Mechanisms of Calmodulin Regulation of Different Isoforms of Kv7.4 K+ Channels*

Calmodulin (CaM), a Ca2+-sensing protein, is constitutively bound to IQ domains of the C termini of human Kv7 (hKv7, KCNQ) channels to mediate Ca2+-dependent reduction of Kv7 currents. However, the mechanism remains unclear. We report that CaM binds to two isoforms of the hKv7.4 channel in a Ca2+-independent manner but that only the long isoform (hKv7.4a) is regulated by Ca2+/CaM. Ca2+/CaM mediate reduction of the hKv7.4a channel by decreasing the channel open probability and altering activation kinetics. We took advantage of a known missense mutation (G321S) that has been linked to progressive hearing loss to further examine the inhibitory effects of Ca2+/CaM on the Kv7.4 channel. Using multidisciplinary techniques, we demonstrate that the G321S mutation may destabilize CaM binding, leading to a decrease in the inhibitory effects of Ca2+ on the channels. Our study utilizes an expression system to dissect the biophysical properties of the WT and mutant Kv7.4 channels. This report provides mechanistic insights into the critical roles of Ca2+/CaM regulation of the Kv7.4 channel under physiological and pathological conditions.

The Ca 2ϩ -binding protein calmodulin (CaM) 2 exemplifies one critical player within the massive interactome of proteins that sculpt Ca 2ϩ -dependent signaling as diverse as cellular motility, gene transcription, and ion channel regulation (1)(2)(3)(4)(5)(6). Indeed, alterations in protein-protein interactions may result in human genetic diseases (7,8). CaM often mediates its functions by interacting with the IQ/CaM binding domains (CaMBDs) of target proteins, which have varying consensus sequences that bind to apoCaM and/or calcified CaM with K d s in subnanomo-lar concentrations (6,9,10). Local placement of Ca 2ϩ sensors allows swift and efficient Ca 2ϩ signaling, whereas global Ca 2ϩ promotes the integration of Ca 2ϩ -mediated functions (3). An example is the tethering of CaM to Ca 2ϩ channels, which serve as the Ca 2ϩ source (11,12). However, in instances where the target protein does not constitute the Ca 2ϩ source, (e.g. Ca 2ϩ / CaM regulation of Na ϩ and K ϩ channels) (4,13), the Ca 2ϩsensing mechanisms are unclear.
In the mouse inner ear, there are at least four functionally distinct splice variants of Kv7.4 that have been identified to be expressed differentially in HCs and spiral ganglion neurons (SGNs) and whose altered expression has been implicated as the potential mechanism for the disease (21,22). A prominent difference between the splice variants of the human Kv7.4 (hKv7.4) occurs at the C termini, where CaMBDs and the coiled-coil domain for subunit oligomerization are featured (18,(23)(24)(25).
The hKv7.4 consists of two major isoforms, a and b, that contain two putative CaMBDs (10). Additionally, the a isoform has a third CaMBD at the C terminus. The Ca 2ϩ -sensing domain for the Kv7.4/CaM complex is not well defined, although earlier studies have suggested that the N lobe is the Ca 2ϩ sensor (10). Here we utilize an expression system to systematically determine the Ca 2ϩ /CaM regulation of the two distinct isoforms of hKv7.4. Moreover, because a known DFNA2 missense mutation (G321S) lies upstream of the CaMBD, we tested the hypothesis that the missense mutation may alter the inhibitory effects of Ca 2ϩ /CaM on the Kv7.4 channel.
Multidisciplinary techniques were used, including patch clamp recordings, biochemical studies, and molecular modeling. Insights into these existing gaps in our knowledge may provide an important framework for understanding how mutations result in autosomal-dominant, progressive, high-fre-* This work was supported by National Institutes of Health Grants R01 DC0100386, R01 DC007592, and DC003826 (to E. N. Y.); R01 HL085844 and R01 HL085727 (to N. C.); and R01 HL085844-S1 (to R. L. W.). This work was also supported by Department of Veterans Affairs Merit Review Grant I01BX000576 (to N. C.) and American Heart Association Western States Beginning Grant-in-Aid 14BGIA18870087 (to X. D. Z.). The authors declare that they have no conflicts of interest with the contents of this article. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. 1  quency hearing loss and, possibly, for the development of strategies to alleviate the symptoms of the disease.

