Glucose is required for brain energy metabolism. Decerebration, aqueduct occlusion, and cannula mapping studies have established that glucose-sensing cells capable of eliciting feeding and adrenal medullary responses to glucoprivation are localized in the hindbrain. Glucoprivation also evokes corticosterone and glucagon secretion, but the location of receptors mediating these responses is unknown. To determine whether glucoreceptive sites controlling these responses are present in the hindbrain, we administered the antiglycolytic agent, 5-d-thioglucose (5TG, 24 μg in 200 nl) into brain stem sites through implanted cannulas and examined plasma concentrations of corticosterone and glucagon. Both hindbrain and hypothalamic sites were tested. Blood was collected remotely from intra-atrial catheters at 0, 30, 60, 90, 120, 180, and 240 min after 5TG or control injection. Caudal hindbrain 5TG injections potently increased circulating corticosterone and glucagon concentrations. For corticosterone, the mean peak response (maximum concentration minus time 0 concentration) elicited at positive sites (23 of 40 sites) was 391 ng/ml (SE = 16). For glucagon, the mean peak response at positive sites (27 of 40 sites) was 46 pg/ml (SE = 6). Glucoprivically evoked glucagon secretion was abolished by the ganglionic blocker, hexamethonium, but not by adrenal denervation. Six of twenty-five hypothalamic sites were positive for corticosterone secretion, yielding plasma levels of 279 ± 23 ng/ml, but none of the hypothalamic injection sites elevated glucagon concentrations. Results demonstrate that receptor cells responsive to glucose deficit and capable of increasing corticosterone and glucagon concentrations exist within the hindbrain, thus further delineating central glucoregulatory neural circuitry.
central glucose deficit stimulates food intake and elicits secretion of corticosterone, glucagon, and adrenal epinephrine. Food intake replenishes depleted energy stores, and the adrenal medullary and endocrine responses mobilize stored energy, stimulate gluconeogenesis, and conserve existing glucose supplies by promoting use of fatty acids. These responses are essential for maintenance of adequate glucose delivery to the brain.
The sensory receptor cells that monitor the brain's glucose supply and elicit appropriate glucoregulatory responses have not been precisely identified. However, results from decerebration (9), cerebral aqueduct obstruction (35), and localized microinjection studies (37) concur in showing that glucoreceptors controlling the feeding and adrenal medullary responses to glucoprivation are located in the hindbrain. Receptor cells capable of eliciting glucagon and corticosterone secretion in response to glucoprivation have been shown to be present within the brain (3, 6), but whether they are present in the hindbrain is not known. Existing evidence, however, suggests that the hindbrain is a likely location for such receptors. Glucagon secretion is increased by both sympathetic and parasympathetic neurons (12, 14–17, 41, 42). Hindbrain catecholaminergic and serotonergic neurons provide heavy innervation of autonomic preganglionic neurons (7, 10, 26) and have been strongly implicated in glucagon secretion (19, 20, 40). Hindbrain catecholamine neurons are also involved in glucoprivic control of corticosterone secretion, since immunotoxic destruction of the hindbrain norepinephrine and epinephrine neurons that innervate the paraventricular hypothalamus (PVH), severely and selectively impairs this response (36, 38). Because both feeding and adrenal medullary responses, which are dependent on hindbrain catecholamine neurons, are controlled by hindbrain glucoreceptors, we hypothesized that both glucagon and corticosterone secretion also may be subject to hindbrain glucoreceptive control.
Our major objective in the present study was to test the hypothesis that glucoreceptive sites controlling corticosterone and glucagon secretion are present in the hindbrain. We produced localized hindbrain glucoprivation by injection of nanoliter volumes of the antiglycolytic agent, 5-d-thioglucose (5TG), through cannulas chronically implanted at sites throughout the hindbrain. We administered a dose of 5TG previously shown to elicit other glucoregulatory responses (feeding and hyperglycemia) from hindbrain sites (31). Hypothalamic sites have also been implicated in glucoregulation (2, 3, 23, 24). Therefore, in the present experiment, we tested both hindbrain and hypothalamic sites with the same dose of 5TG. We reasoned that if a given dose of 5TG was effective in the hindbrain, but not when injected into the hypothalamus, then we could rule out the hypothalamus as the site of action of hindbrain-injected 5TG.
