Comparison of Binding and Block Produced by Alternatively Spliced Kvβ1 Subunits

Abstract Voltage-gated K+ (Kv) channels consist of α subunits complexed with cytoplasmic Kvβ subunits. Kvβ1 subunits enhance the inactivation of currents expressed by the Kv1 α subunit subfamily. Binding has been demonstrated between the C terminus of Kvβ1.1 and a conserved segment of the N terminus of Kv1.4, Kv1.5, and Shaker α subunits. Here we have examined the interaction and functional properties of two alternatively spliced human Kvβ subunits, 1.2 and 1.3, with Kvα subunits 1.1, 1.2, 1.4, and 1.5. In the yeast two-hybrid assay, we found that both Kvβ subunits interact specifically through their conserved C-terminal domains with the N termini of each Kvα subunit. In functional experiments, we found differences in modulation of Kv1α subunit currents that we attribute to the unique N-terminal domains of the two Kvβ subunits. Both Kvβ subunits act as open channel blockers at physiological membrane potentials, but hKvβ1.2 is a more potent blocker than hKvβ1.3 of Kv1.1, Kv1.2, Kv1.4, and Kv1.5. Moreover, hKvβ1.2 is sensitive to redox conditions, whereas hKvβ1.3 is not. We suggest that different Kvβ subunits extend the range over which distinct Kv1α subunits are modulated and may provide a variable mechanism for adjusting K+ currents in response to alterations in cellular conditions.

Voltage-gated K ؉ (Kv) channels consist of ␣ subunits complexed with cytoplasmic Kv␤ subunits. Kv␤1 subunits enhance the inactivation of currents expressed by the Kv1 ␣ subunit subfamily. Binding has been demonstrated between the C terminus of Kv␤1.1 and a conserved segment of the N terminus of Kv1.4, Kv1.5, and Shaker ␣ subunits. Here we have examined the interaction and functional properties of two alternatively spliced human Kv␤ subunits, 1.2 and 1.3, with Kv␣ subunits 1.1, 1.2, 1.4, and 1.5. In the yeast two-hybrid assay, we found that both Kv␤ subunits interact specifically through their conserved C-terminal domains with the N termini of each Kv␣ subunit. In functional experiments, we found differences in modulation of Kv1␣ subunit currents that we attribute to the unique N-terminal domains of the two Kv␤ subunits. Both Kv␤ subunits act as open channel blockers at physiological membrane potentials, but hKv␤1.2 is a more potent blocker than hKv␤1.3 of Kv1.1, Kv1.2, Kv1.4, and Kv1.5. Moreover, hKv␤1.2 is sensitive to redox conditions, whereas hKv␤1.3 is not. We suggest that different Kv␤ subunits extend the range over which distinct Kv1␣ subunits are modulated and may provide a variable mechanism for adjusting K ؉ currents in response to alterations in cellular conditions.
The electrical properties of excitable cells such as neurons and cardiomyocytes are strongly influenced by the K ϩ currents they express. A variety of voltage-gated K ϩ (Kv) 1 channels control the falling phase of the action potentials of excitable cells. Kv channels are also important in many nonexcitable cells, where they contribute to diverse processes such as volume regulation, hormone secretion, and activation by mitogens (1). Functional Kv channels assemble as tetramers of poreforming ␣ subunits (Kv␣). Many mammalian Kv␣ genes have been cloned and assigned to four subclasses based on sequence similarities: Kv1, Kv2, Kv3, and Kv4 (2). In heterologous expression systems, individual Kv␣ subunits confer characteristic properties of gating, selectivity, and ion conduction, but several lines of evidence suggest that native Kv channels are more complex. First, within subfamilies, Kv␣ subunits are able to form functionally distinct heterotetramers, which contribute to increased K ϩ channel diversity (3)(4)(5). Second, accessory or Kv␤ subunits that modify the gating properties of coexpressed Kv␣ subunits have been cloned (6 -12) and found to be associated with Kv␣ subunits in native membranes (6,13).