Experimental Procedures
Generation of Mutant (MT) Forms of hKv7.4 (hKCNQ4) and Epitope-tagged Constructs-WT hKv7.4 clones, isoform a (695 amino acids, gene ID NM_004700) and isoform b (641 amino acids, gene ID NM_172163), were purchased from Qiagen (Valencia, CA). The coding sequences were subcloned into the pIRES2-EGFP plasmid vector (Clontech, Mountain View, CA). Isoleucine and glutamine residues in the IQ domain were substituted by valine and alanine residues to generate IQ mutant channels in WT hKv7.4 (isoform a and b) clones (26) using the QuikChange II mutagenesis kit (Stratagene, La Jolla, CA) and verified by automated sequencing. A missense mutation, G321S, has been shown previously to cause progressive hearing loss (27). The mutant G321S channel (Kv7.4 G321S ) was generated in the WT hKv7.4 (isoform a). pIRES2-EGFP-hKv7.4-WT and mutants were used in electrophysiologic recordings and biochemical studies. We used EGFP expression as a transfection marker.
To study the subcellular localization of WT and MT subunits, two different epitopes, modified HA and c-Myc tags, were inserted into the pCMV-hKv7.4-WT and MT constructs in which the EGFP genes were eliminated. Modified HA and c-Myc epitopes were flanked with the ClC-5 chloride channel D1-D2 loop to increase accessibility and inserted in the end of the S1-S2 loop of hKv7.4, as described previously, in the Kv7.2/7.3 (28) and Kv7.4 (29) channels. S1-S2 loop amino acid sequences were changed to STIQEHQELANENSEHYPYDVPDYAVTFEERDKCPEWNC for HA-tagged constructs and STIQEHQELANENSEH-EQKLISEEDLVTFEERDKCPEWNC for c-Myc tagged constructs. The inserted regions are underlined, and epitopes are shown in boldface. Epitope tags were generated by recombination polymerase chain reaction and verified by automated sequencing.
Heterologous Expression, Tissue Culture, and Transfection-Plasmids (pJG60 and pGJ65) containing CaM and DN MT of CaM (CaMDN) were obtained from Dr. Trisha Davis (University of Washington, Seattle, WA). CaMDN has alanine substitutions of each of the four aspartates in the Ca 2ϩ -binding EF hands (D20A, D56A, D93A, and D129A) (30). CHO cells were obtained from the ATCC. CHO cells were maintained in F12-K containing 10% fetal bovine serum (Invitrogen). Cell cultures were kept at 37°C in a 5% CO 2 incubator. The cells were trypsinized, plated at a concentration of 1.5 ϫ 10 5 cells/ml in 2 ml of culture medium in 35-mm dishes, and transfected with 1 g of total DNA/dish. Transfections were performed using Lipofectamine following the protocol of the manufacturer (Invitrogen). The cells were rinsed in fresh culture medium and incubated for 24 h before patch clamp recordings. Transfected cells were identified for recording by visualization of the EGFP co-expression (BD Biosciences, Clontech).
