Functional differences in Kv1.5 currents expressed in mammalian cell lines are due to the presence of endogenous Kv beta 2.1 subunits.

The voltage-sensitive currents observed following hKv1.5 α subunit expression in HEK 293 and mouse L-cells differ in the kinetics and voltage dependence of activation and slow inactivation. Molecular cloning, immunopurification, and Western blot analysis demonstrated that an endogenous L-cell Kvβ2.1 subunit assembled with transfected hKv1.5 protein. In contrast, both mRNA and protein analysis failed to detect a β subunit in the HEK 293 cells, suggesting that functional differences observed between these two systems are due to endogenous L-cell Kvβ2.1 expression. In the absence of Kvβ2.1, midpoints for activation and inactivation of hKv1.5 in HEK 293 cells were −0.2 ± 2.0 and −9.6 ± 1.8 mV, respectively. In the presence of Kvβ2.1 these values were −14.1 ± 1.8 and −22.1 ± 3.7 mV, respectively. The β subunit also caused a 1.5-fold increase in the extent of slow inactivation at 50 mV, thus completely reconstituting the L-cell current phenotype in the HEK 293 cells. These results indicate that 1) the Kvβ2.1 subunit can alter Kv1.5 α subunit function, 2) β subunits are not required for α subunit expression, and 3) endogenous β subunits are expressed in heterologous expression systems used to study K+ channel function.

cells, and HEK 293 cells (10 -15). Although there are some functional differences between these expression systems, it has been generally assumed that these systems faithfully reproduce native channel activity.
The differences in structure and function between the cloned voltage-gated K ϩ channel ␣ subunits, combined with the finding that they can form functional heteromeric structures, indicate that potassium channel contribution to excitable membrane function is complex (13, 16 -18). In addition, the recent discovery of function-altering ␤ subunits, some of which can convert a delayed rectifier into a rapidly inactivating channel, has added yet another layer of complexity (19). Five different ␤ subunits have been described in detail, one from Drosophila (20) and the others from mammals (19,(21)(22)(23)(24)(25). Three of these subunits, Kv␤1.1, 1.2, and 1.3, induce a variable degree of rapid inactivation to delayed rectifiers (19,(22)(23)(24). The fourth protein, Kv␤2.1, was reported to have no effect on Kv1.1 or 1.4 currents in the Xenopus oocyte system (19). However, most recently Kv␤2.1 has been shown to increase the N-type inactivation of Kv1.4 2.3-fold (26). The Kv␤1.1, 1.2, and 2.1 subunits were originally designated ␤1, 3, and 2, respectively. The Kv␤1.1-1.3 proteins arise by alternative splicing from the same gene, whereas Kv␤2.1 is derived from a distinct gene (24,26). The nomenclature used here reflects a subfamily classification based on genomic structure as previously proposed (24). An issue that must be resolved with all heterologous expression systems is whether they contain endogenous, function-altering ␤ subunits.
The Kv1.5 delayed rectifier has been cloned from heart, insulinoma tissue, gastrointestinal smooth muscle, and skeletal muscle from rat, mouse, canine, and human species (10,11,(27)(28)(29)(30). The present study was undertaken to determine why the human Kv1.5 channel has different properties when expressed in the HEK 293 system compared with L-cells. Specifically, in L-cells, the voltage sensitivity is shifted 10 mV in the negative direction and slow inactivation is increased. The data presented here indicate that the L-cells express a Kv␤2.1 subunit isoform that assembles with the transfected Kv1.5 ␣ subunit. The HEK 293 cells lack a ␤ subunit based on both mRNA and protein analysis. Coexpression of the mouse L-cell ␤ subunit with the Kv1.5 channel in the HEK 293 cells reconstitutes the L-cell current phenotype. These results indicate that the functional differences observed between the L-cell and HEK 293 expression systems are due to the presence of an endogenous L-cell ␤ subunit and represent the first description of functional effects of Kv␤2.1 on delayed rectifier function. . The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  Northern Blot Analysis-Total RNA was isolated by the guanidinium thiocyanate method (31), and 10 g was electrophoretically separated on a formaldehyde denaturing agarose gel and transferred overnight to a Nytran-Plus membrane (Schleicher & Schuell) by capillary action. Ethidium bromide (40 g/ml) was added to the samples prior to electrophoresis, enabling visual confirmation of RNA integrity. The template for the random primer synthesized, [␣-32 P]ATP-labeled probe was a polymerase chain reaction-generated rat brain cDNA fragment of Kv␤1.1 (nucleotides Ϫ6 to ϩ1567). The filters were hybridized overnight at 65°C 10% formamide, 10% dextran sulfate, 4 ϫ SSPE (600 mM NaCl, 40 mM NaH 2 PO 4 , 4 mM EDTA, pH 7.4), 5 ϫ BFP (1 g/liter bovine serum albumin, 1 g/liter polyvinylpyrrolidone 40, 1 g/liter Ficoll, 0.001% sodium azide), 0.1 mg/ml sonicated salmon sperm DNA, 0.2 mg/ml yeast tRNA, and 5% SDS supplemented with 10 6 cpm/ml of the probe. The filter was then washed three times at 65°C for 30 min with 3 ϫ SSC.
