A potassium channel (Kv4) cloned from the heart of the tunicate Ciona intestinalis and its modulation by a KChIP subunit

Voltage-gated potassium channels related to Drosophila Shal are encoded in mammals by the Kv4 gene subfamily (Kv4.1: Pak et al., 1991; Kv4.2: Roberds and Tamkun, 1991; and Kv4.3: Serôdio et al.,1994). Kv4 channels mediate transient outward currents in cardiac myocytes (Dixon et al., 1996),smooth muscle (Amberg et al.,2002) and neurons(Serôdio et al., 1994). The currents mediated by Kv4 channels play important roles in specifying the firing properties of molluscan and lobster neurons(Connor and Stevens, 1971; Baro et al., 1996), suggesting that Kv4 channels are conserved determinants of cellular excitability. The three mammalian Kv4 channel isoforms differ in their biophysical properties. For example, the fast component of inactivation of Kv4.1 channels contributes significantly less to inactivation than the fast inactivating component of Kv4.2 and Kv4.3 channels (Pak et al.,1991; Jerng and Covarrubias,1997). The full significance of these differences in properties between the three Kv4 channel isoforms is not yet understood. However, since expression of Kv4.2 and Kv4.3 isoforms is complementary in brain and heart(Serôdio et al., 1996),it seems likely that the specific properties of each of these isoforms contribute to determining the function of the particular tissue area where it is expressed. These differences between the three Kv4 channel isoforms likely represent adaptive features of these channels. To identify adaptive and conserved features of mammalian Kv4 channels it is necessary to compare these with invertebrate Kv4 channels such as jellyfish Shal(Jegla and Salkoff, 1997),arthropod Shal (Baro et al.,1996) and a tunicate Shal(Nakajo et al., 2003).

KChIPs, a family of Ca²⁺ binding proteins(Burgoyne and Weiss, 2001),are integral components of Kv4 channel supra-molecular complexes(An et al., 2000). KChIP isoforms 1-3, but not KChIP4, increase the density of Kv4 currents(Shibata et al., 2003). All four KChIP isoforms modify the kinetics and gating properties of Kv4 channels(An et al., 2000; Holmqvist et al., 2002). Modulation of Kv4 channels by KChIP subunits is likely a conserved mechanism to modulate tissue excitability, because the effects of human KChIP1 on the lobster and mammalian Kv4 channels are similar(Zhang et al., 2003). Thus,KChIP1 increased current amplitude, slowed the rate of inactivation, and shifted activation and inactivation midpoints of lobster Kv4 channels(Zhang et al., 2003). In the present study, we address the question of how tunicate Shal is modulated by endogenous KChIP subunits.

The vertebrate multi-gene Kv4 and KChIP families are represented by single genes in the tunicate genome. Scaffolds 168 and 457 of the genome database for Ciona intestinalis(http://genome.jgi-psf.org/ciona4/ciona4.home.html,release version 1.0) contain the Kv4 and the KChIP gene,respectively. We cloned the transcripts for a Kv4 channel (CionaKv4)and a KChIP subunit (CionaKChIP) from myocardial tissue of Ciona intestinalis. Repolarization of the action potential in the tunicate heart is complex with several repolarization phases, which are presumably an adaptation for pumping blood by means of peristaltic contractions that can propagate in both directions through the tubular heart(Kriebel, 1967). Since Kv4 channels play a crucial role in determining the excitability properties of cardiac tissues of vertebrates, we suspected that these channels were playing a similar role in the tunicate heart.

We characterized the potassium currents produced by CionaKv4 channels heterologously expressed in Xenopus oocytes, in the presence and absence of CionaKChIP. Our biophysical data for CionaKv4 reinforces and complements what is known for tunicate Shal(Nakajo et al., 2003). Because the N terminus of Kv4 channels appears to be essential for modulation by KChIP(Bähring et al., 2001b),an N-terminal deletion mutant of CionaKv4, lacking amino acids 2-32,was constructed (ntCionaKv4), and the effects of CionaKChIP on this mutant CionaKv4 channel were evaluated. In this paper we describe the modulation of kinetic and gating parameters of a tunicate Shal channel (CionaKv4) by a tunicate KChIP subunit (CionaKChIP),the first KChIP subunit cloned from either an invertebrate or a non-vertebrate chordate. Our data and those provided by Zhang et al.(2003) on modulation of lobster Shal by KChIP1, are the first indications that modulation of Kv4 channels by KChIP subunits might be an ancient mechanism to modulate the excitability of electrically active tissues.

Specimens of Ciona intestinalis (Linnaeus 1767) were purchased from the Marine Biology Laboratory (Woods Hole, MA, USA) and kept in a seawater aquarium at 12°C until used. The heart, enclosed in the pericardium, was excised in diethyl pyrocarbonate (DEPC)-treated artificial seawater whose composition was (in mmol l^-1): NaCl (425), KCl (9),CaCl₂ (9.3), MgSO₄ (25.5), MgCl₂ (23), Hepes(10). This heart complex was pinned onto a base of Sylgard (Dow Corning,Midland, MI, USA) in a Petri dish and the pericardium and raphe (connective tissue that joins the pericardium to the myocardium) were cut away. The isolated myocardium was rinsed in filtered DEPC-treated artificial seawater several times in order to remove any remaining blood cells and cellular debris and then frozen at -80°C. Total RNA was extracted from heart tissue (4-6 hearts per experiment) using the `Totally RNA' kit (Ambion, Austin, TX,USA).