To date, five distinct mammalian Kv␤ subunits have been cloned: Kv␤1.1, Kv␤1.2, Kv␤1.3, Kv␤2, and Kv␤3, according to recently proposed terminology (11,12). Kv␤ subunits have been shown to alter the phenotype of a subset of Kv␣ subunit currents within the Kv1 subfamily (7-10, 12, 14) primarily by introducing inactivation into noninactivating delayed rectifier channels (i.e. Kv1.1 and Kv1.5) and accelerating the intrinsic inactivation of rapidly inactivating channels (i.e. Kv1.4). The Kv␤ N terminus is thought to act as a "ball peptide" to mimic the N-type inactivation characteristic of Shaker and Kv1.4 (7). There are differences between Kv␤-induced and N-type inactivation, however. Whereas N-type inactivation is complete, Kv␤-induced inactivation is generally partial with significant sustained currents remaining at the end of the pulse.
Recent biochemical studies have shown that Kv␤1.1 interacts with the N-terminal domains of the Kv1 subfamily ␣ subunits, Kv1.4, Kv1.5, and Shaker, but not Kv2, Kv3, or Kv4 (15,16). A recently proposed model of Kv␣-Kv␤ interactions suggests that the inactivation conferred on Kv1 currents by Kv␤ subunits is the result of two sequential interactions: 1) physical association of the two subunits through the interaction of the conserved Kv␤1 C terminus with conserved regions in the Kv1␣ N terminus, and 2) plugging of the Kv1␣ pore by functional interaction of the Kv␤1 N terminus with its corresponding receptor site on the Kv1␣ subunit (15,16). One resulting hypothesis is that all Kv1␣ subunits would interact with all Kv␤1 subunits but that the functional effects on a particular Kv␣ might differ depending upon which Kv␤1 N terminus is present.
To test this hypothesis, we have studied the interaction of two human Kv␤1 subunits with a variety of Kv channels. hKv␤1.2 and hKv␤1.3 share identical C-terminal domains but have unique, nonhomologous N termini. Our results indicate that both Kv␤1 subunits interact selectively with Kv1␣ subunits and produce open channel block, but hKv␤1.2 is much more potent than hKv␤1.3. In addition, hKv␤1.2 is sensitive to redox potentials, while hKv␤1.3 is not. The results are consistent with our hypothesis that functional differences originate with distinct Kv␤1 N-terminal domains.

EXPERIMENTAL PROCEDURES
Cloning of hKv␤1.3 by 5Ј-RACE and Reverse Transcription-PCR-5Ј-RACE Ready human heart cDNA (Clontech) was used as a template in a nested PCR with antisense oligonucleotides encoding portions of the Kv␤1 subunit C terminus near the N-terminal junction point in combination with the sense oligonucleotide anchor primer supplied with the 5Ј-RACE Ready cDNA. The first PCR consisted of the ␤ subunit-specific antisense oligonucleotide (R7): 5Ј-AGCATAGACTTCGGCAGTATC-3Ј with the 5Ј-RACE anchor primer. A small aliquot of the first PCR product was used as template in the second reaction with an internal ␤ subunit-specific antisense oligonucleotide (R6): 5Ј-TCCAAATGTCAC-CCATGTTCC-3Ј and the anchor primer oligonucleotide. Without purification, the products from the second PCR were cloned into the pCRII vector (Invitrogen). Resulting clones were sequenced with the Sequenase kit (U. S. Biochemical Corp.), and compared to hKv␤1.2, Kv␤1.1, Kv␤2, and the Drosophila Kv␤, Hk. This protocol generated several overlapping clones, the longest of which resulted in an open reading frame of 90 amino acids (hKv␤1.3). This clone was identical to hKv␤1.2 in the region between oligonucleotide R6 and the arginine (R) residue marking the junction point between the N and C termini. Since this partial clone did not possess a putative initiating methionine residue, another 5Ј-RACE was performed to obtain the full N-terminal coding region. The second 5Ј-RACE reaction was done with the antisense ␤1.3-5Ј-RACE oligonucleotide 5Ј-TGAGGGACTAAGGCTGCTGTC-3Ј in combination with the anchor primer, and generated a clone that contained a putative initiating methionine by virtue of two upstream in-frame stop codons.
To obtain the full-length hKv␤1.3 sequence from human heart, we performed reverse transcription-PCR with human atrial total RNA using oligonucleotides spanning the proposed initiating methionine of hKv␤1.3 and the C-terminal residues plus stop codon of hKv␤1.2. A single band of approximately 1.3 kilobase pairs was obtained. Sequencing of full-length hKv␤1.3 revealed a unique N terminus identical to what had been obtained from the 5Ј-RACE protocol and a C terminus identical to hKv␤1.2.
