Amino-terminal Determinants of U-type Inactivation of Voltage-gated K+ Channels*

The T1 domain is a cytosolic NH2-terminal domain present in all Kv (voltage-dependent potassium) channels, and is highly conserved between Kv channel subfamilies. Our characterization of a truncated form of Kv1.5 (Kv1.5ΔN209) expressed in myocardium demonstrated that deletion of the NH2 terminus of Kv1.5 imparts a U-shaped inactivation-voltage relationship to the channel, and prompted us to investigate the NH2 terminus as a regulatory site for slow inactivation of Kv channels. We examined the macroscopic inactivation properties of several NH2-terminal deletion mutants of Kv1.5 expressed in HEK 293 cells, demonstrating that deletion of residues up to the T1 boundary (Kv1.5ΔN19, Kv1.5ΔN91, and Kv1.5ΔN119) did not alter Kv1.5 inactivation, however, deletion mutants that disrupted the T1 structure consistently exhibited inactivation phenotypes resembling Kv1.5ΔN209. Chimeric constructs between Kv1.5 and the NH2 termini of Kv1.1 and Kv1.3 preserved the inactivation kinetics observed in full-length Kv1.5, again suggesting that the Kv1 T1 domain influences slow inactivation. Furthermore, disruption of intersubunit T1 contacts by mutation of residues Glu131 and Thr132to alanines resulted in channels exhibiting a U-shaped inactivation-voltage relationship. Fusion of the NH2terminus of Kv2.1 to the transmembrane segments of Kv1.5 imparted a U-shaped inactivation-voltage relationship to Kv1.5, whereas fusion of the NH2 terminus of Kv1.5 to the transmembrane core of Kv2.1 decelerated Kv2.1 inactivation and abolished the U-shaped voltage dependence of inactivation normally observed in Kv2.1. These data suggest that intersubunit T1 domain interactions influence U-type inactivation in Kv1 channels, and suggest a generalized influence of the T1 domain on U-type inactivation between Kv channel subfamilies.

The inactivation mechanisms exhibited by different voltagegated potassium (Kv) 1 channels provide important physiological means by which the duration of action potentials in many excitable tissues is regulated at different frequencies and potentials. Inactivation of Kv channels has historically been divided into two categories, fast (N-type) inactivation which involves occlusion of the inner pore by an NH 2 -terminal ball, and slow (C-type) inactivation which involves a concerted constriction of the outer mouth of the channel pore (1)(2)(3). However, recent studies have distinguished a second slow inactivation phenotype termed U-type inactivation, which has been characterized in several voltage-gated K ϩ channels, including Kv2.1 (4) and most recently in Shaker and Kv3.1 (5). U-type inactivation has been named for its characteristic U-shaped inactivation-voltage relationship, showing maximal inactivation at intermediate potentials where only a fraction of channels are open, and less pronounced inactivation at more positive potentials where channel opening has saturated (5). This U-shaped voltage-dependence of inactivation is caused by preferential inactivation from channel closed states, although the conformational changes underlying U-type inactivation remain unclear. Interestingly, whereas C-type inactivation is slowed by elevation of extracellular K ϩ , this condition generally accelerates U-type inactivation, suggesting a distinct mechanism for channel inactivation (4,5). In addition, our laboratory has recently demonstrated that a naturally occurring NH 2 -terminal truncated form of the cardiac potassium channel Kv1.5 (Kv1.5⌬N209) exhibits a U-type inactivation phenotype (6 -8). This finding clearly suggests that Kv1.5 possesses machinery to undergo both C-type and U-type inactivation, and directed our attention toward an investigation of the NH 2 terminus as a potential regulatory site of U-type inactivation in Kv1.5 and other channels.
A number of recent studies have investigated a modular architecture of potassium channel gating machinery, considering the membrane-bound segments of potassium channels as interchangeable pore modules and voltage-sensing modules (9,10). However, structural and biochemical evidence for modularity within Kv channels is strongest with respect to the NH 2 -terminal T1 domain, a roughly 120-amino acid cytosolic region that is highly conserved within Kv channel subfamilies. Crystal structures determined for T1 domains of both Shaker and Shaw family channels demonstrate that the T1 domains are arranged as a rotationally symmetrical tetramer which is thought to lie in alignment with the channel pore (11). Despite significant primary sequence differences between different Kv channel subfamilies, the structural scaffold of the T1 domain is common to all Kv channels (12)(13)(14). Tetramerization of the T1 domains is an early step in channel biosynthesis, and crystallographic, biochemical, and electrophysiological evidence suggest that the T1 tetramer exists as a module distinct from the transmembrane channel core, adopting a "hanging gondola" structure away from the inner pore of intact Kv channels (13,(15)(16)(17)(18). The T1 domain influences numerous fundamental channel functions, including interaction of channels with Kv␤ subunits (19 -22), interaction with many intracellular signaling molecules (23)(24)(25), and prevention of heteromultimerization between different Kv channel subfamilies (14, 18, 26 -28). Although the interactions between the T1 domains and gating elements of intact channels have not been identified, a number of recent studies have characterized the influence of the T1 domain on channel gating. Specifically, it has been demonstrated that deletion mutations and point mutations in the T1 domain can substantially alter the voltage dependence and kinetics of activation in Shaker-related channels, suggesting conformational coupling of the T1 domain and the transmembrane segments of the channel (29 -31). The influence of the T1 domain on channel inactivation has yet to be established.
