Brain GLP-1 Signaling Regulates Femoral Artery Blood Flow and Insulin Sensitivity Through Hypothalamic PKC-δ

Experiments were carried out under protocols approved by the institutional animal care and use committee. Male C57BL/6J (Charles River Laboratories, Paris, France) and Glp1r^−/− C57BL/6J mice (age 16 weeks) from our colony were studied. Groups of 4-week-old mice were fed a normal chow (NC) diet (SAFE, Epinay sur Orge, France; content: 12% fat, 28% protein, and 60% carbohydrate, low nitrate) or an HFD (SAFE; 72% fat [corn oil and lard], 28% protein, and <1% carbohydrate) for 12 weeks, which has been shown to induce diabetes and insulin resistance without overt body weight gain (7,14,24,25). Throughout the study period, mice were housed at 21–22°C with a normal daily cycle and ad libitum access to food and water.

Femoral artery blood flow was continuously monitored during an intracerebroventricular infusion of pharmacological compounds as described previously (5,8,26). Briefly, a catheter (Charles River Laboratories) was inserted into the right lateral ventricle and secured on the top of the skull under anesthesia with isoflurane-oxygen. At 10 days after the cranial surgery, an intravenous catheter was introduced into the left femoral vein, sealed under the skin, and externalized at the back of the neck. The mice were allowed to recover for 3 days before an ultrasonic flow probe (Transonic Systems, Emka Technologies, Inc., Paris, France) was inserted around the right femoral artery. The probe wire was sealed under the skin at the back of the neck where it was secured using surgical thread. After surgery, the mice were returned to their cage for at least 4 days until their body weight recovered. At the end of the recovery period, mice that did not reach the presurgery weights were not studied further (i.e., 15% of the animals).

On the experimental day, the flow probe was connected to a transonic model T403 flowmeter (Transonic Systems, Emka Technologies, Inc.) to continuously record the blood flow (mL/min) of the artery and the heart rate bpm). The basal femoral artery blood flow was recorded for 30 min in overnight fasted, freely moving mice. Intracerebroventricular infusions were performed at a rate of 12 µL/h preceded by a 5-µL bolus with artificial cerebrospinal fluid (aCSF; pH 7.35, Na⁺ 144 mmol/L, Cl⁻ 146 mmol/L, K⁺ 3 mmol/L, Mg²⁺ 1 mmol/L, Ca²⁺ 1.5 mmol/L, PO⁴⁻ 1.2 mmol/L, vehicle/control group) or the test compounds (as described below) for 3 h. To validate that the probe was correctly recording the blood flow, at the end of the insulin infusion, mice were given a flash injection of a rapid nitric oxide donor (sodium nitroprusside, 10 mg/kg i.v., 25–40 µL, purchased from Sigma, Saint-Quentin Fallavier, France), which induced a 100% increase in blood flow and heart rate.

Mice were clamped in hyperinsulinemic conditions to activate whole-body glucose utilization and induce vasodilation. Subgroups of mice were clamped in euglycemic (5.5 mmol/L) or hyperglycemic (20 mmol/L) conditions, as previously described (5,8). Briefly, insulin was infused through the intrafemoral catheter at a rate of 18 mU/kg/min for 3 h. Glycemia was clamped by adjusting an intrafemoral glucose infusion (30%). A control group was infused with NaCl 0.9% (saline) for 3 h at a rate that matched the rate of all infusions to ensure a similar liquid flow (8).

The intracerebroventricular infusions were started 30 min before the beginning of the clamp procedure and continued throughout the whole infusion period. Briefly, a 5 µL bolus was given followed by a continuous infusion with the aCSF or with the GLP-1R agonist exendin-4 (Ex4) or antagonist exendin-9 (Ex9) at a rate of 0.5 pmol/kg/min for 3 h, as described (5,8,13).

