A Mammalian Transient Type K+ Channel, Rat Kv1.4, Has Two Potential Domains That Could Produce Rapid Inactivation*

The “ball and chain” model has been shown to be suitable for explaining the rapid inactivation of voltage-dependent K+ channels. For theDrosophila Shaker K+ channel (ShB), the first 20 residues of the amino terminus have been identified as the inactivation ball that binds to the open channel pore and blocks ion flow (Hoshi, T., Zagotta, W. N., and Aldrich, R. W. (1990)Science 250, 533–538; Zagotta, W. N., Hoshi, T., and Aldrich, R. W. (1990) Science 250, 568–571). We studied the structural elements responsible for rapid inactivation of a mammalian transient type K+ channel (rat Kv1.4) by constructing various mutants in the amino terminus and expressing them in Xenopus oocytes. Although it has been reported that the initial 37 residues might form the inactivation ball for rat Kv1.4 (Tseng-Crank, J., Yao, J.-A., Berman M. F., and Tseng, G.-N. (1993) J. Gen. Physiol. 102, 1057–1083), we found that not only the initial 37 residues, but also the following region, residues 40–68, could function independently as an inactivation gate. Like the Shaker inactivation ball, both potential inactivation domains have a hydrophobic amino-terminal region and a hydrophilic carboxyl-terminal region having net positive charge, which is essential for the domains to function as an inactivation gate.

Aldrich and co-workers have shown that a "ball and chain" model, originally proposed for Na ϩ channel inactivation (4), can also explain the rapid inactivation of a Drosophila Shaker K ϩ channel (1,2). The amino-terminal domain (ball) tethered by the adjacent region (chain) to the channel protein binds to the channel pore after channel activation and blocks ion flow. In the Shaker K ϩ channel (ShB), the initial 20 amino acids have been identified as the inactivation ball. The following region preceding the assembly domain (5) has been identified as the chain tethering the ball to the channel (1). The 20-amino acid inactivation ball is composed of the 11 amino-terminal hydrophobic residues and the following 9 hydrophilic residues containing net positive charge. Both the hydrophobic stretch and the charged region are thought to be involved in the binding of the ball to its receptor via hydrophobic and electrostatic interactions. In contrast, in mammalian transient type K ϩ channels, the ball and chain structure had not been well defined, although it had been shown that deletion of various lengths from the amino-terminal region of Kv1.4 disrupted rapid inactivation suggesting the presence of a "ball" structure (6,7). Tseng and co-workers (3) have studied this issue in more detail by deleting different domains in the amino-terminal region of rat Kv1.4. They have not identified "chain" structure but have shown that deletion of the amino-terminal hydrophobic domain, residues 3-25, resulted in loss of rapid inactivation. Deletion of the following hydrophilic region containing five positive and two negative charges, residues 26 -37, greatly attenuated inactivation. Based on these and other findings, they suggested that the amino terminus of rat Kv1.4 might be similar to that of ShB in having one inactivation ball, which is composed of the initial 37 residues. In the present study, we investigated the structural elements responsible for rapid inactivation of rat Kv1. 4 and have identified another domain that can produce rapid inactivation independently of the proposed inactivation ball.

