The Kv7.2/Kv7.3 Heterotetramer Assembles with a Random Subunit Arrangement*

Background: Kv7.2 and Kv7.3 coassemble to form a heteromeric K+ channel that generates the M-current. Results: Atomic force microscopy (AFM) imaging of Kv7.2/Kv7.3 heteromers in complex with subunit-specific antibodies reveals the channel architecture. Conclusion: The Kv7.2/Kv7.3 heteromer assembles with a random subunit arrangement. Significance: Unusually, heteromeric Kv7 channels have a variable architecture. Voltage-gated K+ channels composed of Kv7.2 and Kv7.3 are the predominant contributors to the M-current, which plays a key role in controlling neuronal activity. Various lines of evidence have indicated that Kv7.2 and Kv7.3 form a heteromeric channel. However, the subunit stoichiometry and arrangement within this putative heteromer are so far unknown. Here, we have addressed this question using atomic force microscopy imaging of complexes between isolated Kv7.2/Kv7.3 channels and antibodies to epitope tags on the two subunits, Myc on Kv7.2 and HA on Kv7.3. Initially, tsA 201 cells were transiently transfected with equal amounts of cDNA for the two subunits. The heteromer was isolated through binding of either tag to immunoaffinity beads and then decorated with antibodies to the other tag. In both cases, the distribution of angles between pairs of bound antibodies had two peaks, at around 90° and around 180°, and in both cases the 90° peak was about double the size of the 180° peak. These results indicate that the Kv7.2/Kv7.3 heteromer generated by cells expressing approximately equal amounts of the two subunits assembles as a tetramer with a predominantly 2:2 subunit stoichiometry and with a random subunit arrangement. When the DNA ratio for the two subunits was varied, copurification experiments indicated that the subunit stoichiometry was variable and not fixed at 2:2. Hence, there are no constraints on either the subunit stoichiometry or the subunit arrangement.

bled multimers can be distinguished from unassembled subunits. The proteins are incubated with antibodies to the tags, and the resulting multimer-antibody complexes are imaged by AFM. Multimers with two bound antibodies are identified, and the angles between the antibodies are measured. A frequency distribution of these angles then reveals the structure of the multimer. In this study, we have used this method to determine the architecture of the Kv7.2/Kv7.3 heteromer. We show that Kv7.2 and Kv7.3 form a heterotetramer with a random subunit arrangement.

EXPERIMENTAL PROCEDURES
Cell Culture-tsA 201 cells (a subclone of human embryonic kidney 293 cells stably expressing the SV40 large T-antigen) were grown in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin in an atmosphere of 5% CO 2 /air.
Channel Constructs-DNA encoding human Kv7.2 bearing a Myc epitope tag at its N terminus was subcloned into the pSRC5 vector (20). Human Kv7.3 DNA, bearing either a double hemagglutinin (2xHA) or a FLAG/HA epitope tag at its N terminus, was subcloned into the same vector. Initial experiments demonstrated that the 2xHA tag could be decorated simultaneously by two anti-HA antibodies (data not shown). To circumvent this complication, we generated a new Kv7.3 construct that had a single HA tag for experiments involving decoration with anti-HA antibodies. (The FLAG tag was not used in these experiments.) It has been shown previously that addition of N-terminal tags does not affect the functional properties of either Kv7.2 or Kv7.3 (12). DNA encoding the rat P2X2 receptor subunit bearing a His 6 epitope tag at its N terminus was subcloned into the pcDNA3.1 vector.
Transient Transfection of tsA 201 Cells-Transient transfections of tsA 201 cells with DNA were carried out using the calcium phosphate precipitation method. A total of 250 g of DNA (usually 125 g for each Kv7 construct) was used to transfect cells in 5 ϫ 162 cm 2 culture flasks. After transfection, cells were incubated for 48 h at 37°C to allow protein expression. Protein expression and intracellular localization were checked using immunofluorescence analysis of small-scale cultures. Cells were fixed, permeabilized, and incubated with appropriate primary antibodies (rabbit polyclonal anti-Myc (Abcam), mouse monoclonal anti-HA (Covance), and mouse monoclonal anti-V5 (Invitrogen) as a negative control), followed by either Cy3-or fluorescein isothiocyanate-conjugated goat secondary antibodies (Sigma). Cells were imaged by confocal laser scanning microscopy.