To evaluate the role of brain PKCs in the regulation of arterial blood flow and glucose turnover rates in response to brain GLP-1R activation, we examined the effects of a nonspecific PKC activator, the phorbol ester PMA (phorbol-12,13-myristic acid; Sigma), or inhibitors (calphostin C and rottlerin) infused for 3 h at a very low rate into the cerebroventricle (intracerebroventricularly). The PMA was diluted in aCSF and then infused at a rate of 0.1 nmol/µL, as previously described (27). In a separate set of mice, calphostin C (Sigma), a pan-specific inhibitor that specifically inhibits the binding of DAG to the regulatory domain of PKCs (28), was diluted in aCSF and then infused at a rate of 3 pmol/μL, as previously described (29). Rottlerin (Sigma), a specific PKC-δ inhibitor, was diluted in aCSF and then infused at a rate of 0.27 nmol/µL. Intracerebroventricular infusions were started 30 min before the beginning of the intravenous insulin/glucose infusions and continued for 3 h. All solutions used contained no more than 2% alcohol to dissolve the pharmacological agents.

At the end of the infusions, cytosolic and membrane fractions were separated according to the procedure described below to evaluate the translocation of PKC. Briefly, when the clamp procedure ended, each hypothalamus was removed from the skull immediately after cervical dislocation and cooled down in a frozen brain frame (World Precision Instruments, Stevenage, U.K.) to stop all endogenous enzyme reactions and then frozen at −80°C in liquid nitrogen. A 3-mm coronal section corresponding to the hypothalamus was sliced out, and a triangle with a side length of 2 mm corresponding to the frozen hypothalamus was separated and kept frozen until studied, as previously described (26). Slices were then lysed in ice-cold 0.25 mol/L sucrose buffer (containing 25 mmol/L Tris-HCl, pH 7.4, 2 mmol/L EDTA, 1 mmol/L EGTA, 10 mmol/L β-mercaptoethanol, 1 μg/mL leupeptin, 1 μg/µL aprotinin, and 1 mmol/L phenylmethylsulfonyl fluoride) and centrifuged at 14,000g for 5 min at 4°C. The supernatant was then collected and centrifuged at 100,000g for 45 min at 4°C to separate the cytosolic fraction (supernatant) from the particulate fraction (pellet). The latter was resuspended in the same buffer, to which 0.1% Triton X-100 had been added and sonicated 12 times in ice before being recentrifuged at 100,000g for 30 min at 4°C. The supernatant was then removed. This last fraction is called the membrane fraction. In some preliminary studies, we validated the presence of the Na⁺-K⁺-ATPase pump, a component of the cytoplasmic membrane, in each membrane fraction. This membrane marker was totally absent from the cytosolic fraction.

PKC activity was assayed in each fraction (cytosolic and membrane) using the PepTag nonradioactive PKC assay system (Promega Corporation, Madison, WI) as described (30). Briefly, 2 μg of each sample and purified PKC enzyme (internal control) were incubated at 30°C for 30 min with a fluorescent peptide substrate that is highly specific for PKCs and then loaded onto a 0.8% agarose gel in 50 mmol/L Tris-HCl, pH 8.0. Phosphorylation by PKCs alters the net charge of the peptide from 1 to −1. This change in the net charge of the substrate allows phosphorylated and nonphosphorylated substrates to be separated on an agarose gel. This assay can detect <10 ng of kinase. The intact agarose gel was then analyzed in a scanning densitometer using a wavelength of 570 nm and data interpreted qualitatively and quantitatively using ImageQuant 5.2 software. In each fraction, activity was quantified and normalized by the positive internal control to allow multiple comparisons between gels. Next, the PKC translocation was calculated for each hypothalamus by the corresponding membrane-to-cytosol ratio. The results of scanning densitometry expressed in arbitrary units (AUs) correspond to the upper band intensity divided by the sum of the upper and lower bands. The upper band corresponds to the phosphorylated and the lower band to the nonphosphorylated substrate.

