20 min, rinsed in PBS, and incubated in 2N HCl for 30 min at 37°C. Sections were rinsed in PBS, blocked for 1 h with 5% normal horse serum in PBS/0.02% Triton X-100, and incubated with sheep anti-BrdU antibodies (1:400; Abcam) overnight at 4°C. Sections were rinsed and incubated with the appropriate secondary antibody for 1 h, rinsed in PBS, and coverslipped. For colabeling analyses, sections first underwent BrdU immunohistochemistry and were then incubated with primary antibodies (neuronal nuclei [NeuN] 1:1,000; Abcam) or glial fibrillary acidic protein (GFAP; 1:1000; Dako) in blocking solution. Sections were rinsed and incubated with the appropriate secondary antibody for 1 h, washed in PBS, and coverslipped.

BrdU⁺ cells in the caudal hypothalamus were counted using stereological principles of quantifying BrdU⁺ cells from systematically collected serial sections with the initial section chosen at random. Per animal, every 4th coronal section (30 µm thickness) throughout the caudal hypothalamus (−1.22 to −2.70 mm from bregma) was analyzed by standard fluorescence microscopy. BrdU⁺ cells within the hypothalamic parenchyma were counted for each section analyzed, excluding cells of the uppermost focal plane to avoid oversampling. The average number of BrdU⁺ cells for any given section of the caudal hypothalamus was calculated as the sum of the counted BrdU⁺ cells divided by the total number of sections counted.

Mice ran on the treadmill at 10 m/min for 2 min (5% grade), and the speed was increased by 2 m/min every 2 min until the mouse reached exhaustion.

An insulin tolerance test (ITT) was conducted 72 h after the last exercise bout. Mice were fasted for 4 h, and venous blood glucose was assessed (Accu-Check glucomenter; Roche) before and after intraperitoneal insulin administration (1 units/kg body weight; Actrapid; Novo Nordisk, Bagsvaerd, Denmark).

After a 3-h fast (0800–1100 h) mice were injected via a tail vein with 0.5 units/kg insulin, 10 μCi of 2-[1-³H]deoxyglucose and 2 μCi of [U-¹⁴C]glucose. Blood samples were obtained from a cut in the tail at 2, 5, 10, 15, and 20 min. Mice were killed by decapitation, and tissue was collected. 2-[1-³H]Deoxyglucose clearance from the blood and into tissues was performed in mice as described previously (28,29).

Phosphorylated Akt (Ser⁴⁷³) and Akt 1:1,000 (catalog #9271 and #4685, respectively; Cell Signaling Technology) were assessed in hypothalamic lysates by standard methods (30).

Results are expressed as the mean ± SEM. Statistical analysis was performed by using the Student t test, or one-way, two-way, or repeated-measures two-way ANOVA with Bonferroni post hoc test, as appropriate. Significance was established a priori at P ≤ 0.05.

Short-term exercise increased the expression of genes involved in cell proliferation, including Egf, Fgf2, Il3, and Vegfa (Table 1). Proneurogenic growth factors, including Artn, Bdnf, Bmp15, Bmp2, and Ndp, were induced by exercise. Levels of regulators of the cell cycle and known transcriptional regulators of neurogenesis, including Arnt2, Hey1, Mef2c, Neurod1, Notch2, Pax3, and Pax6, were also increased after exercise (Table 1). No gene was downregulated by exercise. Thus, short-term exercise promotes a proneurogenic transcriptional program in the hypothalamus of adult mice.

To determine whether regular exercise training induces hypothalamic cell proliferation, mice ran at a moderate intensity on a treadmill for 30 min/day for 1 week. Mice received a continuous ICV infusion of BrdU to label dividing cells and were killed 1 day after the last exercise bout. Hypothalamic cell proliferation in sedentary mice averaged 31 ± 5 BrdU⁺ cells per 30-µm section, and this number increased 3.5-fold in exercise-trained mice (Fig. 1B). This was substantially less than in mice treated with the neurogenic factor CNTF, which was implemented as a positive control (Fig. 1C). The number of BrdU⁺-labeled cells was reduced by 45% in exercise-trained mice at 33 vs. 8 days (Fig. 1B), indicating that approximately half of the newborn cells survived. One caveat to this interpretation is that cells newly labeled with BrdU cannot be distinguished from cells that have incorporated BrdU and then divided. Thus, actual “survival” rates are semiquantitative. Together, these findings show that exercise training increases cell proliferation in the hypothalamus of mice.