EXPERIMENTAL PROCEDURES
In Vitro Mutagenesis- Fig. 1 shows the amino-terminal sequences of Kv1.4 and the mutants investigated in this study. Eleven deletion mutants and one addition mutant were made in the amino-terminal region of Kv1.4. In addition, one mutant in which amino acid residues 40 -68 of Kv1.4 were inverted in ⌬2-39 & ⌬69 -162 was constructed. Fragments for all the mutants except the one with inverted residues were generated by polymerase chain reaction (PCR). 1 The 20 -22-base pair sense primers used for generating ⌬2-25, ⌬2-26, ⌬2-28, ⌬2-30, ⌬2-32, ⌬2-39, and ⌬2-61 corresponded to the appropriate region in Kv1.4 and contained an ApaI site, unique within the multiple cloning site of the vector pBluescript II, and an ATG at the 5Ј-end. The antisense primer (AS1) complementary to nucleotides (nt) 532-551 of Kv1.4 was used for the above seven mutants. The sense primer used for generating ⌬29 -162 corresponded to nucleotides 487-506 of Kv1.4 and contained a XhoI site (which is unique in Kv1.4 at nt 80) at the 5Ј-end; the antisense primer was AS1. For constructing ⌬2-39 & ⌬69 -162, the sense primer (S1), with an ApaI site at the 5Ј-terminus and corresponding to nucleotides Ϫ35 to Ϫ16 of Kv1.4, was used with an antisense primer complementary to nucleotides 180 -199 with a XhoI site at its 5Ј-end. To generate ⌬38 -162 and ⌬2-39 & ⌬61-162, two fragments, amplified by PCR, were ligated into the mutants. The upstream fragment for each mutant was designated fragment I; the downstream fragment was fragment II. Fragment II for both mutants was the same and corresponded to amino acid residues 163-185 of Kv1.4. The sense primer for fragment I of ⌬38 -162 was S1, and that of ⌬2-39 & ⌬61-162 was the same one used for ⌬2-39. The antisense primer for the fragment I contained a StuI site at the 5Ј-end and corresponded to nucleotides 96 -114 for ⌬38 -162 and to nucleotides 161-181 for ⌬2-39 & ⌬61-162. The sense primer for fragment II corresponded to nucleotides 490 -508 and contained a SmaI site at the 5Ј-end; the antisense primer was AS1. Amino acid residues 26 -39 of Kv1.4 were added to the amino terminus of Kv1.4 in the addition mutant. To make the addition mu-* This study was supported by Grants-in-aid 07557170 for Scientific Research and 08268202 for Scientific Research on Priority Areas of "Channel-Transporter Correlation" from the Ministry of Education, Science and Culture, Japan, and by the Sapporo Bioscience Foundation, Nishinomiya Basic Research Fund, Kanae Foundation of Research for New Medicine, and a Japan Heart Foundation Research Grant for 1995. 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.
‡ To whom correspondence should be addressed. Present address: Dept. of Pharmacology, Yamagata University School of Medicine, 2-2-2 Iida-nishi, Yamagata 990-23, Japan. Tel.: 81-236-28-5234; Fax: 81-236-28-5235. tant, two fragments (fragment I was generated by PCR and fragment II was cut out from Kv1.4) were ligated into the mutant. The sense primer for fragment I of the addition mutant corresponded to nucleotides 76 -99 and contained an ApaI site and ATG at the 5Ј-end; the antisense primer was complementary to nucleotides 97-116 and contained a NcoI site at the 5Ј-end. To make amino acid residues 40 -68 (AALAVAAATA-AVEGTGGSGGGPHHHHQTR) invert in ⌬2-39 & ⌬69 -162, the sense oligonucleotide that codes for MRTQHHHHPGGGSGGTGEVAATA-AAVALAA and the antisense oligonucleotide complementary to it were used. They were designed to produce an ApaI site at the 5Ј-end and a MluI site at the 3Ј-end when annealed. The annealed fragment was ligated to Kv1.4, which was digested with ApaI and MluI. In a 100-l PCR reaction, 100 pmol of each primer, 0.2 g of template cDNA (Kv1.4 for all the mutants except ⌬2-39 & ⌬69 -162; ⌬2-39 for ⌬2-39 & ⌬69 -162), and 5.0 units of Taq DNA polymerase (Perkin-Elmer) were used. Reaction temperatures were varied using a thermal cycler (Perkin-Elmer): 94°C, 1 min; 55°C, 2 min; and 72°C, 3 min for 25 cycles. The amplified fragment for ⌬2-25, ⌬2-26, ⌬2-28, ⌬2-30, ⌬2-32, ⌬2-39, and ⌬2-61 was digested with ApaI and MluI and ligated to Kv1.4 between the ApaI and MluI sites. The amplified fragment for ⌬29 -162 was digested with XhoI and MluI and ligated to Kv1.4 between the XhoI and MluI sites. The fragment for ⌬2-39 & ⌬69 -162 was amplified using ⌬2-39 as template and digested with ApaI and XhoI. The digested fragment was ligated to ⌬29 -162, which was digested with ApaI and XhoI. The amplified fragment I for ⌬2-39 & ⌬61-162 and ⌬38 -162 was digested with ApaI and StuI, and the other fragment (fragment II) was digested with SmaI and MluI. These two fragments were ligated to Kv1.4, which was digested with ApaI and MluI. The amplified fragment I for the addition mutant, which was digested with ApaI and NcoI, fragment II, which was cut out from Kv1.4 with NcoI (at nt Ϫ2), and PmaCI (at nt 200) were ligated to Kv1.4 digested with ApaI and PmaCI. Sequences of all the fragments generated by PCR in the mutants were verified on both strands by the dideoxy chain termination method using an A.L.F. DNA Sequencer II (Pharmacia Biotech Inc.).
