Inflammation of the Hypothalamus Leads to Defective Pancreatic Islet Function*

Type 2 diabetes mellitus results from the complex association of insulin resistance and pancreatic β-cell failure. Obesity is the main risk factor for type 2 diabetes mellitus, and recent studies have shown that, in diet-induced obesity, the hypothalamus becomes inflamed and dysfunctional, resulting in the loss of the perfect coupling between caloric intake and energy expenditure. Because pancreatic β-cell function is, in part, under the control of the autonomic nervous system, we evaluated the role of hypothalamic inflammation in pancreatic islet function. In diet-induced obesity, the earliest markers of hypothalamic inflammation are present at 8 weeks after the beginning of the high fat diet; similarly, the loss of the first phase of insulin secretion is detected at the same time point and is restored following sympathectomy. Intracerebroventricular injection of a low dose of tumor necrosis factor α leads to a dysfunctional increase in insulin secretion and activates the expression of a number of markers of apoptosis in pancreatic islets. In addition, the injection of stearic acid intracerebroventricularly, which leads to hypothalamic inflammation through the activation of tau-like receptor-4 and endoplasmic reticulum stress, produces an impairment of insulin secretion, accompanied by increased expression of markers of apoptosis. The defective insulin secretion, in this case, is partially dependent on sympathetic signal-induced peroxisome proliferator receptor-γ coactivator Δα and uncoupling protein-2 expression and is restored after sympathectomy or following PGC1α expression inhibition by an antisense oligonucleotide. Thus, the autonomic signals generated in concert with hypothalamic inflammation can impair pancreatic islet function, a phenomenon that may explain the early link between obesity and defective insulin secretion.

It is predicted that obesity will affect more than one billion people by the year 2020 (1). As the most important risk factor for type 2 diabetes mellitus (DM2), 2 the worldwide rise in the prevalence of obesity is expected to lead to a similar rise in DM2 and all clinical complications that accompany this disease (2).
Over the years, a number of studies have investigated the mechanisms linking obesity to DM2. Currently it is accepted that the generation of inflammatory factors by the enlarged adipose tissue and the high levels of circulating fatty acids activate serine/ threonine kinases and TLR2/4 signaling in insulin-sensitive tissues, impairing insulin signal transduction and, thus, leading to whole body insulin resistance (2)(3)(4). However, DM2 will develop only if the pancreatic ␤-cells become unable to compensate for the increasing demand for insulin in peripheral tissues (5,6). Thus, it has been proposed that the failure of the pancreatic ␤-cells is, in fact, the determining factor for the development of DM2.
Glucose is the main modulator of ␤-cell function; however, other nutrients, hormones, and neural inputs play important regulatory roles in the physiology and pathology of this endocrine tissue (7,8). The autonomic nervous system can either stimulate insulin secretion through muscarinic receptors or inhibit ␤-cell function, through ␣-adrenergic signaling (9,10). Many of the neural inputs that control ␤-cell activity are generated in the hypothalamus (9).
Recent studies have shown that, in animal models of obesity, the hypothalamus is the first tissue to be affected, becoming dysfunctional because of the development of inflammation, which leads to resistance to the adipostatic hormones leptin and insulin and to the loss of the homeostatic control of food intake and energy expenditure (11)(12)(13)(14). Because of the important role played by the autonomic nervous system in the control of insulin secretion, we decided to evaluate whether a low grade, obesity-like inflammation of the hypothalamus would lead to an anomalous control of pancreatic ␤-cell function.
Here, we show that hypothalamic inflammation, generated by either TNF␣ or a saturated fatty acid, activates apoptotic genes and modulates insulin secretion by the pancreatic ␤-cells.

