Expression and Function of Pancreatic β-Cell Delayed Rectifier K+Channels

Voltage-dependent delayed rectifier K+ channels regulate aspects of both stimulus-secretion and excitation-contraction coupling, but assigning specific roles to these channels has proved problematic. Using transgenically derived insulinoma cells (βTC3-neo) and β-cells purified from rodent pancreatic islets of Langerhans, we studied the expression and role of delayed rectifiers in glucose-stimulated insulin secretion. Using reverse-transcription polymerase chain reaction methods to amplify all known candidate delayed rectifier transcripts, the expression of the K+ channel gene Kv2.1 in βTC3-neo insulinoma cells and purified rodent pancreatic β-cells was detected and confirmed by immunoblotting in the insulinoma cells. βTC3-neo cells were also found to express a related K+ channel, Kv3.2. Whole-cell patch clamp demonstrated the presence of delayed rectifier K+ currents inhibited by tetraethylammonium (TEA) and 4-aminopyridine, with similar Kd values to that of Kv2.1, correlating delayed rectifier gene expression with the K+ currents. The effect of these blockers on intracellular Ca2+ concentration ([Ca2+]i) was studied with fura-2 microspectrofluorimetry and imaging techniques. In the absence of glucose, exposure to TEA (1-20 mM) had minimal effects on βTC3-neo or rodent islet [Ca2+]i, but in the presence of glucose, TEA activated large amplitude [Ca2+]i oscillations. In the insulinoma cells the TEA-induced [Ca2+]i oscillations were driven by synchronous oscillations in membrane potential, resulting in a 4-fold potentiation of insulin secretion. Activation of specific delayed rectifier K+ channels can therefore suppress stimulus-secretion coupling by damping oscillations in membrane potential and [Ca2+]i and thereby regulate secretion. These studies implicate previously uncharacterized β-cell delayed rectifier K+ channels in the regulation of membrane repolarization, [Ca2+]i, and insulin secretion.

Glucose-stimulated insulin secretion is initiated via induc-tion of intracellular calcium ([Ca 2ϩ ] i ) 1 transients, regulated in part by a mechanism that couples nutrient metabolism to the membrane potential (1)(2)(3). Increases in cytosolic ATP, derived from the glycolytic reduction of NAD ϩ , results in the block of ATP-sensitive K ϩ channels (K-ATP) (4,5). The resultant membrane depolarization, characterized by bursts of action potentials superimposed on slower oscillations (6,7), activates voltagedependent L-type Ca 2ϩ channels and triggers intracellular Ca 2ϩ release, elevating [Ca 2ϩ ] i and initiating insulin secretion (8 -11). These oscillatory depolarizations are regulated by the interplay of several ion channel types, including various K ϩ channels and a calcium store depletion-activated cation channel (12). The K ϩ currents known to exist in ␤-cells include K-ATP, calcium-activated K ϩ channels, inwardly rectifying K ϩ channels, and delayed rectifier K ϩ channels (2). Previous studies on ␤-cell outward K ϩ currents have demonstrated that they are predominantly of the delayed rectifier type and are sensitive to tetraethylammonium (TEA) with K d values of 1-5 mM (2,(13)(14)(15). It has been suggested that in rodent ␤-cells membrane repolarization results from the opening of delayed rectifier K ϩ channels (16). TEA blockade of K ϩ channels has been shown to increase glucose-dependent electrical activity and insulin secretion in mouse islets (17). Despite their importance in the regulation of ␤-cell signal transduction, detailed knowledge of K ϩ channel isoform expression in ␤-cells is incomplete. Of the 25 K ϩ channel-related families that have been reported, only a subset encode membrane proteins that produce voltagesensitive delayed rectifier outward K ϩ currents similar to those found in ␤-cells (18). They have been named according to the single Drosophila gene locus they are most similar to Kv1.X (Shaker), Kv2.X (Shab), Kv3.X (Shaw), and Kv4.X (Shal) (19). Only the expression of the Kv1.X family has been reported in insulin secreting cells (20), and no previous studies examining K ϩ channel expression have employed purified normal ␤-cells or well-differentiated insulinoma cells for this purpose. Here we report the expression in ␤TC3 insulinoma cells and purified rat and mouse ␤-cells of Kv2.1 and Kv3.2-related transcripts that encode TEA-sensitive delayed rectifier K ϩ channels. Furthermore, we correlate block of these channels with potentiation of glucose-stimulated oscillations in membrane potential and [Ca 2ϩ ] i and augmentation of insulin secretion. These studies indicate that the activation, following nutrient stimulation, of hitherto uncharacterized ␤-cell delayed rectifier K ϩ channels produce membrane repolarization, lowering of [Ca 2ϩ ] i and thereby dampen insulin secretion.

