A High Affinity Glutamate/Aspartate Transport System in Pancreatic Islets of Langerhans Modulates Glucose-stimulated Insulin Secretion*

To examine the role of glutamatergic signaling in the function of pancreatic islets, we have characterized a high affinity glutamate/aspartate uptake system in this tissue. The islet [3H]glutamate uptake activity was Na+-dependent, and it was blocked byl-trans-pyrrolidine-2,4-dicarboxylic acid, a blocker of neuronal and glial glutamate transporters. Islet glutamate transport activity exhibited a V max of 8.48 ± 1.47 fmol/min/islet (n = 4), which corresponds to 102.2 ± 17.7 pmol/min/mg islet protein. The apparent K m of islet glutamate transport activity depended on the glucose concentration used in the assay. In the presence of glucose concentrations that do not stimulate insulin secretion (2.8 mm), the apparent K m was 34.7 ± 7.8 μm (n = 3). However, in high glucose (16.7 mm) the apparent K m increased to 112.7 ± 16.5 μm (n = 3) with little or no change in V max. Like most known plasma membrane glutamate transporters, islet glutamate transporters also transported d-aspartate. Anti-d-aspartate immunoreactivity showed that the islet glutamate/aspartate transport activity was localized to the non-β cell islet mantle. In perifusion experiments with isolated islets in the absence of exogenous amino acids,l-trans-pyrrolidine-2,4-dicarboxylic acid in the presence of 8.3 mm glucose potentiated insulin secretion 23.3 ± 2.3% (n = 3) compared with 8.3 mm glucose alone. This effect was abolished in the presence of the α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid receptor antagonist, 6-cyano-7-nitroquinoxaline-2,3-dione. Furthermore, 6-cyano-7-nitroquinoxaline-2,3-dione alone inhibited glucose-stimulated insulin secretion in isolated islets by 15.9 ± 5.9% (n = 3). Taken together these data suggest that a high affinity glutamate transport system exists in pancreatic islets and that this system contributes to a glutamatergic signaling pathway that can modulate glucose-inducible insulin secretion.

Although the role of glutamate as a signaling molecule is well established in the central nervous system, a similar role in the periphery has only recently been suggested. We (1) and others (2) have detected functional glutamate receptors in the pancreatic islets of Langerhans. These miniature organs, found dispersed throughout the exocrine pancreas, are composed of four major cell types as follows: the insulin-secreting ␤ cell, the glucagon-secreting ␣ cell, the pancreatic polypeptide-secreting PP cell, and the somatostatin-secreting ␦ cell. The electrically excitable ␤ cells are stimulated to secrete insulin in response to changes in serum glucose concentrations. Secretion of insulin, and the three other major peptide hormones found in islets, is also believed to be affected by other metabolic and neuronal signals (reviewed in Refs. 3 and 4). Bertrand et al. (5,6) have shown that ␣-amino-3-hydroxy-5-methylisoxazole-4-propionic acid (AMPA) 1 receptor agonists can potentiate both insulin and glucagon secretion from a perfused pancreas preparation and that oral or intravenous glutamate can increase insulin secretion and glucose tolerance in vivo (7).
We have localized AMPA-type glutamate receptors to ␤, ␣, and PP cells and kainate receptors in ␣ cells using immunohistochemistry and electrophysiology (1). To elucidate the role of glutamatergic signaling in islet physiology, we examined islets for the presence of a high affinity uptake system similar to those described in the central nervous system. Glutamate transporters in the central nervous system allow glutamate to act as a specific signaling molecule despite its relatively high concentration in the cerebrospinal fluid because they reduce the concentration of glutamate in the vicinity of receptors.
Earlier studies have suggested that islets do not possess a high capacity for glutamate uptake and utilization; however, these studies were focused on the possible role of glutamate as a carbon source or as a fuel secretagogue (8). The millimolar concentrations of glutamate used in these studies did not adequately address the possibility of a high affinity glutamate uptake system in islets, the presence of which might serve to support a role for glutamate receptors in islet signal transduction.