L-cell cDNA Library Construction, Screening, and Sequence Analysis-A ZAPII cDNA library was constructed from oligo(dT) purified mouse fibroblast L-cell mRNA using a cDNA synthesis system, as described by the manufacturer. No size fractionation of the cDNA was performed. Polymerase chain reaction-generated cDNA fragments corresponding to nucleotides ϩ435 to ϩ1089 of rat Kv␤2.1 (rKv␤2.1) served as template for the random primer-generated radiolabeled probe that was used to screen ϳ3.5 ϫ 10 5 unamplified recombinants. Two clones, 1.5 and 1.0 kilobases in length, were plaque-purified and recovered by in vivo excision into pBluescript (SK Ϫ ). Nucleotide sequence was determined from both strands of the full-length 1.5-kilobase cDNA by the dideoxynucleotide chain termination method using double stranded templates and appropriate oligonucleotide primers.
Antibody Production-A polymerase chain reaction-generated cDNA fragment corresponding to nucleotides ϩ435 to ϩ1089 of rKv␤2.1 was subcloned into pGEM7 at the EcoRI site. A cDNA fragment corresponding to nucleotides ϩ577 to ϩ1089 (amino acids 193-363) was digested from this construct, subcloned into the NcoI and XhoI sites of the pGEX-2T vector (Pharmacia), and transformed into Escherichia coli strain UT481 as described previously (32). This construct utilized the stop codons of the pGEX-2T vector. The glutathione S-transferase-rKv␤2.1 fusion protein (GST-Kv␤2.1) was prepared and injected into two New Zealand White rabbits as reported (32). The ability of anti-Kv␤2.1 antisera to specifically detect Kv␤2.1 was confirmed by Western blot analysis of in vitro translated full-length hKv␤1.2 and mKv␤2.1 and the conserved COOH-terminal domain of rKv␤1.1 (amino acids 74 -401) (data not shown). Neither of the anti-Kv␤2.1 antisera were able to immunopurify in vitro translated or endogenous Kv␤2.1 subunits nor detect Kv␤2.1 in L-cells or heart cryosections by immunohistochemistry. Polyclonal antibodies directed against the amino-terminal 112 amino acids (anti-Kv1.5 (N-term)) and the first extracellular loop of hKv1.5 (anti-hKv1.5 (S1-S2)) have been described (32).
