Caveolin interaction governs Kv1.3 lipid raft targeting

The spatial localization of ion channels at the cell surface is crucial for their functional role. Many channels localize in lipid raft microdomains, which are enriched in cholesterol and sphingolipids. Caveolae, specific lipid rafts which concentrate caveolins, harbor signaling molecules and their targets becoming signaling platforms crucial in cell physiology. However, the molecular mechanisms involved in such spatial localization are under debate. Kv1.3 localizes in lipid rafts and participates in the immunological response. We sought to elucidate the mechanisms of Kv1.3 surface targeting, which govern leukocyte physiology. Kv1 channels share a putative caveolin-binding domain located at the intracellular N-terminal of the channel. This motif, lying close to the S1 transmembrane segment, is situated near the T1 tetramerization domain and the determinants involved in the Kvβ subunit association. The highly hydrophobic domain (FQRQVWLLF) interacts with caveolin 1 targeting Kv1.3 to caveolar rafts. However, subtle variations of this cluster, putative ancillary associations and different structural conformations can impair the caveolin recognition, thereby altering channel’s spatial localization. Our results identify a caveolin-binding domain in Kv1 channels and highlight the mechanisms that govern the regulation of channel surface localization during cellular processes.

Here, we study the influence of Cav on the lipid raft targeting of Kv1 channels by analyzing the function of a putative CBD motif conserved in the Shaker family. Both Kv1.3 and Kv1.5 target lipid rafts, but only Kv1.3 efficiently interacted with Cav via the CBD where this association is essential for the channel localization in these domains. Moreover, Kv1.3 behavior and activity was conditioned by the presence of Cav1. Therefore, the presence of a CBD near the T1 of Kv1.3 has important functional consequences for Kv1.3 channel physiology.
The caveolin 1 expression induces de novo formation of caveolae structures in caveolin-null cells, thereby increasing the plasma membrane structuration 24,25 . Moreover, caveolae appear as rigid structures in which caveolins show a reduced mobility 26 . Because Kv1.3 and Cav1 physically interact, the caveolae targeting and the membrane dynamics of Kv1.3 were tested in HEK Cav-and HEK Cav1. Electron micrographs indicated that Kv1.3 was recruited into caveolae-like structures in the presence of Cav1 (Fig. 3A,B). Whether Kv1.3 membrane lateral diffusion was altered in the presence of Cav1 was studied by fluorescence recovery after photobleaching (FRAP) analysis ( Fig. 3C-G). The Kv1.3YFP fluorescence recovery was monitored over time in HEK Cav-and HEK Cav1, until a steady state was achieved (Fig. 3E). While the mobile fraction (0.56 ± 0.04 vs 0.48 ± 0.04, for HEK  Cav-and HEK Cav1, respectively, n = 10) was similar, the half-life recovery increased in HEK Cav1 (21.77 ± 1.49 s and 28.87 ± 2.49 s for HEK Cav-and HEK Cav1, respectively, p < 0.05, n = 10) (Fig. 3F), where a lower motion of Kv1.3 in the presence of caveolin was observed. This could be explained not only by the major recruitment of the channels in rigid structures, such as caveolae, but also by an increase of liquid-ordered domains structuring the membrane. Therefore, the Kv1.3 membrane dynamics were also analyzed by single particle tracking (SPT) using Qdots (Fig. 3H,I). Single Kv1.3 molecules at the cell surface were tagged with Qdots and monitored by total internal reflection fluorescence (TIRF) imaging with a temporal resolution of 0.5 s (Supplementary video 1). Qdots were classified according to their behavior in single or multiple units ( Fig. 3H and Supplementary video 1). The last were defined when more than one Qdot motion agroupated for more than 10 s, thereby suggesting aggregated channels. While the abundance of multiple Qdots increased in HEK Cav1 cells (11.15 ± 2.24% vs 23.84 ± 3.26%, p < 0.05), single Qdots decreased (88.85 ± 2.24% vs 76.15 ± 3.26%, p < 0.01), which suggests an aggregated distribution of Kv1.3 channels in the presence of Cav1. Moreover, the trajectories of single Qdots were analyzed by plotting the mean square displacement (MSD) against time 27 . Three types of motion were observed: (i) simple Brownian diffusion (free), (ii) confined diffusion (confined) and (iii) stationary diffusion (stationary) (Fig. 3I). In this context, the diffusion coefficient of free diffusing channels in the presence of Cav1 decreased (0.016 ± 0.002 vs 0.010 ± 0.001 μm 2 /s for HEK Cav-and HEK Cav1, respectively, p < 0.05). Thus, SPT results suggested both the aggregation of Kv1.3 and reduced channel mobility in the presence of caveolin.