To investigate the molecular basis for the modulation of slow inactivation by the Kv1.5 NH 2 terminus, we have characterized the gating properties of a series of NH 2 -terminal deletions of human Kv1.5, and chimeric constructs that we substituted the NH 2 terminus of Kv1.5 with the NH 2 terminus of other Kv1 channels. Our study demonstrates that the NH 2 -terminal region responsible for modulation of slow inactivation in Kv1.5 lies within the T1 domain. Furthermore, we demonstrate that fusion of the NH 2 terminus of Kv2.1 to the transmembrane segments of Kv1.5 imparts a U-shaped inactivation-voltage relationship to Kv1.5, whereas the NH 2 terminus of Kv1.5 attenuates the U-type inactivation properties of Kv2.1.

EXPERIMENTAL PROCEDURES
Cell Preparation and Transfection-Unless otherwise stated, experiments were carried out on transiently transfected HEK 293 cells grown in minimal essential medium with 10% fetal bovine serum, at 37°C in an air, 5% CO 2 incubator. In a few experiments, mouse ltkϪ cells were used when expression of a construct in HEK 293 cells proved difficult. One day before transfection, cells were plated on sterile glass coverslips in 35-mm Petri dishes with 20 -30% confluence. On the day of transfection, cells were washed once with minimal essential medium with 10% fetal bovine serum. To identify the transfected cells efficiently, channel DNA was co-transfected with the vector pHook-1 (Invitrogen). This plasmid encodes the production of an antibody to the hapten phOX, which was expressed and displayed on the cell surface. Channel DNA was incubated with pHook-1 (1 g of pHook, 1-3 g of channel DNA) and 4 l of LipofectAMINE 2000 (Invitrogen) in 100 l of serumfree media, then added to the dishes containing HEK 293 cells in 1 ml of minimal essential medium with 10% fetal bovine serum. Cells were allowed to grow overnight before recording. One hour prior to experiments, cells were treated with beads coated with phOX. After 15 min, excess beads were washed off with cell culture medium, and cells that had beads stuck to them were used for electrophysiological recordings. For some experiments, stable HEK 293 cell lines expressing full-length (FL) Kv1.5, Kv1.5⌬N209, or Kv2.1 were employed. These HEK 293 cells were stably transfected with FL Kv1.5, Kv1.5⌬N209, or rKv2.1 cDNAs in pcDNA3 using LipofectACE reagent (Invitrogen).
Electrophysiological Procedures-Coverslips containing cells were removed from the incubator before experiments and placed in a superfusion chamber (volume 250 l) containing the control bath solution at ambient temperature (22-23°C) and perfused with bathing solution throughout the experiments. Whole cell current recording and data analysis were done using an Axopatch 200A amplifier and pClamp 8 software (Axon Instruments, Foster City, CA). Patch electrodes were fabricated using thin-walled borosilicate glass (World Precision Instruments, Sarasota, FL). Electrodes had resistance of 1-3 M⍀ when filled with control filling solution. Capacity compensation and 80% series resistance compensation were used in all whole cell recordings. No leak subtraction was used when recording currents, and zero current levels are denoted by the dotted lines in the current tracings in Figs. 1C, 2, and 6D. Data were sampled at 10 -20 kHz and filtered at 5-10 kHz. Membrane potentials have not been corrected for small junctional potentials between bath and pipette solutions. Throughout the text the data are presented as mean Ϯ S.E.
Molecular Biology and Channel Mutations-The mammalian expression vector pcDNA3 was used for expression of all channel constructs used in this study. All primers used were synthesized by Sigma-Genosys (Oakville, Ontario, Canada). All constructs were sequenced to check for sequence errors, and to ensure the correct reading frame. The Kv1.5⌬N19, Kv1.5⌬N91, and Kv1.5⌬N162 mutants were generated by Bal31 exonuclease digestion from the 5Ј end of hKv1.5. Resulting fragments were ultimately subcloned into a pcDNA3 vector cut with NcoI (which was blunted to introduce a start codon) and XbaI restriction enzymes. The Kv1.5⌬N119 and Kv1.5⌬N140 mutants were generated by PCR amplification of the cDNA encoding residues 120 or 141 to the COOH terminus of hKv1.5. The 5Ј primers used were 5Ј-CCCAAGCT-TATGCAGCGCGTCCACATCAACATC-3Ј for Kv1.5⌬N119, and 5Ј-CCCAAGCTTATGGGCACCCTGGCGCAGTTTCC-3Ј for Kv1.5⌬N140 (introduced restriction sites are underlined). The resulting channels were ultimately subcloned in a pcDNA3 vector using HindIII and NotI restriction sites. The Kv1.5⌬N188 mutant was generated by removal of the NcoI-HincII fragment of Kv1.5. Kv1.5⌬N209 was generated by removal of the NcoI-NcoI fragment of Kv1.5. Kv2.1⌬N101 was generated by removal of sequence up to the NarI restriction site in rKv2.1.