Metabolic Labeling and Immunopurification of hKv1.5 from Transfected Cells-Preparation and maintenance of stably transfected L-cells expressing hKv1.5 have been reported earlier (14). Immunopurification of transfected hKv1.5 ␣ subunits was performed as described previously for the Kv1.1 channel (33). Briefly, confluent cultures (60-mm dish) of hKv1.5-and sham-transfected cells were incubated overnight in cysteine-and methionine-deficient labeling media containing 150 Ci/ml of Tran 35 S-label [ 35 S]methionine. Following the labeling period, cells were detergent-solubilized with Triton X-100 extraction buffer (1% (w/v) Triton X-100, 150 mM NaCl, 50 mM Tris (pH 7.5), 1 mM EDTA, 0.2% bovine serum albumin, 1 g/ml leupeptin, 2 g/ml aprotinin, 5 mM phenylmethylsulfonyl fluoride, 2 mM benzamidine, 5 mM N-ethylmaleimide, and 1 mg/ml bacitracin; the last six items are protease inhibitors that were added just prior to use) and incubated on ice for 10 min. Insoluble cellular debris was removed by centrifugation at 15,000 ϫ g for 1 min. Cellular extracts were kept at 4°C throughout the solubilization process. Kv1.5 antiserum (1 l/1 ml cell extract) was added to the solubilized extract and incubated at room temperature for 2 h on a rocking platform. Packed Protein A Sepharose beads (3 l) preblocked with 0.2% bovine serum albumin were added, and the incubation was continued for an additional 2 h at room temperature with rocking. Beads were then sedimented and washed. The bound protein was eluted from the beads by boiling for 2 min in SDS sample buffer, analyzed by SDS gel electrophoresis, and fluorographed as described previously (34).
Expression Vector Construction-hKv1.5 (nucleotides Ϫ22 to ϩ1895) was subcloned into the XbaI and KpnI sites of pBK CMV . The mKv␤2.1 clone (nucleotides Ϫ6 to ϩ1237) was digested from pBluescript with NaeI and EcoRI, blunted, and inserted into the SmaI site of pBK CMV . The green fluorescent protein (GFP) behind a cytomegalovirus promoter was used as described previously (35,36).
Transfection of HEK 293 Cells-Recently thawed HEK 293 cells (ATCC #1573-CRL, passage 33) were maintained in minimum essential medium supplemented with 10% HS. Cells were transfected by the calcium phosphate method with the hKv1.5/pBK CMV construct and selected with 500 g of G418/ml minimum essential medium ϩ 10% horse serum until distinct foci appeared (14 days). Cells were treated with trypsin, plated in 60-mm dishes, grown to confluence, and metabolically labeled with Tran 35 S-label [ 35 S]methionine. Radiolabeled Kv1.5 protein derived from the HEK 293 cells was immunopurified as described above.
For functional analysis, transient expression of hKv1.5 with or without mKv␤2.1 in HEK 293 cells was obtained using the lipofectamine method according to suppliers directions. GFP was coexpressed with the channel subunits to assess transfection efficiency (10 -30%) and identify cells for voltage clamp analysis (36). The transient transfection used 0.5 g of hKv1.5/pBK CMV , 2 g of Kv␤2.1/pBK CMV , and 4 g of GFP/pRC CMV mixed with 25 l of lipofectamine reagent. The lipofection mixture was applied for 2-3 h, after which the standard culture medium was restored. The duration of the lipofection was reduced because the standard 6-h exposure routinely resulted in expression levels exceeding 5 nA (at 50 mV). The cells were removed from the dish using brief trypsinization, washed twice with maintenance medium, and stored at room temperature for recordings within the next 12 h. Voltage clamp recordings revealed typical hKv1.5 currents in 100% of the cells expressing GFP. Control cells (transfected without channel subunits or nonfluorescing cells) did not display these currents, although an endogenous current of variable but small size (50 -150 pA) was observed in a subset of these cells. This problem was minimized by using cell lines of low passage number.
Western Analysis of Immunopurified Kv1.5 Channel Proteins-Cells from two confluent 75-cm 2 flasks of hKv1.5-and sham-transfected L-cells were homogenized with a glass/glass Dounce homogenizer (33), the debris was removed by centrifugation at 1000 ϫ g at 4°C for 15 min, and the membranes were sedimented from the supernatant at 17,000 ϫ g at 4°C for 1 h. The final pellet was resuspended in 150 l of sterile phosphate-buffered saline and stored at Ϫ80°C. Of this membrane preparation, 70 l were solubilized in 1 ml of Triton X-100 extraction buffer and immunopurified as described above. The purified protein was fractionated by SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose, and incubated with antibody according to standard protocols as described previously (32,33). The primary Kv␤2.1 antiserum was diluted 1:200, whereas the secondary antibody, horseradish peroxidase-conjugated goat anti-rabbit IgG (Sigma), was diluted 1:10,000. Detection was by the ECL kit from Amersham Corp. according to manufacturers directions. Exposure was for 1-10 s.