The GeneRacer Kit (Invitrogen, Carlsbad, CA, USA) was used to perform a 3′-RACE assay. Briefly, cDNA was synthesized from total RNA using an oligo-dT primer, with a unique sequence at the 3′ end, provided with the kit. Each 50 μl 3′-RACE assay contained: 1× Opti-Prime Buffer(Stratagene, La Jolla, CA, USA) (10 mmol l^-1 Tris-HCl, 1.5 mmol l^-1 MgCl₂,25 mmol l^-1 KCl, pH 8.3), 0.5 unit of Taq polymerase, 0.25 mmol l^-1 of each dNTP, 1 μmol l^-1 each of forward primer and reverse primer and 2 μl cDNA. The sense primers (5′-GACGGATCTACAGCCAAAATCAAAGACA-3′ for Kv4 and 5′-CAGTTGTTGGGATCGCATGT-3′ for KChIP) were bound to an internal sequence of the Kv channel or the KChIP subunit transcript;the anti-sense primer, supplied with the kit, bound to the specific sequence of the oligo-dT primer. The temperature conditions were: 30 s at 94°C,followed by 5× (30 s at 94°C, 30 s at 65°C, 3 min at 72°C),25× (30 s at 94°C, 30 s at 60°C, 3 min at 72°C) followed by 5 min at 72°C. 3′-RACE products were sequenced with the forward primers used in these assays. Next, new forward primers were designed to bind downstream within these sequences in nested 3′-RACE assays. Successive nested RACE assays were done until the first 3′-RACE assay product was fully sequenced.

The full-length open reading frames for CionaKv4 and CionaKChIP were amplified with sense primers containing the start codon and a Kozak consensus site for the 5′ end, and anti-sense primers designed to bind downstream of the translational termination codon(CionaKv4-sense:5′-GGCTCGAGGCCGCCACCATGGCAACAGCAGTAGC-3′; CionaKv4-antisense:5′-GGCCTAGGCAAAGTCCCGCCGCTACAGTGAG-3′; CionaKChIP-sense:5′-GGCTCGAGGCCGCCACCATGTCTCTCGCTATCTTAACCATGGTGAC-3′; CionaKChIP-antisense:5′-GGACTAGTAACGCCAGAAACGCCGCCTTGATGGAGCTATAACGC-3′). Primers also contained restriction sites to allow directional insertion of the PCR products into pXT7 (Dominguez et al.,1995). Each 50 μl PCR reaction contained: 1× Opti-Prime Buffer (Stratagene, La Jolla, CA, USA; 10 mmol l^-1 Tris-HCl, 1.5 mmol l^-1 MgCl_2, 25 mmol l^-1 KCl, pH 8.3), 0.5 unit of Taq polymerase, 1 unit of Pfu, 0.2 mmol l^-1 of each dNTP, 1μmol l^-1 each of forward primer and reverse primer, and 2 μl cDNA. Temperature conditions were: 30 s at 94°C, 25× (30 s at 94°C, 30 s at 60°C, 3 min at 72°C) followed by 5 min at 72°C. The resulting PCR products were then digested with appropriate restriction enzymes, ligated with appropriately digested pXT7 plasmid and transformed into E. coli. Clones were sequenced to verify that no mutation had been introduced by PCR. This construct was linearized with SalI and cRNA was synthesized using the T7 mMessage mMachine kit (Ambion, Austin, TX,USA).

To generate a mutant channel lacking amino acids 2-32 of the N terminus domain, a diluted sample of the CionaKv4 clone was amplified by PCR. The composition of the PCR reactions and the temperature conditions were as described in the previous section. The forward primer(5′-GGCTCGAGGCCGCCACCATGAACCGACGTAAAACAAAAGAC-3′) was designed to bind to the CionaKv4 clone, starting from nucleotide 97. A start codon, Kozak consensus sequence and an XhoI site were added to this sequence. The reverse primer was the same as used for the full-length clone. These PCR products were inserted into pXT7 and transformed into E. coli. Clones were sequenced to verify the deletion and to check that the undeleted sequence had no mutations introduced by PCR.

Amino acid sequences of a set of phylogenetically representative Kv4 channels (18) and KChIP subunits (17) were aligned with T-Coffee software(Notredame et al., 2000). Kv4 channel regions that contained extensive gaps were edited out, leaving a data matrix that included the T1 domain and the membrane-spanning core(trans-membrane domains S1 to S6), giving a total of 405 characters. MrBayes v3.0b (Huelsenbeck and Ronquist,2001) was used to determine the phylogenetic relationships of the channel sequences. The default parameters for protein sequences were used. A total of 200 000 cycles were run; the topology and branch length of every 100th tree was saved. The Bayesian maximum likelihood tree was obtained by taking a consensus of the collected trees after a 101 tree burn-in. The nodes on the tree are labelled with the posterior probability for each node; nodes with a probability of one were left unlabelled. The tree was visualized using Treeview (Page, 1996).

Oocytes from Xenopus laevis were surgically removed and dissociated using 2 mg ml^-1 collagenase 1A (Sigma Aldrich, St Louis, MO, USA) in a solution containing (in mmol l^-1): NaCl (96),KCl (4), MgCl₂ (20) and Hepes (5), pH 7.4. Oocytes were incubated at 18°C in culture medium containing (in mmol l^-1): NaCl (96),KCl (2), MgCl₂ (1), CaCl₂ (1.8), Hepes (5), sodium pyruvate (2.5), pH 7.4 with 100 mg l^-1 gentamycin and 3% horse serum (Gibco BRL, Carlsbad, CA, USA). 24 or 48 h following oocyte isolation,40 nl of CionaKv4 cRNA (∼16 ng) or ntCionaKv4 (∼8 ng) were injected into each Xenopus oocyte. For co-expression experiments, 20 nl (∼8 ng) of CionaKv4 or 20 nl (∼4 ng) of ntCionaKv4 cRNA solution was mixed with 20 nl (∼7 ng) of CionaKChIP cRNA solution, to give a final volume of 40 nl that was injected into each oocyte. The approximate molar ratio was [1 CionaKv4:3 CionaKChIP] or [1 ntCionaKv4:6 CionaKChIP]. Immediately prior to each experiment the vitelline membrane was manually removed with forceps after treatment in a hyperosmotic solution containing (in mmol l^-1): NaCl (96), KCl₂ (2),MgCl₂ (20), Hepes (5) and mannitol (400), pH 7.4.