Western blots for specific PKCs expressed in the hypothalamus were performed on the same fractions used to assess the total PKC activity as described above. We used specific antibodies targeted against α (15), ε (15), βII (23), and δ (19) PKC isoforms. These isoforms are expressed in the brain (31–34) and activated by an increase in DAG in the cytosol or by a phorbol ester such as PMA (35). They are inhibited by calphostin C, a kinase inhibitor specific for PKC, which targets the DAG or phorbol ester binding site (36). After protein determination by the protein assay (Bio-Rad, Marnes-la-coquette, France), using BSA as a standard, samples were separated by 10% SDS-PAGE. For each hypothalamic sample, 20 μg of cytosolic or membrane fraction was loaded onto a 10% acrylamide SDS-PAGE. After separation, the proteins were transferred to a polyvinylidene difluoride membrane. Membranes were then blocked for 90 min with 3% BSA and incubated for 90 min at room temperature with a primary polyclonal antibody against PKC-α, -ε, -βII, or -δ (Santa Cruz Biotechnology, Inc., Santa Cruz, CA; references 937, 208, 214, and 210, respectively, were used with the following dilutions: anti-δ 1/1,000, anti-α 1/2,000, anti-βII 1/5,000, and anti-ε 1/8,000) (19). We used PKC-βII or PKC-ε as an internal control because they did not translocate under different experimental conditions. After washing three times, membranes were incubated with horseradish peroxidase–conjugated anti-rabbit IgG-antibody (Amersham Biosciences Europe, Orsay, France; dilution 1/8,000) for at least 60 min. Immunoreactivity was detected using an enhanced chemiluminescence detection kit (ECL system; Amersham) and exposure to X-ray film. Bands were quantified using ImageQuant software.

At the end of the infusions, blood was collected from the retro-orbital sinus into a tube, mixed with 1 µg/µL aprotinin per 0.1 mmol/L EDTA, and centrifuged at 8,000 rpm for 5 min at 4°C. Plasma was stored at −80°C until assay. Insulin level was measured using an ELISA kit (Mercodia, Uppsala, Sweden).

Data are expressed as means ± SE. Data were analyzed using GraphPad Prism version 5.00 for Windows (GraphPad Software, San Diego, CA) for statistical significance by applying, respectively, a Student t test and a one-way or a two-way ANOVA test for repeated measurements followed by a post hoc test (Bonferroni multiple comparison test) when appropriate. The level of significance was defined as P < 0.05.

To determine whether brain GLP-1 signaling regulates hypothalamic PKC activity and vascular/metabolic functions, we studied mice under hyperglycemic and hyperinsulinemic conditions during a 3-h clamp procedure. Under these conditions, the Ex4 infusion decreased both femoral artery blood flow and whole-body glucose infusion rates (GIRs) compared with the control group receiving the vehicle into the lateral ventricle of the brain (Fig. 1A and C). Brain Ex4 infusion enhanced PKC translocation to the plasma membrane and increased the corresponding membrane-to-cytosol ratio (Fig. 1D–F). The PKC translocation was higher in response to brain Ex4 infusion than in vehicle-treated mice. Basal femoral artery blood flow and GIR were relatively increased in Glp1r^−/− mice and in wild-type (WT) mice infused with Ex9, showing that control of blood flow and GIR were dependent on GLP-1R signaling (Fig. 1B and C). In a similar manner, the Ex4-dependent stimulation of membrane PKC translocation was abolished in mice with pharmacological blockade or GLP-1R genetic disruption (Fig. 1D–F).

To analyze the effect of PKC inactivation (Fig. 1D–F), a nonspecific inhibitor, calphostin C, was infused into the brain under similar experimental conditions. Coinfusion of calphostin C with Ex4 reversed the reductions in femoral artery blood flow and GIR produced by Ex4 alone (Fig. 1A and C). The blood flow and GIR in mice infused with calphostin C were similar to that observed in the control group (Fig. 1A and C).

To activate PKC activity, we infused PMA, a nonspecific activator of PKC, into the brain (Fig. 2A–C) and clamped the mice under hyperglycemic conditions. PMA brain infusion decreased femoral artery blood flow and GIR compared with the vehicle-infused control group (Fig. 2D and E).