For preparation of the Kv1.1N/Kv1.5 and Kv1.3N/Kv1.5 chimeric channels, DNA encoding the Kv1.5 channel core beginning at the NcoI site encoding residue Met 210 was subcloned into homologous NcoI sites in pGBT9 vectors encoding the NH 2 termini of Kv1.1 and Kv1.3 using NcoI and XbaI restriction sites. The resulting fusion protein was then subcloned into pcDNA3 for mammalian expression as an EcoRI/EcoRI fragment, followed by screening for correct orientation of the insert.
For preparation of Kv1.5N/Kv2.1 T19ϩ163/Kv2.1, DNA encoding amino acids 180 to the COOH terminus of rKv2.1 were amplified by PCR, such that a BspEI restriction site at Pro 180 of rKv2.1, and a XbaI site following the termination codon were introduced. The primers used were: 5Ј-ATCGTCCGGAGTCGTCGGTGGCCGCCAAG-3Ј for the 5Ј end, and 5Ј-GCTCTAGACCCTCTGTGGTAGGGAGC-3Ј for the 3Ј end (introduced restriction sites are underlined). The resulting fragment was used to replace the analogous region of Kv1.5 encoding amino acids 243 to the COOH terminus of Kv1.5, in a pcDNA3 vector encoding Kv1.5 or Kv1.5T19ϩ163 using BspE1 and XbaI restriction sites.
For preparation of Kv2.1N/Kv1.5, the NH 2 terminus of rKv2.1 was amplified by PCR, introducing a EcoRI restriction site preceding the start codon, and a BspEI restriction site at 180 of rKv2.1. The primers employed were 5Ј-CGGAATTCGGCATGACGAAGCATGGC-3Ј as the 5Ј primer and 5Ј-GCTGTCCGGATACTCCAGCAGATCCCAGAG-3Ј as the 3Ј primer (introduced restriction sites are underlined). The resulting fragment was used to replace the analogous Kv1.5 sequence encoding amino acids of the Kv1.5 NH 2 terminus up to residue 243, by subcloning into a pcDNA3 vector encoding Kv1.5 using HindIII and BspEI restriction sites.
Site-directed mutagenesis of Kv1.5 was performed using the QuikChange method from Stratagene. For preparation of the Kv1.5AAQL mutant, the primers used were GGGCTGCGCTTTGCG-GCGGCAGCTGGGCACCCTG and its complement.

C-and U-type Inactivation in Kv1
.5-In intact channels, the T1 domain is thought to be structurally dissociated from the transmembrane segments of the channel, forming a hanging gondola (shown schematically in Fig. 1A). Interestingly, however, disruptions of the NH 2 termini of Kv channels can significantly affect the gating properties of the channel. A good example of this phenomenon is Kv1.5⌬N209, the naturally occurring short form of Kv1.5 comprising a deletion of greater than 80% of the cytosolic NH 2 terminus of Kv1.5, which exhibits activation and inactivation properties substantially different from the long-form of Kv1.5 (FL Kv1.5) (8). Throughout this study, we examined the voltage-dependent activation of Kv channels using the double pulse protocol described in Fig. 1B. Cells were stepped from a holding potential of Ϫ80 mV to potentials between Ϫ65 mV and ϩ50 mV in 5-mV steps for 200 ms (P1, Fig. 1B), followed by a brief repolarization to Ϫ40 mV (P2, Fig. 1B). The magnitude of the tail currents observed at Ϫ40 mV was proportional to the number of channels activated during P1 (Fig. 1B). Inactivation-voltage relationships were derived using the triple-pulse protocol described in the legend to Fig. 1C. From a holding potential of Ϫ80 mV, cells were given a 100-ms control pulse to 60 mV (P1, Fig. 1C), rested for 2 s at Ϫ80 mV, then stepped from voltages between Ϫ70 and ϩ60 mV in 10-mV steps for 5 s (P2, Fig. 1C), followed by a brief test pulse to ϩ60 mV (P3, Fig. 1C). The current during the P1 pulse serves to control for any rundown of peak current during experiments. The measured amplitude of the test pulse current in P3 is proportional to the number of available (non-inactivated) channels following P2 (Fig. 1C). FL Kv1.5 channels stably expressed in HEK 293 cells exhibited a half-activation potential of Ϫ10.8 Ϯ 0.8 mV (Fig. 1D), and a half-inactivation potential of Ϫ21.0 Ϯ 1.2 mV (Fig. 1E), which was consistent with previous studies from our laboratory and others (8,32). In the current study, we were particularly interested in the shape of the inactivation-voltage relationship, with FL Kv1.5 exhibiting a flat voltage dependence of inactivation at positive potentials (Fig. 1E).
