Increased GABAergic Output in the Ventromedial Hypothalamus Contributes to Impaired Hypoglycemic Counterregulation in Diabetic Rats

All rats started with a body weight of ~300 g and were individually housed in the Yale Animal Resources Center in temperature-controlled (22–23°C) and humidity-controlled rooms. The animals were fed a standard rodent diet (Agway Prolab 3000, Syracuse, NY) and water ad libitum and were acclimatized to handling and a 12-h light cycle (lights on between 0700 h and 1900 h) before experimental manipulation. Principles of laboratory animal care were followed, and experimental protocols were approved by the institutional animal care and use committee at Yale University.

Diabetic BB/Wor rats (n = 6) were obtained from the University of Massachusetts (Worcester, MA). Blood glucose levels were monitored twice daily using a glucometer (Bayer Contour), and a single subcutaneous injection of protamine zinc insulin was administered to maintain plasma glucose concentrations between ~350 and 450 mg/dL. Once blood glucose levels were stable within this range, the BB rats were maintained at this target for a period of 2 weeks before being studied. The insulin dose was reduced on the day prior to the study to prevent unexpected hypoglycemia during the fasting period.

Diabetes was induced in male SD rats (~300 g; Charles River) using a single intraperitoneal injection of streptozotocin (STZ; 65 mg/kg in citrate buffer). Nondiabetic control animals received a single intraperitoneal injection of an equivalent volume of citrate buffer. A 10% glucose solution was provided in the drinking water for the first 24 h to prevent hypoglycemia. STZ-induced diabetic animals were not insulin treated, and the condition was maintained for a total of 2 weeks before the animals were studied. Blood glucose was monitored twice daily with a glucometer to ensure adequate hyperglycemia (>250 mg/dL) in our diabetic animals. Nondiabetic SD rats served as controls.

Approximately 8–10 days prior to the experiment, the animals underwent aseptic surgery to have vascular catheters implanted into the left carotid artery for blood sampling and right jugular vein for infusion. These catheters were tunneled subcutaneously and exteriorized at the back of the neck. Subsequently, the animals were placed into a stereotaxic frame (David Kopf Instruments, Tujunga, CA) and bilateral stainless steel guide cannulas (Eicom, Kyoto, Japan) for microdialysis and/or microinjection (from bregma: VMH: 2.6 mm posterior, 3.8 mm lateral, and 8.9 mm ventral at an angle of 16°; paraventricular nucleus [PVN]: 1.5 mm posterior, 3.8 mm lateral, and 7.9 mm ventral at an angle of 22°), which were bilaterally inserted into the brain and secured in place with screws and dental acrylic. These coordinates position the 1-mm microdialysis probes into the ventrolateral portion of the VMH. The animals were then allowed to recover for ~1 week in their home cages before being studied.

Animals were fasted overnight in the microdialysis cages to allow sufficient time for acclimation. The next day, the animals were connected to infusion pumps and bilateral microdialysis/microinjection probes were inserted. Artificial extracellular fluid (aECF) was perfused through the microdialysis probes at a constant rate of 1.5 μL/min for 2.5–3 h to allow GABA levels to stabilize prior to the start of the microdialysate and baseline blood sample collection. Microdialysate samples were then collected at 10-min intervals for the duration of the study.

Following a baseline collection period of 35 min, a constant insulin (50 mU/kg/min) and variable 50% dextrose infusion were used to induce baseline euglycemia (115 ± 10 mg/dL) for at least 30 min. Once plasma glucose levels were stable, blood samples were taken at this time to assess plasma hormone concentrations during hyperinsulinemic-euglycemia, and, subsequently, the rats were microinjected with 0.1 μL of aECF vehicle over the course of 1 min using a CMA 402 syringe pump (CMA Microdialysis, North Chelmsford, MA). This microinjection protocol minimizes the spread of the compound and has been shown to localize the injection to the ventromedial nucleus (33). To determine whether the counterregulatory defect could be reversed in diabetic rats by blocking GABAergic inhibition in the VMH, we bilaterally microinjected the GABA_A receptor antagonist, bicuculline methiodide (BIC; 12.5 pmoL per side), into the VMH of one subgroup of STZ-induced diabetic animals just prior to the induction of hypoglycemia. The dose of BIC used here is a subconvulsive dose determined in a previously described pilot study (28). Specificity of the response to VMH BIC injection was assessed in a subgroup of animals that received a similar injection of BIC into the PVN of the hypothalamus. Following microinjection, the glucose infusion rate was decreased and plasma glucose levels were lowered and maintained at ~50 ± 5 mg/dL for 90 min. Blood samples were collected at 30-min intervals throughout the hypoglycemic clamp portion of the study for measurement of plasma glucagon and catecholamine responses and at the end of the study for plasma insulin concentrations. Following each sample collection, the erythrocytes were resuspended in an equivalent volume of artificial plasma (34) and reinfused back into the animal to prevent volume depletion and anemia. At the end of the study, following collection of the final blood and microdialysate samples, the animals were killed with an overdose of sodium pentobarbital, and the brains were removed and frozen on dry ice. Subsequently, accuracy of probe placement was determined histologically by visual inspection of coronal brain sections. Only data obtained from those animals with correctly positioned microdialysis probes were analyzed.