In the basal fasted state, the Ex4 brain infusion increased plasma membrane activity of PKC (Fig. 3A–C), which corresponded to increased levels of PKC-α (Fig. 3D and E) and -δ (Fig. 3D and G) but not -ε and -βII (Fig. 3D and F). During experimental hyperinsulinemia and hyperglycemia, the Ex4 brain infusion activated the translocation of PKC-δ (Fig. 4A and B) but not -α, -ε, or -βII to the plasma membrane. Under hyperinsulinemia and hyperglycemia, levels of PKC-δ were higher in response to Ex4 when compared with vehicle-treated mice. However, this set of data does not demonstrate that control of PKC-δ is strictly glucose dependent.

This effect was prevented by the concomitant infusion of Ex9 into the brain of WT mice, and PKC localization in Ex9-treated mice was similar to the pattern detected in Glp1r^−/− mice (Fig. 4A and B). We did not observe translocation to the membrane fraction of other PKC isoforms (ε, α, and βII) (Fig. 4A). The PKC-δ activation by Ex4 was totally blunted in Glp1r^−/−mice (Supplementary Fig. 1). The performance of both the euglycemic-hyperinsulinemic and the hyperglycemic-hyperinsulinemic clamps are provided in Supplementary Table 1. We illustrate the steady-state glycemia obtained during the clamps during the last hours of the infusion, which shows the quality of the experiments.

Rottlerin, a specific PKC-δ inhibitor, was concomitantly infused into the brain with Ex4 under hyperglycemic-hyperinsulinemic conditions. Rottlerin totally prevented the reduction of femoral artery blood flow and GIR produced by Ex4 (Fig. 5A and B) and inhibited Ex4-induced PKC-δ translocation (Fig. 5C and D). Under these experimental conditions, levels of PKC-βII and the ratio of PKC-ε to PKC-βII and PKC-α to PKC-βII remained unchanged (not shown).

Diabetes is frequently associated with insulin resistance and vasoconstriction; hence, we analyzed whether an increase in brain GLP-1 signaling was associated with this phenotype under fasted and insulin-stimulated conditions. In the fasted state, plasma membrane PKC activity was increased in the hypothalamus of fasted HFD-fed diabetic mice and reduced by Ex9 (Fig. 6A–C).

To assess the effect of chronic hyperglycemia because of the HFD and not simply the effect of acute hyperglycemia as a result of the infusion, as in the first sets of experiments, we clamped both NC and diabetic (HFD) mice under euglycemic conditions. During the euglycemic and hyperinsulinemic clamp procedures, diabetic mice exhibited reduced insulin-induced vascular blood flow and GIR when compared with normal mice. The Ex9 brain infusion improved the GIR and normalized blood flow (Fig. 6D and E). In a similar manner, the inhibition of PKC activity by calphostin C (Fig. 6F and G) had the same effect as Ex9 on blood flow and GIR (Fig. 6E and H). The translocation of PKC-α and -δ (but not of -βII and -ε) was increased in the hypothalamus of diabetic mice (Fig. 7A–F). It is important to note that we normalized data relative to PKC-ε levels but not PKC-βII since PKC-βII is modulated in experimental diabetes (23).

It is interesting that the Ex9 brain infusion could prevent membrane translocation of PKC-δ (Fig. 7E) but not of PKC-α (Fig. 7F). The other isoforms (βII and ε) remained unaffected (not shown). To ascertain the role of PKC-δ in the control of femoral artery blood flow and glucose metabolism, rottlerin was infused into the brain of diabetic mice to specifically inhibit PKC-δ activity (Fig. 7G and H). Rottlerin increased both femoral artery blood flow (Fig. 7J) and GIR (Fig. 7K), rendering values for these two parameters similar to that observed in control WT mice. PKC-δ translocation (and not PKC-α) was concomitantly reduced by the inhibitor (Fig. 7G–I), whereas PKC-ε remained unchanged (Fig. 7G).

