An essential role for cortactin in the modulation of the potassium channel Kv1.2

In Vivo Interaction of Kv1.2 and Cortactin Detected by FRET Microscopy.

In a previous report, we used fluorescence microscopy to show that Kv1.2 colocalizes with cortactin and immunoprecipitation and pull-down assays to show that the proteins interact directly (1). Here, we use FRET to determine whether Kv1.2 and cortactin interact within intact cells and, if so, where those interactions occur. We first confirmed that conjugation of yellow fluorescent protein (YFP) to Kv1.2 did not impair channel function. YFP-Kv1.2, Kvβ2, and cyan fluorescent protein (CFP) cortactin were transiently expressed in HEK293 cells stably expressing the M1 muscarinic receptor. To label surface channel, live cells were exposed to α-Kv1.2e for 10 min at 37°C, fixed, and imaged by using the DeltaVision reconstruction microscopy system (Applied Precision, Issaquah, WA). YFP-Kv1.2 appears throughout the cell, with intense fluorescence at the periphery and leading edge structures (Fig. 1b), and a similar distribution was observed with unconjugated Kv1.2 (Kv1.2wt) (1). Distribution of α-Kv1.2e fluorescence (Fig. 1a) completely overlaps peripheral YFP fluorescence, indicating that peripheral YFP-Kv1.2 is at the cell surface. YFP-Kv1.2 ionic currents lack fast inactivation and have a half activation voltage of approximately −15 mV (Fig. 1 c and d). Using flow cytometry to measure surface expression in cells expressing YFP-Kv1.2 and CFP cortactin, we found that 10 μM carbachol at 10 min evoked a significant loss of surface channel (P < 0.01, n = 11), as did an application of 10% serum at 10 min (P < 0.001, n = 11) (Fig. 1e). Therefore, YFP-Kv1.2 channels exhibit electrophysiological and trafficking properties that are indistinguishable from Kv1.2wt.

To maximize confidence in using FRET to assay in vivo interactions between Kv1.2 and cortactin, we used three distinct detection methods: acceptor photobleaching, donor photobleaching, and microFRET. In acceptor photobleaching, energy transfer from CFP to YFP is reduced or eliminated upon acceptor bleaching, resulting in increased donor fluorescence (9). FRET efficiencies were determined for a region of interest (ROI) inside and outside the bleached zone. Examples of prebleach and postbleach ROIs are shown in Fig. 2 a and b, respectively. The decrease in YFP fluorescence within the bleached region (ROI 1) depicted in Fig. 2 b and c is accompanied by an increase in CFP fluorescence (Fig. 2c), suggesting the existence of FRET. FRET efficiencies were measured in bleach and nonbleach ROIs in cells expressing either CFP cortactin only or YFP-Kv1.2 and CFP cortactin. The distributions of unbleached and bleached values were similar in cells expressing only CFP cortactin; the means of the nonbleach regions (4.07 ± 0.73%, n = 75) and the bleach regions (4.70 ± 0.69%, n = 77) do not significantly differ (P = 0.5299). In contrast, cells expressing both YFP-Kv1.2 and CFP cortactin showed a significant difference (P < 0.0001) in FRET efficiency between bleached (11.21 ± 0.89%, n = 126) and nonbleached (4.46 ± 1.25%, n = 127) regions (Fig. 2d). These results indicate the occurrence of FRET between YFP-Kv1.2 and CFP cortactin.

Donor photobleaching is based on the principle that processes such as FRET, which shorten the excited-state lifetime of a fluorophore, also protect it from photobleaching, making decay kinetics an efficient measure of FRET efficiency (10). Fluoresence decay from cells expressing only CFP cortactin (Fig. 2e) or CFP cortactin and YFP-Kv1.2 (Fig. 2f) was measured, and decay constants were determined by fitting to single exponential curves. Measurements from multiple cells (Fig. 2g) show that the mean decay constant for cells expressing CFP cortactin alone (3.13 × 10⁻² ± 0.04 × 10⁻², n = 37) was significantly greater (P = 0.016) than that for cells expressing CFP cortactin and YFP-Kv1.2 (0.20 × 10⁻² ± 0.02 × 10⁻², n = 37). These data are consistent with the occurrence of FRET between the CFP and YFP fluorophores and again suggest a close in vivo interaction between Kv1.2 and cortactin.