Expression and Current Recording-Electrophysiological measurements were carried out essentially as reported previously (8,9). The pBluescript II vectors containing the constructs were linearized with EcoRI, and cRNAs were prepared from these templates with T7 RNA polymerase (Stratagene). Transcribed RNAs were dissolved in water at a final concentration of 0.2 g/l for oocyte injection. The integrity of the cRNA products was checked by running the samples on formaldehyde containing agarose gels (10). Defolliculated Xenopus oocytes (stage V-VI) were injected with 40 -50 nl (8 -10 ng) of cRNA. The injected oocytes were incubated in Barth's medium supplemented with penicillin G (71.5 units/ml) and streptomycin (35.9 g/ml) at 18°C for 2-4 days before doing electrophysiological measurements. The K ϩ currents were recorded by a conventional two-microelectrode voltage clamp method with 3 M KCl-filled electrodes as described (8,9). The basic bath recording solution consisted of ND 96 (96 mM NaCl, 2 mM KCl, 1.8 mM CaCl 2 , 1 mM MgCl 2 , 5 mM HEPES, pH 7.5). For the bath solution containing high K ϩ (20K), Na ϩ was replaced with K ϩ . All electrophysiological measurements were carried out at room temperature (21 Ϯ 1°C). Current records were low pass-filtered at 3 kHz.
All data are expressed as the mean Ϯ S.E. The statistical significance was evaluated using Student's paired or unpaired t test. A p value smaller than 0.05 was considered to be significant.

RESULTS
Oocytes expressing Kv1.4 and all the mutant channels showed voltage-dependent outward currents upon depolarization (data not shown). They were held at Ϫ80 mV and depolarized to test potentials.
Effects of Inverting the Amino Acid Sequence of the Second Potential Ball-We constructed a mutant in which the second potential inactivation ball was inverted (Inv(40 -68)). The inverted ball has positive charge at the amino terminus and a hydrophobic region at the carboxyl terminus. This mutant showed little inactivation during a 400-ms test pulse to ϩ20 mV, whereas the parent mutant showed rapid inactivation (Fig. 5C, upper panel). The inact of Inv(40 -68) recorded during a 5000-ms pulse was 2616.3 Ϯ 252.8 ms (n ϭ 5). Inverting the second potential ball resulted in the loss of rapid inactivation. There were no differences in rec of the mutant between 2 mM [K ϩ ] o (4.74 Ϯ 0.16 s) and 20 mM [K ϩ ] o (4.88 Ϯ 0.48 s; n ϭ 5) (Fig. 5C, lower panel). DISCUSSION We found that there are two potential inactivation balls in the amino-terminal region of rat Kv1.4. Deletion of amino acids 2-28 resulted in loss of rapid inactivation. This is consistent with the finding of Tseng and co-workers (3), who found that deletion of residues 3-25 disrupted rapid inactivation. Surpris-ingly, deletion of 11 more residues resulted in reappearance of rapid inactivation even though the ⌬2-39 mutant did not have the core hydrophobic region of the inactivation ball. With further deletion of residues 40 -61, rapid inactivation disappeared again. It seems probable that besides the inactivation ball proposed by Tseng and co-workers (the initial 37 residues), there exists a second potential inactivation ball having residues 40 -61 as an essential domain. To confirm the presence of two potential balls, we made deletion mutants that had only one potential ball and lacked most of the amino-terminal region preceding the assembly domain (⌬38 -162 and ⌬2-39 & ⌬69 -162). As expected, the currents of both the mutants showed rapid inactivation, which indicated that the two potential ball, residues 2-37 and residues 40 -68, respectively, could produce rapid inactivation independently. Comparison of ⌬38 -162 and ⌬2-39 & ⌬69 -162, both of which lack most of the possible chain region, gave some information about the characteristic differences between the first and the second ball. In the case of the second ball (in ⌬2-39 & ⌬69 -162), inactivation was more rapid and the recovery from inactivation was slower than in the case of the first ball (in ⌬38 -162) (Figs. 1 and 3). This suggests that the second ball may have a higher affinity for the receptor than the first ball. Compared with Kv1.4, binding between the ball and the receptor seems to be much stronger for the second ball than for the ball in wild type Kv1.4, as ⌬2-39 currents recover significantly more slowly than Kv1.4 currents (Fig. 1). Among the mutants investigated, the ones that have the second ball recovered from inactivation most slowly (⌬2-39 and ⌬2-39 & ⌬69 -162). The recovery rates of their currents were significantly slower than those of the other mutants. The ⌬2-39 & ⌬69 -162 currents recovered faster than ⌬2-39 currents, probably reflecting the influence of the chain region on binding of the ball to the receptor. The presence of residues 69 -162 caused slowing of the recovery from inactivation of the ⌬2-39 current.