To address the issue of pharmacological specificity with the use of bicuculline, we used a gene knockdown approach to locally decrease the expression of the rate-limiting enzyme in GABA synthesis, GAD₆₅, in the VMH of STZ-induced diabetic rats. In two subgroups of STZ-induced diabetic rats, an adeno-associated virus (AAV) that expressed either a short-hairpin RNA (shRNA) against GAD₆₅ or a scrambled RNA sequence that does not adhere to any known sequences listed in Genbank were microinjected into the VMH 1 week after the induction of diabetes at a rate of 0.1 μL/min for 15 min. This method of delivery limits the spread of the virus to the VMH (35). Nondiabetic control animals received an injection of the AAV with the scrambled RNA sequence. One week after microinjection of the AAV, the animals were subjected to a hyperinsulinemic-euglycemic-hypoglycemic clamp, as described above, to assess counterregulatory hormone profiles. The shRNA sequence, which has 100% homology with both rat and mouse GAD₆₅, is 5′-TCTGGTTTGTACCTCCTAGTT-3′ and that for the scrambled sequence is 5′-GGAATCTCATTCGATGCATAC-3′.

Plasma catecholamine concentrations were analyzed by high-performance liquid chromatography, using electrochemical detection, whereas plasma hormone concentrations were determined using commercially available radioimmunoassay kits. VMH GABA concentrations from microdialysate samples were determined using liquid chromatography–tandem mass spectrometry after butylation, as previously described (36).

Frozen tissue micropunches from the VMH (~2.12 to −3.60 mm from bregma) were homogenized in 30 μL of lysis buffer (20 mmol/L HEPES, 50 mmol/L glycerol-2-phosphate, 2 mmol/L EGTA, 1 mmol/L dithiothreitol, 10 mmol/L NaF, 1 mmol/L sodium orthovanadate, 1% Triton X-100, and 10% glycerol), using a plastic pestle and ultrasonicator. Protein content was assessed with the Bradford protein assay (BioRad). Equal amounts of protein (15 μg) were resolved under reducing conditions in a 4–20% SDS-polyacrylamide gradient gel. Protein samples were then transferred onto nitrocellulose membranes. Success of the protein transfer was assessed by exposure of the nitrocellulose membranes to Ponceau S solution. Subsequently, the membranes were thoroughly washed prior to being soaked in 5% blocking solution for 1 h at room temperature. The membranes were then washed with 0.01 mol/L PBS, incubated with primary antibodies directed against GAD and β-actin, washed with PBS, and incubated with an appropriate secondary antibody conjugated to peroxidase. Following incubation with the secondary antibody, the membranes were washed with PBS and immersed into chemiluminescent reagent. The membranes were then exposed to Kodak X-OMAT film. Film analysis was undertaken using a computerized image analysis system running Kodak Digital Science One D software. Protein levels are expressed as a ratio of GAD and β-actin in order to correct for differences in loading and transfer.

Treatment effects were analyzed using one- or two-way ANOVA for independent or repeated measures, as appropriate, followed by post hoc analysis using the Statistica suite of analytical software for personal computers by StatSoft. P < 0.05 was set as the criterion for statistical significance.

Table 1 summarizes baseline levels of glucose and glucoregulatory hormones. Diabetic rats exhibited marked elevations in plasma glucose levels. Despite the presence of hyperglycemia in the diabetic animals, fasting plasma insulin concentrations were inappropriately reduced. Insulin was almost completely absent in the BB rats and lower, albeit not significantly, in STZ-induced diabetic rats. Plasma glucagon levels were more than twofold higher in BB diabetic rats but not significantly increased in the STZ-induced diabetic animals compared with nondiabetic controls. Epinephrine concentrations did not differ significantly between control and diabetic rats.

