Interactions among Inactivating and Noninactivating Kvβ Subunits, and Kvα1.2, Produce Potassium Currents with Intermediate Inactivation*

Experiments were carried out to determine whether coinjection of Kvα1.2 with inactivating and noninactivating Kvβ subunits would produce currents with intermediate kinetics and channel complexes containing a mixture of these subunits. Upon coexpression with a saturating amount of Kvβ1.2 and increasing levels of a noninactivating deletion mutant of Kvβ1.2, we show that macroscopic Kvα1.2 currents have levels of fractional inactivation and inactivation time constants that are intermediate between those obtained with either the inactivating Kvβ1.2 or the noninactivating Kvβ1.2 mutant. We also find that coexpression of Kvα1.2 with saturating amounts of Kvβ1.2 and the deletion mutant produces a population of single channels with properties intermediate to either the inactivating or noninactivating parental phenotype. Our data can best be explained by the presence of an intermediate population of heterooligomeric channels consisting of Kvα1.2 with different combinations of both types of subunits. Since Kvα1.2 subunits coexist in cells with inactivating and noninactivating Kvβ subunits, our findings suggest that heterooligomeric assembly of these subunits occurs to increase the range of K+ current kinetics and expression levels.

Biochemical data suggests dendrotoxin-sensitive K ϩ channels exist as ␣␤ complexes (1). Moreover, both noninactivating (Kv␤2) and inactivating Kv␤ subunits (Kv␤1) have been found in complexes with Kv1.2 ␣-subunits (2,3). These observations suggest that various combinations of inactivating and noninactivating Kv␤ subunits may exist in complexes with Kv1 ␣subunit tetramers. The functional consequences of Kv␤ heterooligomerization on Kv␣1 currents has not been fully described. A recent study showed that the inactivation of a Shaker mutant produced by Kv␤1.2 can be removed by Kv␤2 leading to currents that were either inactivating or noninactivating (4). Intermediate levels of inactivation, although antic-ipated, were not observed. A model was proposed in which the Kv ␣-subunits exist as complexes with either Kv␤1 or Kv␤2 subunits. Binding of Kv␤2 to Kv␤1 was also reported, and it was suggested that this interaction might reduce the effective concentration of inactivating Kv␤1 subunits available for binding to Kv␣1 subunits.
To further investigate whether intermediate levels of inactivation might exist in the presence of both inactivating and noninactivating Kv␤ subunits, we coexpressed Kv␣1.2, Kv␤1.2 and a noninactivating Kv␤1.2 N-terminal deletion mutant in Xenopus oocytes (5). We used the mutant rather than Kv␤2, because differences in ␤-␣ binding could be excluded. Kv␤1 and Kv␤2 are homologous, but not identical, in the C terminus, the Kv␤ region shown to bind to Kv␣1 N termini (6,7). We observed macroscopic currents with inactivation kinetics intermediate to either parental phenotype. Single channel analysis revealed a population of channels with kinetics spanning inactivating to noninactivating phenotypes. We also found that Kv␤1.2 and its deletion mutant could interact in the yeast two-hybrid assay. Thus, the deletion mutant, by reducing the effective concentration of Kv␤1.2, may have contributed to the production of intermediate inactivation as suggested by Xu and Li (4). Because noninactivating (Kv␤2) and inactivating Kv␤ subunits (Kv␤1) have been found in complexes with Kv␣1.2 and all three subunits may coexist cells (2,3,8), we suggest that our single channel results may be the functional counterparts of such complexes. Some of the data have been presented in abstract form (9).

EXPERIMENTAL PROCEDURES
Yeast Two-hybrid Interaction-The Kv␣1.2 cDNA was a gift from O. Pongs (Zentrum fü r Molekulare Neurobiologie, Institut fü r Neurale Signalarbeitung, Hamburg, Germany). Kv␤1.2 was cloned as described previously (10). Protein-protein interactions were tested using the yeast GAL-4 matchmaker two-hybrid system from CLONTECH. In-frame EcoRI and SalI sites were incorporated into the 5Ј-and 3Ј-ends, respectively, of the coding sequences of Kv␤1.2 (amino acids 1-408) and Kv␤1.2-N⌬20 (amino acids 21-408) by polymerase chain reaction for cloning into the yeast shuttle vectors, pGBT9 and pGAD424. The polymerase chain reaction-amplified constructs were sequenced to confirm the correct reading frame for the yeast fusions and check that no unwanted polymerase chain reaction mutations were introduced.
