The Medial Amygdalar Nucleus: A Novel Glucose-Sensing Region That Modulates the Counterregulatory Response to Hypoglycemia

Male Sprague-Dawley rats (mean ± SEM wt, 305 ± 4 g) were housed in the local Animal Resource Center with water and chow pellet available ad libitum. The animal care and experimental protocols were reviewed and approved by Yale University Institutional Animal Care and Use Committee and the Institutional Animal Care and Use Committee of the East Orange Veterans Affairs Medical Center.

The surgical procedures used in this study have been described in detail elsewhere (8,16). In brief, one week prior to each study, all animals were anesthetized with an intraperitoneal injection (1 ml/kg) of a mixture of xyzaline (20 mg/ml; AnaSed, Lloyd Laboratories Inc.) and ketamine (100 mg/ml; Ketaset, Wyeth) at a ratio of 1:2 (vol/vol). Vascular catheters [PE50 tubing with a tip made from silastic laboratory tubing (0.51 mm inner diameter)] were inserted via a neck incision into the internal jugular vein and carotid artery. After catheter implantation, cannula guides were stereotaxically inserted, bilaterally to the MAN (coordinates from bregma, anterior-posterior [AP] = −2.80 mm, medio-lateral [ML] = ±3.3 mm, and dorso-ventral [DV] = 8.9 mm at an angle of 90°) or VMH (coordinates AP = −2.6 mm; ML = ±0.5 mm; DV = 9.4 mm). Guide cannula were designed to reach a point 1 mm proximal to the target nucleus, limiting gliosis in the region where microinjection would take place 7 days after guide catheter insertion.

At the initial surgery, as described above, and instead of guide cannula insertion, each rat received bilateral microinjections to the MAN of 2 pg of ibotenic acid (n = 6) using a 1 μl Hamilton syringe (total volume 200 nl over 30 min), after which the skin was closed with wound clips. Sham-lesion rats received the identical surgical procedure but were administered saline rather than ibotenic acid (n = 9).

Seven days after surgery, overnight fasted rats had their catheters opened and were allowed to acclimatize over 90 min. Bilateral 26-guage injection needles, designed to extend 1 mm beyond the tip of the guide cannula, were then inserted into each MAN or VMH, and each rat received bilateral microinfusions (1 μl given at a rate of 0.033 μl/min for 30 min) of 10 mmol/l 2-deoxyglucose (2-DG), a nonmetabolizable form of glucose that creates local glucopenia, or artificial extracellular fluid (aECF), depending on the study. Venous samples for measurement of plasma glucose were taken every 10 min and for glucagon and epinephrine preinjection (t = 0 min) and at 30 and 60 min after microinjection.

A modified hyperinsulinemic glucose clamp was used to produce a standardized hypoglycemic stimulus, as described previously (17). Thirty min after bilateral MAN microinfusions of 2-DG (n = 6) or artificial aECF (n = 9), a constant 20 mU/kg/min infusion of insulin (Human Regular Insulin, Lilly, IL) was started. Both solutions were mixed with 4% wheat germ agglutinin (WGA) to confirm the location of microinjection (supplementary Fig. 1, available in an online appendix at http://diabetes.diabetesjournals.org/cgi/content/full/db09-0995/DC1). The plasma glucose was allowed to fall to 70 mg/dl (∼3.9 mmol/l), where it was maintained for 90 min using 20% dextrose, with the dextrose infusion rate adjusted every 5 min based on plasma glucose determinations. Blood samples for measurement of epinephrine, glucagon, and insulin were taken at 0, 60, and 90 min. Plasma glucose was measured by glucose oxidase method (Glucose Direct; Analox Instruments, Lunenberg, MA). Catecholamine analysis was performed by high-performance liquid chromatography using electrochemical detections (ESA, Acton, MA); plasma insulin and glucagon were measured by radioimmunoassay (Millipore, Temecula, CA).