Under basal conditions, VMH GAD₆₅ protein levels were elevated by ~50% in both of our diabetic animal models (Fig. 1A). No differences were observed in GAD₆₇ protein levels between control and diabetic animals (Fig. 1B).

Figure 2A shows a schematic of our experimental protocol. Plasma glucose concentrations of diabetic animals were significantly higher than their nondiabetic counterparts (Fig. 2B) at the beginning of the study. After the start of the insulin infusion, plasma glucose levels of nondiabetic animals were maintained at euglycemia, whereas those of the diabetic animals were gradually lowered over the course of 2 h to match the euglycemic levels in nondiabetic controls. Stable euglycemia was then maintained for 30 min in all animals prior to microinjection and the induction of hypoglycemia. Average plasma glucose levels during the hypoglycemic plateau were similar between all of the treatment groups.

The increase in GAD₆₅ corresponded with an approximate doubling of baseline interstitial GABA levels in the VMH of diabetic animals (Fig. 3A). During hyperinsulinemic-euglycemia, extracellular GABA levels did not change from their respective baseline levels. As expected, in response to hypoglycemia, VMH GABA levels decreased by ~50% in nondiabetic animals. This drop was associated with activation of both the glucagon and epinephrine responses (Fig. 3B and C, respectively). In contrast, VMH GABA levels in both diabetic animal models did not fall from their high baseline concentrations in response to hypoglycemia, and this was accompanied by an absent glucagon and an attenuated epinephrine response.

To assess whether increased inhibitory GABAergic output is responsible for the impairment of counterregulatory hormone release, we microinjected the GABA_A receptor antagonist, BIC, into the VMH of STZ-induced diabetic animals to block postsynaptic GABA_A receptors prior to the induction of hypoglycemia. Administration of bicuculline restored both the glucagon and epinephrine responses to normal in diabetic animals, despite the persistence of elevated GABA levels in the VMH (Fig. 3). To establish whether the effects of the bicuculline injection were specific to the VMH, we microinjected bicuculline into another region of the brain that also expresses GABA_A receptors, the PVN of the hypothalamus, and measured the counterregulatory responses to hypoglycemia. In this study, we noted that blockade of GABA_A receptors in the PVN did not affect the release of glucagon or epinephrine in either nondiabetic or STZ-induced diabetic animals (Table 2).

Using an alternative and more specific approach to test our hypothesis, we administered an AAV expressing a shRNA against GAD₆₅ into the VMH of STZ-induced diabetic rats 1 week prior to the glucose clamp procedure. The GAD₆₅ shRNA lowered GAD₆₅ protein levels in STZ-induced rats with diabetes to normal levels, whereas the scrambled shRNA had no significant effect (Fig. 4). Administration of neither the AAV expressing GAD₆₅ shRNA nor the scrambled shRNA had any significant effect on baseline plasma glucose concentrations compared with control and STZ animals in the above study (Fig. 5). However, the reduction in GAD₆₅ expression resulted in normalization of baseline extracellular GABA levels in STZ-induced diabetic animals and, more importantly, restored the drop in VMH GABA levels during hypoglycemia (Fig. 6A). Knockdown of GAD₆₅ in the VMH of STZ-induced diabetic rats also restored the glucagon and epinephrine responses to hypoglycemia (Fig. 6B and C, respectively).

Counterregulatory failure in type 1 diabetes is a well-documented phenomenon. The mechanisms responsible for this defect, however, are not well understood. In previous studies, we showed that inhibitory GABAergic neurotransmission in the VMH is important for modulating the magnitude of glucose counterregulatory responses to an acute bout of hypoglycemia in rats (28). We subsequently reported that marked increases in GABAergic inhibitory neurotransmission within the VMH play an important role in the development of counterregulatory failure following antecedent exposure to hypoglycemia in nondiabetic rodents (29). In the current study, we show that a similar mechanism may be operating in type 1 diabetic rodents and could be a new unrecognized factor contributing to the development of counterregulatory failure, which is a major complication of the disease.