The activation and inactivation properties of Kv1.5⌬N209 differ substantially from those observed in FL Kv1.5. When stably expressed in HEK 293 cells, Kv1.5⌬N209 exhibited a half-activation potential of Ϫ20.3 Ϯ 1.7 mV, which was shifted left-ward by 8 mV relative to the FL Kv1.5 channel (Fig. 1D). The half-inactivation potential in Kv1.5⌬N209 was Ϫ32.8 Ϯ 0.9 mV, which was shifted left-ward by roughly 12 mV relative to FL Kv1.5 (Fig. 1E). In addition, deletion of the Kv1.5 NH 2 terminus appeared to uncover additional pathways for inactivation, as Kv1.5⌬N209 inactivated more completely than FL Kv1.5 over the range of potentials examined. Most importantly, Kv1.5⌬N209 exhibited a U-shaped voltage dependence of inactivation, in which inactivation was maximal at intermediate depolarizations between Ϫ20 and 0 mV and significantly less pronounced with more positive depolarizations (Fig. 1E). This shape of the inactivation-voltage relationship contrasts sharply with the flat voltage dependence of inactivation observed in FL Kv1.5, and more closely resembles the U-type inactivation properties of a channel such as Kv2.1 (Fig. 1E). Kv2.1 expressed stably in HEK 293 cells exhibited a half-activation potential of Ϫ2.8 Ϯ 1.7 mV (Fig. 1D) and a half-inactivation potential of Ϫ26.3 Ϯ 0.7 mV (Fig. 1E), and again was clearly distinguished from typical C-type inactivation by a marked upturn of its inactivation-voltage relationship. These observations clearly distinguish inactivation in FL Kv1.5 from the U-type inactivation phenotype of Kv1.5⌬N209 and Kv2.1, and suggest that the NH 2 terminus of Kv1.5 may influence the inactivation properties of the channel.
The shape of the inactivation-voltage relationship in Kv1.5 was consistent with a C-type inactivation mechanism, generally viewed as a voltage-independent inactivation process that is coupled to channel opening (33). This observation was also consistent with previous studies of Kv1.5 and other Kv1 channels (32,34,35). Kv1.5 exhibits a number of other features consistent with a C-type inactivation mechanism. First, although residue Arg 487 (corresponding to Thr 449 in Shaker) renders wild-type Kv1.5 insensitive to extracellular TEA, inactivation in the R487T mutant of Kv1.5 was inhibited by TEA ( Fig. 2A). The application of extracellular TEA diminished the peak currents observed through Kv1.5 R487T channels, with Kv2.1 were stably expressed in HEK 293 cells and their kinetic properties were examined using whole cell patch clamp methods. B, to study activation, cells were stepped from a holding potential of Ϫ80 mV to voltages between Ϫ65 and ϩ50 mV in 5-mV steps for 200 ms (P1), followed by brief repolarizations to Ϫ40 mV (P2). Tail current amplitudes were measured isochronally (marked with an asterisk), normalized, and fit with single Boltzmann equations. C, to study inactivation, cells were given a 100-ms control pulse to 60 mV (P1), rested for 2 s at Ϫ80 mV, stepped from voltages between Ϫ70 and ϩ60 mV in 10-mV steps for 5 s (P2), followed by a test pulse to ϩ60 mV (P3). The interpulse interval was 35 s. Peak currents during P3 were normalized to peak currents during P1, and fit with single Boltzmann equations. D, V1 ⁄2 values for activation in Kv1.5, Kv1.5⌬N209, and Kv2.1 were Ϫ10.8 Ϯ 0.8, Ϫ20.3 Ϯ 1.7, and Ϫ2.8 Ϯ 1.7 mV, respectively. E, V1 ⁄2 values for inactivation in Kv1.5, Kv1.5⌬N209, and Kv2.1 were Ϫ21.0 Ϯ 1.2, Ϫ32.8 Ϯ 0.9, and Ϫ26.3 Ϯ 0.7 mV, respectively. 10 mM extracellular TEA resulting in a 51 Ϯ 2% (n ϭ 3) block of peak current. Normalized data demonstrating the effect of 10 mM extracellular TEA on the inactivation time course of Kv1.5R487T are shown in Fig. 2A. Currents through the R487T mutant channel inactivate by 38 Ϯ 2% during 5-s depolarizations to 60 mV under control conditions, but inactivate by only 25 Ϯ 3% in the presence of 10 mM extracellular TEA. To confirm a C-type mechanism of inactivation in Kv1.5, and to contrast the inactivation mechanisms in Kv1.5 and Kv1.5⌬N209, we also examined the effects of elevation of extracellular K ϩ on inactivation in both channels. Clearly, elevation of extracellular K ϩ results in deceleration of Kv1.5 inactivation (Fig. 2B), which suggests a C-type mechanism of inactivation in FL Kv1.5, and was consistent with previous studies on Kv1.5 and other Shaker homologues (32,36). In contrast, Kv1.5⌬N209 exhibits an opposite sensitivity to extracellular K ϩ (Fig. 2C), with more rapid inactivation observed in 135 mM extracellular K ϩ . This paradoxical sensitivity of inactivation to extracellular K ϩ appears to be a common feature of channels which exhibit a U-shaped inactivation-voltage relationship (5).