RNase Protection-Fragments encoding the unique N-terminal 91 amino acids of hKv␤1.3 and 79 amino acids of hKv␤1.2 were prepared by PCR and cloned into pCRII (Invitrogen) to be used as the template to prepare ␤ subunit-specific probes for RNase protection assays. The fragments were sequenced to confirm that no mutations were introduced by PCR, and to determine the orientation in the pCRII vector. Plasmid DNA was purified on Qiagen Midi-prep plasmid purification columns and linearized with HindIII. 32 P-Labeled antisense transcripts were prepared from linearized template with T7 RNA polymerase using the MAXIscript transcription kit (Ambion) according to the manufacturer's protocol. Full-length antisense transcripts (385 bases for hKv␤1.3; 420 bases for hKv␤1.2) were gel-purified on a denaturing 5% polyacrylamide gel containing 8 M urea and eluted from the gel slices by overnight incubation at 37°C in the following buffer: 0.5 mM ammonium acetate, 1 mM EDTA, 0.2% SDS.
Total RNA from human atrial appendage tissue, right atrium, right ventricle, and left ventricle was isolated using RNA-STAT-60 (Tel-Test). The ventricular and right atrial samples were obtained in accordance with Tulane University School of Medicine Institutional guidelines from the explanted heart of a 7-year-old female patient with dilated cardiomyopathy undergoing cardiac transplant. Human atrial appendage tissue was pooled from adult patients undergoing aortocoronary bypass surgery. Total RNA from whole adult human brain was purchased from Clontech (catalog no. 64020-1).
The RNase protection experiments were performed using the Hyb-Speed RPA kit (Ambion) following the manufacturer's protocol. Ten g of brain RNA and 50 g of each cardiac tissue sample were incubated with 10 5 cpm of each ␤ subunit probe for 20 min at 68°C. Forty g of yeast total RNA was added to the brain RNA to keep the same total amount of RNA in each tube. As a control the probes were also incubated with 50 g of yeast tRNA only. The hybridization mixtures were then digested for 45 min with ribonuclease T1 at 37°C. Following RNase treatment, the protected fragments were precipitated, resuspended in gel loading buffer, and electrophoresed in a 5% polyacrylamide gel containing 8 M urea. The gel was dried and exposed to Kodak X-Omat AR film for 5 days at Ϫ80°C.
Protein-protein interactions were tested in two host yeast strains, SFY526 and HF7C by cotransformation with pairs of pGBT9 and pGAD424 fusion constructs according to the manufacturer's protocol. Transformations in SFY526 were plated on media lacking tryptophan (trp Ϫ ) and leucine (leu Ϫ ), and grown for 3-4 days at 30°C. Yeast colonies were lifted to paper filters and tested for ␤-galactosidase activity. Colonies turning blue within 8 h were scored as positive. Transformations in HF7C were plated on media lacking trp, leu, and histidine (his). Only yeast transformed with interacting fusion proteins should grow as a result of activation of the HIS3 gene, but ␤-galactosidase assays were performed for confirmation of interaction.
Functional Expression in Xenopus Oocytes-hKv␤1.3 and hKv␤1.2 were subcloned into pSP64 (Promega) for oocyte expression. The sources of the Kv␣ subunit constructs and preparation of cRNA are as described previously (8). The hKv1.4⌬N2-146 construct was prepared by PCR using a sense oligonucleotide incorporating a NotI site, a Kozak consensus sequence for initiation and an initiating ATG linked to 15 bases of Kv1.4 sequence encoding residues 147-151, and an antisense oligonucleotide just 3Ј to the internal BstEII site. After cloning into pCRII to verify the correct construction, the NotI/BstEII fragment was subcloned into hKv1.4-A ϩ -pCRII from which the wild-type fragment had been removed.
cRNA was prepared with the mMESSAGE mMACHINE kit (Ambion) using either SP6 or T7 RNA polymerase after linearization of the plasmids with EcoRI (for hKv␤1.3 and hKv␤1.2) or HindIII (for hKv1.4⌬N2-146). cRNAs were dissolved in 0.1 M KCl, stored at Ϫ80°C, and diluted immediately prior to injection. Stage V-VI Xenopus oocytes were injected with 46 nl of cRNA.