Also of interest is that in the fed state and in Glp1r^−/− mice, PKC translocation to the membrane fraction remained lower than in WT mice in NC-fed (Fig. 8A and B) and in HFD-fed groups (Fig. 8C and D). Diabetic Glp1r^−/− mice were then clamped under hyperinsulinemic conditions. Femoral artery blood flow was similar in NC and HFD Glp1r^−/− mice. They were therefore unaffected by the dietary treatment (Fig. 8E), whereas the GIR remained slightly increased, although more than in the diabetic WT mice (Fig. 8F). In Glp1r^−/− mice, PKC-δ translocation to the plasma membrane remained similar between NC and HFD mice; therefore, the dietary treatment did not increase PKC-δ translocation to the plasma membrane, whereas levels of PKC-α were still increased (Fig. 8G–I) but not levels of PKC-βII (not shown). These findings demonstrate a relationship between PKC-δ, blood flow, and insulin resistance.

In this study, we demonstrated that hypothalamic PKC-δ constitutes a GLP-1–sensitive molecular mechanism required for the coordinated control of femoral artery blood flow and whole-body insulin sensitivity. This mechanism is strictly brain GLP-1– and glucose-dependent and overtly activated in chronic hyperglycemic conditions, which contribute to the development of insulin resistance and vasoconstriction during diabetes.

During the last decade, we and others have shown that the regulation of vascular blood flow and muscle glucose utilization is tightly coordinated. One of the key corresponding mechanisms was the activation by insulin of endothelial nitric oxide synthase, an identified molecular target that linked both the metabolic and vascular phenotypes in the WT mouse (14,37–39). In the quest for physiological endocrine regulators of peripheral glucose homeostasis, we identified a role for brain GLP-1 signaling (4,5,7,8). Hence, it was important to determine the molecular mechanisms activated by GLP-1 in the brain that coordinated the regulation of both vascular and metabolic functions.

Here, we have shown that brain GLP-1 signaling acutely triggers the translocation of cytoplasmic PKC to the plasma membrane of cells in the hypothalamus. The activation of PKC by GLP-1 has been previously reported in β-cells (15), which by analogy supports the hypothesis that PKCs could be part of molecular mechanisms mediating brain GLP-1 signaling. PKCs are serine/threonine kinases that function as second messengers responsible for the phosphorylation of numerous substrates. In the brain, PKCs regulate the synthesis and release of neurotransmitters and electrical activation of neuronal cells (34). In previous studies, the molecular signaling mechanisms associated with GLP-1–induced insulin secretion were via the α and ε PKC isoforms (15). To identify the precise isoforms in the hypothalamus responsible for the regulation of vascular and metabolic homeostasis, we focused our attention on PKC-ε/α/δ and -βII, which are all expressed in the brain (31–34). These isoforms are sensitive to glucose and can be activated by glucoregulatory hormones (15,18,19,23). We showed that hypothalamic PKC-δ, but neither -α, -ε, nor -βII, was activated when GLP-1 was directly infused into the brain. Therefore, our present data strongly support the hypothesis that brain PKC activation could mediate neuronal activity (40), leading to the regulation of vascular and metabolic functions.

To test this hypothesis, we first blocked GLP-1 signaling using an Ex9 infusion into the brain. GLP-1R blockade reduced the plasma membrane translocation of PKC activity by Ex4 under conditions of hyperinsulinemic hyperglycemia and increased insulin-induced vascular blood flow and whole-body glucose utilization. Similar data implicating a role for the hypothalamic GLP-1R system were obtained in studies with Glp1r^−/− mice. In a second set of experiments, the nonspecific PKC inhibitor calphostin prevented the effect of GLP-1, clearly indicating a role for hypothalamic PKCs in brain GLP-1 activity. Furthermore, in a third set of experiments, we activated brain PKCs using the DAG activator, PMA, which reproduced the findings observed with Ex4. Taken together, these data demonstrate that brain GLP-1 signaling regulates PKC activity in the hypothalamus to control vascular and metabolic homeostasis. This mechanism most likely requires PKC-δ since a specific inhibitor, rottlerin, abrogated these effects. It is noteworthy that levels of PKC-α, -βII, and -ε were unchanged by the infusion of rottlerin, further underscoring the specific activation of PKC-δ by GLP-1.

