Class A Calcium Channel Variants in Pancreatic Islets and Their Role in Insulin Secretion*

The initiation of insulin release from rat islet β cells relies, in large part, on calcium influx through dihydropyridine-sensitive (α1D) voltage-gated calcium channels. Components of calcium-dependent insulin secretion and whole cell calcium current, however, are resistant toL-type channel blockade, as well as to ω-conotoxin GVIA, a potent inhibitor of α1B channels, suggesting the expression of additional exocytotic calcium channels in the islet. We used a reverse transcription-polymerase chain reaction-based strategy to ascertain at the molecular level whether the α1Acalcium channel isoform was also present. Results revealed two new variants of the rat brain α1A channel in the islet with divergence in a putative extracellular domain and in the carboxyl terminus. Using antibodies and cRNA probes specific for α1A channels, we found that the majority of cells in rat pancreatic islets were labeled, indicating expression of the α1A channels in β cells, the predominant islet cell type. Electrophysiologic recording from isolated islet cells demonstrated that the dihydropyridine-resistant current was sensitive to the α1A channel blocker, ω-agatoxin IVA. This toxin also inhibited the dihydropyridine-resistant component of glucose-stimulated insulin secretion, suggesting functional overlap among calcium channel classes. These findings confirm the presence of multiple high voltage-activated calcium channels in the rat islet and implicate a physiologic role for α1A channels in excitation-secretion coupling in β cells.

The initiation of insulin release from rat islet ␤ cells relies, in large part, on calcium influx through dihydropyridine-sensitive (␣ 1D ) voltage-gated calcium channels. Components of calcium-dependent insulin secretion and whole cell calcium current, however, are resistant to L-type channel blockade, as well as to -conotoxin GVIA, a potent inhibitor of ␣ 1B channels, suggesting the expression of additional exocytotic calcium channels in the islet. We used a reverse transcription-polymerase chain reaction-based strategy to ascertain at the molecular level whether the ␣ 1A calcium channel isoform was also present. Results revealed two new variants of the rat brain ␣ 1A channel in the islet with divergence in a putative extracellular domain and in the carboxyl terminus. Using antibodies and cRNA probes specific for ␣ 1A channels, we found that the majority of cells in rat pancreatic islets were labeled, indicating expression of the ␣ 1A channels in ␤ cells, the predominant islet cell type. Electrophysiologic recording from isolated islet cells demonstrated that the dihydropyridine-resistant current was sensitive to the ␣ 1A channel blocker, -agatoxin IVA. This toxin also inhibited the dihydropyridineresistant component of glucose-stimulated insulin secretion, suggesting functional overlap among calcium channel classes. These findings confirm the presence of multiple high voltage-activated calcium channels in the rat islet and implicate a physiologic role for ␣ 1A channels in excitation-secretion coupling in ␤ cells.
The metabolism of glucose in ␤ cells is linked to membrane excitation through increases in the components that influence the ATP/ADP ratio (1,2). A local increase in ATP relative to ADP inhibits the ATP-sensitive K ϩ (K ATP ) channel, giving rise to oscillations in K ATP permeability and consequent fluctuations in membrane potential (3,4). ATP-induced depolarizations evoke insulin release through the activation of voltagedependent calcium channels, promoting calcium influx (3)(4)(5)(6). If extracellular calcium is eliminated, glucose-stimulated insulin secretion is abolished, highlighting the important role of cal-cium channels in insulin homeostasis (7)(8)(9)(10)(11).
Of the six genes (A-E, S) encoding the pore-forming ␣ 1 subunits of high voltage-activated calcium channels (12), all are capable of coupling to exocytotic machinery, although differences in tissue distribution and efficacy with which they stimulate secretion vary among the channel classes (for review, see Ref. 13). Pharmacological and heterologous expression studies have identified selective antagonists for certain of these: -agatoxin IV blocks ␣ 1A , -conotoxin GVIA blocks ␣ 1B , and 1,4dihydropyridines block ␣ 1C , ␣ 1D , and ␣ 1S (13). No selective antagonist has been identified to date for ␣ 1E calcium channels, although they are sensitive to nonselective calcium channel antagonists (e.g. -grammotoxin SIA and cadmium).
