Coupling of voltage-dependent potassium channel inactivation and oxidoreductase active site of Kvbeta subunits.

The accessory beta subunits of voltage-dependent potassium (Kv) channels form tetramers arranged with 4-fold rotational symmetry like the membrane-integral and pore-forming alpha subunits (Gulbis, J. M., Mann, S., and MacKinnon, R. (1999) Cell. 90, 943-952). The crystal structure of the Kvbeta2 subunit shows that Kvbeta subunits are oxidoreductase enzymes containing an active site composed of conserved catalytic residues, a nicotinamide (NADPH)-cofactor, and a substrate binding site. Also, Kvbeta subunits with an N-terminal inactivating domain like Kvbeta1.1 (Rettig, J., Heinemann, S. H., Wunder, F., Lorra, C., Parcej, D. N., Dolly, O., and Pongs, O. (1994) Nature 369, 289-294) and Kvbeta3.1 (Heinemann, S. H., Rettig, J., Graack, H. R., and Pongs, O. (1996) J. Physiol. (Lond.) 493, 625-633) confer rapid N-type inactivation to otherwise non-inactivating channels. Here we show by a combination of structural modeling and electrophysiological characterization of structure-based mutations that changes in Kvbeta oxidoreductase activity may markedly influence the gating mode of Kv channels. Amino acid substitutions of the putative catalytic residues in the Kvbeta1.1 oxidoreductase active site attenuate the inactivating activity of Kvbeta1.1 in Xenopus oocytes. Conversely, mutating the substrate binding domain and/or the cofactor binding domain rescues the failure of Kvbeta3.1 to confer rapid inactivation to Kv1.5 channels in Xenopus oocytes. We propose that Kvbeta oxidoreductase activity couples Kv channel inactivation to cellular redox regulation.

DNA sequences amplified by the polymerase chain reaction were verified by sequencing using BigDye terminator cycle sequencing kit (PerkinElmer Life Sciences). The sequencing reactions were analyzed on an ABI 377 automated sequencer (PerkinElmer Life Sciences).
Point mutations in rat Kv␤1.1 were introduced by site-directed mutagenesis using appropriate mutation primers (17). Polymerase chain reaction products were cloned into Kv␤1.1pGEM using a DraIII and a NcoI restriction site. DNA constructs were sequenced prior to use. RNA synthesis was done using the mMessage mMachine in vitro transcription kit according to the manufacturer's protocol (Novagen Inc.).
Recording Techniques and Data Analysis-Xenopus oocytes were prepared and injected with cRNA, and electrophysiological recordings were made as previously described (18). Briefly, oocytes were injected with 50 nl of a solution containing equal amounts (25 ng) of cRNA for rKv1.5␣ and rKv␤ subunits (wild-type or Kv␤chi). Deviations from these cRNA concentrations are indicated in the figure legends. Oocytes were then incubated at 20°C for 24 -48 h in multiwell tissue culture plates (one oocyte/well) containing modified Barth's solution (88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO 3 , 0.82 mM MgSO 4 , 0.33 mM Ca(NO 3 ) 2 , 0.4 mM CaCl 2 , 7.5 mM Tris-HCl, pH 7.6, 10,000 units/l penicillin, 100 mg/l streptomycin). To record expressed membrane currents, the oocytes were held in a recording chamber (50 l volume) and continually perfused (2 ml/min) with Ringer's solution (115 mM NaCl, 2 mM KCl, 1.8 mM CaCl 2 , 10 mM HEPES adjusted to pH 7.2 with NaOH). Membrane currents were recorded with the two-microelectrode voltageclamp technique (microelectrodes filled with 3 M KCl) using a Geneclamp 500 amplifier (Axon Instruments), and signals were filtered at 2 kHz. Twenty 10 mV hyperpolarizing steps (1 s duration, 0.5 Hz) were applied and used to remove leak and capacitance currents. To construct current-voltage (I-V) relationships, the cell was held at Ϫ80 mV, 1-s