All recordings were made from cell-attached macro-patches. The pipettes contained ND96 solution, with the following composition (in mmol l^-1): NaCl (96), KCl (2), MgCl₂ (1), CaCl₂(1.8) and Hepes (5), pH 7.4. During the recordings, oocytes were bathed in a grounded, isopotential solution, thereby ensuring that the potential applied by the recording pipette would be the true membrane potential for the macro-patch. This bath solution contained (in mmol l^-1): NaCl(9.6), KCl (88), EGTA (11), Hepes (5), pH 7.4. Patch pipettes were pulled from aluminosilicate glass, coated with dental wax, and fire-polished before each experiment. Only pipettes that had resistances between 0.7 and 1.4 MΩwere used.

Membrane seals were obtained by applying negative pressure. Voltage-clamp and data acquisition were carried out using an EPC-9 patch-clamp amplifier(HEKA, Lambrecht, Germany) controlled with PULSE software (HEKA) running on a Power MacIntosh G4 computer. Data were acquired at sampling intervals of 50μs and low-pass filtered at 5 kHz during acquisition. Bath temperature was maintained at 12±0.2°C using a Peltier device controlled by an HCC-100A temperature controller (Dagan, Minneapolis, MN, USA). The holding potential was set at -100 mV and leak subtraction was performed with a P/4 protocol. PulseFit (HEKA), Igor Pro (Wavemetrics, Lake Oswego, OR, USA) and Instat (GraphPad Software, San Diego, CA, USA) software were used for analyses and graphing.

Conductances were calculated using the equation G_K=I_peak/(V-E_K),where G_K is the potassium conductance, I_peak is the peak measured amplitude of the K⁺ current to a test pulse, V is the voltage at which the current was measured, and E_K is the measured potassium equilibrium potential, in this case -90 mV. Conductance-voltage relationships and steady-state inactivation curves were fitted in Igor Pro by a sigmoid (Boltzmann)distribution of the form:f(V)=G_max/{1+exp[(V_0.5-V)/K]},where G_max is the maximal conductance, V is the voltage of the depolarizing pulse (for activation) or the prepulse voltage (for inactivation), and K is the slope factor. For a first order Boltzmann, used to fit conductance-voltage relationships and steady-state inactivation curves, V_0.5 is the voltage at which activation or inactivation is half maximal.

Time constants of activation were obtained from a fit of the rising phase of the currents to the Hodgkin-Huxley model of the form: I(t)=I_max[1-exp(-t/τ)]ⁿwhere I_max is the maximum current obtained in the absence of inactivation, τ is the activation time constant, t is time and n=4. To study the decay of the current as a function of time, the decaying phases of the outward currents evoked by test pulses between +30 mV and +80 mV were fitted to the equation: I(t)=I₀+I₁exp(-t/τ₁)+I₂exp(-t/τ₂),where τ₁ and τ₂ represent the fast and the slow time constants of inactivation, respectively, I₁ and I₂ represent the relative contribution of each component to inactivation, and I₀ is the offset.

To investigate the voltage dependence of the kinetics of channel closing,tail currents were evoked at several membrane potentials, from -160 to -40 mV,from a brief (10-15 ms) depolarizing pulse to +50 mV. Tail currents obtained with pulses from -160 to -100 mV were used for analyses.

The voltage dependence of steady-state inactivation was determined by measuring the peak current evoked with a depolarizing pulse to +50 mV as a function of the voltage of a preceding 10 s prepulse test (between -130 and-30 mV). A double-pulse protocol was used to assess the rate of recovery from inactivation at -100 mV. A test pulse to +50 mV of 1 s duration was separated by a recovery period (at -100 mV) of increasing duration (50-2000 ms) from a second test pulse to +50 mV. The currents evoked by the second pulse of a double-pulse protocol were normalized to the currents produced by the first pulse and plotted against the duration of the interpulse interval.

Statistical comparisons were carried out using the Student t-test, or the alternative Welch t-test when there were significant differences between the standard deviations of the two groups. Data are presented as the mean ± s.e.m., and N is the number of macro-patches or oocytes, because recordings from one macro-patch per oocyte were included in the analyses.

The nucleotide sequences (ORF and 3′-untranslated region) of CionaKv4 and CionaKChIP are deposited in GenBank (Accession numbers: AY514487 for CionaKv4 and AY514488 for CionaKChIP). The 3′-RACE assays allowed identification of the stop codon for CionaKv4 and verification of the position of the stop codon for CionaKChIP (Fig. 1A,C). This sequence information was used to design reverse primers to amplify the full ORF for both transcripts. The exon/intron organization of the genes encoding CionaKv4 and CionaKChIP were deduced by comparing their cDNA sequences with the sequences of scaffolds 168 and 457 (Ciona intestinalis genomic database, release version 1.0), which contained the CionaKv4 and the CionaKChIP gene,respectively (Fig. 1B,D). The exon sequences in the genomic database and the corresponding regions of the transcript were 96% identical for both CionaKv4 and CionaKChIP. The CionaKv4 gene is encoded in ∼17.8 kb, of which 2.4 kb encode the ORF. The first exon (exon 0) encoded the first amino acids of the N terminus. The second exon (exon 1), 1070 bp in length, coded most of the channel protein, including the T1 domain, trans-membrane domains S1-S5, and the first part of the pore domain. The second intron interrupted the sequence of the K⁺ selectivity filter motif (GYG), since the last base pair of the second exon and the first two base pairs of the third exon encoded the first G of this motif. The second part of the pore region and the S6 trans-membrane domain were encoded by exon 2. Exons 3-5 encoded parts of the C-terminal cytoplasmic domain. The last exon (exon 6), encoded the rest of the C terminus of CionaKv4, and contained the stop codon (TAG) and the 3′-UTR sequence.

Fig. 2 shows an alignment between the predicted amino acid sequences of CionaKv4 and selected vertebrate and invertebrate Kv4 channels. The six trans-membrane domains(S1-S6), the pore region, and the first ∼23 amino acids of the N terminus of all Kv4 channels showed high sequence similarity. The C-terminal motif PTPP, which is involved in the interaction of Kv4 channels with filamin and is located ∼30 amino acids upstream of the C termini of mammalian Kv4 channels (Petrecca et al.,2000), was not found in CionaKv4, although CionaKv4 shares the first residue of this motif (P) and the preceding residue (I) with all mammalian Kv4 channels (residues 651-652 in CionaKv4, Fig. 2). The conserved di-leucine motif, consisting of 16 amino acids that span positions 474-489 and is involved in the targetting of membrane proteins(Rivera et al., 2003), was also found in CionaKv4 (Fig. 2). The C-terminal domain of CionaKv4 and the N terminus contained, respectively, two and one putative sites for phosphorylation by cAMP-dependent protein kinase A (PKA). Additionally, the N terminus of CionaKv4 contained four sites for phosphorylation by protein kinase C(PKC); another four putative sites for PKC phosphorylation were found in the C terminus; and another such site was found in the intracellular loop between membrane-spanning domains S4 and S5.