Examination of coronal sections of sedentary and exercise-trained mice show that BrdU⁺ cells were diffusely distributed throughout the hypothalamus but were concentrated mostly in regions ∼200–300 µm from the third ventricle, including the ARC and the ventromedial and dorsomedial hypothalamic nuclei (Fig. 1B). Very few BrdU⁺ cells were present in the ependymal lining of the third ventricle. Pairs of newly divided BrdU cells were distributed extensively within the hypothalamic parenchyma (Fig. 1E), which is consistent with the notion that cell proliferation in the hypothalamus is not restricted to a specific zone or nuclear region.

Cell proliferation was not impaired in obese sedentary mice compared with lean sedentary mice, with an average of 43.4 ± 6.6 and 31.5 ± 4.5 BrdU⁺ cells per section, respectively (Fig. 1D). Exercise increased the number of BrdU⁺ cells by 3.3-fold above that observed in obese sedentary mice, and there was no significant reduction in BrdU⁺ cells after 28 days in the obese exercise-trained mice (Fig. 1D), suggesting that the majority of the proliferating cells survived. Thus, the exercise-induced proliferation and survival rates in obese mice were not impaired when compared with lean mice.

Given that aging has been associated with reduced neuroregenerative capacity, comparison was made between groups of aged (∼52 weeks) and relatively young (∼18 weeks) mice that were exercise trained for 1 week. These experiments showed that hypothalamic cell proliferation is reduced by 2.2-fold in aged mice compared with young mice and that exercise increases cell proliferation by 1.5-fold in aged mice, even after 1 year of high-fat feeding (Fig. 1F). The level of cell proliferation in exercise-trained aged mice was similar to that in sedentary young mice.

We next determined the neural contribution to cell proliferation in the ARC by assessing colocalization of BrdU with neuron (NeuN) and astrocyte (GFAP) markers. Only ∼0.5% of the BrdU⁺ cells were colabeled with NeuN in exercise-trained mice (33 days), and there was no BrdU/NeuN colabeling in sedentary mice (Fig. 1G). We applied this same approach to the subgranular zone of the hippocampal dentate gyrus, an area in which neurogenesis is well established. The majority of BrdU-labeled cells were colocalized with NeuN in exercise-trained mice, confirming marked neurogenesis. BrdU/NeuN colabeling was less apparent in sedentary mice (Fig. 1H). There was also limited colocalization of BrdU and GFAP in sedentary and exercise-trained mice, indicating restricted production of new astrocytes (Fig. 1I). Because negligible neurogenesis was detected in the ARC of lean mice (0.5% of proliferating cells), we rationalize that major changes in the neurochemical phenotype of the proliferating cells is unlikely in obesity and have not conducted these analyses.

Given the well-recognized impact of regular moderate exercise on improved metabolic performance and the observation that exercise promotes marked cell proliferation in the hypothalamus (Fig. 1B), the link between these two factors was tested in groups of HFD-fed mice that were exercise trained and exposed to ICV AraC (Fig. 2A). Four weeks of training was selected because this was the minimum volume of training required to detect improvements in whole-body insulin action in high-fat–fed C57BL/6J mice (Fig. 2B). A range of metabolic parameters was measured with and without AraC treatment to test the involvement of cell proliferation in the beneficial effects of exercise training on energy balance and insulin action. AraC blocked cell proliferation in the hypothalamus of mice treated with CNTF, a powerful neurogenic stimulator (Fig. 2C), and in exercise-trained mice (blank images not shown). Blocking cell proliferation did not affect food intake (Fig. 2D) (P = 0.77), body mass (Fig. 2E) (P = 0.23), epididymal fat mass (Fig. 2F) (P = 0.31), plasma leptin levels (Fig. 2G) (P = 0.49), or plasma free fatty acid (FFA) levels (Fig. 2H) (P = 0.36) in sedentary mice.