duration test depolarizations at 0.1 Hz were applied in 10 mV increments from Ϫ70 to ϩ120 mV, and peak and end current amplitudes were measured. I-V curves were fitted with Boltzmann functions of the The inactivation time course was fit by a sum of two exponentials by the least squares technique. Student's t test was used to test for statistical significance. In some experiments cDNA (vector pcDNA3) for Kv1.5 (25 ng/l) together with Kv␤chi cDNA (500 ng/l) were microinjected into Chinese hamster ovary (CHO) cells. Whole-cell currents were measured with the patch clamp technique on the following day using an EPC9 amplifier and PULSE software (HEKA, Lambrecht, Germany). The extracellular solution contained (in mM) NaCl 135, KCl 5, CaCl 2 2, MgCl 2 2, HEPES 5, sucrose 10 (pH 7.4, NaOH), and recording pipettes (2-5 M⍀) were filled with intracellular solution containing (in mM) KCl 125, CaCl 2 1, MgCl 2 1, EGTA 11, HEPES 10, glutathione 2, K 2 ATP 2 (pH 7.2, KOH). All experiments were conducted at room temperature (20 -22°C).  (12). In agreement with previous results (13), wild-type Kv␤3.1 subunits did not cause rapid inactivation when we coexpressed it with Kv1.5 channels in Xenopus oocytes ( Fig. 1A; Table I). Upon depolarization of the oocyte membrane, slowly inactivating outward currents were recorded, and 84% of the initial peak amplitude remained at the end of a 1-s test pulse to ϩ80 mV (Fig. 1, A and D; Table I). By contrast, Kv1.5/Kv␤1.1 channels mediate rapidly inactivating currents both in mammalian cells (14,15) and in Xenopus oocytes (2). In our experiments they decayed to ϳ13% of the initial maximum current amplitude at the end of a 1 s test pulse to ϩ80 mV (Fig. 1, A and D; Table I). The inactivation time course (Fig. 1A) was fitted with two time constants, 1 (13.0 Ϯ 0.9 ms) and 2 (75.0 Ϯ 0.5 ms; n ϭ 14; Table I). The fast time constant 1 accounted for 90 Ϯ 10% of the total current decay. We examined the structural motifs in Kv␤3.1 responsible for its inactivation failure in Xenopus oocytes by constructing chimeras between Kv␤3.1 and Kv␤1.1. Possible structural determinants correlated with the apparent lack of function of Kv␤3.1 may be located in the N-terminal inactivating domain (2,3), the interface for association with Kv␣ subunits (11) and/or the oxidoreductase active site (Ref. 1; Fig. 1, B and C).

Kv␤1.1 but Not
Failure of Kv␤3.1 in N-type Inactivation Linked to C-terminal Domains-We connected in the first chimera the Kv␤3.1 N terminus (residues 1-229), which contains the N-terminal inactivating domain and the conserved catalytic residues (1) of the Kv␤3 oxidoreductase (Fig. 1B), with the Kv␤1 C terminus corresponding to Kv␤1.1 residues 223-401 (Kv␤chiA; Fig. 1C). The N-terminal Kv␤3.1 inactivating domain became fully functional when connected to the Kv␤1 C terminus (Fig. 1, A and  D), in agreement with the previous observation that the Kv␤3.1 inactivating domain becomes functional when connected to Kv␤2.1 (3). The results obtained with Kv␤chiA indicated that C-terminal Kv␤3.1 domain(s) must be responsible for the lack of functional interaction with Kv1.5 channels. To test this hypothesis, we constructed a reverse chimera (Kv␤chiA rev ), in which the Kv␤1 N-terminal half was linked to the Kv␤3 Cterminal half (Kv␤1 residues 1-222 and Kv␤3 residues 230 -404; Fig. 1B). The Kv1.5/Kv␤chiA rev -mediated currents exhibited inactivation kinetics similar to the ones observed for Kv1.5 with wild-type Kv␤3.1, with 80 Ϯ 1% of the maximal current amplitude remaining at the end of a 1-s depolarizing pulse ( Fig.  1A; Table I). These results confirmed the idea that the ability of Kv␤chiA to inactivate Kv1.5 channels was due to the presence of the Kv␤1.1 C terminus.