MATERIALS AND METHODS
Preparation of animals.
Adult male Sprague-Dawley rats weighing 320–380 g at the start of the experiments were obtained from Simonsen Laboratories, (Gilroy, CA). Rats were housed individually in suspended wire mesh cages in a temperature-controlled room (21 ± 1 C) illuminated between 0600 and 1800. Rats had ad libitum access to pelleted rat food and tap water, except as noted. Tests were conducted between 0800 and 1400. All experimental animal protocols were approved by the Washington State University Institutional Animal Care and Use Committee, which conforms to the National Institutes of Health guidelines.
Rats were prepared for experimentation by implantation of intra-atrial catheters and intracranial cannulas. For surgical procedures, rats were induced with a ketamine/xylazine/acepromazine cocktail [5 ml ketamine HCl, 100 mg/ml (Fort Dodge Animal Health, Fort Dodge, IA); 2.5 ml xylazine, 20 mg/ml (Vedco, St. Joseph, MO); 1 ml acepromazine, 10 mg/ml (Vedco); and 1.5 ml 0.9% saline solution] and maintained under anesthesia with 1.5% isoflurane (Isoflo; Abbott Laboratories, N. Chicago, IL). Catheters constructed from Silastic tubing (inside diameter, 0.64 mm; outside diameter, 1.19 mm; Dow Corning, Midland, MI) were implanted intra-atrially through the right jugular vein. When not in use, catheters were locked with heparin saline (0.9% saline with 10 U/ml heparin) and temporarily occluded with a polyvinylpyrrolidone solution (40,000 molecular weight, Sigma-Aldrich), 11 g polyvinylpyrrolidone in 20 ml 0.9% saline containing 1000 U heparin (Elkins-Sinn, Cherry Hill, NJ), and 2 mg/ml Gentamicin (Boehringer Ingelheim, Vetmedica, St. Joseph, MO). Coordinates for stereotaxic cannula implantation were determined using Paxinos and Watson's The Rat Brain in Stereotaxic Coordinates (31), with the intent of achieving a distribution of implantation sites in the caudal hindbrain and hypothalamus. A single cannula was implanted into each rat. Cannulas for implantation in the brain were fabricated from stainless-steel tubing. Guide cannulas (26-gauge tubing) occluded with removable obturators were implanted into the hypothalamus or hindbrain. The obturator was removed and replaced with an injection cannula (33-gauge stainless-steel tubing) that extended 0.2 mm beyond the tip of the guide cannula and was connected by polyethylene tubing to a microinfusion pump for intracranial drug delivery. The drug delivery system was filled with 5TG or control solution. Movement of a 1-mm bubble in the calibrated infusion line was used to confirm drug delivery. Following the injection, the obturator was replaced, and the rat was returned to the blood sampling chamber.
Responses to hypothalamic and hindbrain 5TG injections.
In an attempt to maintain the sensitivity of the tissue at the 5TG injection sites, each rat received only two injections, 5TG and control. The dose of 5TG was determined from a similar study in which we examined 5TG-induced hyperglycemic and feeding responses (37). The injection volume (200 nl) was determined by analysis of dye injection to have a diffusion radius of 0.5 − 1.0 mm 30 min after the injection, depending on the structure of the underlying tissue.
Before blood collection experiments, rats were extensively habituated in 30 cm × 10 cm opaque Plexiglas chambers. On test days, which were at least 1 wk apart, rats were placed in chambers without food 1 h before the first (time 0) blood sample, which was followed immediately by an intracranial injection of 5TG (24 μg/200 nl; Aldrich Chemical, Milwaukee, WI) or saline (0.9%, 200 nl). Additional samples were collected remotely at 30, 60, 90, 120, 180, and 240 min after the 5TG or control injection. Samples were centrifuged, and the plasma was aliquoted and stored at −80°C. Glucose, corticosterone, and glucagon were determined from the same blood samples. Glucose concentration was analyzed using the glucose oxidase method (39). Corticosterone concentrations were determined in duplicate aliquots using RIA kits obtained from Diagnostic Products (catalog no. TKRC-1; Los Angeles, CA). Glucagon concentrations were determined in duplicate aliquots using RIA kits obtained from Linco Research, (catalog no. GL-32K; St. Charles, MO). At each sampling time, blood volume withdrawn was replaced with an equal volume of washed and resuspended erythrocytes obtained from heparinized donor blood. Donor blood was prepared within 24 h of experiments and was stored at 4°C overnight.