Previous work on pancreatic ␤ cells demonstrated that calcium influx and depolarization-evoked insulin release are blocked 60-80% by inhibitors of ␣ 1D channels (14)(15)(16). In addition, -conotoxin GVIA blocks a portion of calcium current (17) and 27% of the second phase insulin secretion from rat pancreatic ␤ cells under conditions of maximal glucose stimulation (18). In the presence of both -conotoxin GVIA and the dihydropyridine antagonist nifedipine, 25% of Ca 2ϩ -dependent insulin secretion remains, suggesting the involvement of another calcium channel type. Because ␣ 1A and ␣ 1B channels are known to colocalize within mammalian central nervous system nerve terminals and play prominent roles in neurosecretion (19)(20)(21)(22), we sought to determine whether the ␣ 1A isoform was also present in ␤ cells and responsible for the current and insulin release that are resistant to dihydropyridines and -conotoxin GVIA.
Here we report the full-length sequences of two unique ␣ 1A splice variants cloned from rat pancreatic islets. We demonstrate their expression in ␤ cells and provide an initial characterization of their pharmacologic and electrophysiologic properties using tight seal whole cell recording. These channels play a role in calcium-dependent insulin secretion. Results may have significant implications for understanding, and perhaps treating, ␤ cell disorders such as diabetes and hyperinsulinemia.

MATERIALS AND METHODS
Islet Isolation-Islets were isolated from male, 200 -225 gm, Spraque-Dawley rats. Animals were anesthetized (Nembutol 50 mg/kg), and the pancreatic bile duct exposed, cannulated, and infused with 2 mg/ml collagenase (Boehringer Mannheim) in medium M199 (Life Technologies, Inc., with 5.6 gm/l HEPES, 2.2 gm/l NaHCO 3 ). The infused pancreas was removed and incubated in a 50-ml tube submerged in a 37°C water bath for 17 min. Tubes were shaken briefly to dissociate tissue, the digestion reaction was stopped by the addition of cooled M199 with 10% newborn calf serum (Life Technologies, Inc.), and the tissue was shaken again to further dissociate while washing free of collagenase. The tissue was spun for 30 s (speed 3) in an ICN table-top clinical centrifuge and washed in fresh medium M199 with serum. This procedure was repeated twice to remove collagenase. After a final spin (1 min. at speed 3) to pellet the tissue, the medium was discarded, and the tissue was suspended in 10 ml of Histopaque (Sigma) and overlaid with 10 ml of M199 (without serum). The tissue was centrifuged in this gradient at 900 ϫ g in a swinging bucket rotor at 4°C for 20 min. The supernatant containing the islets was removed and centrifuged (speed 4, ICN centrifuge) to pellet the islets. The supernatant was discarded and the islets washed with 10 ml of M199 (with serum) twice before use. They were used intact for insulin release assays, further dissociated into single cells for electrophysiology, or homogenized for isolation of RNA.
Cloning-Primers were designed along the full-length of the rat brain class A channel (rbA-1, GenBank TM accession no. M64373). Regions of high homology to the rat brain ␣ 1B channel (rbB) were avoided. Sense and antisense oligonucleotides ranged in length from 18 to 24 bases beginning at the following 5Ј nucleotides (nt) 1 of the rbA-1 sequence: Ϫ20, 540, 1265, 1610, 2091, 2484, 2879, 3309, 3658, 4104, 4620, 5170, 5607, 5961, 6527, 7010, and 7125. Total RNA purified from homogenized rat islets (Trizol, Life Technologies, Inc.) was DNase treated (Life Technologies, Inc., amplification grade DNase) and reverse-transcribed for 30 min at 42°C (reverse transcriptase, Perkin Elmer). GC rich regions were reverse transcribed at 50°C (Superscript, Life Technologies, Inc. or avian myeloblastosis virus reverse transcriptase, Amersham Pharmacia Biotech) and 10% glycerol was used in the PCR amplification. At least two separate PCR amplifications (Taq polymerase, Perkin Elmer, Amersham Pharmacia Biotech) were performed for each primer set and 2 sense and 2 antisense strands sequenced (ABI, fluorescent-labeled automated sequencing) to ensure accuracy of sequence. To verify regions detected to diverge from rbA-1 over a ϳ2-kilobase region, reverse transcription of islet total RNA was performed at 50°C for 1 h using avian myeloblastosis virus reverse transcriptase (Amersham Pharmacia Biotech) and Superscript (Life Technologies, Inc.) to generate longer products over GC rich regions. For this "long" PCR, primers at nt 4620 and from the rbA-1 3Ј UT region were used and generated two ϳ2 kilobases alternatively spliced carboxyl-terminal fragments as described below.