By using a [ 3 H]glutamate uptake assay, we detected glutamate transport activity in isolated rat pancreatic islets. The uptake observed in islets had properties similar to those of central nervous system transporters. It had high affinity for glutamate, was Na ϩ -dependent, and was blocked by L-transpyrrolidine-2,4-dicarboxylic acid (L-trans-PDC), a compound that blocks neuronal and glial transporters. Furthermore, the apparent K m of islet glutamate uptake was markedly increased by increasing glucose concentrations. Visualization of D-aspartate uptake with anti-D-aspartate antibodies showed that the transporter activity was located in the ␣ cell-rich islet mantle. The blockade of glutamate transport in isolated islets by Ltrans-PDC potentiated glucose-induced insulin secretion, whereas blockade of AMPA receptors with 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) resulted in inhibition of insulin secretion. These observations indicate that glutamate transporters are important components of a glutamatergic signaling system within the pancreatic islets of Langerhans.

EXPERIMENTAL PROCEDURES
Materials-Percoll was purchased from Pharmacia Biotech Inc. Culture dishes were purchased from Corning. Ham's F-12 tissue culture media, fetal bovine serum, horse serum, penicillin, and streptomycin were purchased from Life Technologies, Inc. L- [3, H]Glutamate and D- [2, H]aspartate were purchased from NEN Life Science Products. L-trans-PDC, bicuculline methiodide, and CNQX were purchased from Research Biochemicals. L-Cystine was purchased from Sigma. Alkaline phosphatase-conjugated goat anti-rabbit antibody was purchased from DAKO. Alkaline phosphatase-conjugated donkey anti-guinea pig antibody was purchased from Jackson ImmunoResearch. Triton X-100 Surfact-Amps was purchased form Pierce. Electron microscopy grade glutaraldehyde, paraformaldehyde, and tissue freezing medium were purchased from Electron Microscopy Sciences. Poly(A)qua/Mount was purchased from Polysciences. All other chemicals were of reagent grade or higher. Pregnant Sprague-Dawley rats were obtained from Harlan.
Islet Isolation and Culture-Islets were isolated and cultured by a modification of the method described (9). Pancreata were harvested from 5 to 10, 4-or 5-day-old neonatal rat pups and placed in 4 ml of Ham's F-12 medium containing 5% (v/v) fetal bovine serum, 100 IU/ml penicillin, and 100 g/ml streptomycin (culture medium). The pancreata were diced with fine iris scissors until the pieces were approximately 0.5 mm 3 . During the process of dicing, the pieces of pancreas were washed 5 times with fresh culture medium. Diced pancreas tissue was placed in 100-mm diameter polystyrene culture dishes at a density of approximately three diced pancreata per dish in 6 ml of culture medium. The pieces were incubated overnight in a 5% CO 2 incubator at 37°C. The following day, the medium was removed from the pancreas explants and replaced with fresh culture medium. The explants were then incubated for 3 days in the 5% CO 2 incubator after which time the medium was changed again. Two days following the last medium change the islet explants were dislodged from the bottom of the polystyrene dishes with gentle trituration and transferred to a 50-ml polypropylene conical tube. The explants were pelleted by centrifugation at 2000 rpm for 2 min in a Beckman TJ-6 centrifuge. The explant pellet was then resuspended in 5 ml of fresh culture medium. The resuspended islet pellet was layered onto a Percoll gradient prepared as described (10) with the exception that the Percoll solutions were made up in culture medium. The gradients were centrifuged at 5000 rpm for 15 min in a Beckman TJ-6 centrifuge. The islets were harvested from the 1.060/1.082 g/ml Percoll interface, and the remaining acinar tissue was found at the medium 1.060 g/ml Percoll interface. Five ml of culture medium was added to the harvested islets in a 15-ml conical tube, and the sample was mixed by inversion. The islets were pelleted by centrifugation at 500 ϫ g for 3 min in a clinical centrifuge and washed with an additional 10 ml of culture medium. The islets were pelleted as above and resuspended in culture medium. The islets were plated at a density of approximately 1000 -2000 islets per 100-mm diameter polystyrene culture dish in 6 ml of culture medium per dish and subsequently cultured in a 5% CO 2 incubator overnight at 37°C. The following day the islets were harvested from the culture dish using a 5-ml polystyrene pipette leaving behind the remainder of the fibroblasts adhering to the culture dish. The islets were pelleted as described above and resuspended in 6 ml of culture medium where 25% (v/v) horse serum was used in place of 5% (v/v) fetal bovine serum. The islets were incubated overnight in a 5% CO 2 incubator and used for L-glutamate/D-aspartate uptake or insulin release assays the next day.