Whole-cell Two-electrode Recording-Whole-cell macroscopic currents were recorded with conventional two-electrode techniques as described previously (17). Electrodes filled with 3 M KCl had resistance of approximately 0.2-0.8 megohms when measured in the bath solution containing (in mM): NaCl, 100; KCl, 5; CaCl 2 , 0.3; MgCl 2 , 2; and HEPES, 10. pH was adjusted to 7.4 with Tris base. Records were digitized at 10 KHz and filtered at 3 KHz. Experiments were conducted at room temperature (20 -22°C).
Macropatch Recording-Macropatch currents were measured in cellattached and inside-out configuration by using pipettes made from borosilicate glass with tip openings of ϳ10 -15 m. The electrodes were connected to a patch-clamp amplifier (Axopatch1-C, Axon Instrument, Foster City, CA). The pClamp suite of programs was employed for data acquisition and analysis. The bath solution contained (in mM): KCl, 100; MgCl 2 , 1; and HEPES, 10. The pipette (external) solution contained (in mM): NaCl, 100; KCl, 5; CaCl 2 , 1.5; MgCl 2 , 2; and HEPES, 10. pH was titrated to 7.3 with Tris base and readjusted to this value whenever additional components were added to the solution such as glutathione (GSH, 5 mM). H 2 O 2 (0.1%) or GSH was applied to the cytoplasmic surface of the inside-out patches by changing the perfusate. Inside-out patches were positioned in the stream of a large pipette (diameter ϳ2 mm) to achieve faster solution exchange (Ͻ10 s) (18). Data were low pass-filtered at 2 KHz before digitization (Ϫ3 dB, 4-pole Bessel filter). To minimize leak current and capacitive transient, on-line subtraction with a P-4 protocol was used in all recordings. Traces at each potential represent an average of 5-10 sweeps.
Single-channel Recording-Single channels were recorded in both cell-attached and inside-out patches. Fire-polished and Sylgard-coated electrodes, pulled from borosilicate glass on a horizontal Flaming-Brown micropipette puller, had a tip resistance of ϳ10 megohms when filled with the external solution, which had the same composition as the bath solution described for the two-electrode experiments. Here the bath solution contained the following compositions (in mM): NaCl, 5; KCl, 100; CaCl 2 , 0.3; MgCl 2 , 2; and HEPES, 10. Data were digitized at 10 KHz and filtered with a 2-KHz cutoff frequency. Junction potentials were zeroed before formation of the membrane-pipette seal. All experiments were carried out at room temperature (ϳ20 -22°C). Chemicals were purchased from Sigma.
Data Analysis-For whole-cell and macropatch recordings, the peak and steady-state currents were measured as the maximal current amplitude relative to the zero base-line level during a 100-ms pulse or as the current amplitude at the end of the pulse by Clampfit in pClamp. Single-channel data were analyzed with Fetchan in pClamp, and open and closed transitions were detected using a half-amplitude threshold criterion. Group data are presented as mean Ϯ S.E. Student's t tests (paired or unpaired) were used to evaluate the statistical significance of differences between means. A two-tailed probability of 0.5% was taken to indicate statistical significance. For analysis of current kinetics, data points were fitted by Clampfit in pClamp. For other curve-fitting procedures, a nonlinear curve-fitting technique (Marquardt's procedure) was performed using Sigmaplot software (Jandel Scientific).

Expression of hKv␤1.2 and hKv␤1.3 in Brain and Heart-
The Kv␤1 subfamily of Kv␤ subunits consists of three members, which result from alternative splicing of a single gene: Kv␤1.1 (7), hKv␤1.2 (8), and hKv␤1.3 (11). We originally cloned hKv␤1.2 from human atrium and found that it shared an identical C terminus to rat brain Kv␤1 (subsequently renamed Kv␤1.1). When this study was initiated, only two Kv␤ subunits had been cloned: Kv␤1.1 and hKv␤1.2. Anticipating that there might be other members of this family, we searched for distinct Kv␤ subunit family members with a common structural plan, i.e. a conserved C-terminal domain with a variable N terminus, by using 5Ј-RACE on human heart RNA. With this strategy we cloned hKv␤1.3 from human atrium, which proved to be identical to a recently published hKv␤1.3 clone from human ventricle (11). Each Kv␤1 subunit consists of an identical C-terminal domain of 329 amino acids spliced to a unique N-terminal domain. The nonhomologous N termini of the three Kv␤1 subunits are shown in Fig. 1.