The retrograde tracing technique has been described previously (18). Briefly, each rat was microinjected bilaterally to the VMH or MAN (coordinates as above) with the retrograde neuronal tracer WGA or 4% fluorogold dissolved in aECF and delivered through a glass micropipette with tip diameter 10–15 μm by passing 2-μA positive current pulse (7 s on/off) for <10 min. The rats were then allowed to recover for 7 days before undergoing a 90-min hyperinsulinemic (20 mU/kg/min) hypoglycemic (70 mg/dl) or hyperinsulinemic euglycemic (120 mg/dl) clamp, as described above.

The immunohistochemistry protocol used has been described in detail previously (18,19). Briefly, analysis was performed on every six 40-μm-thick frontal sections. Sections were washed in PBS for 10 min and then pretreated in 0.3% H₂O₂ for 1 h to block endogenous peroxidase. They were then incubated overnight at room temperature with 1:1,000 rabbit anti-UCN antibody (1:2000; a gift from Dr. W. Vale) in 0.3% fresh normal donkey serum, goat anti-WGA (1:1,000; Vector) and PBT-azide (0.02% sodium azide and 0.04% Triton X-100 in PBS). The next day, sections were incubated with biotinylated donkey anti-rabbit antibody (Jackson Laboratories; 1:500) for 1 h, followed by incubation with a cocktail of Alexa Fluor 594-conjugated Streptavidin (Molecular Probes, Eugene, OR; 1:1K) and Alexa Fluor 488-conjugated chicken anti-goat antibody (1:400) for 1 h. After mounting on polylysine slides, the sections were cover-slipped with antifade mount medium for fluorescence (Vectashield, Vector, CA). Dual fluorescence images in the MAN from the representative brain were taken with a digital camera. Counting of dual staining for UCN3 and WGA was performed in three brains. All single-labeled UCN3 neurons and dual-labeled neurons for UCN3 and WGA in the MAN were counted. Cytoarchitectonic areas in the amygdala were determined with reference to the atlas of Paxinos and Watson (20). Light-microscope images in the MAN and its adjacent regions from a representative brain were taken with a digital camera. The digital images were arranged in the software Canvas (Deneba System, Miami, FL). The border of the MAN, its adjacent structures, and distribution of single labeling of UCN3 neurons were drawn using Canvas software. For triple staining, immunofluorescence serial sections from the brains of fluorogold-injected rats were incubated overnight with a mixture of 1 μg/ml anti-Fos goat antibody (1:1K; Santa Cruz) and anti-UCN3 rabbit serum (1:2,000). After a rinse with PBS-X, the sections were incubated for 1 h with 10 g/ml biotinylated anti-rabbit IgG donkey antibody (Jackson) and then for 1 h with a mixture of 1 μg/ml Alexa488-conjugated chicken anti-goat antibody (1:400; Molecular Probes) and 1 μg/ml Alexa594 Streptavidin-conjugated antibody (Molecular Probes). Sections were observed under an epifluorescence microscope Olympus BX-50 with appropriate filter sets for Alexa488 (excitation, 450–490 nm; emission, 514–565 nm); Alexa594 (excitation, 530–585 nm; emission, ≥615 nm) and fluorogold (excitation, 359–371 nm; emission, 397–590 nm). Although Alexa488 fluorescence could be partially seen with the filter set for fluorogold, Fos-immunoreactive neuropil with Alexa488 fluorescence was easily differentiated from retrograde labeled cell bodies with fluorogold fluorescence. The counting of triple staining for UCN3, Fos, and fluorogold was performed in three brains. All single-labeled UCN3 neurons and dual-labeled and triple-labeled neurons in the MAN were counted.

Frozen brains were cut on a cryostat at −12°C and placed in RNAlater (Ambion, Foster City, CA) until being micropunched. Micropunches of the MAN were performed by modifications of the method of Palkovits (21), where brain micropunches are made under microscope guidance from brain slices placed on the base of a stereotaxic frame. Micropunched brain areas were sonicated in a guanidinium thiocyanate solution and purified using magnetic beads (Ambion MagMax-96). Quantitation of mRNA was carried out by real-time quantitative PCR (QPCR) as previously described (22). Primer sets for each mRNA were designed by reference to published sequences, and their specificity was verified using Genebank. Primers and their sequence-specific FAM-labeled probes prepared by Applied Biosystems were sequenced and then quantified with an Applied Biosystems 7,700 real-time PCR system set for 40 PCR cycles. Standard curves were generated from serially diluted pooled samples for each probe and for constitutively expressed mRNA (cyclophilin) to control for differences in amplification efficiency and micropunch size. Results were calculated from the standard curves relative to cyclophilin mRNA levels in the same samples.