EXPERIMENTAL PROCEDURES
Growth of Insulinoma Cells-␤TC3 cells were cultured essentially as originally described (21). A clonal subline (␤TC3-neo) was isolated after transfection with pSV2-neo by electroporation (22), and this line was maintained in the continued presence of 1 mg/ml G418 (Life Technologies, Inc.). Cells were seeded for 4 -6 days in Dulbecco's modified Eagle's medium supplemented with 15% normal horse serum, 2.5% fetal calf serum, 50 units/ml penicillin, and 50 g/ml streptomycin. Eighteen hours prior to experiments, medium was changed to RPMI 1640 containing 15% dialyzed horse serum, 2.5% dialyzed fetal calf serum, 1 mM glucose, and antibiotics as above.
Fluorescence-activated Cell Sorting of Rodent ␤-Cells-Fluorescenceactivated cell sorting (FACS) was employed to isolate ␤-cells essentially as described (24). Islets were dissociated in Ca 2ϩ -free Earle's/HEPES medium (EH, Ref. 24) with deoxyribonuclease and trypsin with gentle trituration. The recovered single cell pellet was resuspended in EH containing 1 mM EGTA and introduced into the sample stream of a flow cytometer (Becton Dickinson FACStarPlus) equipped with an argon ion laser in the primary light path (90 mW at 488 nm, Ion Laser Technology, Salt Lake City, UT). Events were analyzed and collected on a forward scatter trigger. Signals and side scatter were displayed and collected via a 530/30 nm band-pass filter following transit through a 560 nm short-pass dichroic filter. The integrity of the cell preparation was checked by propidium iodide exclusion. The propidium iodide fluorescence signal was collected by transit of a 660/20 nm band-pass filter receiving signal from a 610 nm shortpass dichroic filter. Rectangular sort regions were defined by producing Cartesian plots displaying forward light scatter versus autofluorescence to delineate the ␤-cell population. The processing of the fluorescence signals in the linear mode aided in the resolution of two major cell populations; ␤-cells could be distinguished as a set of cells with elevated green fluorescence, elevated forward and side scatter relative to other cells in the preparation. The ␤-cell populations were sorted in two-drop packets into sterile glass collection tubes containing 2 ml of fetal bovine serum. All cell events with values outside of the sort region were diverted to a separate collection tube. The cytometer target fluorescence values were standardized using glutaraldehyde-fixed chicken red blood cells immediately before and after islet cell sample batches were run. In addition, the sorting efficiency of the instrument was evaluated using a mixture of labeled microspheres in the performance of a "test sort." Reverse Transcriptase-Polymerase Chain Reaction Amplification (RT-PCR)-␤TC3-neo and mouse brain poly(A) ϩ RNA was prepared using an oligo-dT binding method (Quickprep micro mRNA kit, Pharmacia Biotech Inc.), and first strand was cDNA reverse-transcribed using random primers (Superscript Plus Transcriptase, Life Technologies, Inc.). The polymerase chain reaction (PCR) was then employed (typical protocol: 94°C, 3 min; then 94°C, 1 min, 52°C, 30 s, 72°C, 30 s for 30 cycles) for each of the primer pairs shown in Table I. In some experiments, the annealing temperature was reduced to 48°C. Amplicons were excised after agarose gel electrophoresis and ligated into pCRII vector (Invitrogen) for sequencing via the dideoxy chain termination method (manually, using Sequenase kit (USB), or automated, using fluorescently labeled oligonucleotide primers (ABI)). The best match of the sequences was determined using the fasta and gap programs (Genetics Computer Group, Madison, WI).

Measurement of [Ca 2ϩ ] i and Membrane
Potential-Cells were loaded with fura-2 for 25 min at 37°C in KRB supplemented with 5 M acetoxymethyl ester of fura-2 (Molecular Probes Inc.). [Ca 2ϩ ] i was estimated as described elsewhere (9). For simultaneous membrane potential and [Ca 2ϩ ] i measurements, single ␤TC3-neo cells, pre-loaded with fura-2, were current-clamped using perforated patch techniques as described (5). In some experiments (Table II), dual wavelength digitized video fluorescent microscopy with fura-2 in single ␤TC3-neo cells was performed using an intensified charge-coupled device (Hammatsu C2400) and Metafluor imaging software (Universal Imaging).