Insulin resistance and vasoconstriction are major features of diabetes, and several lines of evidence link PKC overactivity to chronic hyperglycemia (41). This overactivity has also been suggested to encompass nitric oxide synthase activity in brain cells (42–44), perhaps linked to impairment of the control of femoral artery blood flow and muscle glucose utilization in response to central insulin (14,39,45). However, whether brain GLP-1–induced PKC-δ plays a role in the pathophysiological consequences stemming from chronic hyperglycemia is unknown.

Our data show that total PKC activity was increased in diabetic WT mice both under basal and during hyperinsulinemic euglycemic clamp conditions. This was associated with an increased translocation of PKC-δ and -α to the plasma membrane. It is interesting that only PKC-δ translocation could be inhibited by the GLP-1R antagonist Ex9. In addition, brain infusion of Ex9 and calphostin C normalized the impaired insulin sensitivity and vascular function observed during experimental diabetes. These data demonstrate that excessive hypothalamic PKC activity is deleterious for both metabolic and vascular functions in response to activation of the brain GLP-1R. Experimental studies on diabetic rodents showed that brain GLP-1 signaling is stimulated during diabetes and may perturb the overall control of glycemia. This last observation is supported by previous data from our group and others demonstrating that the expression of the proglucagon gene encoding GLP-1 was increased in the brainstem of diabetic HFD-fed mice (7) and obese Zucker rats (46), which contains cell bodies of GLP-1 neurons projecting to the hypothalamus (9–13). The functional disruption of the GLP-1R in Glp1r^−/− mice prevents PKC activation to the membrane and PKC-δ translocation.

Our data further show that in the fed state, the gut-to-brain axis is activated through a GLP-1–dependent mechanism, as previously described (4). We also previously demonstrated that brain GLP-1 signaling allows hepatic glucose repletion upon fasting by preventing excessive muscle glucose utilization (5) that was associated with reduced muscle blood flow (8). The hypothalamic arcuate nucleus is likely to comprise the central site responsible for this mechanism because direct administration of GLP-1 into this nucleus decreases peripheral glucose uptake and production (47). We attributed this metabolic response to a feedback loop that might arise pursuant to excessive insulin-stimulated glucose utilization, operating to prevent hypoglycemia and favor hepatic glucose storage. We further showed that during diabetes, this mechanism is exacerbated and associated with increased brain GLP-1 signaling (7). We now demonstrate that signals emanating from the hypothalamic GLP-1R require PKC-δ activation, a mechanism continuously stimulated during HFD-induced insulin resistance and vasoconstriction. However, we cannot rule out that the GLP-1R could also mediate part of its effect through a ligand-independent activity since in Ex9-treated mice, blood flow was increased by the treatment. Previous data support this interpretation because in GLP-1R–expressing cell lines, the intracellular cAMP concentration was reduced by Ex9 even in the absence of GLP-1 (48). Hence, our data suggest that hypothalamic PKC-δ is activated in response to brain GLP-1R stimulation, perhaps contributing to the central control of insulin resistance and vasoconstriction. Therefore, current therapeutic treatments of type 2 diabetes using GLP-1R agonists delivered systematically may also target brain GLP-1R signaling with implications for regulation of peripheral blood flow and insulin sensitivity (49,50).

C.C. has received grants from the European Association for the Study of Diabetes and the Amylin Company and from the Société Francophone d’Etude du Diabète and is a member of the European Club for the Study of GLP-1 (EuCSGLP-1; http://www.glp1.eu). R.B. has received subsidies from INSERM and is a member of EuCSGLP-1. D.J.D. was supported in part by Heart and Stroke Foundation of Ontario Grant NA6997 and the Canada Research Chairs program. No other potential conflicts of interest relevant to this article were reported.

C.C. performed experiments and wrote the manuscript. C.V. performed experiments. G.C. and D.J.D. contributed to discussion and reviewed the manuscript. R.B. wrote the manuscript.