Glucose transporter type 4 (Glut4) and insulin receptor (IR) are recruited in caveolae 28 , and Cav1 participates in Glut4 and IR stability. Thus, Cav1 depletion reduces Glut4 and IR protein abundance by their faster degradation 29 . Therefore, we next studied whether Cav1 association altered Kv1.3 stability. Time-course experiments performed in Cav-and Cav1 HEK cells demonstrated that, similar to Glut4 and IR, Kv1.3 persisted for a longer period of time in the presence of Cav1 (Fig. 4A,B). In this context, Cav1 can also affect channel activity 1 . Therefore, for these experiments, Cav1 was reintroduced into HEK Cav-cells, and Kv1.3 electrophysiological properties were analyzed. While the threshold of activation was similar, the presence of Cav1 increased the Kv1.3 current density (Fig. 4C,D). Slow C-type inactivation is a characteristic of Kv1.3. It involves conformational changes of the channel that result in closure of the external mouth of the pore with probable cooperativity between subunits 30 . In this context, the Cav1 interaction enhanced the C-type inactivation of Kv1.3. Thus, the current at the end of a 5 s pulse (+ 60 mV) was lower in the presence than in the absence of Cav1 (Fig. 5E,F).

Caveolin 1 interacts with Kv1.3 via a CBD signature located at the N-terminal of the channel.
The N-terminus of Kv1 channels contains important signatures involving tetramerization and regulatory subunit association. Although caveolin uses a CSD to interact with substrates via a CBD 7,8,31 , this model has been compromised by structural and bioinformatic analysis 32 . In this context, the N-terminus of Kv1.1-Kv1.5 contains putative CBDs lying next to the first transmembrane segment (amino acids 166 to 174 in rKv1.3), right after the T1 tetramerization domain and the Kvβ subunit association signature ( Supplementary Fig. S2G). This CBD is represented by a ФxxxxФxxФ consensus sequence, where Ф is a hydrophobic residue. Our data indicates that Kv1.3 and Kv1.5 localized significantly in rafts but only Kv1.3 directly interacted with Cav1. Therefore, we next focused on whether the Kv1.3 CBD molecular determinant was involved in Cav1 interaction. To do so, we performed coimmunoprecipitation assays with different Kv1.3 mutants and Kv1.3-Kv1.5 chimerical channels (Figs 5 and 6). While Kv1.3ΔCt (no C-terminal domain) coimmunoprecipitated with Cav1, Kv1.3ΔNt (no N-terminal domain) did not. In this scenario, to rule out the effect of an altered tetramer formation on traffic and subcellular localization of the Kv1.3 truncated channels, we analyzed Kv1.3/Kv1.5 chimeras that preserved the full integrity of the channel. Chimeras, containing the Kv1.3 N-terminal domain coimmunoprecipitated with Cav1 to a greater extent than did chimers that contained the N-terminus of Kv1.5 (Fig. 5B,C). To further understand this specific Kv1.3 signature, the putative CBD of Kv1.3 and Kv1.5 was mutated. Thus, aromatic amino acids were substituted by alanine or glycine-generating CBD mutants (Kv1.3: 166 FQRQVWLLF 174 to 166 AQRQVGLLA 174 ; Kv1.5: 232 FQRQVWLIF 240 to 232 AQRQVGLIA 240 ). While the Kv1.3 mutant (Kv1.3CBD) showed a reduced lipid raft partitioning with no Cav1 dependency (Fig. 6A), the Kv1.5CBD showed no lipid raft targeting alterations (Fig. 6B). Moreover, the Kv1.3CBD did not coimmunoprecipitate with Cav1, thereby highlighting the importance of the CBD integrity for Kv1.3 interactions with Cav1 (Fig. 6C).