CionaKChIP was encoded in ∼7 kb of genomic DNA, of which the ORF occupied ∼0.7 kb, distributed in seven exons with sizes between 69 and 162 bp (Fig. 1D). Fig. 3 shows an alignment between the deduced amino acid sequences of CionaKChIP and representatives of the four major isoforms of mammalian KChIPs (KChIP1-4). As with other KChIPs, the sequence of the N terminus of CionaKChIP was the most variable region of the protein (An et al., 2000) whereas the sequence corresponding to the C terminus was conserved, containing four EF-hand Ca²⁺-binding motifs(Kretsinger, 1987). The N terminus of CionaKChIP was encoded by exons 1 and 2. The coding region for the first EF-hand was partitioned between exons 2 and 3. Similarly,the codons for the second and the third EF-hands were partitioned between exons 4 and 5 (second EF-hand), and between exons 5 and 6 (third EF-hand). The fourth EF-hand was encoded by exon 6 and the last section of the C-terminus was coded by exon 7. A Prosite scan of CionaKChIP suggested the location of six putative sites for PKC phosphorylation but no sites for PKA phosphorylation (Fig. 3).

In the phylogenetic tree shown in Fig. 4A, rooting with the cnidarian Kv4 sequences resolved the arthropod and chordate Kv4 channels into sister groups. CionaKv4 and Halocynthia Kv4 (TuKv4; Nakajo et al., 2003) formed a single clade, as would be expected. This tunicate clade is the sister group to the clade containing the Kv4.1, Kv4.2 and Kv4.3 paralogues from vertebrates, indicating that the duplication and divergence that gave rise to the three vertebrate Kv4 channels occurred after the divergence of the vertebrate lineage from the tunicate lineage. In the phylogenetic tree shown in Fig. 4B, the single CionaKChIP branched basal to the four mammalian KChIP isoforms. This tree indicated that there is a 99% likelihood that the KChIP1 isoform is the sister group of the other three isoforms. However, the phylogenetic relationships between the KChIP isoforms 2, 3 and 4 were unclear, with posterior probabilities of approximately 50-60%. There was strong support for the presence of all four isoforms in the common ancestor of modern fish (represented by zebrafish) and the tetrapods (represented by human, mouse, rat and chicken). All of the vertebrate KChIP sequences partitioned robustly into one of the four paralogous families, indicating that the paralogues all originated prior to the origin of the vertebrate clade.

CionaKv4 channels conducted transient A-type currents(Fig. 5A), whereas currents mediated by CionaKv4 channels in the presence of CionaKChIP subunits displayed considerably slower inactivation during depolarizing pulses lasting 0.5 s (Fig. 5B). The amplitudes of the currents produced by CionaKv4 channels co-expressed with CionaKChIP subunits were larger than for CionaKv4 channels alone, even though half the amount of CionaKv4 cRNA was injected for co-expression experiments (Figs 5A,B, 6Ai). For example, the average peak amplitudes of the currents produced by CionaKv4 and CionaKv4/KChIP for a test pulse to +50 mV from a holding potential of-100 mV were 207±27 pA and 867±62 pA, respectively(P<0.0001, Welch t-test). Deletion of the N terminus(amino acids 2-32) of CionaKv4 also increased the average current amplitude significantly (Figs 5C, 6Aii), but to a lesser extent than when this channel was co-expressed with CionaKChIP (Figs 5D, 6Aiii). For instance, average currents produced by ntCionaKv4 channels during a pulse to +50 mV had an average peak of 370±65.2 pA compared with 207±27 pA for CionaKv4 alone (P<0.05, Welch t-test). Addition of CionaKChIP did not affect peak current amplitude of ntCionaKv4 channels, which lacked the N terminus, since the average peak current of ntCionaKv4 channels in the presence of CionaKChIP subunits (432±97) was not significantly different(P>0.05, Student t-test) from the average peak amplitude of currents produced by ntCionaKv4 channels alone (370±65.2 pA), as shown in Fig. 6Aiii.

In all cases, with or without N-terminal truncation and with or without co-expression of KChIP, current levels reached a plateau at stimulation voltages above +50 mV, indicating that conductance decreases above this voltage. The mechanism for this decrease in conductance is unknown. We speculate that the outward current is inhibited at higher depolarization voltages by blocking of the internal face of the ion pore by a positively charged component of the cytoplasm. If the block is similar to Mg²⁺or polyamine blocking of K_ir channels then the affinity of the cytoplasmic blocker must be lower than that seen for polyamine block of K_ir channels since blockage of CionaKv4 does not appear until + 50 mV.

The average time constant of activation of currents produced by CionaKv4 channels during a pulse to +50 mV (3.4±0.2 ms) was not affected by the presence of CionaKChIP (3.3±0.2 ms), as shown in Fig. 6Bi(P>0.05, Student t-test). The voltage-dependence of the time constant for activation was similar for channels expressed alone or with CionaKChIP, increasing by e-fold every 29 or 33 mV, respectively. Deletion of amino acids 2-32 from CionaKv4 channels was also without effect on the rate of activation, since the average time constant of activation of currents produced by this N-terminally deleted CionaKv4 channel was 3.2±0.3 ms (P>0.05, Student t-test). The voltage-dependence of activation kinetics was similar for CionaKv4 and ntCionaKv4 channels, increasing by a factor of e for every 26 or 29 mV of voltage change, respectively(Fig. 6Bii). Activation kinetics of ntCionaKv4 in the presence of CionaKChIP was 3.5±0.3 ms, which was not significantly different from the value for ntCionaKv4 alone (P>0.05, Student t-test). Interestingly, the voltage-dependence of activation kinetics was different for ntCionaKv4 channels alone (e-fold increase every 26 mV) than when co-expressed with CionaKChIP (e-fold increase every ∼37 mV),which suggests that CionaKChIP slightly decreased the voltage sensitivity of activation kinetics of ntCionaKv4 channels(Fig. 6Biii).