MATERIALS AND METHODS
Antibodies, Oligonucleotides, Chemicals, and Buffers-Antibodies against UCP2 (sc-6526, goat polyclonal), TNF␣ (sc-1350, goat polyclonal), BAX (sc-493, rabbit polyclonal), Bcl2 (sc-7382, mouse monoclonal), IL1␤ (sc-7884, rabbit polyclonal), IL6 (sc-1265, goat polyclonal), Akt (sc-1619, goat polyclonal), NFB/p50 (sc-7178, rabbit polyclonal), phospho-JNK (pJNK, sc-12882, rabbit polyclonal), JNK (sc-572, rabbit polyclonal), phospho-IB (pIB, sc-23470, rabbit polyclonal), IB (sc-9130, rabbit polyclonal), and ␤-actin (sc7210, rabbit polyclonal) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Antibodies against PGC1␣ (2178, rabbit monoclo-* This work was supported by grants from Fundação de Amparo à Pesquisa do Estado de São Paulo and Conselho Nacional de Desenvolvimento Científico e Tecnoló gico and in part by the Instituto Nacional de Ciê ncia e Tecnologia-Obesidade e Metabolismo. 1 To whom correspondence should be addressed: DCM-FCM, UNICAMP, 13084-761 Campinas-SP, Brazil. E-mail: lavelloso@fcm.unicamp.br. 2 The abbreviations used are: DM2, type 2 diabetes mellitus; icv, intracerebroventricular(ly); CTL, control; AS, antisense; S, sense; HF, high fat; SA, stearic acid; ipGTT, intraperitoneal glucose tolerance test; ipITT, intraperitoneal insulin tolerance test.   ), stearic acid/ C18:0, bovine serum albumin (fraction V), bovine serum albumin fatty acid free (A-6003), and 2-hydroxypropyl-␤-cyclodextrin (01816LD) were from Sigma-Aldrich. The nitrocellulose membranes were developed using commercial chemiluminescent kits (GE Healthcare). Sense and antisense phosphorothioate-modified oligonucleotide specific for PGC1␣ (sense, TCA GGA GCT GGA TGG C and antisense, GCC ATC CAG CTC CTG A) was produced by Invitrogen. These sequences have been previously tested and reduce the expression of the target by up to 60% in pancreatic islets (15). The selected sequences were analyzed (BLAST, NCBI) for similarities with mRNA of other proteins and presented a 100% intra-species matching only for Rattus norvegicus PGC1␣. ELISA kits for the determination of insulin and leptin were from Linco-Millipore (Billerica, MA). The reagents and primers for real time PCR were from Applied Biosystems (Foster City, CA). Experimental Animals-Male Wistar rats from the University of Campinas Animal Breeding Center were used in all of the experiments. For the experiments with a high fat diet, 4-weekold rats were randomly selected for either control or high fat diets and employed in the experiments after 12 weeks. For the remainder of the experiments, 8-week-old rats were employed. The investigation was approved by the ethics committee and followed the university guidelines for the use of rats in experimental studies and conforms to National Institutes of Health guidelines (48). The rats were maintained on 12-h:12-h artificial light-dark cycles and had free access to water and food. Typically three rats were maintained in each cage. After surgical procedures (sympathectomy or intracerebroventricular cannulation), the rats were housed in individual cages.
Intracerebroventricular (icv) Cannulation-The rats were stereotaxically instrumented under anesthesia (intraperitoneal injection with a mix of ketamin (10 mg) and diazepam, containing 10 and 0.07 mg/ml, respectively, and injecting 0.2 ml/100 g of body weight) with a chronic 26-gauge stainless steel indwelling guide cannula, aseptically placed into the lateral cerebral ventricle at pre-established coordinates, anteroposterior, 0.2 mm from bregma; lateral, 1.5 mm; and depth, 4.2 mm, according to a previously reported technique (16). After a 1-week recovery period, cannula placement was confirmed by a positive drinking response elicited by the administration of angiotensin II (2 l of 10 Ϫ6 M solution) (17); rats that drank less than 5 ml of water within 15 min after treatment were excluded from the studies. Adequately cannulated rats were randomly selected for the experimental groups.
Sense and Antisense Oligonucleotide Treatment Protocols-Phosphorothioate-modified sense (S) and antisense (AS) oligo-nucleotides were diluted to a final concentration of 2 nmol/100 l in dilution buffer containing 10 mmol/liter Tris-HCl and 1.0 mmol/liter EDTA.
Experimental Protocols-In the first part of the study, 4-week-old rats were randomly divided into control or high fat (HF) diet groups for a 12-week period. The compositions of diets have been published elsewhere (18). In some experiments lean rats were treated icv twice a day, in the morning (8:00 a.m.) and in the afternoon (5:00 p.m.), for 5 consecutive days with 90 M BSA or 45, 90, or 180 M stearic acid (SA). Stearic acid used for icv injections was always diluted in ultrapure water containing 2-hydroxypropyl-␤-cyclodextrin detergent (0.1%) and fatty acid free BSA (75 M). In some experiments, a single icv dose of TNF␣ (10 Ϫ12 M) was injected, and the experiments were performed 3, 6, 12, or 36 h later. The volumes injected icv were always 2.0 l/dose. To investigate the role of PGC1␣ in the insulin secretion, rats treated icv with stearic acid, as described above, received via intraperitoneal injection, one daily dose (8:00 am) of 200 l of dilution buffer containing S or AS oligonucleotides for PGC1␣ for 5 consecutive days. The role of the sympathetic nervous system in insulin secretion was investigated in rats submitted to surgical pancreatic sympathectomy, as described previously (19). Thereafter, they were submitted to icv cannulation. On the sixth day after surgery and an overnight fast, the rats were submitted to an intraperitoneal glucose tolerance test (ipGTT) to confirm the effectiveness of pancreatic sympathectomy. A significant reduction in blood glucose levels indicated the effectiveness of sympathectomy.