Insulin Secretion Measurements-␤TC3 cells were plated at a density of 25 ϫ 10 4 /cm 2 and cultured overnight in RPMI 1640 containing 15% horse serum, 2.5% fetal calf serum, 1 mM glucose, and antibiotics (penicillin/streptomycin solution, Life Technologies, Inc.). Insulin secretion measurements were made in the presence of 0 or 1 mM glucose and increasing concentrations of TEA or 4-AP for 1 h. Insulin concentration was determined by SPA assay kit (Amersham Corp.) and calibrated using rat insulin as standard.

RESULTS AND DISCUSSION
While delayed rectifier K ϩ channels are known to regulate the membrane potential of electrically excitable cells (26), detailed knowledge of K ϩ channel isoform expression and function in insulin-secreting ␤-cells is incomplete. A previous study, performed prior to the identification of multiple families of voltage-activated K ϩ channel genes, found transcripts encoding Shaker (Kv1.X)-related K ϩ channels in RIN and HIT insulinoma cells and whole islets isolated from leptin-resistant hyperglycemic ob/ob mice (20). In this investigation we used clonal insulinoma cells (␤TC3-neo, derived from transgenic animals expressing the SV-40 T-antigen driven by the insulin promoter) and purified rodent ␤-cells to analyze the expression of delayed rectifier family genes with RT-PCR. Kv1 family expression was sought in ␤TC3-neo cells utilizing oligonucleotide primers (see Table I) designed to amplify the highly conserved pore region (pair B) or the highly conserved domain upstream from the first membrane spanning domain (pair A) followed by individual primer pairs for each member of the Kv1.1-Kv1.7 family. Identification of other families of delayed rectifier K ϩ channel genes in ␤TC3-neo cells relied on the use of isoform-specific oligonucleotide primers to amplify either the pore region or a section of the 3Ј end of the coding region. DNA amplification experiments were performed using two positive controls, rodent brain cDNA with channel primers and insulinoma cell cDNA with proinsulin primers. We were unable to find evidence of expression of any of the seven Kv1 isoforms in ␤TC3 or ␤TC3-neo cells despite multiple attempts with degenerate and specific primers. Specific amplicons were, however, obtained with primers for Kv2.1 and Kv3.2 (Fig. 1A). DNA sequencing confirmed the closest matches of the amplified sequences were to Kv2.1 and Kv3.2, respectively (Fig. 2), although the sequence of the mouse isoform of Kv3. 2 has not yet been published. An antibody specifically recognizing Kv2.1 was employed to show that a membrane preparation from ␤TC3neo cells contained an immunoreactive band of 108 kDa as reported previously for Kv2.1 (Fig. 1B) (25).
To extend these results to primary cultured cells, we next determined which Kv family channels were expressed in FACS-purified rodent ␤-cells utilizing the RT-PCR strategy as described above for the ␤TC3-neo cells (Table I, Fig. 3). The only delayed rectifier cDNA that could be reproducibly amplified from purified rat and mouse ␤-cells was Kv2.1. In some experiments DNA bands indicating Kv1.6, Kv1.2, and Kv3.2 expression were also faintly detected, but these could not be subcloned and confirmed by DNA sequencing (data not shown). While a previous study reported expression of Kv1.X genes in ob/ob islets, the contribution of RNA from the non-␤-cells present in whole islets, as well as from contaminating pancreatic acinar cells, makes it difficult to interpret in terms of specific expression in ␤-cells (19).
Studied by whole-cell patch clamp techniques, ␤TC3-neo cells have similar delayed rectifier K ϩ currents to normal rodent or human ␤-cells (2,22,27). Exposure of ␤TC3-neo cells to 10 mM TEA caused inhibition (Ͼ90%) of the outward currents (Fig. 4). TEA blocks a variety of K ϩ channel types over distinct concentration ranges. Most sensitive to TEA are calcium-activated K ϩ channels (K d , 0.14 mM), followed by delayed rectifier K ϩ channels (K d , 1.4 mM) and K-ATP (K d , 22 mM) (2,14,27). The K d for block was estimated to be 0.8 mM, approximating the reported K d for block by TEA of the mouse ␤-cell delayed rectifier K ϩ current (15). Exposure to 4-aminopyridine (4-AP), another blocker of delayed rectifier K ϩ channels, also produced a dose-dependent suppression of the outward K ϩ currents with 80% block at 10 mM and an estimated K d of 2 mM (Fig. 4). The K d for block by TEA of Kv3.2 is under 1 mM and that for Kv2.1 is 5.6 mM (28,29). The TEA sensitivity of 0.8 mM observed experimentally in the ␤TC3neo and rodent ␤-cells could thus be achieved by co-expression of Kv3.2 and 2.1 channels. Similarly, the reported K d values for 4-AP block of Kv2.1 and Kv3.2 in the low millimolar range (30,31) are consistent with our experimental observations. By contrast, most Kv1 channels of the type encoding ␤-cell-like delayed rectifier K ϩ currents (Kv1.1-3. 1.5-7) are not sensitive to low millimolar concentrations of TEA (with the exception of Kv1.1 and 1.6) (19, 32). Further-more, Kv1.1 and Kv1.6 are blocked by low concentrations of dendrotoxin (10 -20 nM) (19), an agent that was without effect on ␤TC3 or rodent ␤-cells at 100 nM (data not shown). Also, others have shown that charybdotoxin, a potent blocker of Kv1.3, also has no significant effect on rodent ␤-cell delayed rectifier currents (2). In summary, the combination of molecular biological and pharmacological evidence argues against the significant expression of Kv1 family channels in rodent ␤-cells.
