The formation of KV2.1 macro-clusters is required for sex-specific differences in L-type CaV1.2 clustering and function in arterial myocytes

In arterial myocytes, the canonical function of voltage-gated CaV1.2 and KV2.1 channels is to induce myocyte contraction and relaxation through their responses to membrane depolarization, respectively. Paradoxically, KV2.1 also plays a sex-specific role by promoting the clustering and activity of CaV1.2 channels. However, the impact of KV2.1 protein organization on CaV1.2 function remains poorly understood. We discovered that KV2.1 forms micro-clusters, which can transform into large macro-clusters when a critical clustering site (S590) in the channel is phosphorylated in arterial myocytes. Notably, female myocytes exhibit greater phosphorylation of S590, and macro-cluster formation compared to males. Contrary to current models, the activity of KV2.1 channels seems unrelated to density or macro-clustering in arterial myocytes. Disrupting the KV2.1 clustering site (KV2.1S590A) eliminated KV2.1 macro-clustering and sex-specific differences in CaV1.2 cluster size and activity. We propose that the degree of KV2.1 clustering tunes CaV1.2 channel function in a sex-specific manner in arterial myocytes.


Introduction
Activation of dihydropyridine-sensitive, voltage-gated L-type CaV1.2 channels plays a crucial role in the development of myogenic tone 1 , a process of autoregulation that enables arteries to regulate their diameter in response to changes in intravascular pressure 2 .This mechanism, independent of neural or hormonal influences, is critical to maintain constant blood flow despite changes in blood pressure.
The current model of the myogenic response proposes that mechanical stretch of the membrane leads to the activation of TRPC6 3 , TRPM4 4 , and TRPP1 (PKD2) 5 channels, which depolarizes arterial myocytes, activating CaV1.2 channels 1 .Activation of a single or a small cluster of CaV1.2 channels results in a local increase in intracellular Ca 2+ concentration ([Ca 2+ ]i) termed a "CaV1.2sparklet" [6][7][8] .Summation of multiple CaV1.2 sparklets leads to a global increase in [Ca 2+ ]i that triggers muscle contraction.
CaV1.2 channels form clusters in the plasma membrane via a stochastic self-assembly mechanism 9 .CaV1.2 channels within these clusters can gate cooperatively in response to Ca 2+ -driven physical interactions of adjacent channels 10,11 .CaV1.2 channels in this configuration allow for larger Ca 2+ influx compared to random, independent openings of individual channels.In vascular smooth muscle, cooperative gating of CaV1.2 channels has been estimated to contribute up to ~50% of Ca 2+ influx during the development of myogenic tone 12 .
One route of negative feedback regulation of membrane depolarization and Ca 2+ influx via CaV1.2 channels occurs through the depolarization-mediated activation of delayed rectifier voltage-gated KV2.1 channels 13,14 .
In its canonical role, KV2.1 proteins in arterial smooth muscle cells form ion conducting voltage-gated K + channels whose activation results in membrane potential hyperpolarization, thereby affecting myocyte [Ca 2+ ]i and myogenic tone 13 .Until recently, the accepted role of KV2.1 protein in arterial myocytes was to form K + conducting channels.However, our recent work suggests that only about 1% of the KV2.1 channels in the plasma membrane of arterial smooth muscle are conductive 15 .Indeed, a growing body of evidence suggests that KV2.1 proteins have dual conductive and structural roles in the surface membrane of smooth muscle cells and neurons [15][16][17] .
Little to no channel activity was detectable within KV2.1 clusters, whereas currents could be recorded in areas with diffuse KV2.1 expression 19 .One of the structural roles of KV2.1 is to promote clustering of CaV1.2 channels, thus increasing the probability of CaV1.2-to-CaV1.2 interactions within these clusters 16,17 .
Both, the conductive and structural roles of KV2.1, depend on the level of expression of this protein, which in arterial myocytes varies with sex 15 .In female myocytes, where expression of KV2.1 protein is higher than in male myocytes, KV2.1 has both conductive and structural roles.Female myocytes have larger CaV1.2 clusters, [Ca 2+ ]i, and myogenic tone than male myocytes.In contrast, in male myocytes, KV2.1 channels regulate membrane potential, but not CaV1.2channel clustering.
Based on these data, a model was proposed in which KV2.1 function varies with sex 15 .In males, KV2.1 channels primarily control membrane potential, but in female myocytes KV2.1 plays dual electrical and CaV1.2 clustering roles.Currently, it is unclear whether the conductive and structural functions of KV2.1 in native arterial myocytes rely on its clustering ability, and if this relationship is sex-dependent.
In this study, we tested the hypothesis that conductive and structural roles of KV2.1 channels in male and female arterial myocytes depend on their capacity to form clusters in studies of wild-type (WT) and S586A mutant rat KV2.1 channels expressed in heterologous cells, and in arterial myocytes from a novel gene edited knock-in mouse expressing the S590A mutation.We focused on serine 586 within the PRC domain (amino acids 573-598) of rat KV2.1 because a point mutation changing this amino acid to a non-phosphorylatable alanine decreases the KV2.1 clustering phenotype 24 .This corresponds to a S590A point mutation in the mouse KV2.1 channel.Our data show that KV2.1 is expressed into large macro-clusters composed of micro-clusters that can only be resolved with super-resolution microscopy.The KV2.1S586A point mutation nearly eliminated KV2.1 macro-clusters but had a minimal impact on micro-clusters.Notably, we find that KV2.1 channel function is not dependent on its ability to form macro-clusters in arterial myocytes of either sex.Rather, KV2.1 macroclustering enhances CaV1.2 channel clusters and activity.Based on these results, we propose a new model in which macro-clustering of KV2.1 in arterial myocytes alters CaV1.2 channel organization and function in a sexspecific manner but has no impact on its conductive function.
Our finding that the size distributions of KV2.1WT and KV2.1S586A clusters could be described by exponential functions, the hallmark of a Poisson process, suggests that these clusters are formed stochastically 29 .To test this hypothesis, we implemented the stochastic modeling approach employed by Sato et al. 9 to determine whether we could reproduce our cluster distributions and make testable predictions regarding plasma membrane protein organization.As shown in Supplemental Figure 1, our stochastic self-assembly model effectively reproduced the steady-state size distributions that we measured for KV2.1WT and KV2.1S586A proteins embedded in the surface membrane of HEK293T cells.The parameters used in the model are summarized in Supplemental 1C.These in silico data suggest that in HEK293T cells, KV2.1WT has a higher probability of nucleation (i.e., Pn) and cluster growth (i.e., Pg) than KV2.1S586A channels.The probability of channel removal (PR) was similar.
Analysis of our confocal microscopy data showed that in HEK293T cells, KV2.1WT is expressed in clusters of different sizes but do in fact form large macro-clusters.Using this confocal microscopy analysis, we sought to define the size of a KV2.1 macro-cluster.We began with the mean cluster volume of KV2.1WT generated in