Electrical Recording-Recordings were made with a DAGAN 3900 patch clamp amplifier (Dagan Corp., Minneapolis, MN) using the whole cell configuration of the patch clamp technique. Currents were recorded at room temperature (21-23°C) and were sampled at 1-5 kHz after anti-alias filtering at 0.5-2 kHz. Data acquisition and command potentials were controlled by pClamp software (Axon Instruments, Foster City, CA). To ensure voltage clamp quality, electrode resistance was kept below 3 M⍀; the average resistance was 2.4 Ϯ 0.4 M⍀ (mean Ϯ S.E., n ϭ 18). Junction potentials were zeroed with the electrode in the standard bath solution. Gigaohm seal formation was achieved by suction (10.9 Ϯ 1.4 G⍀, n ϭ 18). After establishing the whole cell configuration, the capacitive transients elicited by symmetrical 10-mV voltage clamp steps from Ϫ80 mV were recorded at 50 kHz for calculation of cell capacitance, access resistance, and input impedance. The average access resistance was 5.1 Ϯ 0.5 M⍀ (n ϭ 18), and after analog compensation the residual access resistance was 1.0 Ϯ 0.3 M⍀. Cells expressing currents in excess of 5 nA were discarded.
Pulse Protocols and Analysis-The holding potential was Ϫ80 mV unless indicated otherwise, and the cycle time for the protocols was 10 s or slower. The standard protocol to obtain current-voltage relationships and activation curves consisted of 250-ms pulses that were imposed in 10-mV increments between Ϫ80 and ϩ60 mV with additional interpolated pulses to yield 5-mV increments between Ϫ35 and ϩ25 mV (activation range of hKv1.5). The steady-state current-voltage relationships were obtained by measuring the current at the end of the 250-ms depolarizations. Between Ϫ80 and Ϫ40 mV, only passive linear leak was observed; least squares fits to these data were used for leak correction. Deactivating tail currents were recorded at Ϫ30 or Ϫ50 mV. The activation curve was obtained from the tail current amplitude immediately after the capacitive transient. For steady-state curves, raw data points were averaged over a small time window (2-5 ms).
The voltage dependence of channel opening and inactivation (activation and inactivation curves) were fitted with a Boltzmann equation y ϭ which k represents the slope factor and E h represents the voltage at which 50% of the channels are open or inactivated, respectively. Because inactivation was incomplete, data were normalized after subtraction of the noninactivating fraction at the test potential. The time course of tail currents and slow inactivation were fitted with a sum of exponentials. Activation kinetics were fitted with a single exponential to the latter 50% of activation to obtain the dominant time constant of activation (37,38). The curve fitting procedure used a nonlinear least squares (Gauss-Newton) algorithm; the results were displayed in linear and semilogarithmic format together with the difference plot. Goodness of the fit and required number of exponential components was judged by statistically comparing 2 values (F-test) and by inspection for systematic nonrandom trends in the difference plot.
The results are expressed as the means Ϯ S.E. Analysis of variance with appropriate post hoc comparisons were used to compare the differences in mean values; p Ͻ 0.05 was considered significant. Specific n values are presented in the text or figure legends. All pooled data were collected from at least three separate transfections.

hKv1.5 Displays Expression
System-dependent Characteristics-When expressed in mouse L-cells or CHO cells, hKv1.5 has a midpoint of activation of Ϫ19 Ϯ 3.0 to Ϫ14 Ϯ 4 mV and an activation time constant of 3.4 to 9 ms at 0 mV (12,14). Expression in HEK 293 cells yields different values with Ϫ0.2 Ϯ 2.0 to Ϫ4.1 Ϯ 2.4 mV for the activation midpoint and 11.8 Ϯ 4.6 to 16.5 Ϯ 1.2 ms for the activation time constant at 0 mV (10). In addition, the degree of slow inactivation is less in the HEK 293 cells relative to the L-cells. It was hypothesized that perhaps the HEK 293 cells have different populations of glycosylating enzymes, kinases, or other post-translational modifying enzymes that produce cell-specific mature forms of hKv1.5 at the plasma membrane. Alternatively, differences in the plasma membrane or cytoskeletal composition could account for the observed functional differences. Finally, another possible explanation for the differences in hKv1.5 expression is that function-altering ␤ subunits are differentially expressed in these cell lines. HEK 293 cells sometimes express endogenous, delayed rectifier type currents (see "Experimental Procedures"), whereas the L-cells always lack voltage-dependent outward currents and in fact have a resting potential near zero (39). Because the HEK 293 cells contain an occasional endogenous ␣ subunit, it was hypothesized that these cells might also contain an endogenous ␤ subunit.