Intraperitoneal Glucose Tolerance Test-An intraperitoneal ipGTT was performed on the fourth day of the experimental period (5 days). Food was withdrawn at 12 h before the experiment, the rats were weighed, and a steady-state blood sample was taken from the tip of the tail (t ϭ 0 min). Subsequently, each rat received a glucose solution load (2 g/kg, intraperitoneally), and additional blood samples were collected at 15, 30, 60, and 120 min after injection. The glucose levels during the test were measured immediately. The area under the curve was calculated from values for each rat.
Intraperitoneal Insulin Tolerance Test (ipITT)-An ipITT was performed on the fourth day of the experimental period (5 days) in fed rats. The rats were weighed, and after collection of a blood sample at time 0, a solution of insulin (2 units/kg of body weight) was injected intraperitoneally. Blood samples were collected from the tip of the tail at 5, 10, 15, 20, 25, and 30 min for serum glucose determination. The constant rate for glucose disappearance (K itt ) was calculated using the formula 0.693/t1 ⁄ 2 . The glucose t1 ⁄ 2 was calculated from the slope of the least square analysis of the plasma glucose concentrations during the linear decay phase.
Metabolic, Hormonal, and Biochemical Measurements-Blood glucose concentrations were measured from the tip tail using a glucosimeter (One Touch; Johnson & Johnson). Insulin and leptin in blood samples were measured by ELISA, following the instructions of the manufacturer (Linco). Insulin secreted in isolated pancreatic islet studies was determined by radioimmunoassay, as described previously (20).
Hypothalamic Dissection-After the 5-day treatment period, the fed rats were killed as described, and the hypothalamus was quickly removed and prepared for either immunoblot or real time PCR analysis.
Islet Isolation-Islets were isolated from fed rats by the collagenase method. Briefly, the pancreas was inflated with Hanks' solution containing 0.8 mg/ml collagenase and then removed from the rat and kept at 37°C for 23 min. After tissue digestion, the islets were collected manually under a microscope using a Pasteur pipette. For each experiment of static or dynamic insulin secretion, we usually isolated the islets from three or four rats. The islets were pooled, and then groups with four (for static insulin secretion studies) or 100 (for dynamic insulin secretion studies) islets were prepared. For statistical analysis, each group corresponds to n ϭ 1.
Static Insulin Secretion-Groups of four islets were first incubated for 30 min at 37°C in Krebs-Ringer bicarbonate buffer containing glucose 5.6 mM and equilibrated with 95% O 2 , 5% CO 2 , pH 7.4. The solution was then replaced with fresh Krebs-Ringer bicarbonate buffer, and the islets were incubated for a further 60-min period with medium containing 2.8, 11.1, 16.7, or 22.2 mM glucose. The incubation medium contained 115 mM NaCl, 5 mM KCl, 24 mM NaHCO 3 , 2.6 mM CaCl 2 , 1 mM MgCl 2 , and 25 mM HEPES, pH 7.4, supplemented with BSA (0.3% w/v). For measurement of the total insulin content, groups of four islets were collected. An alcohol acid solution (1 ml; final concentration of 20% of ethanol and 0.2 mM of HCl) was added to the samples, followed by sonication of the pancreatic islets (three times, 15-s pulses). Insulin in the medium was measured by radioimmunoassay.