CionaKChIP enhanced activation of CionaKv4 by decreasing the voltage necessary to activate half the channels(Fig. 6Ci), since first order Boltzmann fits to the relationships between normalized peak conductance vs voltage had midpoints (V_0.5) of-0.5±1.7 mV for CionaKv4 and -29.2±1.1 mV for CionaKv4/KChIP (P<0.0001, Student t-test). The slope values of the Boltzmann curves were 15.5±0.8 mV/e-fold(CionaKv4) and 13.4±0.3 mV/e-fold (CionaKv4/KChIP)(P<0.05, Student t-test).

Deletion of the N terminus of CionaKv4 also shifted the midpoint of activation of CionaKv4 in the hyperpolarizing direction(Fig. 6Cii). First order Boltzmann fits to normalized peak conductance-voltage relationships had significantly different midpoints of activation (V_0.5) of-0.5±1.7 mV for

CionaKv4 and -12.3±1.6 mV for ntCionaKv4(P<0.0001, Student t-test). The slopes for these curves were 15.5±0.8 mV/e-fold and 14.3±0.4 mV/e-fold, respectively(P>0.05, Welch t-test).

Deletion of the N terminus eliminated the effects of CionaKChIP on activation midpoint (Fig. 6Ciii). The V_0.5 of first order Boltzmann fits to the conductance-voltage relationships were -15.2±1.3 mV for ntCionaKv4 in the presence of CionaKChIP (-12.3±1.6 mV for ntCionaKv4 alone, P>0.05, Student t-test). The slope factors of these fits were 14.3±0.4 mV/e-fold for ntCionaKv4 channels expressed alone, and 13.4±0.2 mV/e-fold for ntCionaKv4 channels co-expressed with CionaKChIP subunits (P>0.05, Student t-test).

All the biophysical parameters determined in this study are summarized in Table 1.

The decaying phase of CionaKv4 currents obtained with test pulses between +20 mV and +80 mV was well fitted by a double exponential function with fast (τ₁) and slow (τ₂) time constants of inactivation. Currents produced by CionaKv4/KChIP complexes inactivated more slowly than currents produced by CionaKv4 channels(Fig. 5B). The decay phases of the currents produced by CionaKv4/KChIP complexes were well fitted with single exponential functions, whose single inactivation time constants were larger than the slow inactivation time constants of currents produced by CionaKv4 channels (Fig. 7Ai and Table 1). For instance, the double exponential fit to the decay of currents produced by CionaKv4 in response to a pulse to +50 mV had a fast time constant of 26±2 ms, which contributed to ∼82% of total current decay, and a slow time constant of 147±26 ms, which contributed to ∼18% of the current decay. However, a single exponential fit to the inactivation phase of currents conducted by CionaKv4/KChIP complexes at the same potential had a time constant of 291±42 ms.

The decaying phase of currents produced by N-terminally deleted CionaKv4 channels was also well fitted by a double exponential function, but the inactivation time constants were slower than for wild-type channels (Fig. 7Aii) and they contributed equally to total inactivation(Table 1). For instance, the decay phase of ntCionaKv4 currents during a test pulse to +50 mV was best described by a double exponential function with a fast time constant of 56±6 ms and a slow time constant of 381±50 ms. Interestingly,co-expression of ntCionaKv4 channels with CionaKChIP subunits slowed inactivation kinetics relative to those for ntCionaKv4 channels alone (Fig. 5D). The decaying phases of these currents were best fitted with single exponential functions, whose time constant was similar to that from untrucated CionaKv4 channels expressed with CionaKChIP(Fig. 7Aiii and Table 1). These results demonstrate that CionaKChIP still forms a functional complex with CionaKv4 channels that lack the N terminus.

Since Kv4 channels inactivate mostly from the closed state (Bahring et al.,2001a), we hypothesized that CionaKChIP slowed inactivation kinetics of CionaKv4 channels by interfering with channel closure. To address this question, we measured and compared the kinetics of tail currents produced by CionaKv4 channels and CionaKv4/KChIP complexes in response to re-polarizing pulses from -160 to -110 mVdelivered after a brief(10-15 ms) depolarizing pulse to +50 mV, which would activate a maximum number of channels. Our results suggest that channel closing of CionaKv4 channels is indeed inhibited by CionaKChIP subunits(Fig. 7Bi). For example, during a re-polarizing pulse to -130 mV, CionaKv4 tail currents deactivated with a time constant of 4.1±0.3 ms, whereas CionaKv4/KChIP tail currents deactivated with a time constant of 8.3±0.4 ms. These values were significantly different (P<0.0001, Welch t-test). Deletion of the N terminus of CionaKv4 also slowed closure kinetics (Fig. 7Bii),indicating that this domain of CionaKv4 has a role in accelerating channel closure. For example, during a pulse to -130 mV, currents produced by ntCionaKv4 deactivated with a time constant of 5.7±0.6 ms,which was significantly slower (P<0.05, Welch t-test)than for wild-type channels (4.1±0.3 ms). The N terminus of CionaKv4 seems to be required for modulation of closing kinetics by CionaKChIP because deactivation time constants of tail currents produced by ntCionaKv4 channels and ntCionaKv4/KChIP complexes were not significantly different(Fig. 7Biii). For example, at the re-polarizing voltage of -130 mV, the deactivation time constant for ntCionaKv4 in the presence of CionaKChIP was 6.5±0.6 ms, which was not significantly different (P>0.05, Student t-test) from the average deactivation time constant for ntCionaKv4 channels expressed alone (5.7±0.6 ms).