Generation of cRNA Probe-Primers were designed to yield a 587-bp product from the 3Ј-end (5961-6548) of coding sequence of rbA-1 (23). Freshly isolated islet RNA for reverse transcription (Perkin Elmer reverse transcriptase), and Taq polymerase (Perkin Elmer) were used for cDNA amplification. Control samples, in which reverse transcriptase was omitted, were included to ensure that products had not been amplified from genomic DNA. Because the region contained a high percentage of GC-rich sequence, 10% glycerol was added to the reaction mix. The annealing temperature was 60°C.
In Situ Hybridization-The original 587-bp PCR product (nt 5961-6548) above was subcloned (vector pCR I, Invitrogen) and sequenced in both directions. Comparison of the sequences revealed 100% nucleotide identity to rbA-1. Using this subclone as a template, digoxigenin-labeled (Boehringer Mannheim) sense and antisense RNA probes were synthesized for in situ hybridization studies of rat pancreas. Paraffin sections were mounted on slides, cleared, and hydrated before treatment with proteinase K. This was followed by equilibration in 0.1 M triethanolamine, pH 8.0, to which acetic anhydride had been added. The probe was added at 4 ng/l to the hybridization mix. Sections were incubated overnight at 60°C in a humid chamber. Post hybridization, the slides were soaked in 2ϫ SSC (300 mM sodium chloride, 30 mM sodium citrate), gently agitated in warm 50% formamide in 2ϫ SSC, and transferred to 60°C. The sections were then treated with RNase A buffer (500 mM NaCl, 10 mM Tris, 1 mM EDTA, and RNase A at 20 g/ml) to remove unhybridized probe. Slides were washed in 0.1ϫ SSC at 65°C. For color detection of labeled cells (Boehringer Mannheim Genius System), slides were blocked, washed, and incubated overnight at 4°C with 200 -500 ml of antidigoxigenin-alkaline phosphatase. The subsequent chromagen substrate (Sigma) consisted of nitroblue tetrazolium salt and X-phos (bromo-4-chloro-3-indoyl phosphate toluidinium salt).
Whole Cell Recording-Islets were isolated as described above, dissociated in trypsin/EDTA (0.05% trypsin, 0.53 mM EDTA) and cultured 1-3 days in M199 (with serum) prior to the experiments. Voltage-de-pendent calcium currents were examined in the whole cell configuration using pipettes fabricated from nonheparinized soda lime glass (VWR, 73811). Pipette resistances averaged ϳ2 megaohms with an internal solution containing 120 mM N-methyl-D-glucamine, 20 mM tetraethylammonium-OH, 10 mM HEPES, 11 mM EGTA, 1 mM CaCl 2 , and 4 mM MgATP (pH 7.0) and an external solution containing 10 mM BaCl 2 , 120 mM NaCl, 20 mM triethanolamine, and 1.2 mM MgCl 2 , 10 mM Na-HEPES, 1 M tetrodotoxin, 1 M nimodipine (pH 7.4). -Agatoxin IVA was applied by pressure ejection from a wide-bore (ϳ2-3 m) pipette positioned within 20 m of the cell. The toxin was stored at Ϫ40°C in 10-l samples of 10 mM in distilled deionized water and diluted to the final working concentration in external solution immediately before use.