We found no interactions between Kv1.5 and Cav1; however, by converting the putative Kv1.5 CBD to that of Kv1.3 (Kv1.5I239L), we observed a positive Cav1 coimmunoprecipitation Fig. S4). Furthermore, evidence suggests that Kv1.5 could interact indirectly with caveolins by the formation of supramolecular complexes, including SAP97 33,34 . SAP97 (synapse-associated protein 97) is a member of the membrane associated guanylate kinase (MAGUK) family that also includes PSD95 (postsynaptic density protein 95). Therefore, we expressed Kv1.5 in the presence and absence of PSD95 in HEK Cav1 cells. In this scenario, Kv1.5 coimmunoprecipitated with Cav1, only in the presence of PSD95 (Fig. S4). Therefore, our data suggest that, while the full integrity of the Kv1.3 CBD is sufficient to interact with Cav1, Kv1.5 requires ancillary proteins.

Discussion
Evidence demonstrates that Kv1.3 targets specific membrane localizations 35,36 , and location displacements entail pathological consequences 37 . Our study highlights the main mechanism of Kv1.3 channel membrane surface partitioning. We report here that, among Kv1 (Shaker) channels, only Kv1.3 and Kv1.5 targeted significantly to lipid rafts. However, while the Kv1.5 floatability was independent of the caveolin expression, Kv1.3 lipid raft targeting increased in a caveolin dose-dependent manner. This is of physiological relevance because this was confirmed in BMDM from Cav1 −/− mice. Furthermore, caveolin-channel colocalization was higher with Kv1.3 than with Kv1.5. Finally, we have clearly identified a CBD that is located at the N-terminal domain of Kv1.3 and is the responsible element for Cav1 interaction and lipid raft localization of the channel. Because lipid raft targeting has been proposed as a mechanism for ion channel regulation 1,38 , our results contribute to this expanding field. Kv1.3 redistribution within the plasma membrane is critical for lymphocyte physiology 37 . Upon activation, T cells spatially reorganize membrane proteins to form the immunological synapse (IS), where lipid rafts accumulate and recruit TCR (T-cell receptor), CD3 (cluster of differentiation protein 3) and Kv1.3 10 . Caveolin is crucial for the IS reorganization of CD8 T cells 39 . Thus, the Kv1.3-caveolin interaction described here could participate in the recruitment of Kv1.3 into the IS orchestrated by caveolin. Moreover, the cell membrane composition and lipid raft integrity regulates Kv1.3 activity 40 . This is in agreement with the functional consequences that are observed when Kv1.3 rearranges into the IS 41 . Similarly, Kv1.3 activity was altered in the presence of caveolin. Caveolins also regulate the activity of other channels such as Nav1.5 42 . In addition, caveolin 3 also coimmunoprecipitates with cardiac Kv11.1 43,44 ; however, a direct interaction between caveolins and channels is not a unique  way to target channels to raft domains. In this context, our Kv1.5 data are in the same line of evidence as that described for Kv1.4. The location of Kv1.4 in caveolar domains is uncertain, but the presence of PSD95 increases the targeting to rafts microdomains 45 . Conversely, the raft localization of Kv2.1 and Kv4.2 seems independent of the presence of auxiliary scaffolding proteins such as caveolins or PDZ-containing proteins 45,46 . In this scenario, much work must be conducted to decipher different partnership associations, thereby conforming specific cell channelosomes, which allows for the spatial localization of channels and the regulation of physiological response.