We examined the expression of hKv␤1.3 in human atrium, ventricle, and brain with RNase protection assays. To distinguish the expression of hKv␤1.3 from other ␤ subunits, a radiolabeled probe covering the unique N terminus of hKv␤1.3 was hybridized to total RNA from human atrium, right ventricle, left ventricle, and brain ( Fig. 2, panel A). Compared to heart RNA, 5 times less brain total RNA was used in the assay. Specific protection of a 273-nucleotide (nt) band was observed in all tissues with the strongest signals coming from left ventricle and brain. In addition, a smaller band of approximately 250 nt was observed in all tissues. The source of this second band is unknown, but could represent a second isoform of hKv␤1.3 with sequence divergence at one or the other end of the N-terminal probe. Interestingly, the relative abundance of the 250-nt band increased in atrium such that the 273-and 250-nt bands were present in at least relatively equal proportions, whereas the 250-nt band appeared to be only a minor component in ventricle and brain. For comparison, the expression of hKv␤1.2 was also examined with an N-terminal hKv␤1.2 probe (Fig. 2, panel B). hKv␤1.2 was expressed in all tissues by virtue of the protected 237-nt band seen in all lanes. Expression was highest in ventricle and brain relative to atrium, however. These data are consistent with higher levels of hKv␤1.2 expression in human ventricle compared to atrium using Northern blotting (10).
hKv␤1.3 and hKv␤1.2 Bind to the N Terminus of Kv1 ␣ Subunits-To determine whether there is any specificity in the interaction of hKv␤1.2 and hKv␤1.3 with members of the Kv1 subfamily, we used the yeast two-hybrid system to identify the Kv1 ␣ subunits to which hKv␤1.2 and hKv␤1.3 bind as well as to map the domains that mediate this interaction. As seen in Fig. 3, hKv␤1.3 interacts with the N terminus of hKv1.4 (aa 1-305) but not the C terminus (aa 562-654), as evidenced by the growth pattern on media lacking histidine. hKv␤1.3 also interacts with the truncated Kv1.4 N terminus, Kv1.4-⌬N2-146, indicating the binding site is within the 159 amino acids immediately preceding the putative S1 transmembrane domain. ␤-Galactosidase filter assays were also performed on these cotransformants and were positive for the expression of the lacZ reporter gene in those that grew on media lacking histidine ( Table I).
Table I also shows that it is the conserved C-terminal core of the Kv␤ subunits which mediates their binding to the Kv1␣ subunit N terminus. Yeast fusion proteins were constructed with the hKv␤1.2 N terminus (aa 1-79), hKv␤1.3 N terminus (aa 1-91), and the conserved Kv␤ C-terminal core region of 329 residues and tested for interaction with the N termini of Kv1.4 and Kv1.5. Only the Kv␤ C-terminal core region interacts with the Kv1␣ N terminus.
Relative Selectivity and Sensitivity of K ϩ Channels to Kv␤  When coinjected with hKv1.2, both ␤ subunits introduced rapid but partial inactivation in the currents (Fig. 4A). In the presence of hKv␤1.2, Kv1.2 peak whole cell currents are reduced and inactivation is introduced at potentials above 0 mV. By contrast, at saturating concentrations of hKv␤1.3, there is a smaller reduction in peak currents and the amount of inactivation, which is only apparent at very positive potentials (Ͼϩ60 mV), is less pronounced. hKv␤1.2 is a more potent modulator of Kv1.2 currents. This is further emphasized in Fig.  4B, in which I-V curves for steady-state Kv1.2 currents, plus and minus Kv␤ subunits, are shown. Clearly, hKv␤1.2 reduces Kv1.2 currents over a much larger potential range than hKv␤1.3.
For Kv1.4, which exhibits intrinsic rapid inactivation, both Kv␤ subunits accelerated the inactivation time course of the currents (data not shown). When intrinsic fast inactivation was removed by deletion of 146 amino acids from the N terminus (hKv1.4⌬N2-146), the channel expressed noninactivating currents that are also sensitive to Kv␤ subunits.