Dynamic Insulin Secretion-Groups of 100 freshly isolated islets were preincubated for 6 h at 37°C in RPMI 2.8 mM glucose. Thereafter, the islets were placed on a Millipore SW 1300 filter (8-m pore) in a perfusion chamber. Islets were continu-FIGURE 1. Metabolic and inflammatory parameters in diet-induced obesity. a-c, body mass variation (a), mean caloric intake (b), and epididymal fat (c) were determined in male Wistar rats fed a CTL (filled squares) or a HF (filled circles) diet for 0 -12 weeks (a and c) or for 12 weeks (b). d-f, TNF␣ (d), IL1␤ (e), and IL6 (f) transcript amounts were determined by real time PCR in hypothalamic samples of Wistar rats fed on CTL or HF diets for 0 -12 weeks. g and h, insulin (g) and leptin (h) were determined by ELISA in plasma samples of fasting Wistar rats fed on CTL (filled squares) or HF (filled circles) diets for 0 -12 weeks. i, glucose consumption rate during a hyperinsulinemic-euglycemic clamp was evaluated in Wistar rats fed CTL or HF diets for 12 weeks. j-l, blood glucose (j) and insulin (k and l) levels were determined during an intraperitoneal glucose tolerance test performed in Wistar rats fed on a control (filled squares, 12 weeks) or a high fat (filled triangles, 4 weeks; filled circles, 12 weeks) diet (j and k) or even submitted to sympathectomy (filled triangles, HFϩS) or treated icv with infliximab (filled inverted triangles; HF diet with insulin, HFϩI) (l). The inset in l depicts the area under the curve (AUC, ng/ml⅐min) for insulin. Insulin (k and l) was determined in plasma by ELISA. In all of the experiments, n ϭ 5. *, p Ͻ 0.05 versus CTL in respective time. In l, §, p Ͻ 0.05 versus HF. ously perfused at a flow rate of 1 ml/min. During the initial 20 min of perfusion, the buffer consisted of a Krebs-bicarbonate solution containing 2.8 mM glucose. Next, perfusion was shifted for a buffer containing 16.7 mM glucose, which was maintained for 40 min. Finally, a final phase perfusion with the 2.8 mM glucose buffer concluded the procedure. Samples of perfusate for quantification of insulin were collected at every second minute, starting at the tenth minute after the onset of perfusion. Insulin in the medium was measured by radioimmunoassay.
Immunoblot-For specific protein determination, groups of 800 -1000 freshly isolated islets or hypothalami from each experimental group were homogenized in a freshly prepared ice-cold buffer (1% Triton X-100, 100 mM Tris, pH 7.4, 100 mM sodium pyrophosphate, 100 mM sodium fluoride, 10 mM EDTA, 10 mM sodium vanadate, 2 mM phenylmethylsulfonyl fluoride, and 0.1 mg of aprotinin), and an immunoblot was performed, as described previously (21). Insoluble material was removed by centrifugation (20,000 ϫ g) for 30 min at 4°C, and 150 g and 30 g of the total protein extracts were separated by SDS-PAGE. After SDS-PAGE, the proteins were transferred from gel to nitrocellulose membrane. The membranes were blocked in 5% nonfat dried milk in PBST (139 mM NaCl, 2.7 mM KH 2 PO 4 , 9.9 mM Na 2 HPO 4 , and 0.1% Tween 20) for 2 h at room temperature and then incubated overnight at 4°C with specific antibodies. After incubation with the specific secondary antibody, immune complexes were detected using the enhanced Super-Signal West Pico Chemiluminescent Substrate (Pierce), as described by the manufacturer, and the visualization was performed by exposure of the membranes to x-ray films. The band intensities were quantified by optical densitometry of developed autoradiographs (Scion Image software; Scion Corporation, Frederick, MD).
Real Time PCR-The expressions of TNF␣, IL1␤, IL6, BAX, BIK, and Bcl2 mRNAs were measured in samples (hypothalamus or islets). Intron-skipping primers were obtained from Applied Biosystems. Glyceraldehyde-3-phosphate dehydrogenase primers were used as a control. Real time PCR analysis of gene expression was performed in an ABI Prism 7700 sequence detection system (Applied Biosystems). The optimal concentrations of cDNA and primers, as well as the maximum efficiency of amplification, were obtained through five-point, 2-fold dilution curve analysis for each gene. Each PCR typically contained 3.0 ng of reverse-transcribed RNA, 200 nM of each specific primer, TaqMan TM (Applied Biosystems), and RNase free water to a final volume of 20 l. Real time data were analyzed using the sequence detector system 1.7 (Applied Biosystems) (22).
Statistical Analysis-All of the numeric results are expressed as the means Ϯ S.E. of the indicated number of experiments. The results of blots are presented as direct comparisons of bands in autoradiographs and quantified by optical densitometry (Scion Image). Student's t tests of unpaired samples and analysis of variance for multiple comparisons were used as appropriate. The post-hoc test (Tukey) was employed when required. The level of significance was set at p Ͻ 0.05.