In a separate experiment, six rats with hindbrain cannula placements positive for 5TG-induced elevation of glucagon secretion were tested to determine whether the glucagon response could be blocked by systemic administration of the sympathetic and parasympathetic ganglionic blocking agent, hexamethonium (Cat. # H-2138, 10 mg/kg ip; Sigma-Aldrich, St. Louis, MO). Hexamethonium was injected 15 min before 5TG administration. Blood samples were collected and processed as described above. Finally, we blocked release of adrenal medullary catecholamines to assess their potential role in central 5TG-induced glucagon secretion. In nine hindbrain cannulated rats, the adrenal glands were bilaterally denervated by transection of the adrenal nerve branches at their point of entry into the adrenal gland. Nine hindbrain-cannulated, sham adrenal-denervated rats were used as controls. 5TG was administered and blood was collected for analysis of glucose and glucagon levels, as described above.
At the conclusion of testing, rats were anesthetized by halothane inhalation and immediately perfused transcardially with fresh 0.1 M PBS (pH 7.4) followed by fixation with 4% formaldehyde in 0.1 M PBS (pH 7.4). Brains were immediately removed and stored at 4°C in 4% formaldehyde until histological analysis. One day before sectioning of tissue, brains were transferred to fresh 25% sucrose (wt/vol) in sterile water. Brains were sectioned along the coronal axis at 40 μm. Sections were directly mounted sequentially onto Superfrost slides and allowed to dry at room temperature for 24 h. Sections were stained with cresyl violet and cover slipped for microscopic analysis of cannula location. The point of deepest penetration of the cannula tip was identified and plotted on the anatomically corresponding section from the Paxinos and Watson stereotaxic atlas of the rat brain (31).
Responses to saline control and 5TG were analyzed for all hindbrain and for all hypothalamic placements using one-way ANOVA with repeated measures. After assignment of a positivity score to each cannula site, responses from positive sites in hypothalamus and hindbrain were compared using a Mann-Whitney rank sum test. Results from the hexamethonium experiment were analyzed using one-way ANOVA on repeated measures followed by a Holm-Sidak multiple pairwise comparison test. Effects of adrenal denervation were analyzed using a t-test.
Plasma values for corticosterone, glucagon, and glucose across the entire sampling periods following saline and 5TG injections into hindbrain or hypothalamic sites are shown in Fig. 1. Hindbrain injection of 5TG significantly elevated plasma glucagon, corticosterone, and glucose concentrations and these responses occurred with time courses similar to those observed when these responses are elicited by systemic glucoprivation. Hypothalamic 5TG injections increased corticosterone secretion modestly, but did not increase glucagon secretion. As we reported previously (37), hypothalamic 5TG injection did not elevate blood glucose.
A statistical criterion was applied to define positive and negative responses so that cannula placements associated with positive responses could be identified. Responses that were ≥2 SDs from the mean peak saline response were considered to be positive, and responses less than this value were designated as negative. Of the rats with hindbrain cannulas, 23 out of 40 had peak corticosterone responses to 5TG that were positive by this criterion, 27 had glucagon responses that were positive, and 36 were positive for glucose. Of the rats with hypothalamic cannulas, 6 out of 25 were positive for corticosterone. Three of these positive cases were associated with cannula sites in or immediately adjacent to the optic tract at different hypothalamic levels. There were no hypothalamic sites positive for glucagon. The mean peak responses for positive responders only are shown in Fig. 2 for hindbrain and hypothalamic cannula sites. The distribution of cannula sites yielding positive and negative responses is shown in Fig. 3 for hindbrain and Fig. 4 for hypothalamus.
Among the rats with hindbrain placements, the same three animals ranked highest for all three 5TG responses. Their cannula sites for these three animals are indicated in Fig. 4. The responses to 5TG (means ± SE) for these three sites were 157 ± 31 mg/dl for glucose, 470 ± 47 ng/ml for corticosterone, and 206 ± 48 pg/ml for glucagon. Responses to saline injection at these three sites were 9 ± 3 mg/dl (glucose), 50 ± 34 ng/ml (corticosterone), and 1 ± 1 pg/ml (glucagon). Among the hypothalamic 5TG injection sites, none were positive for all three responses.