No significant changes in Kv2.1 or Kv3.1 currents were observed with either Kv␤ subunit (data not shown), confirming that hKv␤1.2 and hKv␤1.3 are specific for channels of the Kv1 subfamily.
Mechanism of hKv␤1.2 and hKv␤1.3 Modulation of hKv1.2 Currents: Single-channel Properties-To probe the mechanism by which hKv␤1.2 and hKv␤1.3 introduce inactivation and reduce peak whole-cell currents, we examined the single-channel properties of hKv1.2 with and without coexpressed Kv␤ subunits. Kv1.2 currents were chosen for analysis, since the difference in the magnitude of the effects of hKv␤1.2 and hKv␤1.3 was greatest for this channel.
Kv1.2 single channels recorded under different conditions from the same patches at a test potential of ϩ70 mV, where both Kv␤ subunits introduce inactivation, are presented in   3

and hKv␤1.2 with Kv1 family channel fragments
Yeast strain SFY526 was cotransformed with the BD (pGBT9) and AD (pGAD424) fusion plasmids as outlined. A (ϩ) in the interaction column indicates activation of the lacZ reporter gene as evidenced by blue color development in the ␤-galactosidase assay at 30°. A (Ϫ) indicates no color development up to 8 h. The results for hKv␤1.3 are reported in the left three columns and for hKv␤1.2 in the right three columns. 4. Functional effects of hKv␤1.2 and hKv␤1.3 on coexpressed Kv1␣ currents in Xenopus oocytes. A, analog data from representative experiments. Currents were elicited by 100-ms voltage steps from Ϫ70 to ϩ60 mV with 10-mV increment, delivered from a holding potential of Ϫ80 mV. Families of currents mediated by hKv1.2 alone and hKv1.2 co-expressed with hKv␤1.2 or hKv␤1.3 are shown. The voltage protocols and scales are shown in the inset. B, I-V curves constructed by plotting the current amplitude at the end of the 100 ms pulse as a function of test potential (TP). C, relative sensitivity of Kv␣ subunits to Kv␤ modulation as expressed by percent reduction of currents at the end of a pulse to ϩ60 mV relative to the total peak current amplitude at the same voltage. The difference between the effects of hKv␤1. After recording in the cell-attached mode, the same patch was excised in the inside-out configuration. For Kv1.2 alone, there was no apparent change in the appearance of either the single channels or the ensemble currents. Single Kv1.2 channels coexpressed with hKv␤1.2, however, consistently showed longer, flickery openings, more characteristic of Kv1.2 alone, upon excision (Fig. 5B). By contrast, excision of membrane patches containing Kv1.2 coexpressed with hKv␤1.3 (Fig. 5C, left panel) did not change either the single-channel phenotype or the ensemble current properties. This observation may reflect different responses of the Kv␤ subunits to the redox state of the membrane and was reexamined with macropatch recordings as shown later.
At 3. An average of 400 traces were analyzed. Unitary current amplitude was measured from openings longer than 5 ms. Single-channel conductance was determined by the slopes of linear regression indicated by the lines at potentials between Ϫ40 mV and ϩ30 mV. B, mean open probability determined from recordings of 100-ms pulses to Ϫ30 and ϩ70 mV in patches with only one channel. *, p Ͻ 0.05 (Kv␣ϩKv␤ versus Kv␣ alone, same below). C and D, mean burst duration (C) and mean open time (D) determined from single exponential fits to currents recorded at Ϫ30 and ϩ70 mV. E, closed times were best fitted with a double exponential. mV while the longer closed time 2 was significantly increased (2.90 Ϯ 0.15 ms for Kv1.2 (n ϭ 3) versus 8.27 Ϯ 0.07 ms for Kv1.2 plus hKv␤1.2 (n ϭ 3)). Thus, even at a potential at which no inactivation is apparent, hKv␤1.2 still blocked Kv1.2. hKv␤1.3 had similar effects; the mean open time was significantly decreased (1.86 Ϯ 0.29 ms; n ϭ 4; p Ͻ 0.05), and closed time 2 significantly increased (6.30 Ϯ 0.64 ms; n ϭ 4; p Ͻ 0.05). hKv␤1.3 effects were less dramatic than hKv␤1.2, however. Although the mean P o was reduced (20.3 Ϯ 3.6; n ϭ 4; p Ͼ 0.05), the decrease was not statistically significant and the burst duration was essentially unchanged. This reflects the weaker block by hKv␤1.3. Taken together, these results stress that both Kv␤ subunits produce significant block of Kv1.2 at potentials in the physiological range. They also provide an explanation for the depressed whole-cell currents and partial block conferred by the Kv␤ subunits.