Figure 1E
. Our analysis provided a mean cluster volume of 0.09 μm 3 .Assuming the volume of a cluster is spherical, we extrapolated the diameter of the macro-cluster to be 560 nm.The standard deviation of these measurements was 0.32.Thus, the mean minus 2 standard deviations provides a lower end limit with a 95% confidence and aligned with the lateral resolution of our confocal microscopy.Accordingly, we define the lower limit of a macro-cluster as a cluster that is larger than 0.03 µm 3 or 326 nm in diameter, with clusters smaller than this classified as micro-clusters.

KV2.1S590A mutation does not affect overall KV2.1 channel expression but declusters smooth muscle KV2.1 channels in a sex-specific manner
Next, we investigated whether, as in heterologous expression system (i.e., Figure 1), KV2.1 channels cluster in arterial myocytes and whether this clustering could be disrupted via mutation of critical amino acids in the PRC domain.To do this, we used Crispr/CAS gene editing to generate a knock-in mouse line expressing the S590A point mutation, corresponding to the S586A mutation in rat KV2.1, at the KCNB1 locus of a C57/BL6J mouse (see Methods section for a full description of how these mice were generated).
Using the threshold set from our confocal imaging (i.e., macro-clusters are >0.025μm 3 ), we quantified the number of macro-clusters expressed in myocytes from KV2.1WT and KV2.1S590A mice.Around 62% of clusters in KV2.1WT male myocytes were identified as macro-clusters, with a similar percentage of 58% observed in samples from males with the KV2.1S590A mutation.Approximately 70% of KV2.1 clusters in KV2.1WT female myocytes were classified as macro-clusters.Remarkably, in KV2.1S590A female myocytes, macro-clusters accounted for approximately 49% of the total KV2.1 clusters.
We also quantified KV2.1 micro-clusters.Although the proportion of micro-clusters was similar in male KV2.1WT and KV2.1S590A myocytes (38% and 42%, respectively), female KV2.1S590A myocytes exhibited a much larger proportion of micro-clusters (51%) compared to myocytes from KV2.1WT females (30%).Hence, it can be reasoned that the S590A mutation has a sex-specific effect of reducing the extent of KV2.1 macro-clustering in female but not male arterial myocytes without impacting channel expression.