Effects of peripheral administration of the ganglionic blocker, hexamethonium, on glucagon secretion induced by hindbrain glucoprivation are shown in Fig. 5. Hexamethonium completely blocked the glucagon response (P < 0.006). The Holm-Sidak post hoc test showed that peak responses to 5TG were significantly greater than the baseline response (50.4 ± 4.9 pg/ml) when saline was administered systemically (97.1 ± 12.2 pg/ml, P < 0.05) than when hexamethonium was administered systemically (67.2 ± 2.0 pg/ml). Hexamethonium also completely blocked the hyperglycemic response was to 5TG (Fig. 6), as measured at all sampling times, verifying the effectiveness of the ganglionic blockade.
Adrenal denervation reduced, but did not block, the glucagon response to 5TG. Glucagon secretion was increased to 140.4% of baseline in the adrenal denervated rats and to 170.8% of baseline in intact controls (P = 0.13).
These results show for the first time that glucoreceptive sites controlling glucagon and corticosterone secretion are present in the hindbrain. The fact that the same 5TG dose that produced positive responses when injected into the hindbrain was much less effective or ineffective when injected into the hypothalamus precludes the possibility that the hindbrain-injected 5TG somehow diffused rostrally to exert effects in the hypothalamus. Anatomical results show that sites controlling glucagon, corticosterone, and glucose responses to 5TG are coextensive in the hindbrain. In addition, these sites are located in areas associated with, and essential for, other glucoregulatory responses, including glucoprivation-induced feeding (8, 9, 35–37), adrenal medullary secretion (5, 25, 35–37), suppression of pulsatile LH secretion (28) and estrous cycling (18), and the growth hormone response (32). The confluence in the hindbrain of glucoprivation-sensitive sites capable of exerting widespread action throughout the neuroaxis to elicit diverse behavioral, endocrine, and autonomic responses strongly indicates the importance of this brain region in glucoregulation.
Pancreatic glucose concentration (33, 43) and intraislet insulin levels (11, 34) exert direct control over glucagon secretion by pancreatic alpha cells. An important neural control of glucagon secretion has also been clearly established, based on the strong autonomic influence on secretion (1, 12, 13, 15) and evidence that central glucoprivation increases glucagon secretion (3). Present findings show for the first time that hindbrain glucoreceptors contribute to the central control of glucagon secretion. These results also demonstrate that the hindbrain-stimulated glucagon response is neurally mediated, as it is antagonized by blockade of ganglionic nicotinic receptors. In addition, surgical denervation of the adrenal medulla reduced, but did not block, the glucagon response to 5TG, indicating that the response is not secondary to or dependent on adrenal medullary epinephrine secretion. Our findings therefore complement other data showing that central neural pathways converging on autonomic outputs to the pancreas participate in the control of glucagon secretion.
The existence of a hypothalamic glucoreceptive control of glucagon secretion has been proposed previously, based on the finding that bilateral microdialysis of the antiglycolytic agent 2-deoxy-d-glucose (2DG) into the ventromedial hypothalamus elicits a robust glucagon response (3) and perfusion of glucose into that site blocks glucoregulatory hormone release during systemic glucoprivation (2). However, in these studies, downstream ventricular diffusion of 2DG or glucose was not evaluated, leaving open the possibility that hindbrain glucoreceptive sites were activated inadvertently. Nevertheless, the fact that hypothalamic 5TG injections were not effective in stimulating glucagon secretion in our study must be interpreted cautiously. Because we tested only one 5TG dose, the existence of hypothalamic glucoreceptors controlling these responses cannot be ruled out on the basis of our findings. Higher 5TG doses, larger injection volumes, or different (perhaps bilateral) administration protocols may reveal hypothalamic receptors not detected in this study. Additional investigation will be required to evaluate fully the proposed involvement of the hypothalamus in the glucoprivic control of glucagon secretion.