Immunoprecipitation-CHO cells expressing either Kv7.4a or Kv7.4b were lysed in radioimmune precipitation assay buffer (50 mM Tris-HCl (pH 8), 150 mM NaCl, 1% Nonidet P-40, 200 M Na 3 VO 4 , 0.5% sodium deoxicholate, 0.1% SDS, and protease inhibitor mixture tablets (Roche)). Cell lysates were mixed with protein A-Sepharose beads (Roche) for preclearing and incubated with anti-Kv7.4 antibody (Neuromab) and 2 mM of EGTA or CaCl 2 overnight at 4°C. Immune complexes were pulled down with beads and washed three to four times with radioimmune precipitation assay buffer. Immunoprecipitated proteins were separated by SDS-PAGE and blotted onto nitrocellulose filters after adding protein loading dye and boiled for 5 min. Blocking was performed at room temperature for 1 h in TBS with 5% nonfat milk, followed by incubation with anti-Kv7.4 (1:500 dilution) and anti-CaM (1:1000 dilution) antibodies in TBS. After washing with TBS, the membranes were incubated with horseradish peroxidase-conjugated secondary antibodies. Proteins were visualized using an enhanced chemiluminescence kit (Thermo Scientific, Pittsburgh, PA).
CaM Binding Site Prediction-Kv7.4a and b protein sequences (NP_004691.2) were used to predict CaM binding sites by using web-based tools from a CaM target database.
Electrophysiology-All electrophysiological experiments were performed at room temperature (21-22°C). Reagents were obtained from Sigma-Aldrich (St. Louis, MO) unless specified otherwise. Ca 2ϩ concentrations were determined using custom-written software.
Whole-cell and cell-attached single-channel voltage clamp recordings were performed on CHO cells using an Axopatch 200B amplifier (Molecular Devices, Sunnyvale, CA). Current traces were amplified, filtered (bandpass, 2-10 kHz), and digitized at 5-500 kHz using an analog-to-digital converter, Digidata 1322A (Molecular Devices), as described earlier (26). Fire-polished electrodes (3-5 M⍀) were pulled from borosilicate glass. Several basic criteria were set to ensure optimum quality of recordings and acceptance of data, including the following: initial stable seals for at least 5 min before recordings, elimination of cells with current traces that exhibit signs of voltage inhomogeneities, and exclusion of cells with more than 20% change in series resistance before and during recordings.
Whole-cell K ϩ current amplitudes at varying test potentials were measured at the peak and steady-state levels using a peak and steady-state detection routine. The current magnitude was divided by the cell capacitance (picofarad) to generate the current density-voltage relationship. For single-channel recordings, leakage and capacitative transient currents were subtracted by fitting a smooth template to null traces. Leaksubtracted current recordings were idealized using a halfheight criterion. Transitions between closed and open levels were determined using a threshold detection algorithm that required that two data points exist above the half-mean amplitude of the single-unit opening. The computer-detected openings were confirmed by visual inspection, and sweeps with excessive noise were discarded. Amplitude histograms at a given test potential were generated and then fitted to a single Gaussian distribution using a Levenberg-Marquardt algorithm to obtain the mean Ϯ S.D. At least four voltage steps and their corresponding single-channel currents were used to determine the unitary conductance. Single-channel current-voltage relations were fitted by linear least square regression lines, and single-channel conductances were obtained from the slope of the regression lines. Idealized records were used to construct ensemble-averaged currents, open probability, and histograms for the distributions of open and closed intervals.
Statistical Analyses-Where appropriate, pooled data are presented as means Ϯ S.E. Curve fits and data analysis were performed using Origin software (MicroCal Inc.). Statistical comparisons were performed using the statistical package in the Origin software, with p Ͻ 0.05 considered significant. For multiple comparisons, one-way analysis of variance combined with Tukey test was used. For comparisons between two groups, Student's t test was used.

Inhibitory Effects of CaM on the Long Isoform of hKv7.4 -
Previous studies have documented functionally distinct splice variants of Kv7.4 that are expressed differentially in HCs and SGNs (21,22). We first examined the sequence alignment of two isoforms, hKv7.4a and b, as well as the mouse homologues of Kv7.4 v1 and Kv7.4 v4 , which show stark differential expression in inner ear HCs and SGNs with distinct functional phenotypes (21,22). The C terminus of hKv7.4a contains two CaMBDs, highlighted in red in Fig. 1A and designated here as CaMBD A and CaMBD B , respectively. We identified an additional putative CaMBD in the C terminus of hKv7.4a, highlighted in green in Fig. 1A. In contrast, the C terminus of hKv7b contains only two CaMBDs, CaMBD A and CaMBD B (Fig. 1A).