Hypothalamic Inflammation and Dysfunctional Insulin Secretion in Diet-induced
Obesity-Wistar rats fed on a HF diet become obese after 12 weeks (Fig. 1a); this is accompanied by increased caloric intake (Fig. 1b) and increased adiposity, which is already present 4 weeks after introduction of HF diet (Fig. 1c). During the induction of obesity, evidence of hypothalamic inflammation is detectable as increased expressions of TNF␣ (Fig. 1d), IL1␤ (Fig. 1e), and IL6 (Fig. 1f) as early as 8 weeks after the beginning of HF diet. Obesity is also accompanied by increased blood levels of insulin (Fig. 1g) and leptin (Fig. 1h) and by insulin resistance, as evaluated by the hyperinsulinemic-euglycemic clamp (Fig. 1i). All of these outcomes resulted in no change in glucose tolerance, as evaluated by a glucose tolerance test (Fig. 1j), but were accompanied by dysfunctional insulin secretion, detected as hyperinsulinemia, loss of first phase secretion, and a delayed return to base-line levels (Fig. 1k). Interestingly, the treatment of dietinduced obese rats with the anti-TNF␣ monoclonal anti- body, infliximab, icv, or pancreatic sympathectomy resulted in the reduction of hyperinsulinemia and partial restoration of the first phase insulin secretion (Fig. 1l).

Exogenous TNF␣ icv Injection Affects Pancreatic Islet Gene Expression and Reproduces Some of the Effects of Obesity on
Insulin Secretion-TNF␣ is one of the main mediators of hypothalamic inflammation in obesity (11,13). To evaluate the impact of hypothalamic TNF␣ action on islet functionality in the absence of obesity, lean rats were submitted to icv cannulation and treated with a single dose of TNF␣. Pancreatic islets were isolated after 3, 6, and 12 h, and RNA was prepared for evaluation of gene expression by a macroarray. As shown in Table 1, 43 of 1167 genes were positively modulated following icv TNF␣ treatment. Genes encoding proteins involved in insulin processing and secretion, cell signaling, neurotransmitter production, endoplasmic reticulum stress, ubiquitination, and apoptosis, among others, were affected by the treatment. No changes in systemic levels of TNF␣ were detected after icv injection of the cytokine (Fig. 2a). Changes in pancreatic islet gene expression were accompanied by increased fasting blood insulin levels (Fig.  2b), which resulted in no changes in glucose levels during an ipGTT (Fig. 2c). However, a delayed first phase insulin secretion during the ipGTT in icv TNF␣-treated rats resembled the findings in obese rats (Fig. 2d). This was accompanied by no change in insulin secretion, as measured by the static secretion method (Fig. 2e) and by a significant increase in insulin secretion, as determined by the dynamic secretion method (Fig. 2f). In addition, icv TNF␣ induced an increase in the expression of the pro-apoptotic gene BAX (Fig. 2g) and a reduction in the Bcl2/BAX ratio (Fig. 2h).
Stearic Acid Induces Hypothalamic Inflammation and Dysfunctional Insulin Secretion-icv treatment with stearic acid led to increased TNF␣ (Fig. 3a), IL1␤ (Fig. 3b), and IL6 (Fig. 3c) protein levels in the hypothalamus. This was accompanied by increased inflammatory signaling, as determined by the activation of JNK (Fig. 3d), the phosphorylation of IB (Fig. 3e), and the increased expression of p50NFB (Fig. 3f). No changes in systemic levels of TNF␣ were induced by icv stearic acid (Fig.  3g). Glucose levels during an ipGTT were not affected by icv stearic acid (Fig. 3h); however, treatment with the saturated fatty acid resulted in reduced glucose decay during an ipITT (Fig. 3i), which was accompanied by reduced fasting insulin levels in plasma (Fig. 3j) and by reduced insulin secretion, as determined by the static secretion method (Fig. 3k). The metabolic parameters of rats treated with stearic acid icv are depicted in Table 2.
icv Stearic Acid Leads to Pro-apoptotic Signaling and Increased Expressions of PGC1␣ and UCP2 in Pancreatic Islets-The icv treatment with stearic acid induced an increased expression of BAX (Fig. 4a), a reduced Bcl2/BAX ratio (Fig. 4b), and a reduced phosphorylation of Akt (Fig. 4c) in pancreatic islets. In addition, icv stearic acid increased pancreatic islet expressions of PGC1␣ (Fig. 4d) and UCP2 (Fig. 4e).