Protein-protein interactions were tested in the yeast host strain Y190 by cotransformation with pairs of pGBT9 and pGAD424 fusion constructs. Cotransformants were selected on medium lacking tryptophan (trpϪ) and leucine (leuϪ) after growth for 2-3 days at 30°C. Representative colonies from these transformations were replated on trpϪ/ leuϪ media to allow direct comparison of individual colonies. Transcription of the reporter gene, lacZ, was tested by a ␤-galactosidase filter assay.
Electrophysiology-The standard two-microelectrode voltage clamp * This work was supported by the "Deutsche Forschungsgemeinschaft" (to J. K.), National Institutes of Health Grants NS23877 HL-36930, and HL55404 (to A. M. B.) and HL-57146 and a grant from the American Heart Association, Northeast Ohio Affiliate (to B. A. W.). 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. technique was used for measurement of macroscopic Xenopus oocyte currents three to 5 days postinjection. Two to four days after injection, single channel recordings were performed in the cell-attached configu-ration using pipettes pulled from hard borosilicate glass with resistances of 5-10 M⍀. Pipettes were coated with Sylgard and fire-polished immediately before use. The seal resistance was 50 -500 G⍀. Xenopus oocytes were always patched on the dark hemisphere. Two microelectrode voltage clamp measurements of Xenopus oocytes were performed in a low K ϩ solution containing (in mmol/liter): 5 KCl, 100 NaCl, 1.5 CaCl 2 , 2 MgCl 2 , and 10 HEPES (pH 7.3) or a high K ϩ solution containing 50 KOH, 72.5 NMDG, 122.5 MES, 1 and 10 HEPES (pH 7.3). Single channel measurements were performed as described previously (5,6). The pipette solution for the single channel recordings had the low K ϩ composition. The bath solution in all single channel measurements contained (in mmol/liter): 100 KCl, 1 MgCl 2 , and 10 HEPES (pH 7.3). All measurements were done at room temperature (20°C).
Single channel experiments were carried out as described previously (6). I-V protocols were performed to ensure that the investigated channel had the expected activation threshold and reversal potential. Measurements with only one channel in the patch were used for further investigation. From a holding potential of Ϫ80 mV, we measured 1000 test pulses to a potential of 70 mV at a frequency of two per second. Leak subtraction was done by averaging 10 null records and subtracting it from each trace as described previously (11). Data were low pass-filtered at 2 kHz (Ϫ3 db, 4-pole Bessel filter) before digitalization at 10 kHz. PClamp software (Axon Instruments) was used for generation of the voltage-pulse protocols and for data acquisition. All single channel measurements were leak subtracted in Fetchan and analyzed using TRANSIT software (12). This produced histograms for the open time, ensemble current traces, and values of open probability. The maximum likelihood method is utilized by TRANSIT to determine time constants from open time histograms. The data for open time from TRANSIT are presented on a log scale as described previously (12). 1 The abbreviation used is: MES, 4-morpholineethanesulfonic acid.
FIG. 1. Interactions between Kv␤1.2 and Kv␤1.2-N⌬20 using the yeast two-hybrid assay. Kv␤1.2 binds to itself as well as the N⌬20 deletion. The yeast host strain Y190 was cotransformed with GAL4 binding domain (pGBT9) and activation domain (pGAD424) fusion plasmids as indicated and plated on medium lacking trp and leu. Each plasmid was tested for autonomous transactivation by transforming with the appropriate vector, either pGBT9 or pGAD424, alone. All cotransformants grew on Ϫtrp/Ϫleu medium and were tested for activation of the reporter gene, lacZ, by a filter assay shown in the middle panel. None of the control reactions resulted in the activation of lacZ transcription, but the Kv␤ subunits tested together gave a positive signal.

Interactions between Kv␤1.2 and Kv␤1.2 N-terminal Deletion
Mutants-We tested whether Kv␤1.2 and a noninactivating form of Kv␤1.2, in which the N-terminal 20 amino acids were deleted, could interact using the yeast two-hybrid assay. This deletion had no effect on the ability of Kv␤1.2 to bind to the Kv␣1.2 N terminus (5). We found that Kv␤1.2 was able to interact with itself as well as with the N⌬20 deletion (Fig. 1). These results parallel the findings of Xu and Li (4) (5), ranging from 0 to 100 ng. The Kv␤1.2 deletion mutant was used instead of Kv␤2 to exclude differences in binding affinities between Kv␤s. The Xenopus oocyte expression system also allowed the delivery of mixed cRNAs to be measured accurately. In the presence of Kv␤1.2 alone inactivating currents were measured. The maximal amount of inactivation, calculated as the ratio of peak to steady state current was approximately 0.60 -0.65 and was similar to that found previously (5,6). Addition of increasing amounts of the deletion mutant progressively removed inactivation and slowed the rate of inactivation ( Fig. 2A). A 1:4 ratio of the deletion mutant:Kv␤1.2 cRNAs was sufficient to completely remove inactivation. Addition of the deletion mutant also slowed activation ( Fig. 2A) and increased the rate of deactivation (not shown). Fig. 2, B and C, show the averaged values for rate of inactivation and fractional inactivation in the presence of increasing amounts of the deletion mutant and saturating amounts of Kv␤1.2.