In the first set of studies, we characterized the GABA defect in two different rodent models of type 1 diabetes, the spontaneously diabetic BB rat and the STZ-induced diabetic rats. We observed that under baseline conditions, both diabetic animal models exhibited increased GAD₆₅ protein levels in the VMH, and this was corroborated by a biological marker, the doubling in extracellular GABA levels. The fact that the data are so remarkably similar between these two different animal models of type 1 diabetes suggests that it is likely the diabetic condition itself, and not the potentially toxic side effects of STZ, that is responsible for increasing GABA output within the VMH. Interestingly, no differences were observed with GAD₆₇ protein in the VMH. Our data are consistent with previously published reports showing increased GABA tone in the VMH of STZ-induced diabetic animals. Beverly et al. (37,38) reported increased GABA shunt and GAD activity in the VMH of STZ-induced diabetic rats, whereas Ohtani et al. (39) used microdialysis to show that GABA levels were elevated in the VMH of STZ-induced diabetic rats. Taken together, these data suggest that diabetes raises GABA output in the VMH at least in part by increasing GABA synthesis and release. Although microdialysis provides an assessment of net changes in extracellular GABA levels within the VMH, it does not allow us to specifically identify the source of GABA. Although the majority of GABA is thought to be of neuronal origin, astrocytically derived GABA also has been reported, but its contributions to the VMH interstitial pool and to glucose sensing have not been established (40,41).

In response to a hypoglycemic challenge, VMH GABA levels decrease in our control animals, and this corresponded with activation of both the glucagon and epinephrine responses. In contrast, VMH GABA levels remained elevated during hypoglycemia in both of our diabetic animal models, and this was associated with absent glucagon and attenuated epinephrine responses, establishing for the first time a link between increased inhibitory neurotransmission in the VMH and defective glucose counterregulatory responses in diabetes.

Interestingly, we noted that under hyperinsulinemic-euglycemic conditions, VMH GABA levels in nondiabetic animals did not change significantly from baseline levels. This implies that in normal animals, the release of GABA already may have reached a physiological ceiling under conditions of euglycemia, and the addition of insulin may not provide additional stimulation for GABA release. This is consistent with data from our recent study showing that the administration of increasing doses of glucose into the VMH during peripherally induced hypoglycemia dose-dependently raises extracellular GABA levels. These GABA levels continue to increase until euglycemic levels are reached, at which point they plateau at normal baseline levels (42). Raising VMH glucose concentrations above euglycemic levels does not increase VMH GABA levels further. In the case of the diabetic animals, the fact that GABA levels also remain high under hyperinsulinemic-euglycemic conditions suggests that factors apart from glucose and insulin may be responsible for raising GABAergic output in the VMH.

Although most of GABAergic neuronal cell bodies reside in the perimeter of the ventromedial nucleus, we have shown that a significant number of GAD₆₅-positive nerve terminals exist within the body of the VMH and that a subpopulation of these neurons coexpress ATP-sensitive K⁺ channels (42). More importantly, this subpopulation of VMH GABAergic neurons responds to local changes in glucose levels. What this suggests is that these neurons in particular are likely either glucose-sensing neurons themselves or that they may form synapses with other glucose-sensing neurons to modulate activity of those neurons as part of the glucose-sensing network that regulates the counterregulatory responses to hypoglycemia. In diabetes, increased activation of GABAergic tone may then prevent this network from initiating an appropriate response.

Having demonstrated that diabetes increases GABA levels in the VMH and that this was associated with an impaired counterregulatory response, we examined whether increased GABAergic neurotransmission in the VMH contributed to suppression of counterregulatory responses to hypoglycemia in diabetic animals using two parallel approaches. In the first approach, we microinjected BIC into the VMH of STZ-induced diabetic rats to locally block postsynaptic GABA_A receptors prior to the induction of hypoglycemia. This treatment completely restored both the glucagon and epinephrine defects in the diabetic animals, suggesting that increased GABAergic output in the VMH contributes significantly to attenuating the counterregulatory responses to hypoglycemia. The inhibitory effects on glucagon and epinephrine appear to be specific to GABAergic neurotransmission in the VMH because microinjection of BIC into a neighboring nucleus, the PVN, failed to restore the counterregulatory defects in diabetic animals.