Studies were carried out in 3–4 week old male Sprague-Dawley rats (Charles River). The MAN punched cells were dissociated with papain (2 mg/ml, 30 min, 37°C) and mechanically triturated. Cells were plated onto cover slips and allowed to adhere for 60 min before loading with the Ca²⁺ fluorophore fura-2 acetoxy methyl ester (Molecular Probes) for 20 min in Hank's balanced salt solution buffer containing 2.5 mmol/l glucose, washed twice, and transferred to a microscope chamber held at 37°C. Fura-2 fluorescent images were acquired every 5 s by alternating excitation at 340 and 380 nm using a cooled, charge-coupled device camera at 420–600 nm emission. Changes in glucose were maintained for ∼10 min after addition. Cells were classified as glucose-responsive in >500 neurons by significant ([Ca²⁺]_i) fluctuations (as area under the curve) after changes in glucose concentrations using Origin 7.0 software (OriginLab). Neurons were classified as glucose-excited, glucose-inhibited, and nonglucosensing, as previously described (7). Cytoplasms of individual characterized neurons were then collected for single-cell QPCR (sc-QPCR) as previously described (23); analysis was according to previous standards.

After characterization by Ca²⁺ imaging, cytoplasmic mRNA of individually imaged cells was analyzed by sc-QPCR. Cytoplasm from each neuron was aspirated into a micropipette that was prefilled with DEPC-treated water containing 1 μl RN_ASE OUT RNase inhibitor (Invitrogen, Carlsbad, CA). Subsequently, synthesis was performed with Superscript II first-strand synthesis kit (Invitrogen) following the manufacturer's directions. The RT reaction was incubated at 42°C for 50 min after heat inactivation at 70°C. cDNA was purified to completely remove inhibitory RT components by slight modification of the methods described by Liss (24). After the purification of single-cell cDNA, QPCR was performed using specific primers. Glial fibrillary acidic protein was used to exclude astrocytes, and β-actin was used as an internal control for constitutively expressed mRNA. Primer sequences were designed using Biology WorkBench Primer design software. Amplification was carried out in a LightCycler (Roche Perkin-Elmer, Foster City, CA), using 40 cycles (95°C, 1 s; 56–63°C, 2 s; and 68°C, 30 s) with Advantage 2 Polymerase Mix (BD Biosciences, Palo Alto, CA). Amplified products were run directly on a 1.5% agarose gel and visualized by ethidium bromide staining. Gels were imaged and photo inverted for presentation. To optimize conditions for primer amplification and standardize for the linearity of the amplification process, hypothalamic cDNA was used.

For glucagon and epinephrine, differences between treatments were assessed via repeated-measures ANOVA and based on hormonal readings at 0, 60, and 90 min. Post hoc analysis was made using Bonferroni testing. Mean data were compared using a two-tailed Student t test. For all analyses, significance was assigned at the P ≤ 0.05 level. Data are presented as mean ± SEM.

Of 522 individual MAN neurons examined using fura-2 calcium imaging, 13.5% were glucose-sensing (Table 1). Six percent were glucose-excited, and 7.5% were glucose-inhibited (Fig. 1). By sc-QPCR, 54% of glucose-excited, 42% of glucose-inhibited, and 9% of nonglucosensing MAN neurons expressed glucokinase (Table 2). However, UCN3 was isolated from only 2, 16, and 4% of the glucose-excited, glucose-inhibited, and nonglucosensing neurons, respectively.

Having shown that only a minority of glucose-sensing neurons contained mRNA for UCN3, quantitative real-time PCR was used to examine the spatial distribution of glucokinase, UCN3, and CRH-R2 mRNA expression in MAN micropunches. A rostrocaudal gradient in glucokinase, UCN3, and CRH-R2 gene expression in the MAN was seen. Both glucokinase and CRH-R2 expression showed a similar rostrocaudal distribution, whereas UCN3 showed the opposite rostrocaudal distribution in the MAN (supplementary Fig. 2, available in an online appendix).