The potent effects of hindbrain glucoprivation on corticosterone secretion, shown in this study, and the critical dependence of the corticosterone response on projections of hindbrain catecholamine neurons, shown previously (38), indicate that the glucoprivic control of this response is similar in terms of its organization and circuitry to the feeding (36) and reproductive responses (18). All require activation of hindbrain glucosensory cells and rostrally projecting catecholamine neurons. Corticosterone secretion was also increased by 5TG injections at six hypothalamic cannula sites. The possibility that hypothalamic glucoreceptors may control the HPA axis response to glucose deficit is an interesting one and consistent with previous results showing that glucoprivation is capable of eliciting corticotropin releasing hormone (CRH) from hypothalamic explants (44). Glucose-sensing neurons have been identified in the hypothalamus (22, 29, 30). Possibly, some of these neurons control corticosterone secretion in response to glucose flux. Finally, our PVH DSAP injections to date have not completely eliminated the corticosterone response to glucoprivation, despite the nearly complete catecholamine denervation of the parvocellular CRH-containing areas (38). Roughly 25–30% of the response remains. This portion of the response could be mediated by hypothalamic glucose-sensing neurons.
Although there is a context for interpreting our corticosterone results as evidence for hypothalamic glucoreceptors, it should nevertheless be noted that 3 of the 6 hypothalamic sites where 5TG elicited a positive corticosterone response were positioned in or immediately adjacent to the optic tract. The close association of these 3 sites with the optic tract suggests that the results from these sites should be viewed cautiously. Corticosterone secretion is notoriously sensitive to sensory, emotional, and cognitive perturbations (4). Possibly, the corticosterone response to 5TG injection in these sites was not glucoregulatory in nature but indicative of a response to disruption of visual sensation. Indeed, stimulation of corticosterone secretion by manipulation of any brain site should be viewed cautiously for similar reasons.
On the basis of measurements from dye injections of the same volume (200 nl), we estimate the diffusion radius of the injected 5TG in our experiment was 0.5–1.0 mm, depending on the microstructure and quality of the tissue at the injection site. Assuming this diffusion radius, there was considerable overlap in the anatomical areas affected by 5TG injections at the different cannula sites. Additional studies using lower 5TG doses and smaller injection volumes will be required to determine whether glucoreceptor cells are localized to specific sites or are widely distributed in the hindbrain. Nevertheless, the distribution of positive sites revealed by the present study may suggest phenotypes of neurons that participate in glucoregulation. Known neurotransmitter systems with distributions roughly similar to the distribution of hindbrain glucoprivation-responsive sites include the catecholaminergic and to a lesser extent, the caudal serotoninergic cell groups, both of which extend along the entire extent of the hindbrain. Catecholamine neurons are known to be critical for the corticosterone response to glucoprivation (38), but whether they are involved in the glucagon response is not yet known. Serotonin neurons in the caudal raphe cell groups may be important for glucagon secretion, since injection of kainic acid into nucleus raphe obscurus increases glucagon secretion (20). The fact that some positive cannula sites were located in close proximity to vagal afferent and efferent fiber trajectories may also be significant since central control of glucagon secretion is mediated through the autonomic outflow to the pancreas [(12, 14–17) and this study] and because glucagon secretion is increased by injection of kainic acid into (20) or electrical stimulation of (21) the dorsal vagal complex.
Results of this study add glucagon and corticosterone to the list of key glucoregulatory responses subject to control by hindbrain glucoreceptor cells. Hindbrain catecholamine neurons have been shown to be essential for a number of these responses, including corticosterone secretion (18, 36, 38), but their importance for glucagon secretion has not yet been tested. Whether the hindbrain glucosensors are neuronal (perhaps catecholaminergic) or glial has not been established. However, important new findings indicate that glial, but not neuronal, glucose transporter type 2 expression is a critical component of central glucoprivation-induced secretion of glucagon (27). What specific aspect of diminished energy availability serves as the transduction signal, whether the different glucoregulatory responses are controlled by a common pool or by functionally differentiated pools of receptor cells, and the complete neural circuitry for elicitation and coordination of glucoregulatory responses are issues that remain to be resolved. These are crucial areas for future investigation due to the importance of glucoregulation for normal brain function and because a more complete understanding of glucoregulatory mechanisms is necessary to address problems associated with diabetes and its complications.
This work was supported by Public Health Service Grants DK40498 and NS045520 and Juvenile Diabetes Research Foundation Grant 1-2006-308 to S. Ritter.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society