KV2.1 S590 phosphorylation is higher in myocytes from female versus male KV2.1WT mice
To investigate the potential role of the S590 phosphorylation site in the sex-specific differences in  To ta l K

Expression of clustering impaired KV2.1S590A does not affect channel activity in arterial myocytes
Three studies, one using Xenopus oocytes 28 , one using HEK293T cells 19 and another from our group using arterial myocytes 15 4B).We also simulated the current-voltage relationships in male (Figure 4C) and female (Figure 4D) myocytes assuming 1% and 0.1% of KV2.1 channels are conductive, which are more within the range with previous experimental results in heterologous systems 19,28 and native cells 15 .The magnitude of in silico KV2.1 current densities with 1% or 0.1% functional channels was 70.1 and 5.57 pA/pF in male myocytes and 173 and 16.7 pA/pF in female myocytes.
Next, we recorded voltage-gated K + (KV) currents in male and female KV2.1WT and KV2.1S590A arterial myocytes in response to 500 ms depolarizations to voltages between -50 and +50 mV before and after applying the KV2.1 blocker RY785 (1 μM) 32,33 .This compound decreases KV2.1 currents by blocking the pore of these channels 32 rather than by immobilizing their voltage sensor, as stromatoxin does 34 .As a first step in these experiments, we tested the specificity of the RY785 by recording Kv currents before and after the application of this molecule in male and female KV2.1WT and KV2.1 null (KV2.1 -/-) myocytes (Supplemental Figure 2A-D).Notably, application of 1 μM RY785 decreased the amplitude of K + currents in KV2.1WT but not in KV2.1 -/-myocytes of either sex.This indicates that RY785 is a specific blocker of KV2.1 channels in arterial myocytes.
Having completed these critical control experiments, we recorded KV currents from KV2.1WT and KV2.1S590A myocytes.We noted that the amplitude of the composite K currents were similar in myocytes from KV2.1S590A mice compared to myocytes from sex-matched KV2.1WT littermates (Supplemental Figure 2E.Importantly, for both sexes, RY785-sensitive KV2.1 currents were also similar in male (Figure 4E, G) and female (Figure 4F, H) KV2.1WT and KV2.1S590A myocytes.Indeed, a comparison of the experimental and in silico amplitudes of the macroscopic KV2.1 currents suggests that less than 1% of the channels are functional in myocytes from both male and female KV2.1WT and KV2.1S590A mice.When taken together with our analyses of Kv2.1 clustering detailed above, these findings suggest that in arterial myocytes KV2.1 channel activity is not determined by the extent and nature of its clustering.
Previous work from our group has shown that KV2.1 expression promotes CaV1.2 clustering and activity in neurons 16 and arterial myocytes 15 .Following from this and the data above, we hypothesize that in arterial myocytes KV2.1 plays a sex-specific structural role as an organizer to bring CaV1.2 channels together in female but not male myocytes.We again used PLA to test the hypothesis that the declustering of KV2.1 channels in female but not male myocytes from KV2.1S590A mice would decrease KV2.1-CaV1.2channel interactions in a sex-specific manner.Representative images of KV2.1-CaV1.2PLA puncta show randomly distributed interactions across the cell (Supplemental Figure 3D).
ICa was activated by applying 300 ms voltage Next, we determined the level of expression of CaV1.2 protein in male and female KV2.1S590A and KV2.1WT vessels using immunocytochemistry (Figure 5E, F, H, J) and RT-PCR approaches (Figure 5G and I).
Our analysis suggests that total CaV1.2 protein and mRNA expression is similar in male and female KV2.1S590A and KV2.1WT vessels.This suggests that the smaller ICa in female KV2.1S590A than KV2.1WT myocytes is not likely due to lower CaV1.2expression in these cells.-80 mV +50 mV a le H J