Evidence suggests an increasing number of ion channels, mostly cardiac, are in caveolar rafts 47 . However, our results confirmed, for the first time, that Kv1.3 lipid raft targeting occurs via a direct interaction wherein caveolin recruits the channel inside caveolae structures, thereby restricting the channel's lateral diffusion. The molecular determinant of Kv1.3 that is involved in such interaction is a CBD located at the N-terminus of the channel in close proximity to the T1 tetramerization domain and the Kvβ subunit interaction signature. Although other Kv1 members share similar motifs, none displayed a caveolin-dependent behavior or major lipid raft targeting. Our results demonstrated that few, single point differences within the CBD signature and/or impaired CBD accessibility due to bulky intracellular domains, may impair this interaction. Interestingly, coimmunoprecipitation studies using chimeric Kv1.3/Kv1.5 and Kv1.5(I239L) mutant channels suggest both. Thus, while Kv1.5Nt, containing a CBD motif, was not enough for Kv1.5-Cav1 coimmunoprecipitation, a fairly positive association was observed when the bulky C-terminal of Kv1.5 was substituted by the C-terminal of Kv1.3. In addition, the introduction of a di-leucine motif, within the CBD of Kv1.5 (Kv1.5 I239L), triggered Cav1 co-immunoprecipitation. Our data suggest that the balance of other interacting motifs within Kv1.5 could mask the CBD accessibility and/or effectiveness 18 , which is similar to what has been previously reported for other forward trafficking signals, such as VxxSL or YMVIEE 48,49 . In this sense, Kv1.2, containing the same CBD of Kv1.3, lacks strong trafficking signals and exhibits endoplasmic reticulum retention 50 . Unlike Kv1.3 that colocalizes with caveolin in Golgi 17 , Kv1.2 and caveolin do not share intracellular compartments what would impair the association. However, Kv1.2 cell surface is promoted by PSD95 and Kvβ subunits 51 . It is tempting to speculate that different structural tertiary configurations of bulky domains could condition protein-protein interactions with MAGUK proteins, such as PSD95 or SAP97, that interact differently with Kv channels [37][38][39] . This could be explained by supramolecular complexes formed by Kv1.5, Cav1 and SAP97, which further supports our Kv1.5 and PSD95 data in HEK Cav1 cells 33,52,53 .
In summary, our results help to elucidate the mechanisms that target Kv1 channels to specific surface microdomains that participate in fine-tuning the cellular responses. Unlike other neuronal Kv1 channels, Kv1.3 interacts with caveolin through a CBD placed at the N-terminal domain of the channel adjacent to the first transmembrane segment and in close proximity to the T1 domain and the Kvβ binding site. This association targets Kv1.3 to caveolar structures that regulate both the channel membrane dynamics and activity.

Methods
Expression plasmids and site-directed mutagenesis. Rat Kv1.3 in pRcCMV was provided by T.C.
Cell culture, transient transfections and raft isolation. HEK 293 cells were grown in DMEM containing 10% FBS and 100 U/ml penicillin/streptomycin (Gibco). Transient transfection was performed using MetafecteneTM Pro (Biontex) at nearly 80% confluence. Murine bone marrow derived macrophages were isolated, as previously described 55 . All of the experiments and surgical protocols were performed in accordance with the guidelines approved by the ethical committee of the Universitat de Barcelona following the European Community Council Directive 86/609 EEC.
Low density, Triton-insoluble complexes were isolated, as previously described 18,20 . Cells were homogenized in 1 ml of 1% Triton X-100, and sucrose was added to a final concentration of 40%. A 5-30% linear sucrose gradient was layered on top and further centrifuged (39,000 rpm) for 20-22 h at 4 °C in a Beckman SW41 rotor. Gradient fractions (1 ml) were collected from the top and analyzed by Western blot.

Protein extraction, co-immunoprecipitation and western blot analysis. Cells, washed in cold PBS,
were lysed on ice with NHG solution (1% Triton X-100, 10% glycerol, 50 mmol/L HEPES pH 7.2, 150 mmol/L NaCl) supplemented with 1 μg/ml of aprotinin, 1 μg/ml of leupeptin, 1 μg/ml of pepstatin and 1 mM of phenylmethylsulfonyl fluoride to inhibit proteases. Homogenates were centrifuged at 16,000 × g for 15 min, and the protein content was measured using the Bio-Rad Protein Assay.
For immunoprecipitation, samples were precleared with 30 μl of protein A-Sepharose beads for 2 h at 4 °C with gentle mixing as part of the co-immunoprecipitation procedures. The beads were then removed by centrifugation at 1,000 × g for 30 s at 4 °C. Samples were incubated overnight with the anti-caveolin antibody (4 ng/μg protein) at 4 °C with gentle agitation. Thirty microliters of protein A-Sepharose were added to each sample and incubated for 4 h at 4 °C. The beads were removed by centrifugation at 1,000 × g for 30 s at 4 °C, washed four times in NHG, and resuspended in 100 μl of Laemmli SDS buffer.