H]Glutamate/D-[ H]Aspartate
Uptake Assays-Islets isolated as described above were hand counted using a Gilson P-10 micropipetor and an Olympus CK2 inverted microscope. Thirty to fifty islets were transferred to 1.5-ml polypropylene microcentrifuge tubes containing 100 l of a buffer consisting of (in mM) 140 NaCl, 4.7 KCl, 1.2 KH 2 PO 4 , 1.2 MgSO 4 , 2.2 CaCl 2 , 10 HEPES, pH 7.3 (uptake buffer), containing 11.2 mM glucose. In some studies choline was used in place of Na ϩ and gluconate was used in place of Cl Ϫ in the uptake buffer. After harvest 1 ml of uptake buffer was added to the tubes containing the islets. The islets were pelleted in a swinging bucket microcentrifuge at 500 ϫ g for 15 s. The buffer was removed by careful aspiration, and the islets were resuspended in 150 l of uptake buffer containing varying glucose concentrations and bicuculline, as appropriate, at either 37 or 0°C. The islets were then incubated for 5 min at either 37°C or on ice. Following the preincubation, 150 l of uptake buffer was added containing the appropriate glucose concentration, bicuculline (where appropriate), and twice the desired final concentration of L-[ 3 H]glutamate, D-[ 3 H]aspartate, L-cystine, and L-trans-PDC, as appropriate. Samples were incubated for 5-9 min at either 37°C or on ice. The uptake assay was stopped by the addition of 1 ml of ice-cold uptake buffer. The islets were pelleted by centrifugation as described above. The supernatant solution was aspirated, and the islets were resuspended in 1 ml of ice-cold uptake buffer. This wash procedure was repeated a total of three times. After the final wash, the islets were lysed in 100 l of 0.5 M NaOH for 5 min with occasional vortexing. The samples were centrifuged as above and then transferred to a vial containing 3 ml of aqueous scintillation mixture, mixed by inversion, and counted using a Packard model 1600 TR liquid scintillation counter. Values obtained from the scintillation counter were converted into uptake activity units using L-[ 3 H]glutamate or D-[ 3 H]aspartate standards. For calculation of K m and V max , values obtained at 0°C were treated as a blank and subtracted from the values obtained at 37°C. Uptake data were analyzed using Igor (Wavemetrics), Excel (Microsoft), and Instat (GraphPad Software). Statistical comparisons were paired two-tailed t tests. L-trans-PDC inhibition curves were fit, and L-trans-PDC IC 50 was calculated using the logistic equation. Figures were constructed using Igor.
Immunochemistry-Two to five hundred islets were placed in a 50-l perifusion chamber. The islets were washed for 30 min at a flow rate of 100 l/min with oxygenated uptake buffer containing 2.8 mM glucose at 37°C. Following the wash the islets were perifused for 15 min at the same flow rate and temperature with uptake buffer containing 2.8 mM glucose (control) or uptake buffer containing 2.8 mM glucose and 100 M D-aspartate. The islets were washed for 15 min at 200 l/min with ice-cold uptake buffer. The islets were then fixed at room temperature in the chamber by perifusion with uptake buffer containing 2.5% (v/v) glutaraldehyde and 1% (v/v) paraformaldehyde at 200 l/min for 5 min. The perifusion was halted, and the islets were incubated for 1 h in the fixative at room temperature. At the end of the incubation the chambers were disassembled, and the frits with fixed islets attached were removed. The islet-bearing frits were transferred to 1.5-ml centrifuge tubes containing uptake buffer with 0.25% (v/v) glutaraldehyde and 0.1% (v/v) paraformaldehyde and incubated overnight at 4°C. The frits were then washed 3 times for 20 min in phosphate-buffered saline (PBS) composed of (in mM) 137 NaCl, 2.7 KCl, 10.1 Na 2 HPO 4 , 1.5 KH 2 PO 4 , and 2 times for 30 min in PBS containing 30% (w/v) sucrose. The sucrose equilibrated islet-containing