In order to study the physiological role of delayed rectifier K ϩ channels in stimulus-secretion coupling, we examined the effect of a range of TEA and 4-AP concentrations on [Ca 2ϩ ] i oscillations in populations of ␤TC3-neo cells in the presence of glucose. Unlike normal mouse islets and ␤-cells, ␤TC3 cells do not respond to a step increase in glucose with periodic oscillatory increases in [Ca 2ϩ ] i ; instead, a slow rise in [Ca 2ϩ ] i occurs with occasional intermittent spikes (Fig. 5A) (22,33). However, exposure of ␤TC3-neo cells to 20 mM TEA in the continued presence of 1 mM glucose stimulated large amplitude [Ca 2ϩ ] i oscillations of the type seen in normal ␤-cells exposed to glucose as the sole stimulating agent (Fig. 5A) (22). While 1 mM TEA produced an elevation in [Ca 2ϩ ] i and intermittent oscillations in only 10% of cells, increasing the TEA concentration to 5, 10, and 20 mM caused a dose-dependent increase in the percentage of cells responding to TEA as well as in the amplitude of the oscillatory [Ca 2ϩ ] i response (Table II). The concentration range over which TEA exerted its effects suggested an important role for delayed rectifier K ϩ channels in regulating [Ca 2ϩ ] i oscillations. Since 1 mM TEA exerted only a minor effect on [Ca 2ϩ ] i in ␤TC3-neo cells, this ruled out a significant role for calciumactivated K ϩ channels which should be completely blocked under these conditions (2). Based on the K d estimates for the effect of TEA on K ϩ channels in primary ␤-cells, 1 and 5 mM TEA should block delayed rectifier K ϩ channels but have little effect on K-ATP (2). The TEA-stimulated [Ca 2ϩ ] i oscillations were suppressed in the absence of glucose, further supporting the idea that TEA was not exerting its effects via blocking K-ATP and alterations in the glucose concentration between 0.01 and 1 mM caused graded increases in the frequency of the TEA-stimulated [Ca 2ϩ ] i oscillations (Fig. 5B). A similar effect of 4-AP inducing [Ca 2ϩ ] i oscillations in ␤TC3 cells was also observed (data not shown).
Similar [Ca 2ϩ ] i responses to TEA were observed in rodent islets. As with ␤TC3-neo cells, in rat islets a step increase in glucose concentration induced a tonic rise in [Ca 2ϩ ] i (Fig. 6A).  All sequences are written 5Ј to 3Ј; "F" indicates "forward" primer and "R" indicates "reverse" primer, and the number at the end indicates the 3Ј bp for the sequence indicated.

(F)ATTGGATCCATC(C/T)TC/GTA(C/T)TA(C/T)TA(C/T)CA(G/A)TCIGG (498; hKv1.5) (R)ATTGAATTCG(A/G)TACTTC(G/A)AAIA(T/G)IAGCC (708).
