Localization of Glucokinase Gene Expression in the Rat Brain Ronald M. Lynch, Linda S. Tompkins, Heddwen L. Brooks, Ambrose A. Dunn-Meynell, and Barry E. Levin The brain contains a subpopulation of glucosensing neu- rons that alter their firing rate in response to elevated glucose concentrations. In pancreatic ␤-cells, gluco- ammalian feeding behavior and general energy kinase (GK), the rate-limiting enzyme in glycolysis, medi- homeostasis appear to be regulated by circu- ates glucose-induced insulin release by regulating intra- lating levels of nutrients (glucose) and pep- cellular ATP production. A similar role for GK is pro- tides (e.g., leptin, insulin). Sensors to detect lev- posed to underlie neuronal glucosensing. Via in situ M els of these factors have been found to reside within specific hybridization, GK mRNA was localized to hypothalamic areas that are thought to contain relatively large popu- nuclei of the hypothalamus (1–8), where central regulation of lations of glucosensing neurons (the arcuate, ventrome- energy homeostasis is believed to be coordinated. For exam- dial, dorsomedial, and paraventricular nuclei and the lat- ple, large changes in blood glucose are correlated with cen- eral area). GK also was found in brain areas without trally mediated responses such as thermogenesis through known glucosensing neurons (the lateral habenula, the activation of the sympathetic nervous system. These changes bed nucleus stria terminalis, the inferior olive, the are monitored by the brain (9–11), and such responses are retrochiasmatic and medial preoptic areas, and the thal- altered in obesity-prone animals (11–13). Moreover, lesions amic posterior paraventricular, interpeduncular, oculo- of the ventromedial hypothalamus (VMH) prevent the hypo- motor, and anterior olfactory nuclei). Conversely, GK glycemic activation of the sympathetic response (14). Thus, message was not found in the nucleus tractus solitarius, available data indicate that glucose detection by hypothala- which contains glucosensing neurons, or in ependymal cells lining the third ventricle, where others have mic neurons may play an important role in regulating energy described its presence. In the arcuate nucleus, >75% of homeostasis. neuropeptide Y–positive neurons also expressed GK, and Glucosensing neurons are among the best characterized of most GK+ neurons also expressed KIR6.2 (the pore-form- such metabolic sensors. Unlike most neurons, they use glu- ing subunit of the ATP-sensitive K+ channel). The cose as a signaling molecule to alter their firing rate in anatomic distribution of GK mRNA was confirmed in response to changes in ambient glucose levels (1,2,7,15,16). micropunch samples of hypothalamus via reverse tran- Glucosensing neurons constitute 10–40% of the resident neu- scription–polymerase chain reaction (RT-PCR). Nucleo- ronal populations of the VMH and lateral hypothalamic area tide sequencing of the recovered PCR product indicated (LHA) (1,2,15,17). In addition, glucosensing neurons have identity with nucleotides 1092–1411 (within exon 9 and been discovered in other brain areas including the amygdala 10) of hepatic and ␤-cell GK. The specific anatomic local- ization of GK mRNA in hypothalamic areas known to con- (18), locus ceruleus (19), and nucleus tractus solitarius (20). tain glucosensing neurons and the coexpression of Two distinct types of neuronal responses to changes in glu- KIR6.2 and NPY in GK+ neurons support a role for GK as cose concentrations have been observed, leading to charac- a primary determinant of glucosensing in neuropeptide terization of the neurons as either glucose-responsive or glu- neurons that integrate multiple signals relating to cose-sensitive. Glucose-responsive neurons are defined as peripheral energy metabolism. Diabetes 49:693–700, 2000 those that increase their firing rates when ambient brain glu- cose levels rise; glucose-sensitive neurons decrease their fir- ing rates when glucose levels increase (1,2,7,15,16). From the Departments of Physiology (R.M.L., L.S.T., H.L.B.) and Pharma- Most attention has been focused on the hypothalamus cology (R.M.L.), University of Arizona, Arizona Health Sciences Center, Tuc- because of its important role in energy homeostasis (7,8,21). son, Arizona; Neurology Service (A.A.D.-M., B.E.L.), Department of Veterans Whereas areas such as the VMH contain neurons that are Affairs Medical Center, East Orange; and the Department of Neurosciences (A.A.D.