frits were placed face down in a mold containing tissue freezing medium (OCT compound) and frozen on dry ice. Five-m sections were cut and thaw-mounted onto charge slides. The sections were allowed to dry for 10 min and were rehydrated in PBS. Sections were permeabilized for 10 min in PBS containing 0.2% (v/v) Triton X-100 and blocked for 1 h in 5% (v/v) normal goat serum (for the anti-D-aspartate antibody) or 5% (v/v) normal donkey serum (for the anti-glucagon antibody) in PBS. The blocked sections were incubated overnight at 4°C with a 1:250 dilution of anti-glucagon antibody or a 1:1000 dilution of anti-D-aspartate antibody. The anti-D-aspartate antibody was raised as described (11) and thoroughly tested as described (12). The antibodies were diluted in PBS plus 1% (v/v) normal goat serum (for the anti-D-aspartate antibody) or 1% (v/v) normal donkey serum (for the anti-glucagon antibody) and 0.1% (v/v) Triton X-100. Following incubation the sections were washed 1 time for 10 min in PBS plus 0.1% (v/v) Triton X-100 and 2 times for 10 min/wash in Trisbuffered saline (TBS) composed of (in mM) 25 Tris, pH 8.0, 137 NaCl, 2.7 KCl plus 0.1% (v/v) Triton X-100. For detection of tissue-bound antibody, washed sections were incubated for 1-3 h at room temperature with a 1:100 dilution of alkaline phosphatase-conjugated goat antirabbit antibody (for the anti-D-aspartate antibody) or a 1:500 dilution of alkaline phosphatase-conjugated donkey anti-guinea pig antibody (for the anti-glucagon antibody) diluted in TBS plus 1% (v/v) normal goat serum or 1% (v/v) normal donkey serum, respectively, and 0.1% (v/v) Triton X-100. Following incubation slides were washed 3 times for 10 min/wash in TBS plus 0.1% Triton X-100 and 1 time for 5 min with a buffer containing (in mM) 100 Tris, pH 9.5, 100 NaCl, 5 MgCl 2 (alkaline phosphatase buffer) plus 0.1 mM levamisole. Immunoreactivity was visualized with nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate in alkaline phosphatase buffer containing 0.1 mM levamisole. Slides were mounted with Poly(A)qua/Mount. Photomicrographs were taken using a Leica Laborlux S microscope equipped with a Wild MPS52 camera attachment, a Wild MPS46 photoautomat, and Kodak T max 100 film. Prints were produced on Ilford Multigrade IV RC Deluxe paper.
Insulin Secretion Assay-100 islets per experimental condition were placed in a 50-l perifusion chamber. The islets were perifused for 30 min at a flow rate of 100 l/min at 37°C with a buffer containing (in mM) 128 NaCl, 1.19 MgSO 4 , 18 NaHCO 3 , 2.54 CaCl 2 , 1.19 KH 2 PO 4 , 4.74 KCl, and 0.1% (w/v) bovine serum albumin (perifusion buffer) plus 2.8 mM glucose (bubbled for 1 h prior to perifusion with 95% O 2 /5% CO 2 ). Following this wash, the perifusion was stopped for 5 min. At the end of the 5 min, a fraction was collected by perifusing the islets at 1 ml/min for 30 s. Perifusion was allowed to continue at 100 l/min for 2.5 min after which the flow was stopped for an additional 5 min. Another fraction was collected as above. The islets were then perifused for 2.5 min at 100 l/min with perifusion buffer containing 2.8 mM glucose or perifusion buffer containing 2.8 mM glucose plus either 500 M L-trans-PDC, 10 M CNQX, or both L-trans-PDC, and CNQX. Flow was stopped and fractions were collected as above. Following this treatment, the islets were perifused for 2.5 min at 100 l/min with perifusion buffer containing 8.3 mM glucose or perifusion buffer containing 8.3 mM glucose plus either L-trans-PDC, CNQX, or both L-trans-PDC and CNQX. Flow was stopped and fractions were collected as above. Insulin concentrations in the perifusate were determined by a core facility using the Linco insulin radioimmunoassay with appropriate standards. Data were analyzed using Instat and Excel. Statistical comparisons are from paired two-tailed t tests.