MicroFRET uses short exposure times and standard epifluoresence microscopy, making it the preferred method for detecting FRET in live cells. In this method, FRET is typically measured as the ratio of acceptor emission to donor emission upon excitation of the donor, resulting in a value that is proportional to the degree of physical association between the two fluorphores (11). Correction coefficients are used to compensate for spectral bleed through artifacts (12). To minimize variability inherent in imaging live cells, we began by measuring microFRET in cells that were plated and fixed as described in Methods. Fig. 2 h–s shows examples of images from cells expressing CFP cortactin (h–k), YFP-Kv1.2 (l–o), or both (p–s). In all cases, the uncorrected FRET image (U) shows significant spectral bleed-through fluorescence. However, upon transformation with correction coefficients, FRET fluorescence is observed in cells expressing both fluorophores (Fig. 2s), but not in those expressing only CFP cortactin (Fig. 2k) or YFP-Kv1.2 (Fig. 2o). Therefore, consistent with the acceptor and donor photobleaching results, these microFRET data indicate in vivo interaction between Kv1.2 and cortactin.

A consensus has emerged that cortactin has a key role in vesicular transport (10). Therefore, we were intrigued by the occurrence of FRET not only at the cell surface, but also within subcellular compartments (Fig. 2s). To explore this idea further, we used microFRET to analyze the association between Kv1.2 and cortactin within different cellular regions in living cells treated with a saline control solution or a stimulus capable of producing channel suppression (Fig. 3). Data from previous studies (1) and Fig. 1 indicate that Kv1.2 at the leading edge is at the cell surface. Analysis of FRET ratio intensities from leading edge structures or vesicles revealed the most intense FRET signal at the leading edge, with significantly (P < 0.001, n = 20) weaker FRET within vesicles (Fig. 3a). Because stimuli that evoke channel tyrosine phosphorylation lead to a decrease in the association between Kv1.2 and cortactin (1), we postulated that they also would cause a decrease in FRET. Surprisingly, pooled data from cells that were stimulated with carbachol (n = 4) or serum (n = 4) produced no change compared with saline in FRET within a given region (P > 0.05), although the difference between the regions was preserved (P < 0.001, n = 8) (Fig. 3b). Given that these stimuli cause channel endocytosis, the stimulus-induced decrease in association of cortactin with Kv1.2 detectable in pull-down assays (1) may reflect trafficking of channel from the surface, where the interaction is strong, to intracellular vesicles, where the interaction is weak.

Cortactin Binding to Kv1.2 Is Required for Channel Function.

Having established an in vivo interaction between Kv1.2 and cortactin, we sought to determine the functional consequences of that interaction. We used two approaches to explore that hypothesis. Because tyrosine phosphorylation of Kv1.2 is associated with both decreased interaction with cortactin and decreased presence of channel in the plasma membrane, we hypothesized that depleting intracellular cortactin would cause suppression of Kv1.2 ionic current. Xenopus oocytes were injected simultaneously with antisense DNA oligonucleotides directed against Xenopus cortactin and with cRNA encoding Kv1.2wt. Voltage clamp measurements (Fig. 4a) reveal a significant suppression of Kv1.2 ionic current in oocytes injected with antisense cortactin oligonucleotides relative to control (P < 0.001, n = 25). Coinjection of cRNA encoding mouse cortactin caused a complete rescue of Kv1.2 current, indicating a key role for cortactin in Kv1.2 function.