Similar to the structure of ShB inactivation ball, the two potential balls in Kv1. 4 have an amino-terminal hydrophobic region and a carboxyl-terminal hydrophilic region containing net positive charge, which is thought to be involved in the binding of the inactivation particle to its receptor via electrostatic interactions (2,13,14). Therefore we investigated the contribution of positive charge. Deletion of positive charge from either ball greatly attenuated the inactivation rates and accelerated the recovery rates, which probably reflects the higher affinity of the ball to the receptor site with the carboxyl-terminal positive charge. This result clearly indicates that the positive charge at carboxyl-terminal region of the ball plays an important role. The structural requirements for the inactivation ball were further studied by deleting or adding positive charges in the amino-terminal region of Kv1.4 and by inverting the amino acid sequence of the potential inactivation ball. Since the structure of non-inactivating ⌬2-28 was just like having net positive charge (3 arginine and 2 glutamic acid residues) at the amino terminus of the second potential ball of rapidly inactivating ⌬2-39, the mutants with different numbers of charges were constructed (⌬2-32, ⌬2-30, ⌬2-26 and ⌬2-25). The currents through these mutants hardly inactivated (Fig. 5A, lower panel). These results suggest that one extra positive charge at the amino terminus of the second inactivation ball is enough to disrupt its function. Therefore, the influences of net positive charges at the amino terminus of the inactivation ball of Kv1.4 were studied by adding residues 26 -39 (three net positive charges) at the amino-terminal end of Kv1.4 ((ϩ)Kv1.4). The currents through (ϩ)Kv1.4 inactivated but the rate of inactivation was significantly slower than for wild type Kv1.4 (Fig. 5B). These results indicate that net positive charge at the amino-terminal end of the inactivation ball of wild type Kv1.4 has a profound effect on function. Together the results indicate the structural requirements of the inactivation ball(s) are an amino-terminal hydrophobic region and a carboxyl-terminal hydrophilic region containing net positive charge.
In agreement with the results of Tseng and co-workers, changing [K ϩ ] o had no effects on the recovery rate for our mutants, which did not show rapid inactivation. Elevating [K ϩ ] o accelerated the recovery rate in mutants with rapid inactivation, which might reflect repulsion of the inactivation ball by K ϩ ions (11).
The most striking finding in the present study is that there exist two potential inactivation balls in the amino terminus of rat Kv1.4. It is not known how two inactivation balls could work in wild type Kv1.4. One of the two potential domains might function as the inactivation gate, or one inactivation gate might be composed of both domains. Alternatively, the redundancy of inactivation balls might be a safety device to ensure inactivation. The synthetic ShB inactivation ball peptide has been reported to block several types of K ϩ channels and also cyclic nucleotide gated channels (15)(16)(17)(18)(19). It will be of interest to synthesize the peptides corresponding to the first domain, the second domain and both the domains of Kv1.4, and to compare their effects on currents of the non-inactivating mutant of Kv1.4 and the other channels. Synthetic peptides could give useful information about how the two domains contribute to form the inactivation ball in wild type Kv1.4. Recently, NMR structures of the inactivation peptides of Kv3.4 (the initial 30 residues) and Kv1.4 (the initial 37 residues) have been re-ported. The inactivation peptides have a similar characteristic surface charge pattern with a positively charged, a hydrophobic, and a negatively charged region (20). The inactivation peptide of Kv1.4 whose NMR structure was determined corresponds to our first inactivation ball. It will be of interest to determine the NMR structure of the second domain and both the domains in the amino-terminal region of Kv1.4.