RESULTS
We have identified a high affinity glutamate uptake system in isolated, intact pancreatic islets using a [ 3 H]glutamate uptake assay. Islet glutamate uptake exhibited saturable kinetics (Fig. 1)  3) of that measured at 37°C. Islet [ 3 H]glutamate uptake was also strongly dependent on the presence of Na ϩ in the uptake buffer. In experiments where Na ϩ was replaced with choline, the glutamate uptake activity was 28.7 Ϯ 3.0% (n ϭ 3) that measured in the presence of normal external Na ϩ . A Na ϩindependent, Cl Ϫ -dependent glutamate transport activity has been reported in the periphery (13,14) and in neurons (15). This activity, designated system X c Ϫ , is inhibited by excess cystine. Islet glutamate uptake decreased only modestly (6.1%, n ϭ 2) when Cl Ϫ was replaced by gluconate in the uptake assay buffer. However, the addition of 500 M cystine in Na ϩ -free conditions further decreased [ 3 H]glutamate activity to 11.3% of control. Since system X c Ϫ is believed to be primarily involved in the uptake of cystine, we focused our study on the more prominent Na ϩ -dependent [ 3 H]glutamate uptake activity in islets.
To determine which islet cell types were responsible for the observed glutamate uptake activity, we examined transport of D-aspartate. As in the case of most other known plasma membrane glutamate transporters, islet D-[ 3 H]aspartate transport was quite similar to L-[ 3 H]glutamate transport for both K m and V max (Fig. 2). D-[ 3 H]Aspartate uptake was also sensitive to temperature and to the removal of external Na ϩ (Fig. 2, inset). The similarities between L-[ 3 H]glutamate and D-[ 3 H]aspartate uptake suggested that both were transported by the same transporter proteins and thereby D-aspartate uptake offered a way to determine which islet cells harbored glutamate transporters. In these experiments islets were incubated in the presence and absence of 100 M D-aspartate and fixed with glutaraldehyde and paraformaldehyde to covalently bind amino acids in a form accessible for immunostaining. This technique was first demonstrated as indicated in Ref. 11, and the degree of retention of labeled L-glutamate and D-aspartate by different fixatives was determined (16). Samples were FIG. 1. Islet glutamate transport activity is saturable. Glutamate transport was measured by [ 3 H]glutamate uptake at 37°C in uptake buffer containing 11.2 mM glucose. Points represent the average of triplicate samples (ϮS.E.). Islet glutamate transport activity is Na ϩ and temperature-dependent (inset). Islet glutamate transport was measured at 37°C in standard uptake buffer (control), at 37°C in a buffer where the Na ϩ was completely replaced by choline (Na ϩ -free), or at 0°C in standard uptake buffer (0°C). Data are the average of three determinations from separate islet preparations (ϮS.E.). stained with the selective D-aspartate antibodies (12,17) and with anti-peptide hormone antibodies for the identification of the cell types that concentrated D-aspartate. As shown in Fig.  3A, anti-D-aspartate immunoreactivity was restricted to the cells of the islet mantle which, in our preparation, are primarily glucagon-containing ␣ cells (Fig. 3B). The islet mantle may also contain a small percentage of ␦ and PP cells (data not shown), and thus our data cannot exclude the presence of D-aspartate uptake activity in these cell types. Anti-D-aspartate staining was not observed in islets that had not been incubated in the presence of D-aspartate (Fig. 3C).
Because islets are glucose-sensitive organs, we have also investigated the effect of D-glucose on [ 3 H]glutamate uptake in islets. We found that islet glutamate uptake activity was significantly inhibited (32.0 Ϯ 3.8%, n ϭ 5, p ϭ 0.0013) in the presence of a glucose concentration that stimulates insulin secretion (16.7 mM D-glucose) compared with uptake in low D-glucose (2.8 mM) (Fig. 4A). The inhibitory effect of glucose on glutamate transport was not due to an increase in the osmolarity of the uptake buffer since non-metabolizable L-glucose (16.7 mM) failed to result in an inhibition of glutamate uptake (102.06 Ϯ 4.2% of control) (Fig. 4A).
Replots of the uptake data revealed that inhibition by Dglucose was the result of a change in the apparent K m for glutamate and not a change in the V max (Fig. 4B). At low glucose, islet glutamate uptake had an apparent K m of 34.7 Ϯ 7.8 M (n ϭ 3). However, in the presence of 16.7 mM D-glucose (Fig. 4B) the apparent K m for glutamate shifted to 112.7 Ϯ 16.5 M (n ϭ 3). These data are consistent with a competitive mode of inhibition by D-glucose at a concentration that stimulates insulin secretion.