To address issues associated with the specificity of BIC, we performed a second series of studies in which we locally administered an AAV that expressed either an shRNA against GAD₆₅ to locally knock down GAD₆₅ expression or one that expressed a nonspecific RNA sequence into the VMH of STZ-induced diabetic animals. Administration of the scrambled shRNA AAV did not have any demonstrable effects on VMH GAD₆₅ expression, extracellular GABA levels, plasma glucose, or counterregulatory hormone release. Nondiabetic control animals receiving the scrambled shRNA AAV exhibited the expected fall in baseline VMH GABA levels and activation of the counterregulatory response, whereas STZ-induced diabetic rats injected with the scrambled shRNA AAV showed the expected high VMH GABA levels and impaired counterregulatory responses to hypoglycemia. In stark contrast, when the high VMH GAD₆₅ levels in STZ animals were knocked down into the normal range using a GAD₆₅ shRNA AAV, we not only brought baseline GABA levels back down to normal, but, perhaps more importantly, we also restored the drop in VMH GABA concentrations during hypoglycemia. Knockdown of GAD₆₅ in the VMH restored both the glucagon and epinephrine responses to normal in the STZ-induced diabetic rats. This suggests that 1) a sustained increase in VMH GABAergic inhibition plays a key role in suppressing the glucagon and epinephrine responses to hypoglycemia in diabetes and 2) although the ability to decrease VMH GABAergic tone in response to a hypoglycemic challenge still may be present in diabetes, as evidenced by knockdown of GAD₆₅, it is the abundant overproduction of GABA (most likely stemming from higher GAD₆₅ levels) that prevents GABA levels from dropping below a critical threshold for full activation of counterregulatory responses. This observation underscores the importance of the role of the VMH to counterregulatory failure in type 1 diabetes. Although most studies suggest that the loss of the glucagon response to hypoglycemia in type 1 diabetes is mediated through peripheral mechanisms, including the loss of intraislet factors (10,11,13–15,43) as well as sympathetic innervations to the α-cell (20), less attention has been given to the role of the brain in this regard. We previously reported that activation of the fuel sensor, AMP kinase, in the VMH of BB rats can improve glucagon responses to hypoglycemia (44), and the current data demonstrate a similar restoration in two different models of β-cell injury (and insulin insufficiency) by modulating GABAergic neurotransmission in the VMH. These observations suggest that alterations in the capacity of the VMH, a key glucose-sensing brain region, to recognize fuel deficit may contribute significantly to the glucagon defect in diabetes as well. Thus, the VMH may play a greater role in counterregulatory failure in diabetes than previously appreciated.

Of particular interest is the fact that the data are remarkably similar to what we observed in animals with antecedent hypoglycemia. In both instances, we saw increases in VMH GABAergic output that was associated with impairments in the counterregulatory responses and improvements in these responses with GABAergic blockade. The fact that these two different conditions can lead to similar changes in the central nervous system and similar pathological outcomes was quite surprising. Although the mechanism responsible is uncertain, this finding suggests that a common adaptation within the central nervous system may occur in both of these conditions that acts to increase GAD₆₅ expression and GABA output in the VMH that in turn contribute to counterregulatory failure.

We conclude that diabetes leads to increased GAD₆₅ expression in the VMH with resulting increases in local GABA concentrations in this critical brain glucose-sensing region. In addition, we have shown, using two parallel approaches, that when GABA tone in the VMH is lowered in STZ-induced diabetic animals, either pharmacologically or genetically, we are able to restore the absent glucagon response and also restore the impaired epinephrine response to normal. Thus, uncontrolled or poorly controlled type 1 diabetes increases GABAergic tone in the VMH. Although the specific mechanisms responsible for these changes in the synthesis and release of the inhibitory neurotransmitter, GABA, remain to be established, they appear to contribute, at least in part, to defective glucose counterregulation in diabetes and thus be a therapeutic target for diminishing the risk of this serious complication of insulin treatment.

This work was generously supported by research grants from the Juvenile Diabetes Research Foundation (JDRF), the Yale Diabetes Endocrinology Research Center, and the National Institutes of Health (DK-20495 and DK-45735). O.C. is the recipient of a Career Development Award from the JDRF, and S.P. is the recipient of a postdoctoral fellowship from the JDRF.

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

O.C. conceived and designed the study, researched data, and drafted and revised the manuscript. S.P. researched data and revised the manuscript. D.C. performed mass spectrometry analysis. A.H. researched data and performed mass spectrometry analysis. W.Z., Y.D., and X.F. assisted with the surgeries and studies. M.S. and R.S. revised the manuscript.

The authors thank Aida Groszmann, Andrea Belous, Codruta Todeasa, Maria Batsu, and Ralph Jacob (Yale Center for Clinical Investigations Clinical Research Center) for their invaluable technical assistance with the hormone assays in these studies.