In Situ Proximity Ligation Assay-Cells growing on lysineand collagen-coated glass coverslips in 3.5-cm diameter culture wells were cotransfected with 1 g each of DNA encoding Myc-Kv7.2 and 2xHA-Kv7.3. All transfections also included pEGFP (0.5 g of DNA) to identify transfected cells. In a control experiment, cells were cotransfected with DNA encoding Myc-Kv7.2 and His 6 -P2X2. Cells were incubated for 24 h at 37°C to allow protein expression. Cells were fixed, permeabilized, and incubated with primary antibodies (rabbit polyclonal anti-Myc plus either mouse monoclonal anti-HA or mouse monoclonal anti-His 6 (Invitrogen), all diluted 1:50) for a further 1 h. Cells were washed with phosphate-buffered saline and then incubated for 1 h with anti-mouse (ϩ) and anti-rabbit (-) proximity ligation secondary antibodies, diluted 1:5, at 37°C. These antibodies were obtained as part of a kit from Olink Bioscience that also included ligation and amplification buffers. Cells were washed with phosphate-buffered saline and incubated in T4 DNA ligase diluted 1:40 in ligation buffer for 30 min at 37°C. Cells were washed with TBS-T (150 mM NaCl, 0.05% Tween 20, 10 mM Tris (pH 7.4)), and DNA amplification was performed by incubation with 1:80 DNA polymerase in amplification buffer for 100 min at 37°C. This buffer also contained a fluorescent detection probe (excitation wavelength 554 nm, emission wavelength 576 nm). Cells were washed with 1ϫ TBS (100 mM NaCl, 200 mM Tris (pH 7.5)) and then 0.01ϫ TBS. Coverslips were mounted on slides and imaged by confocal laser scanning microscopy.
Solubilization and Purification of Epitope-tagged Proteins-Transfected cells were solubilized in 1% Triton X-100 for 1 h before centrifugation at 78,000 ϫ g to remove insoluble material. To isolate Kv7, the solubilized extract was incubated with either anti-Myc-or anti-HA-agarose beads (Sigma) for 3 h. The beads were washed extensively, and bound proteins were eluted with Myc or HA peptide (100 g/ml). Samples were analyzed by SDS-polyacrylamide gel electrophoresis, and proteins were detected both by silver staining and by immunoblotting using mouse monoclonal antibodies against Myc (Invitrogen) or HA.
AFM Imaging of Kv7 or Kv7-Antibody Complexes-Isolated channels were imaged either alone or following overnight incubation at 4°C with a 1:2 molar ratio (ϳ0.2 nM channel protein concentration) of anti-epitope tag monoclonal antibody (Invitrogen). Proteins were diluted to a final concentration of 0.04 nM, and 45 l of the sample was allowed to adsorb to freshly cleaved, poly-L-lysine-coated mica disks. After a 10-min incubation, the sample was washed with Biotechnology Performance Certified-grade water (Sigma) and dried under nitrogen. Imaging was performed with a Veeco Digital Instruments multimode atomic force microscope controlled by a Nanoscope IIIa controller. Samples were imaged in air using tapping mode. The silicon cantilevers used had a drive frequency of ϳ300 kHz and a specified spring constant of 40 N/m (Olympus). The applied imaging force was kept as low as possible (A s /A 0 ϳ0.85).
The molecular volumes of the protein particles were determined from particle dimensions on the basis of AFM images. After adsorption of the channels onto the mica support, the particles adopt the shape of a spherical cap. As described previously (21), the heights and radii were measured from multiple cross-sections of the same particle, and the molecular volume was calculated using the following equation: where h is the particle height and r is the radius. Molecular volume on the basis of molecular mass was calculated using the equation where M 0 is the molecular mass, N 0 is Avogadro's number, V 1 and V 2 are the partial specific volumes of particle (0.74 cm 3 /g) and water (1 cm 3 /g), respectively, and d is the extent of protein hydration (taken as 0.4 g water/g protein).