It has been suggested that down-regulation of glucagon secretion in ␣ cells exposed to high glucose may be mediated by GABA A receptors expressed in these cells (18,19). The proposed source of GABA is ␤ cells, which may coordinately release it with insulin (20). Since glutamate/aspartate uptake activity was predominantly in ␣ cells, we tested whether glucose inhibition of glutamate transport might be mediated through this mechanism. To do this we repeated the uptake experiments with 2.8 and 16.7 mM D-glucose in the presence of the GABA receptor antagonist, bicuculline. However, the degree of inhibition (31.7 Ϯ 1.7%, n ϭ 3) was the same as that measured in the absence of bicuculline (Fig. 4A). The addition of 50 M bicuculline alone had no effect on [ 3 H]glutamate uptake (102.7 Ϯ 11.2% of uptake in the absence of bicuculline, n ϭ 3).
Previously, Bertrand et al. (5,7) have demonstrated that AMPA receptor agonists potentiate glucose-induced insulin secretion in a perfused pancreas preparation (5) and in vivo (7). To identify potential roles of glutamate transporters in islets, we examined the effects of glutamate transport inhibition on glucose-stimulated insulin secretion. L-trans-PDC is an inhibitor of glutamate transport in the central nervous system (21) and showed an IC 50 of 129.2 M against islet [ 3 H]glutamate transport. In the presence of 35 M glutamate the maximum amount of inhibition we observed in islets was 66.0% with 2 mM L-trans-PDC. The average inhibition of glutamate transport (35 M) was 55.0 Ϯ 3.9% (n ϭ 4) in the presence of 500 M L-trans-PDC (Fig. 5A). This concentration of L-trans-PDC has been shown to potentiate glutamate receptor activity in hippocampal CA1 neurons (22). Fig. 5B shows the results from a representative experiment where isolated islets, in the absence of exogenous amino acids, were exposed to 2. completely block the AMPA receptor-activated currents in islet cells (1), and 500 M L-trans-PDC has no effect on the activation of islet glutamate receptors. 2 In the presence of 8.3 mM glucose and 500 M L-trans-PDC, insulin secretion increased significantly, 23.3 Ϯ 2.3% (n ϭ 3 separate islet preparations) (p ϭ 0.0099), over the level induced by the 8.3 mM glucose control which averaged 4.8 Ϯ 1.1 ng/ml (n ϭ 3). CNQX (10 M) completely blocked the potentiating effect of L-trans-PDC (75.6 Ϯ 10.8% of control, n ϭ 3), and furthermore, 10 M CNQX alone decreased the level of insulin secretion in the presence of 8.3 mM glucose by 15.9 Ϯ 5.9% (n ϭ 3). Neither L-trans-PDC nor CNQX had any effect on insulin secretion in the presence of 2.8 mM glucose. DISCUSSION We (1) and others (2) have shown that inotropic glutamate receptors are functionally expressed in pancreatic islets and that activation of these receptors may have implications for peptide hormone secretion (5, 6) and glucose tolerance (7). Glutamate receptors in the central nervous system rely on a high affinity uptake system to maintain low extracellular glutamate concentrations so that glutamate can act as a neurotransmitter. To elucidate the physiological relevance of glutamate receptors in pancreatic islets, we have investigated the presence of a glutamate uptake system in isolated, intact islets. The majority of islet [ 3 H]glutamate transport is Na ϩ -dependent, and the kinetic properties of this transport are similar to those described in the central nervous system. Because islet glutamate transporters also used D-aspartate as a substrate, it was possible to use anti-D-aspartate antibodies to localize islet glutamate/aspartate uptake to the non-␤ cells of the islet mantle. Antagonists of glutamate receptors and transporters had inhibitory and potentiating effects on insulin secretion, respectively, even in the absence of exogenous amino acids. Therefore, islets appear to contain an entire glutamatergic signaling system including activity-dependent release of glutamate, specific glutamate receptors, and a high affinity uptake system.