Exercise training improved the endurance exercise capacity of mice by 2.2-fold compared with untrained age-matched mice (Fig. 3A). The endurance capacity was reduced in exercise-trained mice treated with AraC compared with vehicle-treated mice (Fig. 3A) (P = 0.0002). Blocking cell proliferation did not affect food intake (Fig. 3B) (P = 0.44), or body mass (Fig. 2C) (P = 0.48) during exercise training. Epididymal fat mass (Fig. 2D) (P = 0.22), plasma leptin levels (Fig. 2E) (P = 0.78), and plasma triglyceride levels (vehicle: 0.65 ± 0.07 vs. AraC 0.56 ± 0.07, n = 16 per group, P = 0.39) were not different between groups, while plasma FFA levels were mildly increased (Fig. 2F) (P = 0.04).

To test whether hypothalamic cell proliferation (including neurogenesis) contributes to the improvements in insulin action after exercise training, insulin was coadministered intravenously with ³H-2-deoxyglucose (³H-2DG) and ¹⁴C-glucose tracers into conscious mice. Whole-body insulin-stimulated glucose uptake was not affected by AraC administration in sedentary mice (Fig. 4A). Exercise training improved whole-body insulin sensitivity, as demonstrated by a 45% increase in whole-body ¹⁴C-glucose clearance (Fig. 4A) (main effect P = 0.04). ¹⁴C-glucose clearance (Fig. 4A) (P = 0.34) and the reduction in blood glucose levels (Fig. 4B) (P = 0.29) were not different between vehicle-treated and AraC-treated groups of exercise-trained mice. ³H-2DG clearance in the mixed quadriceps (Fig. 3C) (P = 0.74), heart (Fig. 3D) (P = 0.38), brown adipose tissue (Fig. 3E) (P = 0.98), and liver (vehicle 17.8 ± 5.1 vs. AraC 17.8 ± 6.7 K × 1,000, P = 0.91) was not different between groups, whereas ³H-2DG clearance in white adipose tissue tended to be decreased in AraC-treated mice (Fig. 3F) (P = 0.06). Insulin-mediated ¹⁴C-glucose accumulation in liver lipids was also unaffected by AraC treatment (Fig. 3G) (P = 0.80). Finally, AraC treatment did not affect Akt Ser⁴⁷³ phosphorylation in hypothalamic lysates (Fig. 3H) (P = 0.59), indicating that there was no improvement in insulin sensitivity. Together, these data demonstrate that cell proliferation is not required for the exercise-mediated enhancement in insulin action in obese mice.

Adult neurogenesis is most prominent within the subventricular zone of the lateral ventricles and the subgranular zone of the dentate gyrus in the hippocampus (31), and impaired neurogenesis is implicated in the etiology of several neurodegenerative diseases including Parkinson disease, Alzheimer disease, and amyotrophic lateral sclerosis (32). Notably, enhancing neurogenesis through pharmacological or physiological stimuli (e.g., environmental enrichment and voluntary physical activity) attenuates neurodegenerative disease progression and improves cognitive function (27,33–36). Together, these results link the maintenance of adult neurogenesis to normal CNS function and suggest a therapeutic strategy for treating neurodegenerative conditions. The questions are then as follows: to what extent does neurogenesis impact metabolic phenotypes and can regular exercise training modulate this response?