Inhibition of PGC1␣ Expression Reverts the Effects of icv Stearic Acid on Insulin Secretion-The inhibition of PGC1␣
expression was achieved by treating rats with an antisense oligonucleotide specifically targeting the PGC1␣ mRNA. This approach was previously tested (15), and in the present exper-imental setting it led to 45% reduction of PGC1␣ protein expression in pancreatic islets (Fig. 5a). In rats previously treated with stearic acid (icv), the anti-PGC1␣ antisense oligonucleotide completely blunted the increase in PGC1␣ expression (Fig. 5b). This was accompanied by increased Akt phos-  h-j and 45, 90, or 180 M in k), and the experiments were performed at the end of the experimental period. Blood glucose levels were determined during an intraperitoneal glucose tolerance test (h) and an intraperitoneal insulin tolerance test (i); the insets in h and i depict the area under glucose curves (AUC, h, ng/ml⅐min) and the constant of glucose decay (i, K itt ), respectively. Plasma insulin levels were determined by radioimmunoassay (j). Insulin secretion by isolated pancreatic islets was determined by the static insulin secretion method (k); islets were exposed to either 2.8, 11.1, or 22.2 mM glucose. In all of the experiments, n ϭ 5; *, p Ͻ 0.05 versus CTL. In h and i, CTL, filled squares; BSA, filled circles; SA, filled triangles. In islet studies (k), n refers to the number of islet groups obtained from a pool of islets isolated from three or four rats. phorylation (Fig. 5c) and correction of UCP2 expression to levels similar to those of lean controls (Fig. 5d). The inhibition of PGC1␣ produced a significant reduction in glucose area under the curve during an ipGTT (Fig. 5e) and an accelerated glucose decay during an ipITT (Fig. 5f), not only in the icv stearic acid-treated rats but also in the control rats. These effects were accompanied by increased insulin secretion, as determined by the static secretion method (Fig. 5g) and by increased fasting insulin levels in plasma (Fig. 5h).
Sympathectomy Reduces Pancreatic Islet PGC1␣ and UCP2 Expression and Reverts the Effects of icv Stearic Acid on Insulin Secretion-Sympathectomy completely blunted the ability of icv stearic acid to increase pancreatic islet PGC1␣ expres-sion (Fig. 6a). This was accompanied by increased Akt phosphorylation (Fig. 6b) and by reduced UCP2 expression (Fig.  6c). Pancreatic sympathectomy also resulted in the inhibition of stearic acid-induced BAX expression and reversal of the pro-apoptotic ratio Bcl2/BAX (Fig. 6d). This was accompanied by partial inhibition of stearic acid-induced BIK (Fig.  6e) and complete inhibition of the increased expression of the active (cleaved) form of caspase-3 (Fig. 6f) in pancreatic islets. In addition, sympathectomy reduced the glucose area under the curve during an ipGTT (Fig. 6g), and increased the glucose decay during an ipITT (Fig. 6h) in rats treated icv with stearic acid. Finally, sympathectomy increased insulin secretion by isolated pancreatic islets, as determined by the static secretion method (Fig. 6i), and also increased fasting insulin levels in the plasma of rats treated icv with stearic acid (Fig. 6j).