The Kv1.5 T1 Domain Influences U-type Inactivation-To confirm and extend these findings, we attempted to define the NH 2 -terminal region involved in altering the inactivation phenotype of Kv1.5. We began by constructing a series of NH 2terminal truncated forms of Kv1.5 (Fig. 3A). These were transiently expressed in HEK 293 cells, and their activation and inactivation properties were examined. The activation and inactivation curves of FL Kv1.5 and Kv1.5⌬N209 have been included for comparison. We examined three constructs comprising progressive deletions of the Kv1.5 NH 2 -terminal residues up to the T1 boundary (Kv1.5⌬N19, Kv1.5⌬N91, and Kv1.5⌬N119, see Fig. 3A). None of these deletion constructs exhibited any remarkable differences in activation or inactivation gating from FL Kv1.5, although the half-inactivation voltages were slightly right-shifted in these 3 constructs relative to FL Kv1.5 (Fig. 3, B and D). In addition, only very slight differences were observed in the level of inactivation resulting from 5-s inactivating pulses (Fig. 3D). Kv1.5⌬N19, Kv1.5⌬N91, and Kv1.5⌬N119 exhibited half-activation potentials of Ϫ11.0 Ϯ 0.6, Ϫ12.0 Ϯ 0.5, and Ϫ11.5 Ϯ 1.5 mV, and half-inactivation potentials of Ϫ15.9 Ϯ 1.3, Ϫ16.0 Ϯ 0.5, and Ϫ17.0 Ϯ 1.5 mV, respectively. Importantly, all three constructs exhibited a flat voltage dependence of inactivation at positive potentials, which was consistent with the appearance of the inactivation-voltage relationship characteristic of FL Kv1.5. These data suggest that the first 120 amino acids of Kv1.5 exert little effect on the activation or inactivation properties of the channel.
In contrast, a longer NH 2 -terminal deletion into the T1 do- main (Kv1.5⌬N188, Fig. 3A) resulted in left-ward shifts of both activation and inactivation, and a U-shaped inactivation-voltage relationship closely resembling that reported for Kv1.5⌬N209 (Fig. 3, C and E). These observations were consistent with a U-type inactivation phenotype in both Kv1.5⌬N188 and Kv1.5⌬N209, and demonstrate that deletions into the T1 domain were required to produce the U-shaped inactivation-voltage relationship characteristic of Kv1.5⌬N209. Kv1.5⌬N188 exhibited half-activation and half-inactivation potentials of Ϫ24.7 Ϯ 1.3 and Ϫ29.9 Ϯ 4.0 mV, respectively. To further characterize the influence of the T1 domain, we generated several deletion mutants intermediate to Kv1.5⌬N119 and Kv1.5⌬N188, namely Kv1.5⌬N140 and Kv1.5⌬N162 (Fig.  3A). Interestingly, despite fairly robust expressions of both shorter and longer deletion constructs, repeated transfections with these intermediate deletion mutants failed to generate any detectable macroscopic currents. Nevertheless, the data collected suggested that the first 70 amino acids of the Kv1.5 T1 domain (between residues 119 and 188) exert a critical influence on U-type inactivation in Kv1.5.
We also examined the activation and inactivation properties of a construct in which we fused the first 19 amino acids of the Kv1.5 T1 domain to Kv1.5⌬N162 (see Kv1.5T19ϩ163 in Fig.  3A). Surprisingly, this manipulation of the T1 structure both restored channel expression and resulted in a U-shaped voltage dependence of inactivation (Fig. 3, B and C). Kv1.5T19ϩ163 exhibited a strongly left-ward shifted half-activation potential of Ϫ23.8 Ϯ 0.7 mV, and a half-inactivation potential of Ϫ32.1 Ϯ 1.9 mV. These data suggested that disruption of the first 42 residues of the Kv1.5 T1 domain was sufficient to reproduce the U-type inactivation phenotype observed in Kv1.5⌬N209. We examined the location of homologous residues in the recently solved structures of the T1 domains of aplysia Kv1.  Fig. 4A), and examined activation and inactivation properties after transient expression in HEK 293 cells (Fig. 4). Sequence alignment illustrates that channels within the Kv1 family exhibit roughly 85% identity between their T1 domains, and essentially no homology in the remainder of their cytosolic NH 2 termini ( tively (Fig. 4C). Most importantly, both the Kv1.1 and Kv1.3 NH 2 termini were sufficient to rescue the inactivation properties observed in FL Kv1.5, as both chimeric channels exhibited a flat voltage dependence of inactivation at positive potentials, and similar fractional inactivation to that observed in FL Kv1.5 after 5-s depolarizations (Fig. 4D). The half-inactivation potentials of Kv1.1N/Kv1.5 and Kv1.3N/Kv1.5 were Ϫ19.0 Ϯ 2.0 and Ϫ21.2 Ϯ 1.3 mV, respectively. Together with the data presented in Fig. 3, and the lack of NH 2 -terminal homology outside the T1 domain of Kv1 channels, these data provide strong evidence that amino acids within the T1 domain are critical for modulating U-type inactivation in Kv1 channels.