Identification of Endogenous Kv␤ Subunits in Heterologous Expression Systems-In order to determine if endogenous ␤ subunits were present in commonly used expression systems, total RNA from Xenopus oocytes, L-cells, and Madin-Darby canine kidney, LLC-PK, CHO, and HEK 293 cells was subjected to low stringency Northern analysis as illustrated in Fig. 1. The probe contained the complete Kv␤1.1 coding sequence, which was predicted to hybridize with all known mam-malian ␤ subunit isoforms under the conditions used. A 2.5-3.0-kilobase ␤ subunit mRNA species was detected in CHO cells and L-cells at levels similar to that in rat brain, whereas ␤ subunit mRNA was not detected in Xenopus oocytes or Madin-Darby canine kidney, LLC-PK, or HEK 293 cell lines. However, the difference in nucleotide sequence between Xenopus and mammals may have prevented hybridization. Neither dexamethasone nor the presence of the ␣ subunit significantly altered the level of L-cell ␤ subunit mRNA (data not shown). The presence of ␤ subunit message in CHO and L-cells suggested that previously unrecognized endogenous ␤ subunits may be responsible for the functional differences described above.
In order to identify the ␤ subunit isoform present in L-cells, a cDNA library was constructed and screened as described under "Experimental Procedures." Two clones (1.5 and 1.0 kilobases) were purified to homogeneity and excised into pBluescript. The longer clone was sequenced in both directions, yielding an open reading frame of 1101 nucleotides. An in-frame 5Ј stop codon was found at Ϫ30, thus ensuring that the full coding region was obtained. Translation of the coding region predicted a 367-amino acid 41-kDa protein identical to rat Kv␤2.1 with 96.4% nucleotide identity in the coding region (19). The nucleotide and amino acid sequence of the mouse Kv␤2.1 (mKv␤2.1) is depicted in Fig. 2. The sequence between the arrows represents the region used to generate the Kv␤2.1-specific antisera. The shaded residues represent amino acids within this region that differ between the Kv␤2.1 isoform and the Kv␤1.1, 1.2, and 1.3 proteins. Despite 81% sequence identity in this region between all known ␤ subunits, the antiserum raised was specific for the Kv␤2.1 subunit as determined by Western analysis as described under "Experimental Procedures." Analysis of the Kv1.5 Subunit Composition in the L-cell and HEK 293 Expression Systems-Analysis of 200 base pairs of the 5Ј-untranslated region detected two upstream start codons, each followed by an in frame stop codon (Fig. 2). Because such upstream start sites can inhibit protein synthesis (40), it was necessary to demonstrate that the Kv␤2.1 cloned from the L-cells was translated. It was also necessary to demonstrate a direct interaction between the Kv1.5 and Kv␤2.1 subunits in order to conclude that the endogenous subunit is responsible for the observed differences between expression systems. Immunopurifications were performed from metabolically labeled L-cells expressing the hKv1.5 ␣ subunit to address these concerns.
L-cells transfected with either hKv1.5 or a sham vector were labeled with [ 35 S]methionine, solubilized, and immunopurified with anti-hKv1.5 antisera as described under "Experimental Procedures" (Fig. 3, lanes 1 and 2, respectively). Comparison of the two lanes indicates that two prominent bands with electrophoretic mobilities corresponding to molecular weights of ap-  Fig. 3 demonstrates that the 40-kDa protein copurifying with Kv1.5 is Kv␤2.1. The Kv1.5 channel was immunopurified from nonlabeled hKv1.5-or sham-transfected L-cell membranes (Fig. 3, lanes 3 and 4, respectively), fractionated by SDS gel electrophoresis, transferred to nitrocellulose, and incubated with the anti-Kv␤2.1 antiserum. The 40-kDa band was specifically detected by anti-Kv␤2.1 antibodies in the material purified from the hKv1.5-expressing cells, demonstrating that this band represents the mKv␤2.1 protein, as opposed to a proteolytic fragment of the ␣ subunit.