We have found similar effects on both fractional inactivation and inactivation time constants by simply decreasing Kv␤1. When we used used open time, open probability, burst duration, and closed time to distinguish between groups, 14 of 29 channels measured fell into three separate populations (Fig. 4). Fifteen channels had individual parameters that could not be classified neatly into any of the three defined groups possibly, because they possessed a combination of Kv␤ subunits that altered certain kinetic parameters more than others. Fig. 5 shows three examples of channels where all four kinetic parameters lie within one of the three defined groups described above. The line of fit was generated using a maximum likelihood method as described previously (3). Note that the x axis is on a log scale and most of the data are at the longer time points. The line of fit follows the data very closely at these longer times. G-I, corresponding open probabilities. Oocytes were held at Ϫ80 mV, and 1000 traces were recorded in the cell-attached configuration with 5 mM K ϩ in the pipette, during pulses to ϩ70 mV of 100-ms duration at a frequency of 2 Hz. These individual channels were also representative of three separate groups based on burst duration (channel A ϭ 16.55 ms; channel B ϭ 3.58 ms; channel C ϭ 1.47 ms) and closed time (channel A ϭ 0.67 ms; channel B ϭ 3.58 ms; channel C ϭ 7.50 ms).  1.2N ⌬ 20). The model depicts two channel complexes fully occupied by Kv␤1.2 or the Kv␤1.2 deletion mutant (C and D) and homo/heterooligomeric ␤-␤ complexes (E) as described for Kv␤2 and Kv␤ by Xu and Li (4). In addition, the model also shows two channel complexes not fully occupied by Kv␤1.2 (A and B), which could be responsible for the single channels with intermediate kinetics observed. units exist in a 1:1 stoichiometry (1), the data suggest that the intermediate kinetics were produced by one, or a combination, of the following three channel complexes: 1) a fully occupied mixed channel (Kv␤1.2 ϩ deletion mutant ϭ 4); 2) a partially occupied mixed channel (Kv␤1.2 ϩ deletion mutant Ͻ 4); or 3) two fully occupied populations of channels with either Kv␤1.2 or Kv␤1.2-N⌬20.
Single channel analysis showed that three groups of single channels with distinct inactivation kinetics could be distinguished when Kv␣1.2 was coexpressed with equal amounts of Kv␤1.2 and the deletion mutant. Two groups had kinetics identical to those of single channels formed from the expression of Kv␣1.2 with a saturating amount of either Kv␤1.2 or the deletion mutant. A third group had inactivation kinetics significantly different from those of either parental group. The presence of an intermediate group of channels rules out the possibility that the intermediate kinetics of macroscopic currents were produced solely by two populations of fully occupied nonmixed channels. The single channels with intermediate kinetics must have been produced by a fully or partially occupied mixed channel. This extends the scheme of Xu and Li (4) to include channel complexes not only fully occupied by either inactivating or noninactivating Kv␤ subunits (Fig. 6, C and D), but also complexes fully or partially occupied by mixed combinations of ␣ and ␤ subunits (Fig. 6, A and B). Because both noninactivating (Kv␤2) and inactivating Kv␤ subunits (Kv␤1) have been found in complexes with Kv␣1.2, and all three subunits are likely to exist in the same cell (2,3,8), we favor the view that Kv␣1.2 channel complexes fully occupied with a mixture of both inactivating and noninactivating Kv␤ subunits (Fig. 6B) underlie the single channels with intermediate kinetics. However, further experiments will be required to determine precisely the subunit stoichiometry responsible for the intermediate single channel kinetics observed here.
In summary, our results clearly demonstrate the existence of a population of single channels with intermediate inactivation kinetics. This functional evidence suggests multiple combinations of inactivating and noninactivating Kv␤ subunits exist in complexes with Kv1 ␣-subunits. Since Kv␤ subunits are also able to variably modify expression levels of Kv1 channels (5,13,14), the additional level of Kv␤-mediated regulation described in the present study likely contributes to the large range of K ϩ current kinetics and expression levels found in native tissues (15,16).