DISCUSSION
During the installation and progression of obesity, the blood levels of insulin rise in direct proportion to body mass (23). Although the pancreatic ␤-cells can cope with the peripheral needs for insulin, glucose homeostasis will prevail (23). However, depending on genetic and environmental factors, insulin  production and secretion may decline, and the installation of DM2 becomes inevitable (24). At diagnosis, DM2 patients present a significant decrease in ␤-cell function, which can be further compromised during the progression of the disease (24). Several mechanisms have been shown to play a role in this process, such as glucotoxicity, lipotoxicity, the damaging effect of increased leptin levels, the deposition of amyloid, and the activation of inflammation, all contributing to accelerated apoptosis (23,25), which results in the reduction of up to 60% of pancreatic islet mass in the pancreata of DM2 patients (25). The first evidence of defective ␤-cell function in obese subjects is hyperinsulinemia, accompanied by the loss of the first phase insulin secretion (26,27). In experimental dietinduced obesity, we reproduced these features, showing that obese rats, albeit not diabetic, are clearly insulin-resistant and hyperinsulinemic, presenting a loss of the first phase insulin secretion. Interestingly, changes in insulin levels parallel the installation of hypothalamic inflammation, which is an important mechanism leading to leptin and insulin resis-tance in the hypothalamus and subsequently to the loss of the perfect coupling between energy consumption and expenditure (11,13,28). The connection between diet-induced hypothalamic inflammation and defective insulin secretion was further evidenced by the inhibition of hypothalamic TNF␣, which produced no effect on body mass but partially corrected insulin secretion. In addition, sympathectomy was capable of partially restoring pancreatic ␤-cell function in obesity. In obesity, increased insulin under the curve in the ipGTT is a result of a number of distinct mechanisms that have been thoroughly studied over the years. Among these mechanisms we can quote the increased peripheral demand for insulin and incipient ␤-cell lipotoxicity leading to defective insulin secretion (29), among others. Therefore, we were not expecting that the inhibition of inflammation in the hypothalamus would completely correct the defective homeostasis of insulin, as measured by the ipGTT. In fact, we were quite surprised by the findings that this approach could lead to ϳ50% reduction of the difference in the area under the curve detected between control and HF , and UCP2 (d) were determined by immunoblot (IB) of isolated pancreatic islet total protein extracts separated by SDS-PAGE and transferred to nitrocellulose membranes. Blood glucose levels were determined during an intraperitoneal glucose tolerance test (e) and during an intraperitoneal insulin tolerance test (f); the insets in e and f depict the area under glucose curves (AUC, e, ng/ml⅐min) and the constant of glucose decay (f, K itt ), respectively. Insulin secretion by isolated pancreatic islets was determined by the static insulin secretion method (g); the islets were exposed to either 2.8, 11.1, or 22.2 mM glucose. Fasting blood insulin levels were determined by radioimmunoassay (h). In all of the experiments, n ϭ 5; *, p Ͻ 0.05 versus CTL; in e, §, p Ͻ 0.05 versus CTLϩAS. In islet studies (g), n refers to the number of islet groups obtained from a pool of islets isolated from 3-4 rats.
animals. Thus, it seems that hypothalamic inflammation plays a rather important role in this phenomenon.
Because obesity can impair pancreatic islet function by a number of peripheral mechanisms that could act as confounding factors for our original objective, we induced hypothalamic inflammation in lean rats by two mechanisms that have been characterized previously: (a) icv TNF␣ and (b) icv saturated fatty acids (13,28). TNF␣-injected icv induces hypothalamic resistance to leptin and insulin (30) and modulates neurotransmitter expression toward an obesity-like phenotype (28,31). icv TNF␣ produced no change in the peripheral levels of this cytokine; nevertheless, insulin in blood increased significantly, whereas a loss of the first phase secretion was observed during an ipGTT. In addition, isolated islets from icv TNF␣-treated rats behaved similarly in the secretion assays to islets obtained from obese rats (15). Thus, we conclude that the induction of hypothalamic inflammation with TNF␣ induces functional changes in pancreatic islets that are similar to the ones found in obesity. In fact, not only the functional changes induced by TNF␣ resemble the effects of obesity but also changes in gene expression occurred in the same direction. The evaluation of 1167 mRNA specificities revealed that icv TNF␣ modulated 43 genes positively, representing 3.6% of the total analyzed. Genes such as GAD65/67, IL8 receptor, ␤2-microglobulin, prostaglandin receptor, TK1, PKC inhibitor, phosphotyrosine phosphatase-phosphohydrolase, and phosphatidylinositol 4 kinase are involved in cell signaling and inflammation and have been reported to be, at least in part, engaged in defective pancreatic islet function (32)(33)(34)(35)(36)(37)(38)(39). Furthermore, genes involved in energy metabolism and mitochondrial function, such as mitochondrial CK, GLUT, and cytochrome P450, play important roles in ␤-cell activity and are potentially involved in dysfunctional insulin secretion (40 -42). However, genes involved in apoptosis (Bax, Bclx, and Bcl2) and genes involved in endoplasmic reticulum function (HSP27, eIF2␣, and ribosomal proteins) are of major interest, because of the roles played by endoplasmic reticulum stress and apoptosis in pancreatic islet pathology in DM2 (43,44). It is important to emphasize that all of these changes in gene expression occurred in lean rats treated icv with TNF␣, suggesting that hypothalamic inflammation can, per se, modulate pancreatic islet function.