Shown in Fig. 1, B and C, are whole-cell hKv7.4a and b current traces. Heterologous expression of Kv7.4a and b yielded voltage-dependent K ϩ currents similar to the native current I KL in HCs and SGNs (39 -42) (Fig. 1, B and C). Because CHO cells have endogenous CaM (22), we co-expressed hKv7.4a with CaMDN constructs to knock down endogenous CaM effects. The resulting current was ϳ3-fold greater than the current from hKv7.4a alone, and the mid-point of activation (V 1 ⁄ 2) was shifted leftward by ϳ37 mV (Fig. 1D). In surprising contrast, expressing hKv7.4b alone yielded a current density that was at least ϳ2-fold larger than hKv7.4a but remained insensitive to CaM modulation after co-expression with CaMDN (Fig. 1, C and E). Indeed, the V 1 ⁄ 2 of hKv7.4b currents and the current density remained unchanged in the presence of CaMDN (Fig. 1E).
Inhibitory Effects of CaM on the Open Probability and the First Latency of Kv7.4a Unitary Currents-To assess the biophysical mechanisms for CaM-dependent modulation of hKv7.4, we recorded single-channel fluctuations in CHO cells expressing hKv7.4a alone and in combination with CaMDN (Fig. 2). Co-expression of hKv7.4a with CaMDN enhanced the probability of the channel opening by ϳ4-fold, but the single-channel unitary conductance remained unaffected (Fig. 2, A-E). Analyses of the first latency revealed that CaMDN decreased the waiting time to first opening of hKv7.4a by ϳ2-fold (Fig. 2F).

An Intact IQ Motif in CaMBD A Is
Required for the Regulation of hKv7.4a-Next, we examined the mechanisms for CaM regulation of hKv7.4 channels by performing site-directed mutagenesis of the IQ motif to valine and alanine residues, respectively, as illustrated in Fig. 3A. Expression of hKv7.4a IQ-VA in CHO cells yielded current densities that were greater than those of control hKv7.4a (Fig. 3, B and C). The current densities at the step voltage of 0 mV for hKv7.4a and hKv7.4a IQ-VA were 62 Ϯ 7 and 102 Ϯ 9 pA/pF, respectively (p Ͻ 0.01, n ϭ 19). In contrast, similar mutations of hKv7.4b to hKv7.4b IQ-VA produced current densities that were statistically unaltered (Fig. 3, B and  C). The current densities at the step voltage of 0 mV for hKv7.4b and hKv7.4b IQ-VA were 150 Ϯ 25 and 155 Ϯ 19 pA/pF, respectively (p ϭ 0.1, n ϭ 19). The findings from the current density suggest that an intact IQ motif is required for the observed inhibitory effects of hKv7.4a by CaM.
We probed whether the other putative CaMBDs, besides the IQ domain, can confer additional modulation of the channels. We co-expressed hKv7.4a IQ-VA with CaMDN ( Fig.   FIGURE 5. Tandem channel subunits clarify the effect of the MT subunit. A, schematic of expected channel combinations when cells express the hKv7.4 WT and G321S MT subunit or tandem clones. B, the tandem clone has two epitopes, HA and myc, located at the extracellular loop between the S1 and S2 transmembrane domains of Kv7.4a WT and G321S MT, respectively. Anti-HA and anti-myc antibodies were used for the immunofluorescence assay with the conditions either permeabilized or non-permeabilized. Scale bars ϭ 10 m. C, representative traces showing the whole-cell recording from CHO cells transfected with tandem clones alone or with CaMDN. The current traces were recorded from a holding potential of Ϫ70 mV to step potentials ranging from Ϫ90 to 60 mV using voltage increments of 10 mV. Tail currents were recorded at Ϫ30 mV. D, plots of current (I) density (picoampere/picofarad)-voltage relations of currents were derived from CHO cells expressing tandem clones with either empty vector or CaMDN. E, CHO cells were expressed with either hKv7.4a or tandem clones for 24 h. After preparation of cell lysates, immunoprecipitation was accomplished with Kv7.4 antibody to test CaM binding to the channel subunit. 3E). CaMDN did not produce a sizable increase in current density, nor did it shift the voltage-dependent activation of hKv7.4a IQ-VA currents, consistent with the lack of association between hKv7.4a IQ-VA with CaM (Fig. 3, F-H). These findings suggest the necessity for an intact IQ motif to ensure the regulatory role of the other CaMBDs (1).