The absence of detectable ␤ subunit mRNA in HEK 293 cells does not guarantee a complete absence of ␤ subunits. Another family of subunits could exist that does not cross-hybridize, even at low stringency, with the cDNA probe used in the Northern analysis. Therefore, the hKv1.5 protein was immunopurified from transfected and radiolabeled HEK 293 cells, looking specifically for a copurifying protein of 30 -50 kDa. As shown in Fig. 3 (lanes 5 and 6), when HEK 293 cells transfected with hKv1.5 were immunopurified (lane 5), a unique doublet, probably representing the immature and glycosylated forms of hKv1.5, appeared at 66 and 75 kDa. These ␣ bands were absent when nontransfected cells were used as the starting material (lane 6). Low molecular weight bands representing putative ␤ subunits are noticeably absent from lane 5, demonstrating that  ␤ subunits capable of associating with hKv1.5 are absent from the HEK 293 cells. Because functional channels are obtained after transfection of the ␣ subunit alone, ␤ subunits are not required for the synthesis of functional channels. It is also unlikely that the ␤ subunit plays a role in regulating cell surface channel density because the current densities are similar between the L-cell and HEK 293 expression systems (10,14).
Coexpression of mKv␤2.1 with hKv1.5 in HEK 293 Cells-The immunopurification data document a direct association between hKv1.5 and the endogenous mKv␤2.1 subunit. Although Kv␤2.1 increases the rate of Kv1.4 N-type inactivation (26), the literature suggests that this ␤ subunit has no functional effect on delayed rectifier ␣ subunit function (19). To test whether the mKv␤2.1 subunit was responsible for the functional differences between L-cells and HEK 293 cells, hKv1.5 was transiently transfected with and without mKv␤2.1 into the HEK 293 cells using the lipofectamine approach as described under "Experimental Procedures." Fig. 4 shows typical recordings from HEK 293 cells transiently transfected with hKv1.5 alone or with mKv␤2.1, (A and B, respectively). Tracings for depolarization from Ϫ80 mV to potentials between Ϫ30 mV and ϩ50 mV were superimposed for comparison. No voltage-activated current was observed between Ϫ100 and Ϫ40 mV, but a progressively larger outward current was recorded with depolarization above Ϫ30 mV. In both A and B of Fig. 4, the activation of the current proceeded with a sigmoidal time course, and the rate of activation increased with depolarization. Upon repolarization to Ϫ30 mV, outward tail currents were recorded. These hKv1.5 currents were in qualitative agreement with observations in various expression systems (10 -14). However, when hKv1.5 was coex-pressed with Kv␤2.1, substantially more current was activated at the lowest depolarizations as indicated by the arrows in A and B. Although slow inactivation was limited in both cases, the current declined more when hKv1.5 was coexpressed with mKv␤2.1, especially at the strongest depolarizations.
To directly quantitate the effects of mKv␤2.1 coexpression, we analyzed the voltage-dependence of activation from the amplitude of the decaying tail currents. Fig. 4C shows that coexpression with mKv␤2.1 results in a negative displacement of the activation curve. The sigmoidal voltage-dependence was fitted with a single Boltzmann equation resulting in half-activation voltages of Ϫ0.2 Ϯ 2.0 mV (n ϭ 8) and Ϫ14.1 Ϯ 1.8 mV (n ϭ 9; p Ͻ 0.01) without and with mKv␤2.1, respectively. The slope factors were not significantly different (6.2 Ϯ 0.6 mV without and 5.6 Ϯ 0.4 mV with mKv␤2.1, p Ͼ 0.1). Because the steady-state voltage dependence and the kinetics of activation are determined by the same underlying rate constants, we determined whether a similar shift existed in the activation kinetics. For comparison with previous results, the dominant time constant of activation was determined (see "Experimental Procedures"). Fig. 4D shows that coexpression of mKv␤2.1 resulted in a faster activation rate between Ϫ10 and ϩ60 mV, resulting in a parallel shift of the same magnitude as for the activation curve of C. The results obtained in the absence of mKv␤2.1 correspond closely to those of Fedida et al. who expressed hKv1.5 in HEK 293 cells (10). The dashed lines in C and D of Fig. 4 illustrate the corresponding results previously obtained in L-cells under identical conditions (14). In both cases, the results for coexpression of hKv1.5 with mKv␤2.1 correspond closely to hKv1.5 expressed in L-cells.