Disruption of the Intersubunit T1 Interface Influences Slow Inactivation in Kv1.5-Characterization of our deletion mutants of Kv1.5 (Fig. 3), together with recent reports that disruption of intersubunit T1 contacts influences the activation properties of Kv1.2 (31), prompted us to examine the effects of similar disruptions on the inactivation properties of Kv1.5 (Fig.  5). Fig. 5A depicts the interface formed between two adjacent T1 domains, generated from the crystal structure determined for the T1 domain of rat Kv1.2 (Molecular Modeling Database accession number 14393), using RasMol version 2.7.1.1. The interface illustrated in Fig. 5A is repeated four times in an intact channel, because of the rotational symmetry of the T1 tetramer. As shown in Fig. 5A, intersubunit contacts were formed between residues throughout the T1 domain, although primarily within the first 45 residues of T1, and were typically formed between the side chains of polar amino acids (14,31). We generated a double point mutant, Kv1.5AAQL, by mutating two adjacent residues, Glu 131 and Thr 132 , to alanines (backbone and side chain bonds of these residues are colored purple in Fig. 5A). Mutation of the residues homologous to either Glu 131 or Thr 132 in Kv1.2 results in significant shifts of activation gating (31), and our double mutation results in the disruption of a hydrogen-bonding network of intersubunit T1 contacts near the NH 2 terminus of the T1 domain (Fig. 5A). Upon transient expression in HEK 293 cells, Kv1.5AAQL exhibited a half-activation potential of Ϫ21.4 Ϯ 1.2 mV (Fig. 4B), and a half-inactivation potential of Ϫ32.2 Ϯ 3.6 mV (Fig. 5C). Both parameters were left-shifted with respect to FL Kv1.5 channels. These results were somewhat surprising, as individual mutations of the homologous residues to alanine in Kv1.2 have been shown to result in a right-ward (depolarizing) shift of the activation relationship (31). Most interestingly, Kv1.5AAQL channels consistently exhibited a notably upturned inactivation-voltage relationship (Fig. 5C). Kv1.5AAQL channels inactivated maximally by 59 Ϯ 3% at Ϫ20 mV, but only inactivated by 48 Ϯ 2% at ϩ50 mV. Although not as pronounced as the U-shaped inactivation-voltage relationship seen with more complete disruption of the T1 domain (e.g. Kv1.5⌬N209), this result clearly suggests that individual point mutations within the T1 domain can influence the inactivation phenotype in Kv1.5.
Reciprocal Influences of the Kv2.1 and Kv1.5 NH 2 Termini-The data presented thus far strongly suggests that residues within the T1 domain function to prevent U-type inactivation in FL Kv1.5. To investigate the specificity of this interaction between the NH 2 terminus and the transmembrane core of voltage-gated Kv channels, we generated a chimeric channel in which the NH 2 terminus of Kv1.5 was replaced with the NH 2 terminus of Kv2.1, a voltage-gated K ϩ channel that exhibits features of U-type inactivation (Fig. 6A, Kv2.1N/Kv1.5). The structural scaffold of the T1 domain is conserved among all Kv channels, and sequence alignment shows that the Kv2.1 T1 domain primary sequence exhibits roughly 37% identity with the T1 domain of Kv1.5 (Fig. 6B) (12,37). Again, we transiently expressed this construct in HEK 293 cells and characterized its activation and inactivation properties. Kv2.1N/Kv1.5 exhibited half-activation and half-inactivation potentials of Ϫ25.5 Ϯ 1.1 and Ϫ30.6 Ϯ 0.5 mV, respectively (Fig. 6, C and E). Importantly, Kv2.1N/1.5 also exhibited a marked U-shaped voltage dependence of inactivation (Fig. 6E). The inactivation-voltage relationship of Kv2.1N/Kv1.5 was very similar to those observed in the T1-deleted forms of Kv1.5 (Figs. 1E and 3C) and Kv2.1 (Figs. 1E and 6E), however, some slight differences were apparent. In particular, Kv2.1N/Kv1.5 did not inactivate as completely as Kv2.1 or T1-deleted forms of Kv1.5 during the 5-s pulses examined (Fig. 6E).