To examine whether the MAN contributed to the generation of a counterregulatory response during acute hypoglycemia in vivo, the MAN of male Sprague–Dawley rats was lesioned by direct microinjection of ibotenic acid. During the subsequent hyperinsulinemic hypoglycemia study, plasma glucose levels did not differ between MAN lesioned and control rats (70 ± 1 vs. 69 ± 1 mg/dl, respectively; Fig. 2,A). Glucose infusion rates (GIR) required to maintain the hypoglycemic plateau were ∼18% higher in MAN-lesioned rats (mean GIR over 60–90 min, 24.7 ± 1.7 vs. 20.3 ± 1.4 mg/kg/min), although the overall interaction was not significant (F = 4.0; P = 0.07; Fig. 2,B). However, MAN lesioning did result in significantly impaired plasma epinephrine (F = 6.0, P < 0.05) and glucagon (F = 6.9, P < 0.05) responses to the hypoglycemic challenge (Fig. 2 C and D).

To determine whether the MAN might respond directly to a local glucoprivic challenge, the MAN was locally perfused with 2-DG over 30 min. MAN glucoprivation did not result in a significant rise in plasma glucose or change in plasma levels of glucagon and epinephrine (Table 3). Application of 2-DG at an identical concentration and rate to the VMH also failed to elicit a glucoprivic response (Table 3). Additional studies injecting 5-Thioglucose (a more potent glucoprivic agent) into the MAN or VMH also had no significant effect on glucose or counterregulatory hormones (data not shown).

In contrast to the lack of effect of local MAN glucoprivation during euglycemia, comparable MAN glucoprivation during mild systemic hypoglycemia did alter the subsequent counterregulatory response. Despite equivalent hypoglycemia [mean (SEM) 60–90 min glucose levels, 70 ± 2 versus 69 ± 1 mg/dl (3.9 ± 0.1 vs. 3.9 ± 0.1 mmol/l); Fig. 3,A], GIR (60–90 min) were threefold lower in the MAN 2-DG infused rats compared with the aECF control group (3.8 ± 1.1 versus12.3 ± 2.2 mg/kg/min, respectively; F = 24.6, P < 0.01) (Fig. 3,B). Consistent with the lower GIR, 2-DG microinjection to the MAN significantly amplified the glucagon (F = 4.8, P < 0.05) and epinephrine (F = 7.2, P < 0.05) responses during hypoglycemia (Figs. 3 C and D).

To determine the origin of UCN3 neurons innervating the VMH, the retrograde neuronal tracer, WGA, was microinjected into the VMH. While UCN3 neurons are present in the medial preoptic nucleus and the rostral perifornical area as expected, by double-label immunohistochemistry, that most of the UCN3 neurons in the MAN project to the VMH (cell bodies positive for [WGA+UCN3]/UCN3 = 492/520). Moreover, microinjection of a retrograde tracer to the MAN demonstrated direct neural pathways in the reciprocal direction from cell bodies in the VMH to nerve terminals in the MAN (supplementary Fig. 4). A schematic representation of the location of UCN3 cell bodies innervating the VMH is shown in Fig. 4, whereas microphotographs of cells showing dual labeling of UCN3 and WGA in the MAN are shown in supplementary Fig. 3.

Hypoglycemia activates VMH-projecting MAN UCN-3 neurons. To determine whether the UCN3 cell bodies in the MAN were activated during acute hypoglycemia, rats underwent hyperinsulinemic hypoglycemic (∼70 mg/dl) or hyperinsulinemic euglycemic (∼120 mg/dl) clamp studies for 120 min as described above. Using triple fluorescence immunostaining for cFOS, UCN3, and the retrograde tracer fluorogold, we found that ∼30% (155/520 neurons) of those UCN3 neurons in the MAN that innervate the VMH coexpressed cFOS, a marker of neuronal activation, during acute hyperinsulinemic hypoglycemia when compared with hyperinsulinemic euglycemia (1/655; P < 0.05). This suggests that at least one-third of the MAN UCN3 neurons are activated during a mild hypoglycemic stimulus (Fig. 5).