Manfras et al. (23) have detected glutamate transporter mRNA in human islets. However, attempts to measure glutamate uptake in islets have been focused on the possible role of glutamate as a metabolite or fuel secretagogue and not as a signal for cell to cell communication (8). The millimolar concentrations of glutamate used in this study were inappropriate to address the existence of glutamate uptake systems like those described in the central nervous system. Traditionally these transporters have been characterized based on their sensitivity to temperature, external Na ϩ concentration, and on their affinity for glutamate. The almost complete dependence of islet glutamate uptake on temperature suggests that the activity that we report was not the result of [ 3 H]glutamate binding to glutamate receptors. Islet glutamate uptake activity also displayed a dependence for external Na ϩ ; however, this dependence was incomplete (71.3 Ϯ 3.0%). The presence of a Na ϩindependent, Cl Ϫ -dependent, system X c Ϫ , has been reported in the periphery (13,14) and in neurons (15). In islets we observed a 6.1% decrease in [ 3 H]glutamate uptake in the absence of external chloride, and we found that cystine, an inhibitor of system X c Ϫ glutamate transport, substantially blocked the Na ϩindependent uptake activity. We chose not to characterize further the system X c Ϫ activity since it is believed to be primarily involved in cystine uptake and since the majority of the [H 3 ]glutamate transport activity in islets is Na ϩ -dependent.
The apparent K m for glutamate in islets (34.7 Ϯ 7.8 M) is within the range measured in preparations from the central nervous system as follows: 42 M in cerebellar granule cells, 58 and 66 M in cortical astrocytes, and 21 M in C6 glioma cells (24). The V max value that we report for islet glutamate transport 102.2 Ϯ 17.7 pmol/min/mg protein is also similar to those measured for synaptosomal glutamate transporters from the cerebellum (530 Ϯ 280 pmol/min/mg protein) and brain stem (570 Ϯ 290 pmol/min/mg protein) (25). Based on our immunohistochemical data, islet glutamate transporters are only found in 25-30% of islet cells. Therefore, our measurement of V max is an underestimation. Our data suggest that islet glutamate transporters, like their central nervous system counterparts, are capable of lowering extracellular glutamate concentrations to produce an environment favorable for glutamatergic signal transduction.
Most known plasma membrane glutamate transporters are also capable of transporting D-aspartate (26). This observation has provided a means to visualize uptake in fixed cells after incubation in the presence of D-aspartate using stereoselective anti-D-aspartate antibodies (12,17). Since D-aspartate is not normally found in, metabolized by, or released from most known cell types, it makes a convenient exogenous marker for glutamate transport activity. Islet L-[ 3 H]glutamate and D-[ 3 H]aspartate transport activities exhibited similar properties suggesting that the same transporter was responsible for both uptake activities. Incubation of islets in the presence of D-aspartate followed by fixation and probing with anti-D-aspartate antibodies revealed that islet D-aspartate (L-glutamate) transport was restricted to the islet mantle that consists primarily of ␣ cells in our preparation. Islets that were not incubated in the presence of D-aspartate prior to immunostaining failed to show reactivity with the D-aspartate antibodies. These data demonstrate that islet cells, like most other cells examined thus far, contain little or no detectable endogenous Daspartate. Furthermore, since the anti-D-aspartate antibodies used in this study do not generally detect concentrations of fixed D-aspartate below 500 M (17), the D-aspartate signal detected in the mantle cells after exposure of the islets to 100 M D-aspartate is due to the concentration of D-aspartate by these cells.
One of the chief roles of islets is to act as physiological glucose sensors. The mechanism by which ␤ cells sense and respond to serum glucose levels is linked to glucose metabolism. The coordinate regulation of peptide hormone secretion from other islet cell types is poorly understood. We have observed that D-glucose inhibits [ 3 H]glutamate uptake in the non-␤ cells of intact islets. Our results clearly show that the effects of glucose on glutamate uptake were not due to increased osmolarity in the presence of 16.7 mM glucose since the non-metabolizable L-glucose had no effect on islet glutamate uptake. These data suggest that changes in glutamate uptake activity in non-␤ cells are linked to the metabolism of glucose. Although the response of ␣ cells, the major mantle cell type in our preparation, to changes in glucose concentrations is poorly described, it has been suggested that GABA released from ␤ cells confers glucose sensitivity to ␣ cells (18,19). To test the hypothesis that GABA mediates the glucose effect on glutamate uptake, we used the GABA A receptor antagonist bicuculline to block ␣ cell GABA receptors. Blockade of these receptors had no effect on the glucose modulation of glutamate uptake. Therefore, non-␤ cells are either capable of linking glucose metabolism directly to changes in their physiology or effects of glucose in ␤ cells are being transduced to the cells of the islet mantle by an undescribed mechanism. In either case these observations represent a novel way that glucose metabolism may be related to changes in islet physiology.