For transmission electron microscopy, PML were treated as performed for immunocytochemistry, but visualized with different secondary antibodies. Kv1.3 was recognized by a mouse anti-Kv1.3 antibody (1/20, Neuromab). Goat anti-mouse and anti-rabbit secondary antibodies, conjugated to 10 nm and 15 nm gold particles, recognized Kv1.3 and Cav1, respectively. Briefly, processed samples were further fixed with 2.5% glutaraldehyde in PBS for 30 min at RT. Next, samples were subjected to freeze-drying, washed and cryoprotected with 10% methanol. Samples were then cryofixed using slam-freezing (BAF-060, Bal-Tec) for 90 min at − 90 °C and 10-7 mbar pressure. Replicas were obtained by rotationally (136 rpm) evaporating 1 nm platinum through electron cannon (at an angle of 23°). This was reinforced by evaporating 10 nm carbon (at an angle of 75°). Replicas were separated from the sample using 30% fluorhydric acid. Finally, samples were washed and mounted over Formvar coated grilles.
Scientific RepoRts | 6:22453 | DOI: 10.1038/srep22453 Föster resonance energy transfer (FRET). FRET was performed in the acceptor photobleaching configuration. Samples were imaged with a Leica SP2 confocal microscope. Images were acquired before and after YFP bleach using 63 × oil immersion objective at zoom 4. Excitation was via the 458 and 514 nm lines of the Ar laser, and 473-495 and 535-583 bandpass emission filters were used. FRET efficiency (FRETeff) was calculated using the equation: where, F D after: donor fluorescence (Cerulean) after and F D before before acceptor (YFP) bleach. Analysis was performed using ImageJ.
Experiments were performed as previously described 46,54,56 . For FRAP experiments, an Olympus FV1000 microscope was used. Briefly, YFP was bleached during 250 ms with a 515 nm line Ar laser at 30% and was fluorescence monitored before and after bleach with a PLAPO 60x NA 1:1,40 oil objective at zoom 4 acquiring every 1.108 s. Acquisition was performed with the 515 nm Ar laser line at 1% and a 525-560 nm bandpass emission filter. SPT analysis was performed as previously described 46 . Briefly, cells co-expressing Kv1.3LoopBAD and BirA for 24 h were incubated for 5 min at RT with 0.1 nM Streptavidin Qdots655 (Invitrogen, Oregon) in 1% BSA HIS (146 mM NaCl, 4.7 mM KCl, 2.5 mM CaCl 2 × 2H 2 O, 0.6 mM MgSO 4 × 7H 2 O, 0.15 mM NaH 2 PO 4 × 2H 2 O, 0.1 mM ascorbic acid, 8 mM Glucose, 20 mM HEPES, pH 7.4) and washed five times with HIS at RT. Cells were imaged in the following hour at 37 °C in a 5% CO 2 atmosphere. Imaging was performed with Nikon Eclipse Ti PerfectFocus equipped TIRF (Total internal reflection fluorescence) microscope with a 100xPlanApoTIRF, 1.49 NA, oil objective. YFP was excited with 488 nm line Ar laser at 2% and Qdots with 561 nm laser at 20%. Emission was collected through a Sutter Lambda 10-3 filter wheel and recorded with an Andor iXon EMCCD DUD897 camera. For TIRF acquisition, the incident angle was 63.3°. Imaging acquisition was approximately 10 MHz. Videos were processed using Volocity (PerkinElmer Software). SPT was performed manually. The tracks were then analyzed using Sigmaplot to obtain mean square displacement (MSD) and the Diffusion Coefficient.
Electrophysiology. Patch-clamp whole-cell configuration experiments were performed, as performed in 49 .
To evoke voltage-gated currents, cells were stimulated with 250 ms square pulses ranging from − 60 to + 80 mV in 10 mV steps. C-type inactivation was studied by applying a long pulse of 5 s at + 60 mV. The peak amplitude (pA) was normalized using the capacitance values (pF). Data analysis was performed using FitMaster (HEKA) and Sigma Plot 10.0 software (Systat Software). All recordings were performed at RT.