Insulin Secretion Evaluated by RIA-Islets were isolated from exocrine pancreas as described above. They were selected for uniform size (150 -200 m), washed in Krebs-Ringer buffer (KRB, equilibrated with O 2 for 30 min), evenly divided into KRB with 2.8 mM glucose (with or without 500 nM -agatoxin IVA) and allowed to recover from isolation. The islets were incubated at 37°C (95% O 2 , 5% CO 2 ) during equilibration. After 90 min of preincubation in low glucose KRB, the islets were divided into groups of 10 islets/vial in 0.5 ml of fresh KRB, three vials for each condition (2.8 or 11 mM glucose with or without -agatoxin IVA). All samples were incubated in a 37°C water bath for 1 h to promote glucose-stimulated insulin secretion. Samples were centrifuged at 700 rpm, and 300 l of supernatant was removed and viewed under a dissecting microscope to ensure no islet tissue was present. This sample was assayed for insulin content using ICN insulin radioimmunoassay reagents. Briefly, 100 l of each sample or standards were added to duplicate anti-insulin antibody-coated tubes followed by 900 l of 125 I-labeled insulin diluted into buffer. In addition, control samples of toxin alone were included to ensure that the -agatoxin IVA did not interfere with insulin antibody binding. The samples and standards were incubated for 18 h at room temperature. The supernatant was removed, and tubes were washed with double-distilled H 2 O and counted in a gamma counter. Sample counts were compared with the standard curve for determination of insulin concentrations. Results are reported as mean Ϯ S.E. of the mean, and statistical significance was evaluated by Student's t test.

RESULTS
Cloning of Islet ␣ 1A Splice Variants-Given our goal of identifying the class A (␣ 1A ) variants in pancreatic islets, a PCR strategy was followed using primers designed along the fulllength of the cloned ␣ 1A sequence from rat brain, rbA-1 (23,24). Each product was generated twice, and both the sense and antisense strands were sequenced to ensure accuracy. Reverse transcriptase was eliminated from control samples to ensure that amplification products were from islet mRNA and not of genomic origin. The entire coding sequence was determined. From the 5Ј-UT to base 4806, the sequence was 100% identical to the type "a" splice variant (␣ 1A-a ) of rbA-1 (25). At nt 4807, in a putative extracellular domain, there is a six base in-frame insertion in both variants described below cloned from rat islets. This insertion results in the addition of amino acids asparagine and proline (Figs. 1A and 2), as previously reported for ␣ 1A in rat kidney cortex (26) and ␣ 1A-b in rat brain (25). In addition, two carboxyl-terminal variants are present in the islet that have not been reported elsewhere. One, riA-1, diverges over the bases 5385-5477, altering 10 amino acids throughout the region (Figs. 1 and 2). Further 3Ј, the bases corresponding to 6160 -6195 (in rbA-1) are deleted in riA-1. The riA-1 variant is also significantly longer than rbA-1 due to a 5-base (GCCAG) insertion before the stop codon, resulting in a frameshift (Figs. 1 and 2). This same insertion has also been described in human brain ␣ 1A channels (27). The contiguity of the observed variations in riA-1 was confirmed by generating and sequencing Ͼ2 kilobase fragments using primers at 4620 and in the 3Ј-untranslated region. Although combinations of these carboxyl-terminal modifications have been described in human brain partial cDNA clones (27), riA-1 is unique among known full-length clones.
In the other product, riA-2, the reading frame and stop codon of rbA-1 are maintained, but bases 4922-4987 and 5256 -5480 are deleted (Fig. 1). The latter deletion corresponds to the region of nucleotide divergence found in riA-1 and is homologous to the proposed calcium-binding, EF-hand motif in ␣ 1C (25,28). Thus, rat pancreatic islets appear to express two unique forms of class A channel. This represents the first molecular evidence of class A channels in islets. We have explored their expression and localization at the levels of RNA and protein, provided an initial description of their pharmacology and electrophysiology, and evaluated their role in insulin secretion. Cellular Localization of Rat Islet ␣ 1A Channels-In situ hybridization methods were employed to examine the cellular localization of ␣ 1A mRNA in islets. A riboprobe was synthesized by PCR of reverse transcribed islet total RNA with class A calcium channel primers, generating a 587-bp carboxyl-terminal fragment (bp 5961-6548). No product was obtained if reverse transcriptase was omitted. This fragment, 100% identical to rbA-1 (23), was subcloned and labeled with digoxigenin-UTP to generate both sense and antisense RNA probes. The majority of the cells, representing the islet core, were labeled with the antisense probe. This demonstrates that the ␣ 1A calcium channel mRNA is present in ␤ cells, which comprise the core and ϳ80% of islet cells (Fig. 3A). Islets remained unlabeled when the digoxigenin-sense probe was used (Fig. 3B).