Declustering KV2.1 in myocytes with the Kv2.1S590A mutation decreases CaV1.2 cluster sizes in female but not male arterial myocytes
In a previous study 15 , we suggested a model that differences in ICa amplitude between female and male arterial myocytes were due to sex-specific differences in (median = 2219 nm 2 ) was similar to the KV2.1S590A male mean of 2345 ± 82 nm 2 (median = 2354 nm 2 ) (P = 0.173) (Figure 6C), suggesting that declustering KV2.1 in male myocytes does not affect CaV1.2 channel clustering.However, CaV1.2 cluster sizes were significantly smaller in KV2.1S590A female myocytes with a mean area of 2381 ± 91 nm 2 (median = 2251 nm 2 ) compared to KV2.1WT ) Cluster Area (nm 2 female myocytes whose mean area was 3098 ± 164 nm 2 (median = 3117 nm 2 ) (P = 0.0001).Taken together with our electrophysiological data, our findings suggest that the clustering and activity of CaV1.2 channels is modulated by the degree of KV2.1 clustering.
As shown in Supplemental Figure 4A and B masking of the Ca 2+ currents by K + .We found that ICa was larger at most membrane potentials in our postconditioning IV protocol with a peak ICa at 0 mV showing an increase by about 51% (Figure 7A).
We propose this small increase is due to the intrinsic ability of CaV1.2 channels to interact with one another.
However, this increase in Venus fluorescence was significantly lower than that seen in KV2.1P404W transfected cells (Figure 7D).Together these data further support the structural role KV2.1 channels play in modulating CaV1.2-CaV1.2interactions and activity.

The activity of CaV1.2 channels is reduced in KV2.1S590A female but not male arterial myocytes
We next examined whether variations in the activity of CaV1.2 channels could explain the differences in ICa observed in myocytes from KV2.1WT and KV2.1S590A male and female mice.CaV1.2 channel activity was determined by recording CaV1.2 sparklets using TIRF microscopy as previously described 6,7,12,17,[38][39][40] (Figure 8).TIRF microscopy of near-plasma membrane intracellular Ca 2+ levels provides a powerful tool for recording Ca 2+ entry via individual or small clusters of CaV1.2 channels, as it enables the activity of individual channels to be recorded from a relatively large membrane area allowing for the identification of discrete sarcolemma signaling domains.In this analysis, CaV1.2 sparklet activity is expressed as nPs, where n is the number of quantal levels reached by the sparklet site and Ps is the probability of sparklet occurrence.
As previously reported 38 , detailed analysis of CaV1.2 sparklets sites revealed heterogeneity in activity at different sites.Therefore, CaV1.2 sparklets sites were separated into low and high activity sites, using an nPs cutoff of 0.2.
Representative CaV1.2 sparklet traces are provided (Figure 8 A, B, F, G) from low activity sparklet sites.Of note, the majority of the sparklet activity that occurs in male myocytes is produced by a signal that corresponds to a single channel opening (one quantal unit) (Figure 8A, B).The strength of the coupled gating is denoted by the κ value, and in these traces, the κ values are close to or equal to 0, indicating no or weak coupling between the channels.In contrast, the KV2.1WT female trace (Figure 8F) from a low activity site exhibited coordinated multi-channel openings, of up to 3 channels with a κ value of 0.466.Interestingly, the activity of sparklet sites from KV2.1S590A female myocytes (Figure 8G) were similar to those of KV2.1WT and KV2.1S590A male myocytes (Figure 7A, B), exhibiting mostly single channel openings and few coupled gating events.
We found that in low activity sparklet sites, the average nPs was not significantly different between KV2.1WT and KV2.1S590A male myocytes (Figure 8C).KV2.1WT sparklet sites had an average nPs of 0.06 ± 0.02 (median = 0.05) compared to KV2.1S590A where the nPs average was 0.07 ± 0.03 (median = 0.06) (P = 0.41).In male myocytes of either genotype, we rarely observed cells exhibiting high activity sparklet sites, except for a single site in a KV2.1WT male myocyte (Figure 8C).
Furthermore, we did not observe a difference in the number of CaV1.2 sparklet sites between KV2.1WT and KV2.1S590A male myocytes, with most cells exhibiting just one site (Figure 8D).
Similarly, when we compared nPs in low activity sites in KV2.1WT and KV2.1S590A female myocytes, we could not discern a difference in their average nPs values (Figure 7H).Previous work 40 showed that CaV1.2 sparklet sites appeared to arise from the simultaneous opening and/or closing of multiple channels suggesting that small groups of channels may be functioning cooperatively.To examine such coupling, we employed a coupled Markov chain model to determine the coupling coefficient (κ) among CaV1.2channels at Ca 2+ sparklet sites.The κ value ranges from 0 for channels that gate independently  Using this analysis, we found that the average κ value of 0.28 ± 0.09 (median = 0.30) in KV2.1WT male myocytes was not significantly different from 0.20 ± 0.09 (median = 0.22) in KV2.1S590A male myocytes (P = 0.27) (Figure 8E).However, the average κ value of 0.36 ± 0.05 (median = 0.38) in female KV2.1WT myocytes, was significantly higher than 0.14 ± 0.07 (median = 0) (P = 0.0082) in female KV2.1S590A myocytes, suggesting more coupled events (Figure 8J).Taken together, these data indicate increased CaV1.2 channel activity and coupled gating in myocytes from KV2.1WT females compared to those with the KV2.1S590A mutation, suggesting that clustering of KV2.1 modulates CaV1.2 channel activity.