Ca 2ϩ -dependent Modulation of Kv7 Channels in a Diseasecausing MT-A previous study has provided evidence that patients with a missense mutation of G321S (glycine-to-serine mutation at position 321) suffer from progressive hearing loss (27). Because the residue Gly-321 lies within the C terminus and is upstream of the CaMBD A , we took advantage of this known mutation to further examine the inhibitory effects of Ca 2ϩ /CaM on the Kv7.4 channel. Site-directed mutagenesis was used to generate the mutation in the Kv7.4 isoform a (Kv7.4 G321S ) and the mutant subunit was co-expressed with hKv7.4a WT subunits. Fig. 4A outlines the possible combinatorial assembly of a tetrameric channel. Neither the MT chan- nel nor addition of CaMDN yielded measurable currents (Fig.  4B, first and second panels). These findings are in keeping with a previous report that determined that Kv7.4 G321S , when expressed alone, had impaired membrane trafficking. However, co-expression of the MT with the WT subunits facilitated membrane trafficking of the MT subunits (26). As shown in Fig. 4B (third and fourth panels), expression of Kv7.4 G321S :Kv7.4 (at a 1:1 ratio) yielded a detectable current that was enhanced significantly in the presence of CaMDN (Fig. 4C). Analyses of the voltage-dependent activation of the tail currents revealed CaMDN-mediated leftward shifts in the V 1/2 of activation, consistent with CaM modification of the ensuing heteromeric channel assembly (Fig. 4, D  and E). However, compared with homomeric WT subunits, currents expressed from Kv7.4 G321S and Kv7.4 WT subunits exhib-ited smaller leftward shifts in the voltage-dependent activation (Fig. 4D-E). Co-immuno-precipitation experiments of CaM binding to the WT Kv7.4 and Kv7.4 G321S subunits revealed weak binding of the mutant subunit to CaM (Fig. 4F).
To further test the functionality of Kv7.4:Kv7.4 G321S heteromeric channels, we generated tandem Kv7.4:Kv7.4 G321S subunits ( Fig. 5A) with HA and c-Myc tags for Kv7.4 and Kv7.4 G321S subunits, respectively. We first assessed whether channels formed from the tandem subunits were expressed on the plasma membrane and identified the expression of the tandem channels using anti-HA and anti-c-Myc antibodies under non-permeabilized conditions (Fig. 5B). Expression of the tandem Kv7.4: Kv7.4 G321S subunit yielded outward K ϩ currents that were enhanced upon co-expression with CaMDN (Fig. 5, C and D). The co-immunoprecipitation experiments shown in Fig. 5E suggested that CaM binds more weakly with the tandem channel than the WT channel. At the single-channel level, CaMDN increased the probability of the channel opening without altering the unitary conductance (Fig. 6).