Selection of Binding Events-Several criteria were used to identify Kv7-antibody complexes. Heights and radii were measured for all particles, and the particle volumes were calculated. Bound particles needed to have between half and twice the predicted volume of 285 nm 3 to be accepted as antibodies. A crosssection was drawn through the junction between the antibody and the adjacent Kv7 channel, and the height of the lowest point between antibody and channel was measured. This height needed to be greater than 0.3 nm for the antibody to be considered bound. Any particle was rejected if its length was greater than twice its width. To be considered a double binding event, all particles and both binding events needed to meet all of the above criteria.
Statistical Analysis-Histograms were drawn with bin widths chosen according to Scott's equation: where is an estimate of the standard deviation, and n is the sample size (22). Where Gaussian curves were fitted to the data, the number of curves was chosen so as to maximize the r 2 value while giving significantly different means using Welch's t test for unequal sample sizes and unequal variances (23).

RESULTS
Initially, tsA 201 cells were transiently cotransfected with DNA encoding Myc-tagged Kv7.2, and 2xHA-tagged Kv7.3. Protein expression and localization was confirmed by immunofluorescence using appropriate anti-tag antibodies. As shown in Fig. 1A, cells gave positive immunofluorescence signals with anti-Myc and anti-HA antibodies, indicating the presence of the two Kv7 subunits. In contrast, an anti-V5 antibody gave only a background signal. The anti-Myc and anti-HA signals in doubly labeled cell populations overlapped extensively, indicating that the majority of transfected cells expressed both subunits. The reticular staining patterns suggest that proteins were localized predominantly in the endoplasmic reticulum.
To establish whether Kv7.2 and Kv7.3 coexpressed in the cells interact with each other, in situ proximity ligation assays were carried out (24). The assay uses two secondary antibodies, each bearing a short DNA strand. When the secondary antibodies are brought into close proximity (Ͻ 40 nm) by binding to their relevant primary antibodies, the DNA strands hybridize with an additional circle-forming oligodeoxynucleotide. Ligation then creates a complete circularized oligodeoxynucleotide, and rolling circle amplification increases the amount of circular DNA several hundred-fold. The DNA is then visualized using a fluorescent probe. Cells were cotransfected with Myc-Kv7.2 and 2xHA-Kv7.3, and pEGFP (enhanced green fluorescent protein) was used to reveal the transfected cells. Primary antibodies used to tag the subunits were rabbit anti-Myc (Kv7.2) and mouse anti-HA (Kv7.3). As shown in Fig. 1B, strong proximity signals were given by Myc-Kv7.2 ϩ 2xHA-Kv7. 3. In contrast, when cells expressed Myc-Kv7.2 ϩ His 6 -P2X2 (plus pEGFP), only a very weak signal was generated using anti-Myc and anti-His 6 antibodies (Fig. 1C). This result indicates that Kv7.2 and Kv7.3 come into close proximity within the cotransfected cells.
Cotransfected cells were solubilized in Triton X-100 detergent (1% w/v), and proteins were isolated through the binding of the 2xHA tag on Kv7.3 to anti-HA-agarose beads. A silverstained gel of the isolated protein ( Fig. 2A) shows a doublet at about 100 kDa, the expected size of the Kv7 subunit (12). The gel also shows an additional band at about 70 kDa that is likely a degradation product. Both the crude solubilized fraction and the isolated protein were subjected to immunoblotting using either anti-Myc (Fig. 2B, left panel) or anti-HA (Fig. 2B, right Cotransfected cells were fixed, permeabilized, and incubated with appropriate anti-tag antibodies, as indicated, followed by fluorophore-conjugated secondary antibodies. Cells were imaged by confocal laser scanning microscopy. B, in situ proximity ligation assay for Kv7.2/Kv7.3 interaction. Cells were cotransfected with DNA encoding Myc-Kv7.2 and 2xHA-Kv7.3. Transfections also included pEGFP to identify transfected cells. Cells were fixed, permeabilized, and incubated with primary antibodies (rabbit polyclonal anti-Myc and mouse monoclonal anti-HA) followed by anti-mouse (ϩ) and anti-rabbit (-) proximity ligation secondary antibodies. The proximity ligation assay was then carried out, and treated cells were imaged by confocal laser scanning microscopy. The left panel shows the proximity signal, and the right panel shows the EGFP signal. C, control experiment in which cells were transfected with Myc-Kv7.2 plus His 6 -P2X2 receptor. In this case, primary antibodies were rabbit polyclonal anti-Myc and mouse monoclonal anti-His 6 . Transfected cells are indicated by the EGFP signal (right panel). There is a considerable reduction in the proximity ligation signal (left panel). Scale bars ϭ 20 m. panel) antibodies. Both antibodies detected bands at about 100 kDa in both fractions, indicating that both Myc-Kv7.2 and 2xHA-Kv7.3 were successfully captured via the 2xHA tag on Kv7.3. This coisolation provides further evidence for an intimate interaction between the two subunits.