Discussion
In this study, we show that arterial smooth muscle cells from mice expressing a gene-edited point mutation of the KV2.1 channel that selectively eliminates its characteristic macro-clustered localization have properties remarkably like those from KV2.1 knock-out mice.This leads us to formulate a new model in which KV2.1 expressionby itself -is not sufficient for this channel to exert its structural functions on modulating Cav1.2 clustering and activity, but rather depends on KV2.1 channel's capacity to form macro-clusters.Notably, the presence of KV2.1 macro-clusters in female, but not male myocytes underlie sex-specific differences in Ca 2+ influx via CaV1.2 channels in arterial smooth muscle.Our data suggest a new paradigm whereby the clustering of ion channels underlies their physiological functions, independent of their ability to conduct ions.
Analysis of super-resolution images indicates that clustering of KV2.1 and CaV1.2 channels is random and hence does not involve an active process.This stochastic self-assembly mechanism leads to micro-and macro-clusters of varying sizes that represent the default organization of KV2.1 and CaV1 channels expressed endogenously in neurons and smooth muscle cells or exogenously in heterologous cells 9 .Furthermore, we found that KV2.1 macro-clusters are composed of groups of micro-clusters.This is consistent with a recent study showing that in developing neurons KV2.1 macro-clusters formed from the coalescence of numerous micro-clusters 41 and suggests that the organization of KV2.1 clusters is hierarchical.
An important finding in this study is that KV2.1 clustering is more prominent in female than in male arterial myocytes, with female myocytes expressing a larger proportion of macro-clusters.In this context, the development of the KV2.1S590A mouse allowed us to investigate the separatable structural clustering and ion conducting roles of this channel.We found that expression KV2.1S590A nearly eliminated macro-clustering in female myocytes but had no impact on KV2.1 micro-clusters in cells from male or female myocytes.Because the S590A mutation eliminated a phosphorylation site in the PRC domain that causes macro-clustering, these findings suggested that the potential mechanism of these sex-specific differences in KV2.1 clustering was differential phosphorylation of this specific serine in male and female myocytes.
Indeed, KV2.1 phosphorylation and macro-clustering is regulated by a myriad of protein kinases such as CDK5 and protein phosphatases such as calcineurin 42 .These kinases and phosphatases work as a rheostatic mechanism to regulate the phosphorylation status of KV2.1 based on physiological demands 26,27,42 .Accordingly, we found that that the phosphorylation state of KV2.1 in arterial myocytes differs between the two sexes, specifically, that KV2.1 in male myocytes is phosphorylated to a much lower degree.
It is intriguing to consider the potential clustering mechanisms that are impacted by inhibiting phosphorylation at the 590/586 specific site of the PRC domain by the serine to alanine point mutation.One hypothesis is that VAP proteins act to modulate the probability of macro-cluster formation.Studies show that KV2.1 clusters are expressed at sites where the endo/sarcoplasmic reticulum is brought into close juxtaposition to the plasma membrane (ER/SR-PM junctions) 41 and this interaction and accumulation of channels relies on the tethering of KV2.1 to VAP proteins 21,43 43 .Interestingly, the model proposed in these prior papers 21,43 suggested that the phosphorylated PRC domain is necessary and sufficient for macro-clustering of KV2 channels.This is consistent with prior studies showing that mutations disrupting or eliminating the PRC domain 22,37,43 or treatments that impact KV2.1 phosphorylation 26,27,42 47 , and skeletal muscle Ca 2+ channels at SR Ca 2+ release units 48 .In the case of ventricular myocytes, it is concentrating voltage sensors at specific sites in the junctional dyad 49,50 .We propose that KV2.1 clustering is distinct in playing a role in modulating the localization and activity of an otherwise seemingly unrelated ion channel: CaV1.2 channels.This functional impact of KV2.1 is due to the density-dependent cooperative gating that is an intrinsic property of CaV1.2 channels 51 .
Remarkably, the overall impact of KV2.1S590A expression is that the differences between the ICa amplitude of wild-type male and female myocytes were eliminated in myocytes expressing the KV2.1S590A mutation, similar to what we observed in homozygous KV2.1 knockout mice 15 .Thus, declustering KV2.1 channels appears to have the same impact as fully eliminating KV2.1 expression on CaV1.2 clustering and activity in male and female myocytes.As our work also suggests that in arterial myocytes the conductive function of KV2.1 channels is independent of the degree of its clustering, in our model it is the extent of KV2.1 clustering that is the key determinant of the sex-specific differences in Ca 2+ influx observed in these cells.
To conclude, we propose a model by which KV2.