To identify the C-terminal domains responsible for the fail-  Table I). The small reduction would be compatible with a subtle difference in the affinities of Kv␤1.1 and Kv␤3.1 subunits for the Kv1.5␣ subunits (11). Nevertheless, the results demonstrated that the Kv␤3.1 domains for N-type inactivation and the subunit interface for complex formation with Kv1.5␣ subunits were not responsible for the observed inactivation failure of Kv␤3.1. The exchange of Kv␤1.1 residues by those of Kv␤3.1 was extended to residues 283-346 (Kv␤chiC, -D), which may cover the entire substrate binding site of the Kv␤ oxidoreductase (Ref. 1; Fig. 1, B and C). The respective Kv1.5/Kv␤chiC and Kv1.5/Kv␤chiD currents showed a significant reduction in the extent of inactivation (Fig. 1, A and D) because of an increase in 2 as well as an alteration in the relative weights of the fast ( 1 ) and slow ( 2 ) components (Table I). We extended the exchange of Kv␤1.1 by Kv␤3.1 residues further to residues 347-374 (Kv␤chiE), which encompassed a part of the Kv␤ oxidoreductase active site that is essential for binding the cofactor NADPH (Ref. 1; Fig. 1, B and C). The resulting Kv1.5/ Kv␤chiE currents showed an ineffective and slow inactivation (Fig. 1, A and D) due to an increase in both 1 and 2 ( Table I).
The results suggested that C-terminal Kv␤3.1 domains encompassing substrate and cofactor binding sites were responsible for the failure of Kv␤3.1 to inactivate Kv1.5 channels.
Inactivating Activity of Kv␤3.1 Rescued by Swapping with Kv␤1.1 Domains-With chimeras Kv␤chiF -I, we tried to rescue the inactivation failure of Kv␤3.1 by exchanging Kv␤3.1 domains with appropriate Kv␤1.1 domains (Fig. 1, B and C). Exchanging Kv␤3.1 residues 354 -381 for those of Kv␤1.1 produced Kv␤chiF, which conferred to the slowly inactivating Kv1.5 currents a rapid inactivation (Fig. 1, A and D; Table I). By contrast, replacement of Kv␤3.1 residues 328 -353 by those of Kv␤1.1 (Kv␤chiG) had no significant effect on inactivation (Fig. 1, A and D; Table I). Supplementation of Kv␤chiF with an additional Kv␤1.1 sequence replacing Kv␤3.1 residues 278 -309 (Kv␤chiH) did not markedly alter the activity of Kv␤chiF toward Kv1.5 channels. However, extending the replacement in Kv␤chiH by an additional 21 amino acids (Kv␤chiI) generated Kv1.5/Kv␤chiI currents with rapid and nearly complete inactivation (Fig. 1, A and D; Table I). At the end of the 1-s test pulse to ϩ80 mV the current was reduced to ϳ6% of the initial maximum current amplitude similar to Kv1.5/Kv␤1.1 currents (Fig. 1D). The data demonstrated that the exchange of two C-terminal Kv␤3.1 domains (domains I and II in Fig. 1C) (Table I). This is typically observed upon assembly of Kv␣ with Kv␤ subunits (3, 14, 19 -21). Voltages of half-maximal current activation (V 0.5 ϭ ϩ17 to ϩ22 mV) and slope  (Table I)

currents in the presence of wild-type and chimeric Kv␤ subunits
The decay phase of outward currents during a 1-s pulse to ϩ80 mV was fit by a double-exponential function, which yielded 1 , 2 , and the percentage of the total decay accounted for by 1 . The amount of inactivation was accessed by the fractional current remaining after 1 s (I 1s / I peak ). V 0.5 values and respective slope factors (k) for steady-state activation were obtained as described under "Experimental Procedures." Numbers of oocytes (n) are given for both kinetic analysis and voltage dependence of activation. In all cases 25 ng of Kv1.5 and Kv␤ cRNA was injected per oocyte. the combinations gave rise to rapidly inactivating currents ( Fig. 2A). We used a high excess of Kv␤chi cDNA (20-fold; see "Experimental Procedures") to ensure a high expression level of ␤ subunits expected to yield a maximal fraction of Kv␣/␤ complexes (22). Therefore, we also performed control experiments in Xenopus oocytes with a similar excess of Kv␤3.1 cRNA over Kv1.5 channel cRNA. The results are shown in Fig. 2, B and C. They were not significantly different from what we obtained in the experiments with equal amounts of ␣ and ␤ subunit cRNA (see Fig. 1). The results suggested that improper folding or insufficient protein expression represented the most unlikely explanations for the functional inactivity of Kv␤ chimeras in Xenopus oocytes.