The principal finding of the current study is the novel discovery that the MAN contains glucose-sensing neurons that can influence the magnitude of the counterregulatory response to insulin-induced hypoglycemia. The amygdala is a complex structure, containing a number of discrete nuclei involved in a wide range of behavioral and physiological functions. In the rodent, the MAN is strongly connected with the olfactory system, has numerous interconnections with other amygdalar nuclei (enabling it to integrate the neural outputs from these different regions), and, like the VMH, has also been shown to integrate with the neural circuits linked to feeding and body weight control (25,26). Interestingly, a recent study using [¹⁸F]-fluorodeoxyglucose positron emission tomography, in 13 men with type 1 diabetes, reported reduced [¹⁸F]-fluorodeoxyglucose uptake in the amygdala during hypoglycemia in those subjects with hypoglycemia unawareness failure, suggesting a potential role for the amygdala in the development of hypoglycemia unawareness (27).

The VMH is integral to glucose-sensing during acute hypoglycemia (for a review, see ref [28]). The VMH contains specialized glucose-sensing neurons (23), and 14–19% of these are glucose-excited, whereas 3–14% are glucose-inhibited in type (29). As in the pancreatic β-cell, glucokinase appears to be a critical regulator of glucose-sensing in VMH neurons (6), where glucokinase is expressed in ∼65% of glucose-excited and ∼45% of glucose-inhibited neurons (23). In fact, it is most likely that all neurons that express glucokinase are actually glucose-sensing and that the finding of glucokinase mRNA in non-glucose-sensing neurons is due to the stringent criteria used to classify neurons. In the present study, we have been able to show, using fura-2-calcium imaging, that the expression of glucokinase mRNA in the MAN is associated with the presence of specialized glucose-sensing neurons. Of these, ∼6% were glucose-excited whereas 7.5% were glucose-inhibited. Moreover, 54% of glucose-excited and 42% of glucose-inhibited, compared with only 9% of nonglucose-sensing neurons, contained mRNA for glucokinase. These findings clearly parallel those of the VMH, providing support for the hypothesis that the MAN may represent a novel glucose-sensing brain region.

To determine whether the MAN plays a functional role in glucose-sensing, we initially used the selective neurotoxin, ibotenic acid, to lesion the MAN and then performed a hyperinsulinemic hypoglycemic study. Consistent with a previous study where ibotenic acid was used to lesion the VMH (8), lesioning the MAN was shown to suppress the counterregulatory hormonal response to subsequent hypoglycemia. Subsequently, we sought to determine whether localized glucoprivation in the MAN of euglycemic animals would induce a glucoprivic response, characterized by a rise in plasma glucose and in the counterregulatory hormones glucagon and epinephrine, as previously demonstrated in the VMH (9). However, neither 2-DG nor 5-thioglucose infusions into either the MAN or VMH over 30 min raised plasma glucose, glucagon, or epinephrine levels. We then sought to determine whether combining local MAN glucoprivation during a moderate systemic hypoglycemic stimulus might influence counterregulatory responses to assess whether this additional glucoprivic stimulus might amplify the counterregulatory response. In fact, the combination of local MAN and systemic glucoprivation did have this effect. Taken together, these in vivo studies suggest that the MAN contributes to the counterregulatory response induced by systemic hypoglycemia but that local glucoprivation in the MAN alone is insufficient to generate a counterregulatory hormone response.

A previous study that used unilateral microinjection to localize glucose-sensing brain regions also failed to elicit a glucoprivic response in the majority of hypothalamic regions tested, although it did produce glucoprivic responses with unilateral hindbrain microinjections (30). More recently, bilateral VMH 5-TG microinjections were shown to stimulate food intake; however, no glucose readings were assessed in this study (31). Food intake was not measured in our study, but we did not find an increase in blood glucose after VMH 5-TG or 2-DG. The reasons for these discrepancies are not clear. Borg et al. (9) used microdialysis to deliver 2-DG locally to the VMH, and it is possible that, under these conditions, the stimulus to glucose-sensing neurons is greater. Interestingly, recurrent glucoprivation impairs hypothalamic but not hindbrain responses to subsequent hypoglycemia (32), suggesting that repeated hypoglycemia restrains hindbrain glucose-sensing via an upstream (hypothalamic and or other brain region) mechanism. On the basis of the present findings and those of Ritter et al. (30), we would speculate that glucose-sensors in the hindbrain may form part of a classical sensory reflex response, whereas glucose-sensors in higher centers are more integrated in their response to a glucoprivation, that is, hypoglycemia might need to be present in a number of glucose-sensing brain and/or peripheral sensors before counterregulation is initiated.