Structural Modeling of the hKv7.4 WT and G321S MT-To start exploring the experimental data on the G321S MT channel from a structural perspective, we generated preliminary structural models of the hKv7.4 WT and G321S MT channels as described under "Experimental Procedures." Fig. 7A shows the sequence alignment between the pore-forming domains of hKv7.4 channel and the rKv1.2 channel that was used as a template. Our modeling suggests that, in the WT Kv7.4 model (Fig.  7, B-D), native Gly-321 (in the S6 segment) allows the side chain of His-234 (in the S4-S5 linker) to come in direct contact with S6 and potentially stabilize a conformation of the C-terminal structure that favors the binding by CaM. In contrast, in the Kv7.4 G321S mutant model, the side chain of Ser-321 displaces the side chain of His-234 (in the S4-S5 linker) from its direct interaction with S6 observed in the WT model (Fig. 7E). On the basis of our model, the absence of a measurable current in the Kv7.4 G321S mutant (Fig. 4B) can be explained by the decoupling of the S4-S5 linker-S6 segment interaction that plays a key role in voltage-gated ion channel gating (37, 44 -46). Decoupling of the S4-S5 linker-S6 segment interaction can also affect the conformation of the C terminus, which may, in turn, affect CaM binding to the CaMBD region, as demonstrated by our experiments. However, predictions from our structural models will need to be tested in future studies.

Discussion
Kv7 channels underlie the neuronal M current and are constituents of the low voltage-activated K ϩ channels in auditory and vestibular HCs and neurons, where they enhance temporal fidelity and control the regularity of trains of spikes to sculpt the coding of information (39,42,47,48). Here we demonstrate that CaM binds to two different isoforms of hKv7.4 channel in a Ca 2ϩ -independent manner but that only the long isoform (Kv7.4a) is down-regulated by CaM. Ca 2ϩ /CaM mediate reduction of hKv7.4a currents by decreasing the channel open probability and altering activation kinetics. We demonstrate that G321S, a known missense mutation linked to progressive hearing loss, may destabilize CaM binding, leading to a decrease in the inhibitory effects of Ca 2ϩ on the channels.
The hKv7.4 CaMBD-Ca 2ϩ Complex-The co-immunoprecipitation data suggest that CaM binds to both isoforms of hKv7.4 in a Ca 2ϩ -independent manner. Because of the abundance of CaM in cells, the outcome of the immunoprecipitation assay may be due to the nonspecific bindings from CaM. To directly test the specific binding of hKv7.4 and CaM, we generated the hKv7.4a IQ-VA mutant channel. As demonstrated in Fig.  3F, co-immunoprecipitation of the hKv7.4a IQ-VA mutant channel failed to pull down CaM. The findings suggest that the observed association of WT hKv7.4 to CaM is not due to the nonspecific bindings from the abundance of CaM in the cells.
One surprising aspect of this finding is that the binding of CaM to isoform b, which contains two CaM-binding motifs separated by 127 residues, appears to be functionally inconsequential. In stark contrast to Kv7.1-3 and 5, which also harbor two CaM-binding motifs in the intracellular C terminus separated by ϳ135 residues (1), CaM interaction with these channels confers functional down-regulation of the current (5,14,49).
Physiological Relevance of the Findings-This study demonstrates that point mutations upstream of the CaMBDs, such as G321S, may destabilize CaM binding, potentially weakening or abolishing the predicted self-sustaining action potential firings, promoted by CaM modulation of hKv7.4 currents in SGNs. Reduction and/or removal of the capacity to generate excitability will not only alter information coding but also reduce growth and survival of SGNs as well as survival of neurons in the cochlear nucleus (50,51). This, conceivably, may result in progressive degeneration of upstream neurons.
Future Studies-Future studies are required to determine the effects of a local and global increase of intracellular Ca 2ϩ directly in SGNs. Furthermore, it is not yet known whether the G321S mutation affects both the a and b isoforms of Kv7.4. Additionally, CaM binding to Kv7 channels regulates channel assembly (25), trafficking (52,53), and function (5,49). The expressions of multiple splice variants that are resistant to functional modulation by CaM may further our understanding of the regulation of Kv7 channels in native cells.
Author Contributions-C. R. S., H. J. K., R. L. W., V. Y. Y., P. C. Y., and J. X. performed the experiments. C. R. S., H. J. K., R. L. W., V. Y. Y., C. E. C., X. D. Z., N. C., and E. N. Y. designed the experiments and prepared the manuscript. All authors analyzed the results and approved the final version of the manuscript.