In this study, we have observed cell proliferation in the hypothalamus of lean mice that is most abundant in the region immediately surrounding the third cerebral ventricle. Newly formed cells, identified by the incorporation of BrdU, were not localized to anatomically distinct nuclei but were graded in abundance with distance from the ventricle, although there were instances where isolated cells were present deep in the lateral hypothalamus. While we have demonstrated that exercise upregulates a subset of transcripts involved in cell proliferation and neurogenesis, very few newborn cells take a neuronal fate, indicating limited hypothalamic neurogenesis with exercise training. Hypothalamic cell proliferation was not dysregulated in diet-induced obese mice and was substantially increased after exercise training. While exercise-induced cell proliferation occurs in the hypothalamus, a CNS region essential for the control of energy balance and insulin action, our data do not support an important role for hypothalamic cell proliferation in the exercise-induced improvements in insulin sensitivity and energy balance. By extension, our data question the importance of hypothalamic cell proliferation and neurogenesis in the regulation of energy balance and insulin action in the setting of obesity.

While new hypothalamic cells are produced during adulthood, their relevance to physiology remains unclear. The genesis for this, and previous work in the field, was derived from the seminal observation that CNTF administration into the CSF of mice leads to rapid and pronounced weight loss that is maintained for at least 1 month after the cessation of CNTF administration. Importantly, when neurogenesis was blocked by the coadministration of AraC, the sustained weight loss in CNTF-treated mice was abrogated, indicating that neurogenesis was required for the sustained anorectic effects of CNTF (2). These long-lasting effects were attributed to neurogenesis within the hypothalamic feeding circuits, specifically NPY-expressing and proopiomelanocortin-expressing neurons, with each playing crucial antagonistic roles in the regulation of energy balance. We asked whether cell proliferation, and by extension neurogenesis, could be enhanced by a physiological stimulus and whether this was critical for metabolic regulation. We then exploited two well-known phenomena (that obesity induces insulin resistance and that regular exercise training improves insulin action) to test the hypothesis that cell proliferation is important in homeostatic control of metabolism. Blocking cell proliferation in exercise-trained mice did not affect fasting blood glucose levels, whole-body insulin-stimulated glucose uptake, or tissue-specific insulin sensitivity. In addition, hypothalamic insulin signaling was unaffected by AraC administration, indicating that hypothalamic insulin signaling is not impacted despite improvements in whole-body insulin action. While exercise training reduces body mass gain in HFD-fed mice (37), the blocking of cell proliferation has no effect on body mass or food intake, thereby indicating no effect on daily energy expenditure. The simplest conclusion from our studies is that cell proliferation, and by extension neurogenesis, does not play a major role in modulating insulin action and energy balance in obese mice. However, regular exercise training may induce other changes to the hypothalamic circuitry to improve insulin action independent of cell-proliferative effects.

Obesity induced by high-fat feeding results in altered synaptic plasticity in mice (38), which is postulated to lead to a relative aging of the hypothalamic neuronal population (39). Such a disruption in hypothalamic circuitry may be predicted to impair the sensitivity of ARC neurons to nutrient and hormonal signals associated with metabolic challenges and to compromise energy homeostasis. While previous studies have reported neurogenesis in the ARC of the hypothalamus in lean mice, it is currently unclear whether neurogenesis is actually compromised in obesity or whether neurogenesis impacts the development of obesity and its related disorders. On the one hand, both high-fat feeding (24,26) and inhibitory κB kinase/nuclear factor-κB activation in NSCs lead to reduced proliferation and survival of new-born hypothalamic neurons, which is associated with overeating and insulin resistance (26). Conversely, neurogenesis in the median eminence of the hypothalamus is enhanced by high-fat feeding in young mice and is associated with energy overconsumption and increased fat mass (23). The same authors (23) also showed that localized irradiation in the mediobasal hypothalamus of HFD-fed mice reduced neurogenesis, which was associated with increased energy expenditure and decreased weight gain. While, these conflicting data relating to the effect of diet-induced obesity on hypothalamic neurogenesis are difficult to reconcile, there are a number of variables across the studies that are likely to impact outcomes. These include the age of mice, the timing of the dietary intervention, the timing of BrdU administration and subsequent analyses, the hypothalamic localization of progenitors that would invariably impact the specific neurogenic microenvironment (e.g., median eminence ependymal layer [23]), discrete zones of the mid–third ventricle wall (40), ARC, mediobasal hypothalamus (2,23), and the cell of origin being examined (e.g., β₂-tanycytes vs. NSCs). Despite the lack of clear direction from previous studies, the present work shows that high-fat feeding does not impact cell proliferation in the hypothalamus of adult mice, a result that is in keeping with data derived from rats (41).