Like many delayed rectifiers, hKv1.5 displays partial and slow inactivation, presumably of the C-type (41). Often, hKv1.5 Outward currents were elicited by a series of 250-ms step depolarizations from a holding potential of Ϫ80 mV to potentials between Ϫ30 and ϩ50 mV in 10-mV increments. Note that the 0-mV stimulus is indicated in both panels. C shows the voltage activation curve, in the presence and the absence of the mKv␤2.1 subunit. D illustrates the voltage dependence of the activation time constant plus and minus mKv␤2.1. E illustrates the voltage dependence of inactivation with and without the mKv␤2.1 subunit. Pulse protocols are as described previously (14). F summarizes effects of the mKv␤2.1 subunit on hKv1.5 slow inactivation. The data are plotted as the percentage of inactivation (relative to peak current) at the indicated pulse duration. The asterisks represent significant differences (p Ͻ 0.01). In A, B, C, and E, representative data from a single paired experiment are shown, i.e. the curves in C and E were obtained from the same cells represented in A and B. For D and F, data were derived from five or six separate cells from three independent transfections. The dashed lines in C and D represent data obtained from hKv1.5 expression in mouse L-cells (14). expressed in oocytes displays less extensive inactivation than in the L-cells. Therefore we compared both the voltage dependence of slow inactivation and the degree of slow inactivation at 250 ms, 1 s, and 5 s during depolarizations to 50 mV. Fig. 4E shows the results from the cells shown in A and B. Coexpression with mKv␤2.1 resulted in a hyperpolarizing displacement of the normalized inactivation curve with average values E h ϭ Ϫ9.6 Ϯ 1.8 mV (n ϭ 6) and Ϫ22.1 Ϯ 3.7 mV (n ϭ 5; p Ͻ 0.01) without and with mKv␤2.1, respectively. Slope factors were not significantly different (5.2 Ϯ 0.4 and 5.1 Ϯ 0.2, p Ͼ 0.1). Because the voltage dependence of C-type inactivation appears to be linked to that of activation (41), these results are consistent with the effect of Kv␤2.1 on activation. Indeed, when the difference in the midpoints of activation and inactivation were determined for each cell individually, we observed a similar displacement between both curves of 9.8 Ϯ 1.3 mV (n ϭ 6) and 8.9 Ϯ 2.8 mV (n ϭ 5) for hKv1.5 alone or with mKv␤2.1, respectively (p Ͼ 0.1). This displacement is consistent with the 8.6-mV displacement observed in L-cells (14). The presence of the mKv␤2.1 subunit also appeared to enhance the degree of slow inactivation as illustrated in Fig. 4 (A and B). Fig. 4F shows that the average amount of inactivation (with respect to the peak current) at 250 ms, 1 s, and 5 s was greater when hKv1.5 was coexpressed with mKv␤2.1. The percentage of inactivation in the HEK 293 cells expressing both hKv1.5 and mKv␤2.1 compares well with the results for hKv1.5 in L-cells (69% at 60 mV) (14). It should be noted that the comparison is somewhat complicated by the high temperature sensitivity of this process (14). Nevertheless, the results from the HEK 293 cells reported here were obtained at the same temperature. The mKv␤2.1 subunit had no effect on ion selectivity based on the finding that the reversal potential was unchanged in the presence of mKv␤2.1: Ϫ82.3 Ϯ 1.4 mV (n ϭ 4) with ␤ and Ϫ83.7 Ϯ 0.9 (n ϭ 8) without ␤. No effect of mKv␤2.1 on channel surface density was detected, making it unlikely this subunit plays a regulatory role in channel biosynthesis, although such a role may be apparent only in the native cell.