To determine whether ␣ 1A channel protein was also present in ␤ cells, immunohistochemistry was employed with an antipeptide antibody, CNA-1 (29). This antibody recognizes amino acids 865-881 of rbA-1 in the putative second intracellular loop. As the islet clones code for identical amino acids in this domain, we used CNA-1 to detect the expression of class A channels in islets. Intense staining was observed on the majority (Ͼ75%) of cells in the core of the islet, again indicating the presence of class A channel protein in ␤ cells (Fig. 4, A and B). No staining was seen in control samples in which CNA-1 anti-body had been omitted (data not shown).
Whole Cell Recording of ␤ Cell Calcium Channel Currents-To test whether the rat islet class A transcripts give rise to functional channels in ␤ cells, we recorded from dissociated islet cells using tight seal whole cell methods. Calcium channel currents were isolated with standard intra-and extracellular solutions (30,31). Nimodipine (1 M) was added to all extracellular solutions in order to suppress dihydropyridine-sensitive (class C/D) currents. Ba 2ϩ (10 mM) was used as a charge carrier to enhance calcium channel currents over those observed in 1.2 mM Ca 2ϩ (the normal extracellular concentration). ␤ cells could be identified by the presence of small processes, as has been reported elsewhere (32).
Following whole cell access, the cells were held at Ϫ80 mV, and 20 ms rectangular test pulses were applied. Inward Ba 2ϩ current activated near Ϫ30 mV and was maximal at ϩ10 mV. Peak currents (at ϩ10 mV) varied from 50 to 400 pA. Activation was rapid (reaching peak in ϳ5 ms at 0 mV), inactivation was slow (with no decay in 20 ms), and deactivation was rapid (complete in ϳ300 s). In these ways, the current resembled ␣ 1A currents expressed in other cell types (33,34).
To determine whether these currents were, in fact, generated through class A calcium channels, we applied -agatoxin IVA (the selective class A channel antagonist). In 8 of 12 cells tested, -agatoxin IVA (1 M) produced a significant reduction in inward current (Fig. 5, A and C). Inhibition ranged from 15 to 80% in the responsive cells with an average reduction of 48.4 Ϯ 8.5%. The onset of toxin-induced blockade was slow ( ϭ 62.5 s) even at 1 M; the dissociation rate was also slow with no recovery in 3 min (Fig. 5B). These results are consistent with the action of -agatoxin IVA on class A calcium currents in other cell types (35)(36)(37) and demonstrate the presence of functional class A channels in ␤ cells. In addition, the percentage of responsive cells in the sample tested electrophysiologically is similar to that of the islet cells shown to express ␣ 1A mRNA and protein (Figs. 3 and 4).
A Portion of Insulin Secretion Is Inhibited by -Agatoxin IVA-The presence of significant class A current in ␤ cells suggested the possibility that these channels play a role in exocytosis. To test this, we employed a radioimmunoassay to measure glucose-stimulated insulin release from islets, and evaluated the involvement of class A calcium channels using -agatoxin IVA. Islets were equilibrated for 90 min at 37°C in KRB Ϯ 500 nM -agatoxin IVA. Addition of 11 mM glucose evoked insulin release to a level of 1127% of that seen in basal glucose (2.8 mM) during a 60-min incubation period. -Agatoxin IVA (500 nM) diminished this glucose-stimulated insulin release to 811% of basal secretion (Fig. 6A). The inclusion of nimodipine (1 M) in all solutions to block dihydropyridinesensitive calcium channels reduced 11 mM glucose-evoked in-sulin release to a level 140% of that seen in basal glucose (2.8 mM). Thus, a portion of glucose-stimulated insulin release is dihydropyridine-resistant. All of the latter is mediated by Ca 2ϩ influx through class A channels, since it is eliminated by 500 nM -agatoxin IVA (Fig. 6B). Control samples without islets, with -agatoxin IVA alone, had no counts above background, indicating the toxin does not interact with anti-insulin antibodies. These results demonstrate that class A calcium channels are effective co-regulators of insulin secretion.