Generation of the CRISPR/Cas9-edited KV2.1S590A (KCNB1 S590A) knock-in mouse
The KCNB1 S590A mutation changes a AGC codon to GCC in Exon 2, thus converting a serine to an alanine (S590A) in the KV2.1 polypeptide.The knock-in mouse was generated in collaboration with the UC Davis Mouse Biology Program by using Crispr/CAS mediated homology directed repair.KCNB1 S590A mice were generated by introducing a mixture of gRNA (15 ng/L), single-stranded oligodeoxynucleotide (ssODN) repair template and Cas9 protein (30 ng/μL) by pronuclear microinjection into C57BL/6J mouse zygotes.Twenty zygotes were injected and implanted into the oviducts of one surrogate dam.A total of 6 pups were born, and genomic DNA was extracted from tail biopsies followed by PCR amplification using a specific primer set to identify a single male founder (F0).DNA-Seq analysis was used to confirm the mouse genotype.The correctly integrated single mutant F0 male mouse was further backcrossed with WT C57BL/6J female mice to produce offspring (F1) followed by intercrossing for two additional generations to obtain KCNB1 S590A heterozygotes which were used for breeding.Heterozygous and homozygous mutants were identified by a PCR genotyping protocol.

Animals
Mice were euthanized with a single, lethal dose of sodium pentobarbital (250 mg/kg) intraperitoneally.All experiments were conducted in accordance with the University of California Institutional Animal Care and Use Committee guidelines.

Arterial myocyte isolation
Third and fourth order mesenteric arteries were carefully cleaned of surrounding adipose and connective tissue, dissected out, and placed in ice-cold dissecting solution (Mg 2+ -PSS; 5 mM KCl, 140 mM NaCl, 2mM MgCl2, 10 mM glucose, and 10 mM HEPES) adjusted to pH 7.4 with NaOH.Arteries were first placed in dissecting solution supplemented with 1.23 mg/ml papain (Worthington Biochemical, Lakewood, NJ) and 1 mg/ml DTT at 37˚C for 14 min.This was immediately followed by a five-min incubation in dissecting solution supplemented with 1.6 mg/ml collagenase H, 0.5 mg/ml elastase (Worthington Biochemical), and 1 mg/ml trypsin inhibitor from Glycine max at 37˚C.Arteries were rinsed three times with dissection solution and single cells obtained by gentle trituration with a wide-bore glass pipette.Myocytes were maintained at 4˚C until used.To accomplish the Gaussian blur, the GSD generated pixel in an image was replaced by a weighted average of 200 nm of its neighboring pixels.The amount of blur applied to the image was controlled by the size of the kernel, which determines the radius of the neighboring pixels used in the calculation, such that the larger the kernel, the more pixels are included in the calculation, and the stronger the blur effect.

Quantitative PCR
Total RNA was isolated using the RNeasy Mini Kit (Qiagen) as per manufacturer's instructions.Isolated mRNA was then reverse transcribed using the AffinityScript qPCR cDNA Synthesis Kit (Agilent) following manufacturer's protocol.Quantitative PCR (qPCR) analysis was performed using a QuantStudio 7 Pro Realtime PCR System (Applied Biosystems), using PowerUP SYBR Green Master Mix (Thermo Fisher Scientific) as the fluorescence probe.The cycling conditions were 50°C for 2 minutes and 95°C for 10 minutes, followed by 40 cycles of 95°C for 15 s and 56°C for 1 minute.A dissociation curve protocol (ramping temperatures between 60°C and 95°C) was added at the end to verify amplification specificity of each qPCR reaction.