Proteins isolated from cells expressing Myc-Kv7.2 and 2xHA-Kv7.3 were imaged by AFM. A representative low-magnification AFM image is shown in Fig. 2C. Several particles of approximately equal sizes can be seen. The isolated protein was then incubated with anti-Myc antibody and again imaged by AFM. In the representative images shown in Fig. 2D, large particles can be seen, some of which have been decorated by either one (arrowheads), two (arrow) or three (asterisk) smaller particles. The small particles had a molecular volume of around 240 nm 3 , close to the expected volume of 285 nm 3 for an immunoglobulin G molecule, with a molecular mass of 150 kDa. Hence, Multiple binding events were identified, and the volumes of the central decorated particles were determined using Equation 1. A frequency distribution of these volumes had two peaks: a small peak at 489 Ϯ 12 (n ϭ 13) nm 3 and a larger peak at 848 Ϯ 36 (n ϭ 42) nm 3 (Fig. 2E). According to Equation 2, a Kv7 subunit with a molecular mass of 100 kDa should have a molecular volume of 190 nm 3 , so a Kv7 tetramer should have a volume of 760 nm 3 , close to the higher-volume peak. The two volume peaks, therefore, likely correspond to tetramers and dimers that may represent precursors or may have been produced by dissociation of the tetramers.
We identified Kv7 channels that had been decorated by two antibodies and measured the angles between the bound antibodies. This was done in each case by joining the highest point on the central particle (the Kv7 channel) to the highest points on the peripheral particles (the antibodies) by lines and then determining the angle between the two lines. Representative doubly decorated large particles are shown in Fig. 2F, and a frequency distribution of the angles obtained is shown in G. The angle distribution has two peaks: a large peak at 96 Ϯ 4°( n ϭ 30) and a smaller peak at 166 Ϯ 16°(n ϭ 14). The ratio of the numbers of particles in the two peaks is 2.1:1. This angle profile, with two peaks at around 90°and 180°, indicates that the Kv7 heteromer is indeed a tetramer. In contrast, a trimer would give a single angle peak at 120°(17), whereas a pentamer would give equally sized peaks at 72°and 144° (18).
The fact that we are isolating the channel via the 2xHA tag on Kv7.3 and decorating the Myc tag on Kv7.2 ensures that we are only decorating heteromers. However, the results shown in Fig.  2G do not allow us to distinguish between two possible subunit arrangements: the one shown in supplemental Fig. S1A, where the subunit stoichiometry is 2:2 but the subunit arrangement is random, and the one in supplemental Fig. S1B, where the subunit stoichiometry is 3:1, Kv7.2:Kv7.3. In both cases, the predicted ratio of 90°/180°angles is 2:1. Hence, the 2.1:1 ratio observed is consistent with either scenario.
To distinguish between the two possible assembly states, we carried out the reverse experiment, isolating the channel via the Myc tag and decorating the HA tag. Cells were cotransfected with DNA encoding HA-Kv7.3 along with Myc-Kv7.2. Proteins were isolated through the binding of the Myc tag on Kv7.2 to anti-Myc-agarose beads. A silver stain of the isolated protein shows a doublet of bands at around 100 kDa (Fig. 3A), along with two contaminating bands. (Note that the asterisk indicates a band that appeared even in blank lanes that is likely a contam-inant in the gel and not in the protein sample.) Immunoblot analyses using anti-Myc and anti-HA antibodies show decoration of bands at around 100 kDa (Fig. 3B). Low-magnification AFM images of isolated proteins showed a relatively homogenous distribution of particles (Fig. 3C), and when proteins were incubated with anti-HA antibodies, some particles became decorated by two antibodies (arrows in D).