DISCUSSION
Our results demonstrate two new variants of the class A calcium channel in pancreatic islets and establish the presence of both mRNA and protein in ␤ cells. We have begun to characterize the electrophysiology and demonstrate that currents through these channels resemble currents recorded through class A channels in other cells (33)(34)(35)(36)(37)(38). Additionally, we show that, as in the nervous system, calcium influx through islet class A channels is coupled to exocytosis. Despite identification of several ␣ 1A splice variants and missense mutations (described below), little is known about the biophysical properties conferred by these changes. With the molecular information described here, we can begin to ascertain how specific amino acid changes influence electrical and functional properties of ␣ 1A channels.
Multiple ␣ 1A Isoforms-Certain of the variable regions reported here occur in clones described elsewhere, but the islet clones as a whole are unique. Zhuchenko et al. (27) demonstrated five partial human clones from brain containing the GCCAG insertion, producing a frameshift and a longer carboxyl terminus. Unlike the islet clones described here, the human variants contain a CAG repeat in the extended carboxyl terminus that is responsible for a form of familial ataxia, SCA6 (spinocerebellar ataxia). This repeat is not expressed in the riA-1 carboxyl terminus (Fig. 2). Two partial clones from human had the 36-base pair deletion, and two others had the 94-base pair divergent sequence present in riA-1 (27). In addition, the 6-base insertion at nt 4607 in both islet isoforms has been described in a ␣ 1A PCR cDNA fragment from rat kidney (26) and in rat brain ␣ 1A-b (25). The two deletions in riA-2 (nt 4922-4987 and 5256 -5480) have not been described previously. Interestingly, the deletion beginning at nt 5256 in riA-2 corresponds to the 94-bp COOH-terminal region of divergence found in riA-1 and in two of the partial human clones. This region has homology to EF hand motifs in dihydropyridinesensitive calcium channels demonstrated to bind calcium and promote inactivation (25,28), but also see Ref. 39. Rabbit BI-I ␣ 1A calcium channels are 92% identical to riA-1 but differ substantially in the intracellular loop between repeats II and III. Of note, there are also several 3 and 6 base insertions in BI-I ␣ 1A isoform not seen in riA-1 (40).
The functional consequences of these structural variations in ␣ 1A are largely unexplored, but even small changes in sequence can lead to significant biophysical differences among the channel variants. For example, four missense mutations in a human ␣ 1A calcium channel gene are responsible for hemiplegic migraine (41) and a single nonconservative amino acid substitution in a mouse ␣ 1A gene causes seizures and ataxia in the tottering mutant (42). The alterations in calcium channel function produced by these mutations, as well as by the structural variations described for the islet class A calcium channels described here, remain to be examined. Work on voltage-dependent sodium channels suggests that such studies on calcium channels will be fruitful. For example, single amino acid substitutions in sodium channels have dramatic effects on inactivation, giving rise to such disorders as paramyotonia congenita and hyperkalemic periodic paralysis (43)(44)(45). Pharmacological properties of class A channels differ among FIG. 6. Class A channels regulate insulin secretion. A, glucosestimulated insulin release measured by radioimmunoassay in control cells or cells exposed to 500 nM -agatoxin IVA (as marked). Release in 11 mM glucose normalized to that in 2. the variants. Zamponi et al. (25) reported that single amino acid differences in the I-II linker regions of ␣ 1A-a , ␣ 1A-b , and ␣ 1A-c bring about alterations in channel blockade by local anesthetic and antipsychotic drugs. Such naturally occurring structural changes may explain some of the variability reported for -agatoxin IVA-sensitive currents in different cell types or when different ␣ 1A isotypes are expressed in heterologous systems (13, 46 -48). Estimates of -agatoxin IVA binding affinity also vary, ranging from the low nM (for cerebellar Purkinje neurons) (35) to high nM for class A channels in Xenopus oocytes (46). Unique pharmacological phenotypes among the variants offer a means to test their differential involvement in calcium-dependent physiological processes and to determine the consequences of the structural changes for functional phenotype.