Figure
Figure 1A shows confocal maximum intensity projection images of 3D reconstructions of representative HEK293T cells expressing KV2.1WT (left) and KV2.1S586A (right).To the right of each image, we show single plane images from the bottom (top panel) and center (bottom panel) of each cell.Figure 1B-E shows a quantitative analysis of the number and volume of KV2.1 clusters from these 3D confocal images.The frequency distributions of KV2.1WT (black) and KV2.1S586A (blue) cluster volumes could be fit with a single exponential function.Of note, the number of clusters were smaller in most volume bins for the KV2.1S586A mutation compared to KV2.1WT.For example, in the same number of cells (n = 7), we detected a total of 6,344 KV2.1WT clusters, but only 3,335 KV2.1S586A clusters.The number of clusters per cell was 594.4 ± 52.7 in KV2.1WT Figure 1F provides representative superresolution ground-state depletion (GSD) TIRF images (lateral resolution ≈ 40 nm) from representative cells expressing KV2.1WT or KV2.1S586A.Note that data are provided on an area basis since we are imaging a single plane footprint of the cell in contrast to capturing multiple Z-slices.We also provide two regions of interest (boxed areas) by each image.KV2.1WT and KV2.1S586A channels are organized into clusters of varied sizes (Figure 1G), and consistent with the lower resolution confocal data, the distribution of Kv2.1S586A channel clusters was more diffuse than that of KV2.1WT.It should be noted that the regions of interest (ROI) of our GSD images reveal that the macro-clusters observed at the confocal level are, in fact, made up of numerous microclusters.
distances less than 200 nm (Figure1F, insets).Notably, our GSD image with a Gaussian blur accurately reproduced the clustering phenotype of KV2.1WT at the confocal level, providing further evidence that macro-clusters are composed of micro-clusters.Although we do not observe the large macro-clusters in KV2.1S586A expressing cells, we can still resolve KV2.1 micro-clusters.

Figure
2A and 2G show representative 3D images of fixed mesenteric smooth muscle cells labeled with WGA488 and immunolabeled for KV2.1.We investigated whether the KV2.1S590A mutation leads to altered expression levels of the channel in arterial myocytes.Our analysis showed that total, males, the frequency distribution of KV2.1 cluster sizes were similar in terms of relative values in all volume bins and could be fit with an exponential decay function (Figure 2C).The mean cluster volume in KV2.1WT males was 0.07 ± 0.01 μm 3 (median = 0.07 μm 3 ) compared to a mean of 0.07 ± 0.01 μm 3 (median = 0.06 μm 3 ) in KV2.1S590A male myocytes.(P = 0.338) (Figure 2D).Additionally, total clusters per cell of 551.6 ± 71.1 clusters, (median = 505 clusters) in KV2.1WT male myocytes were not significantly different from total clusters per cell of 444.2 ± 33.4 clusters (median = 419 clusters) in KV2.1S590A males (P = 0.10) (Figure 2E).The percentage of the membrane occupied by clusters in KV2.1WT
ge (mV) Vo l ta ge (mV) step depolarizations from a holding potential of -80 to +60 mV.We show ICa traces recorded during a depolarization to 0 mV from representative male (Figure 5A) and female (Figure 5B) KV2.1WT and KV2.1S590A arterial myocytes.Note that the amplitude and kinetics of ICa in these male KV2.1WT and KV2.1S590A arterial myocytes were similar.By contrast, we found that peak ICa was smaller in female Kv2.1S590A myocytes compared to those in KV2.1WT cells.In Figure 5C and D, we show the voltage dependence of the amplitude of ICa from all the cells examined over a wider range of membrane potentials.This analysis shows that the amplitude of ICa is similar in KV2.1WT and KV2.1S590A male myocytes at all voltages examined.However, in female myocytes, ICa was smaller in KV2.1S590A than in KV2.1WT at all voltages examined.Indeed, at 0 mV, ICa amplitude in KV2.1S590A cells was approximately 50% of that of WT females.