hKv␤1.2 and hKv␤1.3 Differ in Response to Redox State-To test whether hKv␤1.2 and hKv␤1.3 respond differently to changes in the redox state, cell-attached and excised macropatch recordings were made in order monitor currents from the same patch under different conditions (reducing versus oxidizing). As shown in Fig. 7, Kv1.2 currents elicited by depolarizing steps did not change significantly from cell-attached to insideout patches or after addition of either H 2 O 2 (to enhance oxidation) or GSH (to enhance reduction). In the presence of hKv␤1.2, however, hKv1.2 currents in the cell-attached configuration showed clear inactivation, which largely disappeared after membrane excision to an inside-out patch in an oxidizing environment (0.1% H 2 O 2 ). In addition, current levels also increased at least 2-fold under oxidizing conditions. In some cells, membrane excision into a bath solution without H 2 O 2 was sufficient to convert inactivating to noninactivating currents as was observed in the single-channel recordings. This difference may be due in part to variable macropatch sizes and amount of cytoplasmic components adhering to the patch. Inactivation was restored and current levels decreased after perfusion with GSH (10 -20 min) (Fig. 7). In sharp contrast to hKv␤1.2, hKv1.2 currents in the presence of hKv␤1.3 did not change appreciably in response to alterations in redox conditions (n ϭ 6). Although H 2 O 2 increased current levels slightly, the extent of inactivation was unaffected.
An important finding is that with removal of inactivation in Kv1.2 currents coexpressed with hKv␤1.2, the current amplitude markedly increased under oxidizing conditions. The difference in current amplitudes between oxidizing and reducing conditions in the same patch approximates the amount of Kv1.2 current blocked by hKv␤1.2. The ratio of currents in oxidizing versus reducing conditions measured at the end of the pulse (5.2 Ϯ 1.3 at Ϫ20 mV and 4.0 Ϯ 0.5 at ϩ60 mV (n ϭ 4)) shows that Kv1.2 currents are reduced by 75-80% in the presence of hKv␤1.2 at a wide range of potentials. These values are consistent with the reduction in whole-cell current amplitudes observed upon coexpression of Kv1.2 with hKv␤1.2.
These experiments also indicate that the binding of the Kv␤ to the ␣ subunit is very stable. In some cases, excised patches in which Kv␤ inactivation was stable (hKv␤1.3) or restored upon application of GSH (hKv␤1.2) were maintained for over 1 h without loss of Kv␤ functional effects. DISCUSSION hKv␤1.2 and hKv␤1.3, both cloned from human heart, derive from alternative splicing of a single gene and contain identical C-terminal (329-amino acid) domains spliced to unique N-terminal (79 and 91 amino acids, respectively) regions (11,14). By examining the interaction of hKv␤1.2 and hKv␤1.3 with Kv1␣ subunits both biochemically and electrophysiologically, we have been able to assess the roles of both the Kv␤ C terminus and the variable N termini.
With the yeast two-hybrid system, we have shown that Kv␤ subunits are able to interact with the N terminus of each Kv1 ␣ subunit tested: Kv1.1, Kv1.2, Kv1.4, Kv1.5. This interaction is mediated by the C termini of the Kv␤ subunits, as the individual N termini alone do not interact with the Kv␣ subunits. These results are consistent with recent reports, which also assign Kv␤ binding to the N termini of Kv1 ␣ subunits (15,16), as well as with a recent report indicating association of Kv␤1.1 with all Kv1␣ subunits upon transient transfection in mammalian cells (19).