A frequency distribution of the molecular volumes of doubly decorated particles (Fig. 3E) shows a single peak at 925 Ϯ 23 (n ϭ 58) nm 3 , close to the expected volume of a tetramer (760 nm 3 ). Zoomed images (Fig. 3F) show angles between pairs of bound antibodies of around 90°(left panels) and around 180°( right panels), and the angle distribution (G) had two peaks at 94 Ϯ 6°(n ϭ 44) and 166 Ϯ 21°(n ϭ 18). The ratio of the numbers of particles within the two peaks is 2.4:1. The fact that a very similar angle distribution is observed irrespective of which antibody is used to isolate or decorate the heteromer strongly argues for the situation illustrated in supplemental Fig.  S1A, where the channel generated by cotransfection with equal amounts of DNA for the two subunits has a predominant subunit stoichiometry of 2:2 and a random subunit arrangement.
Two additional explanations of the apparent variability in subunit arrangement indicated by the results should be considered. One is the possibility that Kv7 tetramers might be selfassociating during purification so that a Myc-Kv7.2 homotetramer, for instance, might be copurified on anti-HA-agarose beads with a Myc-Kv7.2/2xHA-Kv7.3 heterotetramer or, indeed, a 2xHA-Kv7.3 homotetramer. This possibility seems unlikely because there was no sign of dimers of Kv7 tetramers in the AFM images (unlike the situation for P2X receptors in a previous study, for example, where receptor dimers were relatively common (25)). Another possible scenario involves dissociation of Kv7 tetramers during the isolation procedure, followed by reassembly into a novel tetramer that does not exist in vivo. We carried out a further experiment to check these possibilities. Two sets of cells were singly transfected with DNA encoding either Myc-Kv7.2 or HA-Kv7.3. The two populations of transfected cells were mixed at the start of the purification procedure, and the combined detergent extract of the cells was later split into two batches for purification using either anti-Myc or anti-HA immunobeads. The two eluates were then immunoblotted using anti-Myc and anti-HA antibodies. If either of the above scenarios was occurring, then anti-Myc beads should have captured HA-Kv7.3 along with Myc-Kv7.2, and anti-HA beads should have captured Myc-Kv7.2 along with HA-Kv7. 3. In fact, anti-Myc beads captured only Myc-Kv7.2, and anti-HA beads captured only HA-Kv7.3 (Fig. 4). We there-fore conclude that the Kv7.2/Kv7.3 tetramer does indeed assemble with a variable subunit arrangement.
The fact that we observed triply decorated channels (as shown in Fig. 2D) must mean that some channels had a 3:1, Kv7.2:Kv7.3 stoichiometry. However, when equal amounts of DNA were used for the two subunits, these triple decoration events were very rare, indicating that the predominant stoichi-ometry is 2:2. The possibility does exist, however, that a greater proportion of heteromers of 3:1 stoichiometry can be generated when the expression levels of the two proteins are different. To test this possibility, we transfected cells with DNA encoding Kv7.2 and Kv7.3 in ratios of 4:1 and 1:4, which resulted in the expected differences in expression of Myc-Kv7.2 and HA-Kv7.3 in the transfected cells (Fig. 5, left panels). Protein was isolated from the cells by sequential capture on anti-HA and anti-Myc immunobeads. This procedure will necessarily result in the isolation of heteromers. The protein concentrations in the eluted samples from the two cotransfections were adjusted to produce bands of approximately equal intensity on an anti-HA (Kv7. 3) immunoblot. If the Kv7.2:Kv7.3 stoichiometry was fixed at 2:2, one would expect the two anti-Myc (Kv7.2) bands also to be of equal intensity. In fact, the band in the 4:1 Kv7.2:Kv7.3 sample is significantly more intense than the band from the 1:4 sample (Fig. 5, right panels). This result indicates that the Kv7.2:Kv7.3 subunit stoichiometry is not fixed at 2:2 but rather varies in response to changes in relative expression levels.