KV2. 1 -
mediated CaV1.2 clustering that impacted the probability of cooperative gating of these channels.Our data above show differences in ICa amplitude between female KV2.1WT and KV2.1S590A myocytes in the absence of differences in CaV1.2 expression levels.Thus, we investigated whether KV2.1S590A expression altered CaV1.2 channel clustering in a sex-specific manner using ground state depletion (GSD) super-resolution microscopy (Figure 6).We show ground-state depletion run in TIRF mode super-resolution images from representative male (Figure 6A) and female (Figure 6D) myocytes from KV2.1WT and KV2.1S590A mice.The insets show expanded views of two regions of interest (1 µm 2 ) within each cell image.Our TIRF images show that CaV1.2 clusters of various sizes are expressed throughout these cells.The frequency distribution of CaV1.2 cluster areas in KV2.1WT and KV2.1S586A male and females could both be fit with an exponential function (Figure 6B, E).The mean area of CaV1.2 clusters in male KV2.1WT of 2259 ± 55 nm2 , we set out to determine if a decrease in KV2.1 clustering would lead to a reduction in CaV1.2cells expressing CaV1.2-VN and CaV1.2-VC and co-expressing either KV2.1WT or KV2.1S586A (Figure7).The voltage protocols used for these experiments are similar to those used in two recent studies10,15 and are described in detail in the Methods section of this paper.We first transfected HEK293T cells with CaV1.2-VC,CaV1.2-VN, and the non-conducting but clustering competent rat KV2.1P404W channel 37 tagged with red-shifted fluorescent protein dsRed.The P404W mutation confers a non-conductive KV2.1 phenotype, allowing us to study the structural clustering role of KV2.1 without
channels that are tightly coupled and open and close simultaneously.A detailed description of this model is provided in the expanded Methods section.

HEK293T cell culture and
photons/event) using the GSD software and exported as binary TIF images.Particle analyses were determined in ImageJ.Representative images were rendered down to 2 nm for visualization purposes.

5. ICa is reduced in KV2.1S590A female myocytes but unaffected in male arterial myocytes. ICa
blue) mice.Summary data from real-time quantitative PCR experiments of CaV1.2 mRNA expression relative to β-actin in male (G) and female (I) myocytes.Quantification of immunofluorescence of labeled CaV1.2ɑ subunit in male (H) and female (J) myocytes.Error bars indicate mean ± SEM.
, our stochastic self-assembly model effectively reproduced the steady-state size distributions that we measured for CaV1.2 clustering in KV2.1WT and KV2.1S586A arterial myocytes.The parameters used in the model are summarized in Supplemental Figure 4C.These in silico data . The transmembrane ER/SR VAP proteins (VAPA and VAPB) interact with the phosphorylated Kv2.1 PRC domain and have been proposed to function to increase the local concentration of KV2.1 channels at ER/SR-PM junctions resulting in KV2.1 macro-clustering.Consistent with this, Kirmiz et al., 21 found that knock-out of VAPA in RAW664.7 macrophage cells resulted in a decrease in KV2.1 channel clustering.Knockdown of endogenous VAP proteins similarly impaired clustering of KV2.1 heterologously expressed in HEK293T cells 45pact KV2.1 clustering.It is presumed that the phosphorylation of multiple serine residues, including S590, within the PRC domain provide the negative charges needed to generate a functional VAP-binding FFAT -two phenylalanines in an acidic tract -motif, as has been shown for numerous other proteins that exhibit phosphorylation-dependent binding to VAPs44.Therefore, one possible mechanism for the decrease in macro-clustering in the S590A mutant is the inability of VAP proteins to recognize the PRC domain of mutated channels preventing cluster growth.The similarity in the patterns of cluster sizes and densities between HEK293T cells and arterial myocytes of both WT and S590A channels is noteworthy, indicating the possibility of a shared set of mechanisms.Further research will be necessary to uncover the underlying factors that govern these clustering patterns.Prior studies have suggested that the bulk of KV2.1 channels heterologously expressed in Xenopus oocytes28or HEK293T cells19as well as endogenous KV2.1 in hippocampal neurons45and arterial myocytes 15 are in a nonconducting state.The prevailing view is that aggregation of KV2.1 channels into high density clusters is what renders most of these channels incapable of conducting K +(45).Although our study does not address this issue comprehensively, at a minimum, our data suggest that KV2.1 conduction is not dependent on macroclustering formation.Future studies should investigate whether the formation of KV2.1 micro-clusters may be sufficient to electrically silence these channels.This is the first study to definitively demonstrate the structural role of KV2.1 clustering in regulating CaV1.2 channel clustering and activity that occurs in channels of native cells.This is significant because the generally accepted view is that the functional impact of ion channel clustering is to exclusively concentrate ion conducting roles at specific sites.For example, Na + channel clustering at nodes of Ranvier 46 , neuronal Ca 2+ channel clustering at active zones in presynaptic terminals