Mutations of Putative Catalytic Residues Attenuate Kv␤1.1 Inactivating Activity-Because the differing inactivating activities of Kv␤1.1 and Kv␤3.1 were correlated to sequence differences in their oxidoreductase active sites, we mutated the putative catalytic residues in Kv␤1.1 (Asp-119, Tyr-124, and Lys-152), which are highly conserved in the superfamily of oxidoreductase enzymes (1). We investigated the effect of the mutations on Kv␤1.1-mediated N-type inactivation. Coexpression of the mutants Kv␤1.1D119A, Kv␤1.1Y124F, and Kv␤1.1K152A with Kv1.5 at cRNA ratios of 1:1 (Fig. 3A) and 1:5 (Fig. 3C) showed that the respective outward currents did not decay as rapidly as Kv1.5/Kv␤1.1 currents (Table II). Each mutation affected rapid inactivation behavior to a different degree. The most marked effect was observed with Kv␤1.1D119A, which did not confer rapid inactivation to Kv1.5 channels. However, 15-s test pulses revealed that Kv␤1.1D119A did accelerate the slow inactivation of Kv1.5 currents (Fig. 3, B and D), which most likely represents a C-type inactivation (23,24). As N-type and C-type inactivation are coupled (25), the observed acceleration of Kv1.5 inactivation was probably due to some residual inactivating activity of the Kv␤1.1D119A subunit.

DISCUSSION
It has been shown that rapid N-type inactivation of Shaker Kv channels requires the presence of an N-terminal inactivating domain and of a receptor close to the inner entrance of the Shaker channel pore (4). Upon depolarization the inactivating domain rapidly binds to the receptor and thereby occludes the pore. Kv␤3.1, like Kv␤1.1, contains a functional inactivating domain (3) and thus may confer rapid N-type inactivation to Shaker type channels. However, the inactivating activity of Kv␤3.1 depends on the in vitro expression system; in CHO cells Kv␤3.1 confers rapid inactivation to Kv1.5 channels (12) but fails to do so in Xenopus oocytes (3,13). Our results demonstrate that the lack of function of Kv␤3.1 in Xenopus oocytes was correlated with two C-terminal Kv␤3.1 domains encompassing the NADPH and substrate binding sites, respectively, of the Kv␤ oxidoreductase active site. Domain I in Fig. 1C and Fig. 4 contributes to the Kv␤ NADPH cofactor binding site (Ser-325, Gln-329, Glu-332, and Asn-333 in Kv␤2.1; Ref. 1). Seven of the eight domain I residues that differ between Kv␤1.1 and Kv␤3.1 (Fig. 4A) are near or at the Kv␤ adenosinebinding pocket (Fig. 4, B and C). Residues in domain II (Kv␤1.1 amino acids 303-323; Fig. 4A) have been proposed to participate in substrate binding, in particular Kv␤1.1 residue Trp-306 that corresponds to Trp-272 in Kv␤2 (1). When domains I and II in Kv␤3.1 were replaced by those in Kv␤1.1, the resulting Kv␤3.1/Kv␤1.1 chimeras were able to confer rapid inactivation to Kv1.5 channels in the Xenopus oocyte expression system. The results indicated that the functional activity of Kv␤3.1 in Xenopus oocyte could be rescued by replacing Kv␤3.1 amino acid residues in the oxidoreductase site by the ones of Kv␤1.1. In agreement with the assumption that C-terminal Kv␤3 domain(s) are responsible for the observed lack of function, Kv␤1.1 was rendered non-functional when the C-terminal half of the protein was replaced by Kv␤3.1 sequences.
Three main alternatives may be considered to understand the results: i) Kv␤3.1 is not active because the oxidoreductase site is not properly folded; ii) Kv␤3.1 activity is inhibited in Xenopus oocytes by an as yet unknown factor; and iii) Kv␤3.1 is not active because the Xenopus oocytes do not provide a substrate for the Kv␤3.1 oxidoreductase. The results showed that Kv␤3.1 and Kv␤ chimeras are functionally active in CHO cells. This demonstrates that active and properly folded Kv␤3.1 protein can be expressed in in vitro expression systems. Most likely, translation and folding of Kv␤3.1 protein in Xenopus oocytes is not different from that in CHO cells. Therefore, it is unlikely that Kv␤3.1 (and the Kv␤ chimeras) is not properly folded when expressed in Xenopus oocytes. The existence of a putative inhibitory factor in Xenopus oocytes, which is specific for Kv␤3.1, cannot be rigorously excluded but seems to be also unlikely. Thus, we assume that Kv␤3.1 fails to confer rapid inactivation to Kv1.5 channels because its oxidoreductase activity is not functioning in Xenopus oocytes. In agreement with this assumption we find that a replacement of the NADPH and the putative substrate binding domains by those of Kv␤1.1 The decay phase of outward currents during a 1-s pulse to ϩ80 mV was fit by a double-exponential function ( 1 , 2 , and percentage of the total decay accounted for by 1 ). The amount of inactivation was accessed by the fractional current remaining after 1 s (I 1s /I peak ). The fitting parameters for Kv1.5 alone and for Kv1.5 ϩ Kv␤1.1D119A were obtained from 15-s pulses. V 0.5 values and respective slope factors (k) for steady-state activation were obtained as described under "Experimental Procedures." The values in brackets are from oocytes, which were injected with a 1:5 ratio of Kv1.5 versus Kv␤1.1 cRNA and pulsed to ϩ60 mV for kinetic analysis. In all cases 3 ng of Kv1.5 cRNA was injected per oocyte. Numbers of oocytes (n) are given for both kinetic analysis and voltage dependence of activation.