Both hKv␤1.2 and hKv␤1.3 accelerate or introduce partial inactivation into all Kv1 channels tested: Kv1.1, Kv1.2, Kv1.4, and Kv1.5. We had previously reported that hKv␤1.2 had no effect on Kv1.1 and Kv1.2 currents upon heterologous expression in Xenopus oocytes (8). The difference between these two studies is likely to be due to the levels of Kv1␣ currents that were expressed. In the first study whole-cell Kv1. 1  we attribute these differences to the actions of the unique N-terminal domains. We cannot exclude the possibility, however, that the translational efficiency of hKv␤1.3 cRNA is lower than hKv␤1.2 or that hKv␤1.3 protein is less stable than hKv␤1.2. All factors that result in lower local concentrations of hKv␤1.3 available for interaction with Kv1␣ subunits may contribute to weaker effects of hKv␤1.3 relative to hKv␤1.2.
In this report, we provide the first single-channel recordings of Kv1␣ currents in the presence of Kv␤ subunits. These data clearly show how hKv␤1.2 and hKv␤1.3 produce only partial inactivation of noninactivating Kv1␣ subunit currents as seen with whole-cell and macropatch recordings. The sustained currents remaining at the end of the pulse are reflected in the reopening of single channels observed during the pulse period. Thus, complete inactivation is not achieved since channels are able to open sporadically in the presence of the Kv␤ subunits. This is in contrast to the action of the ball peptide of inactivating Kv1 subunits like Kv1.4 in which short channel openings during the initial phase of the pulse are followed by long closings thereafter (21)(22)(23). Kv␤1.1 also introduces a more complete block than either of these Kv␤ subunits (7). By comparison, our data suggest that hKv␤1.2 and hKv␤1.3 act like "bouncing balls"; they move in and out of the channel pore to produce flickery channel openings, suggesting a less stable interaction between Kv␤ subunits and the Kv1␣ receptor site.
We have consistently observed a decrease in peak current amplitudes in whole-cell and macropatch recordings in the presence of ␤ subunits, particularly hKv␤1.2. These changes occur at potentials as negative as Ϫ30 mV. This decrease could result from either 1) Kv␤ induced reduction in the expression of Kv1␣ subunit proteins through impaired assembly and/or targeting of channel proteins to the membrane or 2) Kv␤ open channel block of Kv1␣ subunits. Our single-channel data support the latter. In the presence of the Kv␤ subunits, a decrease in the mean open probability is responsible at least in part for the decrease in macroscopic currents. Thus, even with weak depolarizations, which do not produce inactivating currents, channels are blocked by Kv␤ subunits. Our macropatch data also support an open channel block by Kv␤ subunits. For hKv␤1.2, which is sensitive to the redox state, currents recorded under oxidizing conditions increase in amplitude by at least 50%, suggesting that hKv␤1.2 depresses currents. We cannot exclude, however, that Kv␤ subunits also have effects on assembly or targeting of Kv1␣ subunits. In fact, a recent report suggests that Kv␤ subunits promote the cell surface expression of coexpressed Kv1.2 subunits in transfected mammalian cells (20).
hKv␤1.2 and hKv␤1.3 exhibit differential responses to redox conditions. The effects of Kv␤1.1 on RCK1 are also sensitive to the redox state, and this sensitivity has been mapped to the cysteine residue in position 7 of Kv␤1.1 (7). hKv␤1.2 has a cysteine in a similar position (residue 8) and shows a similar sensitivity to oxidation. Deletion of 10 amino acids, including the cysteine at position 8, from the N terminus of hKv␤1.2 removes the redox sensitivity. 2 By contrast, hKv␤1.3 does not have a cysteine at this position and is not sensitive to changes in the redox state. Differential sensitivity of Kv␤ subunits to the redox state may be important in some pathophysiologic conditions such as ischemia. With an abnormal increase in cellular oxygen radicals, hKv␤1.2 and Kv␤1.1 effects would be lost while hKv␤1.3 could still modulate Kv1 currents due to its insensitivity to oxidation.
The major effects of Kv␤ subunits as assessed electrophysiologically in heterologous expression systems are to control Kv1␣ current levels. It is intriguing that a number of different Kv␤ subunits that are able to modulate Kv1 currents to different extents have evolved. Different cell types may modulate Kv1 currents in unique ways by assembling various combinations of Kv1␣ and Kv␤ subunits. Clearly, a characterization of Kv1␣-Kv␤ interactions in native cells will be critical for determining the molecular composition of individual Kv currents and the physiological relevance of these interactions.