While the majority of hypothalamic neurons are generated during embryogenesis, it is thought that postnatal neurogenesis is essential for maintaining functional plasticity and adaptability in adulthood. The question arises as to what are the determinants of such neurogenic activity. Voluntary physical activity increases neurogenesis in several brain regions in mice (42–44), and as such the current study sought to determine whether structured regular exercise training, similar to what would be prescribed for maintaining healthy weight, would increase hypothalamic neurogenesis. We observed a generalized upregulation of transcripts involved in cell proliferation and survival, transcriptional regulation of neurogenesis, and growth factors in the hypothalamus after a short bout of exercise. Thus, the hypothalamic niche is rapidly altered after exercise to promote neurogenesis. Despite these molecular responses and the finding of marked hypothalamic cell proliferation after exercise training in both lean and obese mice, exercise did not increase neurogenesis per se, and there was evidence of limited astrocyte generation. The absence of hypothalamic neurogenesis is perplexing, especially in light of the coordinated proneurogenic transcriptional response to exercise and the marked hippocampal neurogenesis observed in the same exercise-trained mice (using the same BrdU/NeuN colabeling approach), essentially acting as a positive control. The negligible hypothalamic neurogenesis could be explained by slow neuronal differentiation of NSCs (24,26) or proliferation of NSCs that is matched by decreased survival (24). Further studies that delineate the upstream drivers and requisite molecular signaling that facilitates the proliferation and differentiation of NSCs will provide valuable insights into why ARC neurogenesis is severely limited when compared with the dentate gyrus of the hippocampus. In addition, the phenotype of the new exercise-generated ARC cells remains undetermined, but, based on the differentiation potential of NSCs (26), these BrdU⁺ cells might be NSCs undergoing self-renewal. Alternatively, previous studies of brain regions other than the hippocampus (45,46), such as the substantia nigra and cerebral cortex, report an abundance of cells with oligodendrocytic precursor characteristics that differentiate into mature oligodendrocytes with exercise. In this regard, oligodendrocytes constitute the majority (∼55%) of newly generated cells in the medial prefrontal cortex with exercise training, with evidence of limited neurogenesis (∼5%) (47). Moreover, the majority of BrdU⁺ cells in the hypothalamus of sedentary mice are of oligodendrocyte lineage, and only a small percentage of these become neurons (∼8%) (48,49).

In conclusion, these data demonstrate that exercise induces a rapid transcriptional response in the hypothalamus that drives substantial cell proliferation, even in obesity and aging. In contrast to studies using pharmacological doses of mitogens, exercise does not induce significant neurogenesis in the hypothalamus, and blocking cell proliferation does not change feeding, energy balance, or insulin action in the setting of rodent obesity. These results question whether physiological cell proliferation and neurogenesis are major regulators of insulin action and metabolic phenotypes.

Acknowledgments. The authors thank Clinton Bruce (Deakin University, Burwood, Victoria, Australia) and Vanessa Haynes (Monash University) for technical assistance.

Funding. These studies were supported by grant APP1005053 from the National Health and Medical Research Council (NHMRC) of Australia (to B.J.O., Z.B.A., and M.J.W.). M.L.B. was supported by an Australian Postgraduate Award. Z.B.A. is supported by a Future Fellowship from the Australian Research Council (FT100100966), and B.J.O. and M.J.W. are supported by research fellowships from the NHMRC.

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

Author Contributions. M.L.B. and M.J.W. researched the data and wrote the manuscript. M.L., A.R., and A.S. researched the data. B.J.O. contributed to the discussion, and reviewed and edited the manuscript. Z.B.A. researched the data, contributed to the discussion, and reviewed and edited the manuscript. M.J.W. 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.