Addition of CionaKChIP shifted the midpoint of steady-state inactivation of CionaKv4 to the left(Fig. 7Ci). Midpoints of single order Boltzmann fits to steady-state inactivation curves of CionaKv4 and CionaKv4/KChIP were -72±2 mV for CionaKv4 and-88±2 mV for CionaKv4/KChIP (P<0.001, Student t-test). The slope factors of these Boltzmann fits to steady-state inactivation were also significantly different: 3.6±0.2 mV/e-fold for CionaKv4 and 4.3±0.2 mV/e-fold for CionaKv4/KChIP(P<0.001, Student t-test). The N terminus of CionaKv4 does not appear to contribute to steady state inactivation(Fig. 7Cii) since midpoints of inactivation of N-terminally deleted channels (-73±2 mV) were not significantly different (P>0.05, Student t-test) from inactivation midpoints of wild-type channels (-72±2 mV). The slopes of the steady-state inactivation curves were not significantly different(P>0.05, Student t-test): 3.6±0.2 mV/e-fold for CionaKv4 and 3.7±0.2 mV/e-fold for ntCionaKv4. However, the effect of CionaKChIP on steady-state inactivation parameters was dependent on the N terminus of CionaKv4, since deletion of this domain abolished these effects(Fig. 7Ciii). Midpoints of inactivation were -73±2 mV for ntCionaKv4 and -74±2 mV for ntCionaKv4/KChIP. Slope factors were 3.7±0.2 for ntCionaKv4 and 4.1±0.3 for ntCionaKv4/KChIP(P>0.05, Student t-test, for both midpoint and slope values).

The rate of recovery from inactivation of CionaKv4 channels at-100 mV was also affected by CionaKChIP(Fig. 7Di). Recovery from inactivation was measured as the fractional recovery of current as a function of time at -100 mV. This relationship was well fitted by a single exponential function with recovery time constants from inactivation (τ_rec)of 387.4±46.6 ms for CionaKv4 and 330.6±49.5 ms for ntCionaKv4 (P>0.05, Student t-test), indicating that the N-terminal peptide sequence alone is not a significant factor in recovery (Fig. 7Dii). The recovery of unmutated CionaKv4 channels was significantly slowed by the presence of CionaKChIP, with a time constant of recovery of 927±47 ms (Fig. 7Di and Table 1). The recovery of the ntCionaKv4 channel was also slowed significantly by the presence of CionaKChIP, with a time constant of recovery of 587±76 ms(Fig. 7Diii and Table 1), although the effect was less than on the unmutated CionaKv4.

Since orthologous transcription factors control the early stages of cardiac morphogenesis in invertebrates (Bodmer,1993), early chordates (Komuro and Izumo, 1993; Holland et al., 2003) and vertebrates(Davidson and Levine, 2003),it is reasonable to suppose that hearts are homologous across chordate phylogeny (Romer and Parsons,1986) and also within certain invertebrate lineages. Therefore, it was not surprising that the message for a Kv4 channel was present in RNA extracted from the heart of the tunicate Ciona intestinalis, since these channels are known to mediate a transient current responsible for the`notch' in cardiac action potentials from fish to mammals(Dixon et al., 1996;Brahmajothi et al., 1999; Nurmi and Vornanen, 2002; Roden et al.,2002). Similarly modulation of Kv4 mediated currents by KChIP and analogous calcium-binding accessory subunits is common across the Metazoa(Rosati et al., 2001; Zhang et al., 2003). Although genes for KChIP subunits have been identified by BLAST searches in the genomes of Drosophila and Anopheles, CionaKChIP is the first KChIP subunit from an invertebrate or non-vertebrate chordate that has been cloned and characterized.

The genomic structure of CionaKv4(Fig. 1B) was similar to the genomic structures of the three mammalian Kv4 channel genes, which have been previously characterized (Isbrandt et al.,2000). Interestingly, the CionaKv4 gene contains an additional intron that partitions the codons for the first 17 amino acids onto an additional exon. For comparison, exons 1-7 of CionaKv4 were numbered 0-6 because the boundaries and relative sizes of exons 2-7 correspond to exons 1-6 of mammalian Kv4 channels. The exon that encodes the first 17 amino acids of CionaKv4 is referred to as `exon 0' and is located∼8 kb upstream of exon 1 (Fig. 1B). As in mammalian Kv4 channel genes, exon 1 of CionaKv4 encodes the transmembrane domains S1-S5 and the first part of the pore domain of CionaKv4, exon 2 encodes the second part of the pore and transmembrane domain S6, and exons 3-6 encode the C terminus. The intron between exons 1 and 2 in CionaKv4 interrupts the coding sequence of the K⁺ selectivity sequence GYG at exactly the same position as the equivalent intron of mammalian Kv4 genes. The sixth exon of CionaKv4 is larger in this gene than in its mammalian counterparts,encoding a comparatively longer C terminus.

Comparison between the genomic structures of CionaKChIP(Fig. 1D) and the KChIPs cloned from several mammalian species revealed that most exon/intron boundaries of KChIPs are highly conserved. For example, the last exon of CionaKChIP, KChIP2 and KChIP3 are identical in length(Decher et al., 2004; Spreafico et al., 2001). Each of the EF-hands 1 to 3 of CionaKChIP is encoded by two exons, as in KChIP3, and the partition of these three EF-hands into two exons occurs in the same positions in CionaKChIP and KChIP3. The fourth EF-hand is similarly encoded in a single exon in CionaKChIP, KChIP2 and KChIP3(Decher et al., 2001; Spreafico et al., 2001). The number of exons that encode the different mammalian KChIPs is variable, since numerous splice variants from a single KChIP gene are possible; for example,the gene encoding KChIP2 subunits can be spliced differently to produce at least eight isoforms that differ in their N terminus: KChIP2a(An et al., 2000), KChIP2b(Bähring et al., 2001b),KChIP2c (Decher et al., 2001),KChIP2d (Patel et al., 2002),KChIPs 2e-g (Decher et al.,2004), and KChIP2t(Deschênes et al.,2002). The fact that CionaKChIP is encoded by several exons, in a similar fashion to the mammalian KChIP isoforms(Fig. 1D), suggests that several splice variants of CionaKChIP are possible, although only one variant was detected in our experiments. Although most splice variants of KChIPs differ in their N termini, variants that differ in their C termini have been also cloned (Decher et al.,2004). Our results suggest that splice variants of CionaKChIP that differ in their C termini are not expressed, or are expressed at very low levels in the heart of Ciona intestinalis,because otherwise these would have been detected with the 3′-RACE assay. Since we did not perform a 5′-RACE assay it is possible that alternatively spliced variants of CionaKChIP with one or more alternative N-terminal exons, variants containing the information encoded by possible additional exons, or splice variants without an N terminus, similar to the minimal isoform KChIP2d (Patel et al., 2002), are expressed in the heart of C. intestinalis. However, we can suggest that splice variants that do not contain the information of one or more of exons 2-5 are not expressed, or expressed at relatively low levels, otherwise it is likely that these would have been detected when amplifying the ORF of CionaKChIP.