The findings of the first part of this study raised the question as to how signals generated in the hypothalamus may reach the pancreatic islets to modulate their gene expression and function. Neural mechanisms controlling insulin secretion were thoroughly studied during the 1960s and 1970s (9). The pan-creatic islets are richly innervated by parasympathetic postganglionic fibers originating from the vagus nerve and providing stimulatory signals through m3-muscarinic receptors, and by sympathetic postganglionic fibers originating at the celiac and paravertebral ganglia, providing, predominantly, inhibitory signals through the ␣2-adrenergic receptors (9). In addition to its classical signaling through reduced cAMP formation, the sympathetic signals can also modulate pancreatic ␤-cell function through the induction of PGC1␣ expression, which leads to increased UCP2 gene transcription and uncoupling of the mitochondrial respiration (15,45). The uncoupling of the mitochondrial respiration provides a mechanistic basis for the sympathetic signal-induced ␤-cell hyperpolarization (45).
In lean rats treated icv with stearic acid, hypothalamic inflammation is induced through the activation of TLR4 signaling and endoplasmic reticulum stress (13). This leads to reduced blood insulin levels and reduced glucose-stimulated insulin secretion, which is accompanied by an increased expression of markers of apoptosis and also an increased expression of PGC1␣ and UCP2 in pancreatic islets. In coldexposed rats, the reduction of insulin secretion depends, at least in part, on the sympathetic activation of PGC1␣ and UCP2 expression (19). In that model, either sympathectomy or the inhibition of PGC1␣ expression resulted in the restoration of insulin secretion (19). Given that the hypothalamus integrates most of the actions aimed at adapting metabolism during cold exposure, we decided to evaluate whether a similar mechanism could occur during hypothalamic inflammation, with regard to the function of the endocrine pancreas. As suspected, both the inhibition of PGC1␣ and sympathectomy corrected UCP2 expression levels in pancreatic islets, resulting in the partial restoration of insulin secretion. In addition, Akt, a protein involved not only in the control of insulin secretion but also in the protection against apoptosis was activated to levels similar to those of lean controls. All of these outcomes were accompanied by restoration of the expression of apoptotic proteins to base-line or nearly base-line levels. Thus, the disruption of the sympathetic signal can revert, at least in part, the effect of fatty acid-induced hypothalamic inflammation on pancreatic islet function.
In conclusion, this study provides the first evidence for a direct connection of the inflamed and dysfunctional hypothalamus with the pancreatic islet, leading to a defective insulin secretion and the modulation of gene expression. Given that hypothalamic dysfunction and central resistance to the anorexigenic hormones leptin and insulin are early events in the development of obesity (46) and that correction of hypothalamic inflammation reverts peripheral insulin resistance (47), the FIGURE 6. Effect of sympathectomy on hypothalamic inflammation-induced pancreatic islet dysfunction. In a-j, icv cannulated Wistar rats were submitted to either sham surgery (SH) or sympathectomy (SY) and, after recovery, were treated twice a day for 5 days with 2 l of vehicle (CTL), BSA, or SA (90 M). a-f, the expressions of PGC1␣ (a) phospho-Akt (b), UCP2 (c), Bcl2 (d), BAX (e), and cleaved caspase-3 (c-Casp3) (f) were evaluated by immunoblot (IB) of isolated pancreatic islet total protein extracts separated by SDS-PAGE and transferred to nitrocellulose membranes. The expression of BIK in pancreatic islets was evaluated by real time PCR (e). g and h, blood glucose levels were determined during an intraperitoneal glucose tolerance test (g) and during an intraperitoneal insulin tolerance test (h); the insets in g and h depict area under glucose curves (AUC, g, ng/ml⅐min) and the constant of glucose decay (h, K itt ), respectively. i, insulin secretion by isolated pancreatic islets was determined by the static insulin secretion method (i). j, islets were exposed to either 2.8, 11.1, or 22.2 mM glucose. Fasting blood insulin levels were determined by radioimmunoassay. In all of the experiments, n ϭ 5; *, p Ͻ 0.05 versus CTL; in d-f, #, p Ͻ 0.05 versus SA in nonsympathectomized rats. In g, h, and j, §, p Ͻ 0.05 versus SA. In i, §, p Ͻ 0.05 versus respective conditions in SH. In islet studies (i), n refers to the number of islet groups obtained from a pool of islets isolated from three or four rats. In g and h, CTL, filled squares; SA, filled triangles; SAϩSH, filled circles; SAϩSY, filled inverted triangles.
present study provides the basis for a unifying hypothesis that the hypothalamus coordinates body mass, peripheral insulin resistance, and dysfunctional insulin secretion.