We also conducted the inverse experiment, in which the NH 2 terminus of Kv2.1 was replaced with the NH 2 terminus of Kv1.5 (Kv1.5N/Kv2.1, Fig. 6A). The half-activation and halfinactivation potentials of Kv1.5N/Kv2.1 were Ϫ9.3 Ϯ 3.3 and Ϫ12.1 Ϯ 3.3 mV, respectively (Fig. 6, C and E). Most importantly, fusion of the Kv1.5 NH 2 terminus to Kv2.1 resulted in complete attenuation of the U-shape normally observed in the Kv2.1 inactivation-voltage relationship (Figs. 1E and 6E). In addition, Kv1.5N/Kv2.1 exhibited inactivation kinetics substantially slower than those seen in full-length Kv2.1, illustrated by sample traces elicited with 5-s pulses to ϩ60 mV in cells expressing either Kv2.1 or Kv1.5N/Kv2.1 (Fig. 6D). As a result, wild-type Kv2.1 was inactivated maximally by 69 Ϯ 2% at 0 mV, whereas Kv1.5N/Kv2.1 was inactivated maximally by only 34 Ϯ 4% at ϩ50 mV (Fig. 6E). We also examined an NH 2 -terminal deleted form of Kv2.1 (Kv2.1⌬N101), to confirm that the effects of the Kv1.5 NH 2 terminus in the Kv1.5N/Kv2.1 chimera were not simply because of deletion of the Kv2.1 NH 2 terminus. We had difficulty expressing NH 2 -terminal deletions of Kv2.1 in our HEK cell line, however, expression of Kv2.1⌬N101 in mouse ltkϪ cells confirmed previously published data collected from Xenopus oocytes (38). In particular, deletions of the T1 domain of Kv2.1 result in extreme slowing FIG. 5. Point mutations at the intersubunit T1 interface influence U-type inactivation. A, diagram illustrating specific residues forming intersubunit T1 contacts, generated from the crystal structure of the human Kv1.2 T1 domain (31) using RasMol version 2.7.1.1. C␣ backbone diagrams are shown for selected residues with the blue and green colors indicating different subunits. Side chain atoms of interacting residues are illustrated as ball and stick models. Intersubunit T1 contacts at residues Glu 131 and Thr 132 (side chain bonds and backbone are colored purple) were disrupted by mutating these residues to alanine. B, activation, and C, inactivation curves were constructed as described in the legend to Fig. 1, yielding a V1 ⁄2 of activation of Ϫ21.4 Ϯ 1.2 and a V1 ⁄2 of inactivation of Ϫ32.2 ؎ 3.6 mV for Kv1.5AAQL. of the kinetics of activation and essentially complete ablation of slow inactivation of the channel, evident in Fig. 6D, and consistent with a previous publication (38). This contrasts sharply with currents recorded from the Kv1.5N/Kv2.1 chimera, which exhibit markedly faster kinetics of activation, and substantial inactivation during a 5-s pulse (Fig. 6D). As a further control, we characterized a chimeric channel consisting of the NH 2 terminus of Kv1.5T19ϩ163 (Fig. 3A) fused to the transmembrane domains of Kv2.1. This NH 2 -terminal construct results in a U-shaped voltage dependence of inactivation in Kv1.5 (Fig.  3), but results in activation and inactivation kinetics in Kv2.1 that were effectively indistinguishable from the NH 2 -terminal deletion of Kv2.1 (Fig. 6D). This suggests that the inactivation phenotype observed in the Kv1.5N/Kv2.1 chimera results from a specific effect of the intact Kv1.5 NH 2 terminus. DISCUSSION The current hanging gondola model of voltage-gated K ϩ channel structure suggests that the T1 tetramer was structurally dissociated from membrane-bound segments of the channel (13,15,18). Paradoxically, it has been demonstrated that deletion mutations, and even highly conservative point mutations, can exert strong effects on channel activation (30,31). These observations were particularly interesting in light of several studies demonstrating the expression of truncated isoforms of several Kv channels in vivo, including Kv1.5 and KvLQT1 (6,7,39). NH 2 -terminal truncated channels likely exhibit an altered sensitivity to modulation by signaling mech-anisms as the T1 domain has been identified as a region of interaction with G␤␥ subunits, tyrosine kinases, and Kv␤ subunits (23)(24)(25)35). However, the influence of the T1 domain on channel gating, and particularly inactivation, has not been widely studied.
Our characterization of T1 deletions of Kv1.5 together with recent studies in Shaker (5) suggest that Shaker family K ϩ channels are able to exhibit phenotypes characteristic of both C-type and U-type inactivation. In addition, the data presented in our study suggests that the cytosolic NH 2 terminus of Shaker homologues, such as Kv1.5, strongly influences the extent of U-type inactivation, and hence determines the balance between C-and U-type inactivations. FL Kv1.5 channels exhibit no features consistent with U-type inactivation (32), and deletions up to the T1 border (residue 120) exerted no significant effects on gating (Fig. 3). However, deletions within the T1 domain consistently disrupted both activation and inactivation properties of Kv1.5, resulting in a shift to a U-type inactivation phenotype in the T1-deleted forms of the channel (Fig. 3). Using approaches of progressive channel deletion mutations (Fig. 3), chimeric constructs of Kv1 channels (Fig. 4), and site-directed point mutations (Fig. 5), we have demonstrated that this function is localized within the Kv1 T1 domain.
The results of our studies in Kv1.5 suggest that the amino acids forming intersubunit T1 contacts may be involved in regulating the U-type inactivation phenotype, although the U-shaped inactivation-voltage relationship of the Kv1.5AAQL construct is not as marked as that observed in Kv1.5⌬N209 or other T1 deleted forms of Kv1.5 (Fig. 3). This may be because the Kv1.5AAQL mutation only disrupts a fraction of the intersubunit T1 interactions, as intersubunit contacts were made between several residues throughout the T1 domain (see Fig.  5A). It should be noted that mutations throughout the T1 domain have been demonstrated to exert varying effects on the activation gating of Kv1 channels. This has been most carefully examined among residues at the intersubunit T1 interface, but residues lining the central T1 pore, and even residues in the T1-S1 linker region have been shown to affect channel gating (29,31,40,41). In addition, point mutations within T1 have been shown to cause both hyperpolarizing and depolarizing shifts of the activation relationship, depending on both the position and properties of the substituted amino acid. In summary, we can say with certainty that disruption of the intersubunit T1 contact residues examined in our study alter the slow inactivation properties of Kv1.5. However, without a more complete mutagenic analysis of the T1 domain, we cannot rule out the possibility that more global conformational changes in T1 are involved in the influence of this domain on channel gating.