The addition of 20 mM TEA induced a transient rise in [Ca 2ϩ ] i , followed by rapid oscillations, again similar to the responses seen with ␤TC3-neo cells (cf. Fig. 5A). Likewise, in mouse islets TEA augmented the [Ca 2ϩ ] i oscillations induced by glucose, in agreement with previous results on the potentiating effects of TEA on mouse islet electrical activity (Fig. 6B) (17). The [Ca 2ϩ ] i oscillations induced by K ϩ channel block were immediately abolished by exposure to 500 nM nitrendipine, indicating their dependence on Ca 2ϩ influx through L-type calcium channels (34). To confirm that the induction by TEA of glucose-stimulated [Ca 2ϩ ] i oscillations was in fact due to facilitating depolarization-dependent increases in Ca 2ϩ influx, we carried out simultaneous measurements of membrane potential and [Ca 2ϩ ] i in single ␤TC3-neo cells. Application of 1 mM glucose caused a 20-mV depolarization and low amplitude spike activity (Fig. 7). Addition of 20 mM TEA caused a further abrupt depolarization followed by sinusoidal oscillations of membrane potential that were perfectly synchronized with the oscillations in [Ca 2ϩ ] i (Fig. 7). The oscillations, although small in magnitude, were centered around 0 mV, the peak membrane potential for Ca 2ϩ permeation through L-type Ca 2ϩ channels (2). Thus, it is likely that the oscillations in [Ca 2ϩ ] i induced by TEA stem from augmentation of membrane potential oscilla- . In whole-cell clamp configuration, the membrane potential was clamped in 20-mV increments from Ϫ60 to ϩ60 mV from a holding potential of Ϫ60 mV. Same cell in both panels is shown before and after drug application (representative of five separate experiments). Lower panels: dose-response curves showing percentage block of outward delayed rectifier K ϩ current (measured during the clamp pulse to ϩ40 mV after leak subtraction) and insulin secretion as a function of drug concentration.
tions as a result of TEA-induced modulation of a K ϩ channel conductance.
Finally, the effects of delayed rectifier K ϩ current block on stimulating glucose-dependent insulin secretion were examined. In mouse islets, 20 mM TEA augments glucose-stimulated insulin secretion, with a threshold of about 2 mM (17). In the ␤TC3-neo cells, TEA produced a dose-dependent stimulation of insulin secretion with a threshold of about 1 mM, and 20 mM TEA induced a 4-fold increase in insulin release compared with glucose alone (Fig. 4). TEA in the absence of glucose had no effect on insulin secretion, in agreement with its secretagogue effects depending on block of delayed rectifier K ϩ channels alone (34). Insulin secretion was also stimulated by 4-AP in the presence of glucose over the same concentration range (3-10 mM) where suppression of delayed rectifier K ϩ currents occurred in ␤TC3-neo cells (Fig. 4). It is important to consider possible interactions of TEA with other membrane proteins. Although TEA can block K-ATP, the concentrations required for complete block are much higher than used here, and K-ATP is already substantially blocked in extracellular solutions with elevated glucose concentrations (data not shown). TEA has also been reported to block Cl Ϫ channels (35). However, the higher concentrations required (K d 11.8 mM, with a maximum block of only about 60% at 60 mM (35) cannot account for the dramatic stimulation of signal transduction shown (Figs. 5-7). Furthermore, 4-AP, which produced effects similar to TEA on insulin secretion, has no effect on Cl Ϫ channels (35). In fact, several studies have shown that Cl Ϫ channel blockers actually inhibit, rather than stimulate, insulin secretion (36,37).
Our results confirm and extend previous observations on the importance of delayed rectifier K ϩ channels in repolarization of the ␤-cell plasma membrane and implicate specific delayed rectifier genes in this process. We report for the first time the expression in rat and mouse ␤-cells of Kv2.1 and in ␤TC3-neo insulinoma cells of Kv2.1 (Shab)-and Kv3.2 (Shaw)-related transcripts. The characteristics of the ␤-cell delayed rectifiers are in fact most consistent with those of Kv2.1 currents, as opposed to other known delayed rectifier genes. These findings extend the distribution of the Kv2.1 channel to hormone-secreting cells, in addition to striated muscle and neurons (19). The dose-response relationships for TEA-and 4-AP-stimulated insulin secretion correlate directly with the block of Kv2.1-like voltage-dependent outward K ϩ currents, as do the effects of TEA on glucose-dependent [Ca 2ϩ ] i oscillations. It is likely, therefore, that the observed oscillations in membrane potential and [Ca 2ϩ ] i in ␤TC3 cells induced by TEA and 4-AP stem directly from block of voltage-dependent K ϩ channels. Furthermore, our observations in rodent islets where similar induction of [Ca 2ϩ ] i oscillations was observed point to the importance of delayed rectifier K ϩ channels in normal ␤-cells as well. Taken together these studies have also revealed an important insight into the underlying mechanism regulating ␤-cell oscillations: essentially all of the outward K ϩ currents have to be suppressed in order to unmask the endogenous ␤-cell oscillator, indicating that the changes in conductance(s) underlying this mechanism could be very small.