-M., B.E.L.), New Jersey Medical School, Newark, New Jersey. clearly inhibited by pathological lowering and stimulated by Address correspondence and reprint requests to Ronald M. Lynch, PhD, raising of glucose (14,22,23), both the VMH and LHA also con- Department of Physiology, University of Arizona, Arizona Health Sciences tain neurons that are inhibited as brain glucose increases Center, Tucson, AZ 85724. Received for publication 21 July 1999 and accepted in revised form over a physiologic range of 1.5–4 mmol/l (15,17). Of the 2 types 10 January 2000. of glucosensing neurons, the glucose-responsive neurons R.M.L. has received an honorarium for consulting (speaking) from Beta- have been best characterized physiologically because of their Gene, and B.E.L. has received honoraria for speaking from Merck and similarity to the glucosensing pancreatic ␤-cell (7,24). Alter- Novo Nordisk. ␣ ARC, arcuate nucleus; GK, glucokinase; HKI, hexokinase I; KATP channel, natively, the islet -cell may be the endocrine equivalent of the ATP-sensitive inward rectifier K+ channel; LHA, lateral hypothalamic area; glucose-sensitive neuron, since glucagon secretion is NPY, neuropeptide Y; PBS, phosphate-buffered saline; POMC, pro-opi- decreased in response to elevated glucose (25). omelanocortin; PVN, paraventricular nucleus; RT-PCR, reverse transcrip- ␤ tion–polymerase chain reaction; SSC, sodium chloride–sodium citrate; The pancreatic -cell, in turn, is the most extensively stud- VMH, ventromedial hypothalamus; VMN, ventromedial nucleus. ied model of cellular glucose sensing (24–26). The enzyme DIABETES, VOL. 49, MAY 2000 693 BRAIN GLUCOKINASE DISTRIBUTION glucokinase (GK; hexokinase IV) (27) appears to play a cen- been performed. This issue is of great importance, since GK tral role in the ability of ␤-cells to sense changes in blood glu- has emerged as a central component for glucosensing by the cose (24,28,29). ␤-Cells are activated by increases in blood glu- ␤-cell (24,26,32) and may well serve the same function in cose over a wide range (5–20 mmol/l). GK has a relatively low both glucose-responsive (7) and glucose-sensitive (15) neu- affinity for glucose (Km ~10 mmol/l [30,31]), which provides rons in the brain. For this reason, we postulated that GK the ␤-cell with the ability to increase the rate of glucose phos- would be highly localized to those brain areas in which glu- phorylation in proportion to increases in blood glucose over cosensing neurons reside. In the current studies, we used an the physiologic range (26,28). In ␤-cells isolated from rat analysis of GK mRNA expression in sequential sections of pancreas, the absolute sensitivity of individual cells to glucose entire rat brain by in situ hybridization. In addition, reverse appears to be correlated with the level of GK expression transcription–polymerase chain reaction (RT-PCR) was used (26), and overexpression of GK enhances secretion of insulin to confirm the localized expression in micropunches from spe- in response to elevated glucose (29). Moreover, modulation cific hypothalamic nuclei and to sequence the code for the of GK activity in insulin-secreting cell lines modulates the rate expressed GK. of glucose utilization and the sensitivity of these cells to glu- cose (30). A similar role for GK in glucose sensing by human RESEARCH DESIGN AND METHODS ␤-cells also has been demonstrated (31). Thus, it is clear that Animals. For the in situ hybridization studies, 8 male Sprague-Dawley rats the presence of GK is required for normal sensing of glucose weighing 300–400 g were used. For the RT-PCR studies, 5 female 300-g Sprague- by ␤-cells, and the relative level of GK expression may set Dawley rats were used. Animals were kept on a 12 h:12 h light-dark schedule at 22–23°C with food and water provided ad libitum. Rats were anesthetized by CO overall sensitivity or set point for activation by glucose (26). 2 and decapitated, and their brains were removed quickly for further study. To explain the coupling between changes in glucose phos- In situ hybridization. Probes were generated from a 1.4-kb fragment of the phorylation rate and ␤-cell activation, it is proposed that ATP mRNA sequence for rat GK (nucleotides 1–1422). This fragment was subcloned into produced by GK-regulated glycolysis inactivates the ATP- the BamHI site of a pBS2SK+ vector. Because this portion of the GK gene also con- + tains a small area of partial homology to rat hexokinase I (HKI), a truncated probe sensitive inward rectifier K channel (KATP channel), leading 2+ that shared no homology with HKI (nucleotides 963–1422) was used in a separate to depolarization of the cell limiting membrane, Ca influx, set of experiments to confirm the identity of GK. For prepro-neuropeptide Y (NPY), and insulin release (24,32). The pore-forming subunit of the a 511-bp fragment was subcloned into the EcoRI site of a bluescript SK(+) vector ␤ 35 specific KATP channel expressed in -cells is the KIR6.2, (36). Riboprobes labeled with S, digoxigenin, or fluorescein were generated using which, together with a sulfonylurea receptor, confers activ- standard techniques (37,38). Probes were hydrolyzed for 20 min at 60°C in 0.06 mol/l Na CO and 0.04 mol/l NaHCO , and probe fragments <20 nucleotides were ity to the channel (32). In the brain, this same channel 2 3 3 removed using microspin G-50 columns. Cryostat sections of rat brain were cut at appears to modulate neuronal firing in select glucose-respon- 10 µm, fixed in paraformaldehyde, and dehydrated.