CionaKv4 produced A-type currents that activated and inactivated with relatively rapid kinetics. The relative contribution of the fast inactivation component (∼82%) was comparable to that of lobster Shal (Baro et al.,1996), Kv4.2 (Bähring et al., 2001a) and Kv4.3 (Wang et al., 2002), but was significantly greater than the <20%recorded for Kv4.1 channels (Jerng and Covarrubias, 1997). The fast inactivation time constant of CionaKv4 at 12°C was about 26 ms, similar to lobster Shal (∼31 ms at 16°C; Baro et al., 1996) and mammalian Shal (∼23-23°C; Pak et al., 1991). The fact that this inactivating component has a similar time constant at very different temperatures suggests that inactivation kinetics of Kv4 channels are not constrained by temperature. Low Q₁₀ values are common for marine animals in shallow temperate waters.

The midpoint of activation of CionaKv4, ∼0 mV(Fig. 6Ci and Table 1), is considerably depolarized with respect to that of the Kv4 channel cloned from another tunicate (Halocynthia), around -20 mV, as measured from Fig. 2 of Nakajo et al.(2003) and with respect to most mammalian Kv4 channels, approximately -10 to -15 mV (see records for Kv4 family in http://vkcdb.biology.ualberta.ca/). The first N-terminal amino acids of CionaKv4 are likely involved in regulating membrane expression, since deletion of these residues resulted in increased current amplitude relative to wild-type channels when expressed in Xenopus oocytes (Fig. 6Ai). This result mimicked the effect of deleting a similar fragment from mammalian Kv4 channels(Bähring et al., 2001a; Shibata et al., 2003). It has been shown that the hydrophobicity of most of the first ∼30 amino acids of Kv4 channels impedes the trafficking of these channels to the membrane(Shibata et al., 2003).

Deletion of amino acids 2-32 of CionaKv4 slowed the fast component of inactivation approximately twofold (Fig. 7Aii) and slowed deactivation kinetics(Fig. 7Bii). However, the N terminus of CionaKv4 does not play a role in steady-state inactivation (Fig. 7Cii). These data suggest that the role of the N terminus of CionaKv4 in inactivation is comparable to the role of the N terminus of Kv4.2 channels,because deleting the first 40 amino acids of Kv4.2 channels had similar effects on the rate of inactivation and steady-state inactivation(Bähring et al., 2001a). The role of the N terminus of Kv4.1 channels is divergent, as its deletion resulted in loss of the fast component of inactivation of this isoform(Pak et al., 1991; Jerng and Covarrubias, 1997). Interestingly, the N terminus of CionaKv4 also contributes to the voltage-dependence of activation, since deletion of this domain resulted in a significant shift of the activation midpoint by about 12 mV in the hyperpolarizing direction relative to wild-type channels(Fig. 6Cii and Table 1).

Like the N terminus of mammalian Kv4 channels (Bähring et al., 2001),the N terminus of CionaKv4 is essential for modulation of some channel properties by KChIP subunits, since in the absence of this domain, CionaKChIP did not increase current amplitude, shift midpoint values of activation or inactivation or affect the kinetics of deactivation of CionaKv4 channels (see below). However, the macroscopic decay of ntCionaKv4/KChIP currents was slower than the decay of ntCionaKv4 channels alone. It is likely that CionaKChIP, in addition to interacting with the proximal amino acids of the N terminus of CionaKv4, also binds to a second domain of the N terminus that was not affected by the deletion performed on CionaKv4 channels. Mammalian KChIPs interact with at least two domains of the N terminus of Kv4 channels, one within residues 7-11 and another comprising residues 71-90(Scannevin et al., 2004), and this second binding site is sufficient for binding to KChIPs.

As do mammalian KChIP isoforms 1-3 (An et al., 2000), CionaKChIP increased the amplitude of CionaKv4 currents (Figs 5B, 6Ai). The mammalian KChIP isoforms do not increase the Kv4 current amplitude by a direct effect on single channel conductance (Beck et al.,2002), but by preventing aggregation of their hydrophobic N terminus in the ER that would interfere with folding. Proper folding favours the insertion of Kv4 channels into the plasma membrane and increases their half-life in the membrane (Shibata et al.,2003). This study did not determine whether CionaKChIP increases current amplitude by facilitating proper insertion of these channels in the plasma membrane or by increasing the conductance of CionaKv4 channels. CionaKChIP shifted the midpoint of activation of CionaKv4 in the hyperpolarizing direction(Fig. 6Ci), which is consistent with the effect of most mammalian KChIPs(An et al., 2000; Van Hoorick et al., 2003).