DISCUSSION
Our results indicate that the Kv7.2/Kv7.3 heteromer assembles as a tetramer with a random subunit arrangement. A variable subunit stoichiometry and arrangement was suggested several years ago on the basis of an electrophysiological analysis of channels expressed in tsA 201 cells (26). In a more recent study, Kv7.3-Kv7.2 concatemers were found to produce channel characteristics almost identical with those formed after coexpression of individual subunits (16). In this case, presumably, the channel would have had an alternating subunit arrangement. However, the properties of channels with a constrained adjacent subunit arrangement were not tested.
The variability in subunit arrangement seen with the Kv7.2/ Kv7.3 channel is unusual. More often we have seen an invariant subunit arrangement in both ionotropic receptors, such as the 5-HT 3 receptor (18) and the GABA A receptor (19), and ion channels such as the transient receptor potential (TRP) channels. Specifically, TRPP2 forms heteromers with both TRPC1 (27) and TRPV4 (28), which have an alternating subunit arrangement.
As mentioned above, the majority of the Kv7 proteins were localized in the endoplasmic reticulum, likely because of the overexpression occurring after transient transfection. We must, therefore, bear in mind the possibility that a quality control mechanism operates within the cells to restrict access to the cell surface to heteromers of a particular subunit arrangement. It has recently been shown that Kv7.3 homomers are retained in the endoplasmic reticulum (12). Whether such a mechanism could discriminate between different subunit arrangements within a heteromer remains to be seen.
Functional analysis of the behavior of point mutants and chimeras of Kv7.2 and Kv7.3 has identified the cytoplasmic C terminus as a major region involved in oligomerization of Kv7 subunits (11, 29 -31). All Kv7 subunits contain two coiled coil domains, known as the A-domain head and tail, and the tail in particular is believed to be important for intersubunit recognition, channel assembly, and functional expression. Recently, the A-domain tail has been shown to crystallize as a left-handed coiled coil (32). This structure suggests a possible explanation for the inability of Kv7.3 to efficiently form homotetramers. Specifically, it would be problematic to pack four phenylalanine residues (position 622) within the core of the tetramer, and in addition, interhelical hydrogen bond and salt bridge interactions required to stabilize the tetramer would also be disrupted in Kv7.3. Replacement of the phenylalanine by leucine, as in Kv7.2, and improvement of the interhelical interactions would help to stabilize the tetramer.
Intriguingly, evidence has recently been presented that the pore region in Kv7.3, rather than the A domain, plays a key role in controlling channel assembly and surface expression (12). The Kv7.3 subunit is unusual in having an alanine residue in place of threonine in the pore vestibule (position 315). Although wild-type Kv7.3 is retained in the endoplasmic reticulum and requires coexpression of Kv7.2 to exit this compartment, an A315T Kv7.3 mutant is efficiently trafficked to the plasma membrane by itself and is functionally active and, therefore, fully assembled. Similar to Kv7, the C-terminal domain of TRPP2 again contains two coiled coil domains, but here mutational analysis has shown that second of these domains functions as a heterodimerization domain essential for polycystin 1 binding but not for self-oligomerization of TRPP2 (33). In addition, there is evidence for the involvement of the cytoplasmic N terminus in TRPP2 channel assembly (34) and an intramembrane disulfide bond (35). Whether the distinct functions of the cytoplasmic domains in Kv7 and TRPP2 helps to explain the different levels of stringency in channel assembly in the two cases remains to be seen.
In conclusion, the variable subunit stoichiometry and arrangement demonstrated here raises the possibility that the overall behavior of the Kv7 channel in neurons can be manipulated through endogenous changes in subunit expression lev-  3. The two populations of transfected cells were mixed at the start of the purification procedure, and the combined detergent extract of the cells was later split into two batches for purification using either anti-Myc or anti-HA immunobeads. The two eluates were then immunoblotted using anti-Myc and anti-HA antibodies. Immunoreactive bands were visualized using a horseradish peroxidase-conjugated goat anti-mouse secondary antibody followed by enhanced chemiluminescence. The arrowheads indicate molecular mass markers (kDa). immunoblot. Immunoreactive bands were visualized using a horseradish peroxidase-conjugated goat anti-mouse secondary antibody followed by enhanced chemiluminescence. els, potentially adding an extra layer of subtlety to the manner in which neurons control their excitability.