Heterogeneous ␤ Cell Responses-The density of -agatoxin IVA-sensitive current in ␤ cells varied considerably from cell to cell. The majority of the cells expressed detectable toxin-sensitive currents (as well as ␣ 1A mRNA and protein detected by in situ and immunochemical methods); in those sensitive to toxin blockade, the inhibition varied from 15 to 80%. As -agatoxin IVA has been demonstrated to inhibit a variety of ␣ 1A calcium channels, both in primary and heterologous cells, the heterogeneity found for the ␤ cells likely reflects variations in expression levels for both class A channel types. Quantitating this relationship will require single-cell comparisons between the mRNA or protein levels for the calcium channel subunits and amplitude of -agatoxin IVA-sensitive currents. In undifferentiated PC12 cells, both the sensitivity of the calcium current to -agatoxin-IVA and the rate of toxin-induced inhibition vary from cell to cell, as observed for neurons (37,38), raising the possibility that pharmacological differences between class A channel variants might underlie the heterogeneity of ␤ cell responses to -agatoxin IVA.
Our results suggest the possibility that variations among ␤ cells may allow them to perform different roles within the context of integrated islet function. In this way, our results support those of Moitoso de Vargas et al. (49), which demonstrate that glucose-stimulated activation of insulin gene expression varies among ␤ cells in intact islets. Such differences signal the presence of functional subsets of ␤ cells that are differentially active as physiological conditions vary. It will be important for future work to test such ideas.
Class A Channels in Other Tissues-Exocytosis from mammalian central neurons relies heavily on class A and B channels and very little on dihydropyridine-sensitive class C/D channels (13). By contrast, secretion from neuroendocrine cells is largely dihydropyridine-sensitive. Our results demonstrate that such differences are not absolute. That is, although the majority of insulin release is dependent on dihydropyridinesensitive channels, a portion is mediated through class A channels. Class A and B channels have been demonstrated by pharmacology and electrophysiology to be present in adrenal chromaffin cells, (50 -52), carcinoid cells of the gut (53), pituitary corticotropes (54), the pituitary cell line AtT20 (55), GH4C1 pituitary cells (56), and in RINm5f and HIT insulinoma cells (57,58). Dihydropyridine-resistant channels play a role in exocytosis for certain of these cells (e.g. carcinoid and HIT cells) (51, 58) but not all. Release of adrenocorticotrophic hormone from AtT20 cells, for example, is evoked entirely through dihydropyridine-sensitive channels, despite the fact that Ͼ50% of the calcium current in the cells is resistant to blockade by dihydropyridines (55). The mechanisms that determine specificity of coupling between calcium channels and the exocytotic machinery remain to be defined.
Significance of ␣ 1A Calcium Channels in the Islet-Recent statistical analyses suggest that more than 15 million people in the United States suffer from diabetes (59). A population-based retrospective study indicates the age-adjusted prevalence of diabetes cases rose 65% for men and 37% for women between 1970 and 1990 (60), and a "massive increase" in the prevalence of noninsulin-dependent diabetes is predicted globally, as countries succumb to the "Westernization" of diet and activity patterns (61). It is important, therefore, to continue to expand our understanding of the mechanisms by which the ␤ cell controls the secretion of insulin. Knowing the molecular structure of the ␣ 1A isoforms present in the islet and the roles that they play within the context of the functioning islet may allow the development of clinically useful agents to modulate calcium entry into ␤ cells through these channels, providing another tool to regulate insulin release. In addition, defining the entire sequence of splice variants will facilitate investigations of the ways in which changes in primary structure alter the pharmacology, biophysical, and/or exocytotic properties of class A voltage-gated calcium channels.