Given the importance of the VMH to the detection of hypoglycemia and our previous studies showing a role for UCN3 in the VMH in modulating the counterregulatory response to hypoglycemia, we then sought to examine whether UCN3 might link these two glucose-sensing regions. Previous neuroanatomical studies have shown that the MAN projects topographically to the VMH (25), findings confirmed in our study using retrograde neural tracers to show reciprocal pathways between these two regions. Additionally, we have been able to demonstrate that most of the UCN3 projections to the VMH arise in the MAN. Moreover, using triple-staining immunohistochemistry, we also found that ∼1/3 of these activated during acute hypoglycemia. In this context, our previous finding that pharmacological manipulation of CRH-R2 receptors in the VMH markedly altered counterregulatory responses to acute hypoglycemia is very suggestive of a functional role for this neural network (13). It is perhaps counterintuitive that UCN3 activation in the MAN during hypoglycemia might lead to suppression of glucose-sensing neurons in the VMH. However, it is notable that few of the individual MAN glucose-sensing neurons identified by Ca²⁺ imaging also expressed mRNA for UCN3. On the other hand, many did express mRNA for its receptor, CRH-R2. In addition, gene expression analysis from serial MAN micropunches showed that glucokinase and CRH-R2 gene expression had the same rostrocaudal distribution, but the opposite rostrocaudal distribution as UCN3 mRNA. This implies that UCN3 neurons may represent a discrete neuronal population in the MAN that are not in themselves glucose-sensing and, thus, may also directly regulate glucose-sensing neurons locally, as they do in the VMH. This would lead us to speculate that UCN3 neurons may regulate or coordinate the output, in terms of the counterregulatory responses, from these two discrete glucose-sensing regions.

In summary, in the current study we have identified the MAN as a novel limbic glucose-sensing region that contains characteristic glucokinase-expressing glucose-sensing neurons that respond directly to manipulations of glucose availability. In addition, manipulation of the MAN by lesion or through provision of an additional glucoprivic stimulus modulates the counterregulatory response to moderate systemic hypoglycemia. Finally, we have shown that both these glucose-sensing regions are linked by UCN3 neurons, which potentially provides a mechanism for fine-tuning and integrating the stress response during a hypoglycemic challenge.

This work was supported by National Institutes of Health Grants (DK-069831, R.J.M.) and (DK-53181, B.E.L.), the Diabetes and Endocrinology Research Center (DERC) at Yale, and the Juvenile Diabetes Research Foundation (R.J.M.). L.Z. is a Juvenile Diabetes Research Foundation Postdoctoral Research Fellow. B.E.L. is supported by the Research Service of the Veterans Administration.

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

L.Z. conducted the in vivo studies and performed IHC; N.P. performed the fura-2 calcium imaging and sc-QPCR; Z.S. helped with the in vivo studies and IHC; Y.D. performed rodent surgeries and helped with in vivo studies; X.F. helped with in vivo studies; Q.T. contributed to the discussion; B.E.L. supervised fura-2 calcium imaging and sc-QPCR, helped with manuscript preparation, and contributed to the discussion; R.J.M. wrote the manuscript and designed and supervised the in vivo studies and IHC.

The authors are grateful to Ralph Jacob and Aida Groszmann, Yale University, for their support and assistance. The authors thank Dr. Wylie Vale, Salk Institute, CA, for kindly providing the Urocortin 3 antibody used in these studies, and Dr. Brad Lowell, Beth Israel Deaconness Medical Center, Boston, MA, for his helpful comments.