A generalized role for the NH 2 terminus in regulation of U-type inactivation was also supported by our chimeric studies of Kv1.5 and Kv2.1. These experiments examined the effects of switching the NH 2 termini of a C-type inactivating channel (Kv1.5) and a U-type inactivating channel (Kv2.1). Substitution of the Kv1.5 NH 2 terminus with the Kv2.1 NH 2 terminus resulted in channels that exhibited a U-shaped inactivationvoltage relationship, similar to the inactivation phenotype observed in Kv1.5⌬N209 (Fig. 6). However, the inverse substitution resulted in a deceleration of inactivation compared with Kv2.1 (Fig. 6). Our interpretation of these observations was that the Kv2.1 NH 2 terminus adopts a conformation or interacts with channels in a manner that promotes U-type inactivation. In contrast, the Kv1.5 T1 domain may interact with channels in a manner that prevents U-type inactivation, which was consistent with the absence of U-type inactivation in Kv1.5, and the deceleration of inactivation in Kv1.5N/Kv2.1. Interestingly, the influence of the Kv1.5 NH 2 terminus on the inactivation of Kv2.1 was extremely similar to the influence of Kv2.3 (also known as Kv8.1) (42). In particular, coexpression of Kv2.1 and Kv2.3 results in deceleration of both Kv2.1 activation and inactivation, and studies of chimeric constructs of Kv2.1 and Kv2.3 have localized this effect to the NH 2 terminus of Kv2.3, specifically within the COOH-terminal side of T1 and the T1-S1 linker (42). Similar effects are seen upon coexpression of Kv2.1 and Kv6.1, although the regulatory domains in Kv6. 1 have not yet been determined (43). In addition, as observed in our Kv1.5N/Kv2.1 channel, chimeric channels consisting of this NH 2 -terminal segment of Kv2.3 and the transmembrane domains of Kv2.1 exhibit deceleration of Kv2.1 inactivation (42). These findings raise the possibility of a generalized mechanism of interaction between the NH 2 termini and membrane-bound segments of voltage-gated K ϩ channels. More importantly, in the context of our study these observations support the notion that the modulation of the U-type inactivation phenotype by the NH 2 terminus may be a general feature of voltage-gated K ϩ channels.
Studies to date have suggested several potential mechanisms for the conformational coupling between T1 and the transmembrane domains of the channel. The COOH-terminal region of the T1 domain is thought to face the membrane-bound channel segments, and certain T1 point mutations that modulate activation gating have been shown to alter the conforma-tion of this region of the channel, as does interaction with Kv␤ subunits (13). This suggests that the conformation of the COOH-terminal region of the T1 domain exerts functionally significant effects on gating, and this is consistent with the region of Kv2.3 involved in disruption of Kv2.1 inactivation (36). However, other T1 point mutations resulting in little global change of T1 structure have also been shown to modulate activation gating, suggesting an important role for active rearrangements of T1 during channel gating (31). A further possible mechanism for the influence of T1 is through interaction with the C terminus of Kv channels, as the NH 2 -terminal end of T1 (which faces away from the membrane-bound channel segments) and the COOH-terminal extension of the sixth transmembrane segment are sufficiently close to form disulfide bonds in intact channels (44). Minor et al. (31) hypothesized that point mutations within the T1 domain of Kv1.2 influence activation gating by altering the stability of channel closed states. Interestingly, kinetic modeling of inactivation in Shaker, Kv2.1, Kv3.1, and Kv1.5⌬N209 suggest that the apparent U-type voltage dependence of inactivation arises from accelerated inactivation from closed states of the channel (4,5,8). In our experiments, we observed that all of the T1 deletions (Kv1.5⌬N188, Kv1.5⌬N209, and Kv1.5T19ϩ163) and point mutations (Kv1.5AAQL) which impart U-type inactivation features to Kv1.5 also result in significant hyperpolarizing shifts of channel activation (Fig. 3, Fig. 5). It is possible that the influence of the T1 domain on the stability of closed states of the channel can account for both hyperpolarizing shifts of activation and acceleration of inactivation from closed states, resulting in a U-type inactivation phenotype.
There is a growing body of evidence suggesting that mechanisms underlying a U-type inactivation phenotype play an important role in ion channel function. This inactivation phenotype was first described in Kv2.1, but has been demonstrated in other widely studied voltage-gated K ϩ channels including Shaker and Kv3.1, suggesting that U-type inactivation may be a more generalized feature of ion channels than previously thought (4,5). Although the fundamental conformational changes underlying U-type inactivation remain unclear, the cytosolic T1 domain of Kv channels appears to be intimately involved in the regulation of U-type inactivation properties. As a docking site for a number of intracellular regulatory molecules, the T1 domain may provide an important mechanism for transient alteration of both activation and inactivation properties of Kv channels.