Comparison of Kv1.5 Function in Commonly Used
Expression Systems-The functional characteristics of hKv1.5 expressed in mouse L-cells, HEK 293, and CHO cells and Xenopus oocytes are summarized and compared in Table I. Comparison of the L-cell and HEK 293 data indicate that the primary effects of the Kv␤2.1 subunit are to alter the voltage sensitivity of activation as demonstrated by the leftward shift in the midpoint of the activation and inactivation curves and the voltage dependence of the activation time constants. The activation midpoint did not change in L-cells expressing mutants of hKv1.5 lacking the conserved TDV motif and up to 57 COOHterminal amino acids, indicating that this region is not involved in the Kv␤2.1 modulation (15). Another notable effect of Kv␤2.1 is that slow inactivation is increased approximately 2-fold. Although the Kv␤1.2 subunit has a marked effect on hKv1.5 deactivation (23), Kv␤2.1 did not alter the kinetics of the tail currents at Ϫ30 mV. As Table I indicates, the proper-ties of hKv1.5 expression in CHO cells suggest that the endogenous ␤ subunit in this system is the Kv␤2.1 isoform. No ␤ subunit mRNA was detected in the Xenopus oocytes, as shown by the Northern analysis of Fig. 1. The midpoint of activation in the oocyte system is closer to that found in HEK 293 cells without Kv␤2.1 coexpression. The degree of slow inactivation also appears to be reduced in the oocyte relative to CHO and L-cells (13), although this parameter is very temperature-dependent and difficult to compare between laboratories. Although further experiments are required to determine if endogenous ␤ subunits are present in oocytes, the present data suggest that they are not. Pharmacological studies on the Kv1.5 channel have been performed by several laboratories using the L-cell, CHO, and HEK 293 expression systems. Comparison of data obtained with these systems suggests that no pharmacological effect can be assigned to the Kv␤2.1 subunit. For example, verapamil and terfenadine modification of Kv1.5 currents are similar in HEK 293 and L-cells (42)(43)(44)(45)(46).

CONCLUSIONS
The data presented here are the first to indicate that Kv␤2.1 modifies delayed rectifier potassium channel function. Immunopurification of hKv1.5 from HEK 293 cells indicated that these cells do not contain a ␤ subunit that assembles with hKv1.5. Because these cells are capable of expressing hKv1.5 current when transfected with only the ␣ subunit, the ␤ subunit is not required for expression. Coexpression of Kv␤2.1 with Kv1.5 in HEK 293 cells resulted in a 10-mV leftward shift in the activation curve and an almost 2-fold increase in the degree of slow inactivation. The resulting HEK 293 cell current closely mimics that recorded from hKv1.5 transfected CHO and L-cell lines (which contain endogenous ␤ subunits). In addition to demonstrating a functional effect of Kv␤2.1 on Kv1.5, these data illustrate the need to test for and identify the ␤ subunits present in heterologous expression systems.
The Kv1.5 K ϩ channel is expressed most abundantly in cardiac myocytes and vascular smooth muscle (32). Because the Kv␤2.1 subunit has also been cloned from cardiac tissue (23), the potential exists for Kv1.5 and Kv␤2.1 coassembly in heart. Such in vivo assembly would result in an earlier activation of the rectifying potassium current, perhaps serving as a regulatory mechanism controlling action potential duration. The Kv␤2.1 subunit has several potential phosphorylation sites that could regulate assembly and/or interaction with the ␣ subunit, allowing for a rapid response system controlling heart rate. A search for potential pathophysiologic conditions resulting from altered Kv␤2.1 expression must wait until the cellspecific expression and in vivo ␣ subunit association is confirmed.
Acknowledgments-We thank Dr. Lou Philipson for the generous gift of the anti-hKv1.5 (S1-S2) antisera, Paul Bennett for review of the manuscript, and Holly Shear, Debbie Mays, and Ian Hopkirk for excellent technical support.