reconstitutes the inactivating activity of Kv␤3.1 in Xenopus oocytes.
Because Kv␤3.1 oxidoreductase activity is apparently important for conferring rapid inactivation to Kv1.5 channels, we explored the possibility that mutations of catalytic residues in Kv␤1.1 (Asp-119, Tyr-124, Lys-152) may attenuate the Kv␤1.1 inactivating activity. The catalytic residues are highly conserved among the active sites of oxidoreductases (1), and comparable mutations in established oxidoreductase enzymes have been shown to impair catalytic activity (26,27). The results showed that the mutations severely affected the ability of Kv␤1.1 to confer rapid inactivation to Kv1.5 channels. Although the putative Kv␤ enzymatic activity could not be tested directly, it is likely that the mutations of catalytic residues in Kv␤1.1 affected its putative oxidoreductase activity. We propose that Kv␤ oxidoreductase catalytic activity is required for the inactivating activity of Kv␤1.1. Thus, manipulations of the putative Kv␤1.1 and Kv␤3.1 oxidoreductase active sites were correlated with a loss and, respectively, gain of inactivating activity in the Xenopus oocyte expression system.
Although we have not carried out biochemical experiments to test directly the binding of Kv␤3.1 to Kv1.5, the effects of Kv␤3.1 and the different Kv␤1.1/Kv␤3.1 chimeras on the voltage-gating properties of Kv1.5 are a clear indication that Kv␤3.1 binds to Kv1.5. Changes in the voltage-gating properties of Kv1 channels are typically observed upon assembly of Kv␣ with Kv␤ subunits (3, 14, 19 -21). In conclusion, Kv␤3.1 assembles with Kv1.5 channels, but the activity of the Nterminal Kv␤3.1 inactivating domain is impaired. Apparently, the effects of Kv␤3.1 on the voltage-dependent activation of Kv1.5 channels are distinct from those leading to rapid inactivation. This is in agreement with the previous observation that removal of 10 amino acids from the Kv␤1.3 N terminus eliminated the inactivation activity but not the voltage shift of activation of Kv1.5 channels (28).
Kv␤2 subunits do not have an N-terminal inactivating domain. When coexpressed with Kv␣ subunits, Kv␤2 may also alter the voltage-gating properties of Kv channels and, in addition, enhance trafficking of Kv channels to the plasma membrane. In agreement with our results, it has been shown in a recent report (29) that mutating active site residues in Kv␤2 did not interfere with its binding to Kv1.4 channels. These Kv␤2 mutants still affected the voltage-gating properties of Kv1.4 channels like wild-type Kv␤2, but the enhancing effects on Kv1.4 channel surface expression were attenuated. Apparently, the putative Kv␤ oxidoreductase activity is important for distinct aspects of Kv␤ function.
Previously, we have shown that N-type inactivation can be prevented by a NIP-domain in Kv1.6 subunits (15). Now, we show that N-type inactivation of Kv channels may be coupled to the putative Kv␤1.1 and Kv␤3.1 oxidoreductase activity. This observation indicates that the gating mode of Kv channels linked to N-type inactivation may be regulated by a variety of cellular mechanisms. We propose that the presence of an oxidoreductase activity in Kv channels may couple cellular redox regulation to the gating mode of Kv channels allowing the channels to switch between a rapidly inactivating and a noninactivating mode. Identifying the Kv␤ oxidoreductase substrate(s) will bring us closer to understanding the cellular function of such potential energetic coupling.