The replots of glutamate uptake in the presence of 2.8 and 16.7 mM glucose revealed that increased glucose concentration resulted in a change in the apparent K m of glutamate transport. These data are consistent with a competitive mode of then exposed to 8.3 mM glucose alone or 8.3 mM glucose plus L-trans-PDC, CNQX, or both L-trans-PDC and CNQX. Bars represent the average of two 0.5-ml fractions collected for each of the experimental conditions.  E.). B, shown are the results of a representative experiment where Ϸ100 islets/experimental condition were exposed first to 2.8 mM glucose followed by 2.8 mM glucose alone or 2.8 mM glucose plus either 500 M L-trans-PDC, 10 M CNQX, or both 500 M L-trans-PDC and 10 M CNQX. The islets were inhibition. ␣ cells are known to possess phosphate-activated glutaminase (2), and this enzyme is a characteristic constituent of glutamate-releasing nerve endings in the central nervous system (27). Therefore, ␣ cells may be a possible source of releasable intra-islet glutamate. Glucose-inducible release of this glutamate may explain the apparent competitive inhibition of glutamate uptake and may provide a mechanism by which ␣ cells can communicate with ␤ cells through their AMPA-type glutamate receptors. Preliminary data obtained using high pressure liquid chromatography to detect released glutamate suggest that glucose does indeed increase the release of glutamate from isolated islets.
Other possible explanations for the observed glucose effect on glutamate uptake include glucose-mediated depolarization of non-␤ cells resulting in a decreased driving force for glutamate or modulation of transporter activity by posttranslational modification. In other systems, however, neither a decreased driving force for glutamate nor posttranslational modification have been shown to result in kinetics that resemble competitive inhibition (24, 28 -30). Finally, these observations could result from glucose-induced stimulation of non-␤ cell metabolism leading to the metabolism of glutamate and the release of the 3 H label outside of the cell.
The data reported by Bertrand and colleagues (5,7) in perfused pancreas preparations (5) and in vivo (7) have pointed to a possible role for glutamate in modulating peptide hormone secretion through its action at AMPA receptors. Their data do not address the source of glutamate acting at these receptors. Our data show that the blockade of glutamate uptake increased insulin secretion and confirm that AMPA receptors mediate the actions of glutamate. Since these experiments were performed in isolated islets in the absence of any exogenous amino acids, the effects on insulin secretion caused by L-trans-PDC and CNQX must be due to modulation of a glutamatergic signal that originates in the islets themselves. These observations are intriguing in light of the fact that while islets isolated from both the dorsal and ventral portions of the pancreas contain the same amount of insulin, dorsal islets have been reported to secrete 40 -50% more insulin in response to 16.7 mM glucose than ventral islets (31,32). Dorsal islets contain up to 25% ␣ cells, but ventral islets may only contain 5% ␣ cells. As mentioned above, ␣ cells, but not the other islet cell types, express phosphate-activated glutaminase. The presence of this enzyme in ␣ cells, taken together with our data that suggest islets release glutamate from an intra-islet source, provides a possible explanation for the discrepancy in the amount of glucosestimulated insulin secretion from dorsal and ventral islets.
Multiple components of a glutamatergic signaling pathway appear to exist in islets: 1) specific glutamate receptors, 2) a high affinity uptake system to control the levels of glutamate, and 3) suggestions based on uptake behavior that glutamate is released by relevant physiological stimuli. The precise role of a glutamatergic signaling system in islet physiology or pathology is not completely understood, but a number of possibilities present themselves. Glutamate may play a similar role to the one postulated by Rorsman et al. (18,19) for GABA in mediating communication between ␣ and ␤ cells. Glutamate may also subserve communication between islets and the central nervous system. Our demonstration of glucose-sensitive glutamate transport in the non-␤ cells of pancreatic islets adds to the growing evidence that glutamate is an important signaling molecule in this endocrine tissue.