CionaKChIP slowed inactivation of the transient currents produced by CionaKv4. This was reminiscent of the effects of the mammalian isoform KChIP4a on mammalian Kv4 channels(Holmqvist et al., 2002). The N terminus of KChIP4a [K inactivation suppressor domain (KIS)] is required for its effect on inactivation kinetics(Holmqvist et al., 2002). To determine whether the N terminus of CionaKChIP also has the KIS domain, we aligned the first 40 amino acids of CionaKChIP with the first 40 amino acids of KChIP4a using T-Coffee software(Notredame et al., 2000). The results of this alignment are shown in Fig. 8A. For comparison, we also aligned the first 40 amino acids of CionaKChIP with the first 40 amino acids of representatives of the other vertebrate KChIP subfamilies (KChIP1-3). Interestingly, the N terminus of CionaKChIP was more similar to the N terminus of the KChIP4a isoform than to the N termini of other KChIP isoforms. The alignment between CionaKChIP and KChIP4a revealed that two leucines, at positions 3 and 6 in both subunits, and a methionine, at position 8 also in both subunits, are conserved between KChIP4a and CionaKChIP(Fig. 8A). Other residues that are conserved between the N terminus of KChIP4a and CionaKChIP are a glutamic acid residue and a glycine residue at positions 19 and 22, in CionaKChIP (positions 23 and 26 in KChIP4a), an alanine-glycine motif located at position 28 in CionaKChIP (position 30 in KChIP4a), and a valine residue at position 34 in CionaKChIP (position 32 of KChIP4a; Fig. 8A). These conserved residues may be crucial to the modulation by KChIP subunits on inactivation kinetics. The alignments between CionaKChIP and KChIP2a or KChIP3a had extensive gaps (Fig. 8C,D). The alignment between CionaKChIP and KChIP1a has fewer and smaller gaps for other isoforms, perhaps revealing a closer relationship between the N termini of these two subunits (Fig. 8B).

Most KChIP isoforms either do not affect the midpoint of inactivation (e.g. KChIP1a; Nakamura et al.,2001; Van Hoorick et al.,2003) or shift the midpoint of inactivation in the depolarizing direction (e.g. KChIP2b; Bähring et al., 2001b). However, the splice variant KChIP2g, which has an alternative exon 1 (Decher et al.,2004) and KChIP 1b, which contains the information of an additional exon in its N terminus (Van Hoorick et al., 2003), shifts the midpoint of inactivation in the hyperpolarizing direction as for CionaKChIP(Fig. 7Ci). However, the N termini of these mammalian KChIP splice variants and CionaKChIP are not similar.

CionaKv4 channels required ∼1 s to fully recover from inactivation at -100 mV, with a recovery time constant of ∼387 ms(Fig. 7D). Similar values have been reported for jellyfish and fly Shal(Jegla and Salkoff, 1997) and Kv4.2 (Serôdio et al.,1994) for the same membrane potential. Kv4.1 channels recover fully from inactivation within 300-400 ms at -100 mV(Serôdio et al., 1994),while Kv4.3 channels recover from inactivation relatively rapidly, with recovery time constants of less than 200 ms, also at -100 mV(Dixon et al., 1996). In conclusion, invertebrate and non-vertebrate chordate Kv4 channels, including CionaKv4, exhibit lower rates of recovery from inactivation than the mammalian Kv4.1 and KvV4.3 channel subtypes, suggesting that relatively slow kinetics of recovery is a primitive feature of Kv4 channels.

Since the mammalian genome contains three Kv4 paralogues and four KChIP paralogues, while the genome of Ciona intestinalis contains only one gene for Kv4 channels and a single gene for KChIP subunits, it is probable that both genes duplicated and diverged in the vertebrate lineage during evolution. Current theory states that there were two major genome duplications early in the evolution of the vertebrates(Ohno, 1970). Most vertebrate gene families (81%) are composed of two or three paralogues that derive from a single ancestral gene (Escriva et al.,2002), as is the case for the Kv4 channel gene family. Gene families formed by three paralogues are believed to have lost a fourth paralogue after the second large-scale genomic duplication(Spring, 1997). Since the KChIP gene family has four paralogues in vertebrates, it is unlikely that gene loss occurred after the second large-scale duplication. The pattern of KChIP clades in the vertebrate lineage (Fig. 4B), [A(B(CD))], is not consistent with two successive genome duplications. The fact that this pattern is common within vertebrate gene families composed of four paralogues has been used by Hugues(1999) to argue against two rounds of genomic duplication, as this duplication pattern would have rendered a topology of the form: (AB)(CD). Sidow(1996) has suggested that duplications of single genes or fragments of the genome that were independent of large-scale genomic duplications have played a role in the evolution of the vertebrates and might explain these discrepancies. However, the support values for the two internal branching events in the vertebrate KChIP clade are so low that the relationships of these four paralogues are most reasonably described by an unresolved polytomy rather than either of the specific evolutionary models.

Until this study, invertebrate KChIPs had not been cloned. However it has been shown that crustacean Kv4 channels can be modulated by a mammalian KChIP(Zhang et al., 2003) in a similar fashion to that shown by the effect of CionaKChIP on CionaKv4 shown in this study. This is the first cloning of a non-vertebrate KChIP, and since tunicates are the earliest diverging chordate clade, these results show that modulation of Kv channels by calcium-binding proteins in vertebrates is an ancient and conserved mechanism.

We conclude that tunicate KChIP subunits modulate expression, voltage sensitivity of gating, and kinetic behavior of tunicate Kv4 channels. Modulation by CionaKChIP requires an intact N terminus of CionaKv4, as in mammals. CionaKChIP dramatically slowed the rate of inactivation of CionaKv4 channels, similarly to the vertebrate KChIP4a isoform (Holmqvist et al., 2004) and our comparative analyses may have provided clues as to the N-terminal amino acids in KChIP that are responsible for this action. Our results are also the first report on the molecular nature of excitability in the tunicate myocardium, which may help further our understanding of its physiology and the evolution of hearts. We suggest that modulation of Kv4 channels by KChIP subunits is a conserved mechanism to modulate cardiac excitability.

List of symbols and abbreviations
 
• E_K

  measured potassium equilibrium potential

 
• G_K

  potassium conductance

 
• G_max

  maximal conductance

 
• I_max

  maximum current

 
• I₀

  offset current

 
• I_peak

  peak measured amplitude

 
• K

  slope factor

 
• K_ir

  inward rectifier potassium channel

 
• KChIP

  potassium channel interacting protein

 
• Kv4

  voltage-gated potassium channel, Shal subtype

 
• t

  time

 
• V

  voltage

 
• V_0.5

  voltage at which activation/inactivation is half maximal

 
• τ

  activation time constant.

 
• τrec

  recovery time constant