Although these findings are consistent with the idea that cortactin is important for Kv1.2 function, global depletion of cellular cortactin could have multiple effects. To address this notion, we used a mutagenesis approach that focused on the channel. Earlier studies revealed that truncation of the Kv1.2 C terminus (Kv1.2-K445stop) completely eliminated cortactin binding as well as Kv1.2 ionic current (1). If loss of cortactin binding diminishes Kv1.2 ionic current, reestablishing cortactin binding by artificial means should, at least in part, restore channel function. To accomplish this goal, we fused a cortactin-specific SH3 domain-binding motif, KPPIPQKPTWL (13), to the Kv1.2-K445stop channel (Kv1.2-K445-SH3). Voltage clamp analysis summarized in Fig. 4 c and d shows that addition of this motif produced a partial restoration of Kv1.2 current relative to Kv1.2-K445stop (P < 0.01, n = 10). To ensure that this effect resulted from the cortactin-binding properties of the fused SH3-binding motif, and not simply from the added structure, we introduced a point mutation replacing a lysine required for cortactin SH3 domain recognition with alanine (KPPIPQAPTWL) to ablate cortactin binding (Kv1.2-K445-SH3KA) (13). The difference between Kv1.2-K445stop and Kv1.2-K445-SH3KA was not significant (P > 0.05, n = 10), indicating that the point mutation completely eliminated the restorative effect of the SH3 domain on Kv1.2 function (Fig. 4d). GST-fusion protein pull-down experiments confirm that addition of the cortactin SH3-binding motif to Kv1.2-K445stop restores cortactin binding and that mutation of the central lysine with alanine ablates that binding (Fig. 4d Inset). These data show that the direct physical interaction between cortactin and the C terminus of Kv1.2 is necessary and sufficient for at least partial Kv1.2 function.

Cortactin Has a Role in Plasma Membrane Trafficking of Kv1.2.

The electrophysiological activity of Kv1.2 is governed in part by stimulus-induced channel endocytosis. Cortactin contributes to endocytosis through its actin-regulating properties and association with dynamin (10). We therefore asked whether the functional dependence of Kv1.2 on cortactin demonstrated in Fig. 4 is related to an ability of cortactin to regulate Kv1.2 trafficking. We used RNAi to deplete endogenous human cortactin in HEK293 cells. In some experiments, the depleted cortactin was replaced with mouse cortactin. Immunoblot of cells transfected with cortactin-specific RNAi confirms a strong reduction of endogenous cortactin relative to cells transfected with a nontargeting RNA oligo (Fig. 5 a and b), but shows no corresponding reduction in Kv1.2 protein levels (Fig. 5a). Flow-cytometry analysis (Fig. 5c) revealed a significant (P < 0.001) reduction in surface Kv1.2 levels in cortactin-depleted cells (n = 117) relative to cells transfected with a nontargeting RNA (n = 115). Cotransfection of RNAi with full-length (FL) murine cortactin restored surface Kv1.2 to nontargeting RNAi levels (n = 110; P > 0.05). These findings parallel data shown in Fig. 4, in which cortactin knockdown suppresses Kv1.2 ionic current. Therefore, cortactin influences both Kv1.2 surface expression and Kv1.2 ionic current. Carbachol significantly reduced Kv1.2 surface levels in cells transfected with nontargeting RNAi (P < 0.001). Carbachol also reduced Kv1.2 surface levels in cortactin-depleted cells (P < 0.001). However, the percent suppression evoked by carbachol in cortactin-depleted cells (26%) was significantly less than that produced in nondepleted cells (35%) (P < 0.0001) or cells cotransfected with FL cortactin (36%) (P < 0.001).

Kv1.2 Trafficking Is Modulated by C-Terminal Tyrosines Within Cortactin.

Among the proteins capable of binding to the SH3 domain of cortactin is dynamin, a protein with key roles in vesicle trafficking and endocytosis. Having established a role for cortactin in Kv1.2 trafficking, we explored the specific hypothesis that cortactin contributes to Kv1.2 endocytosis by complexing with dynamin. To do so, endogenous cortactin was depleted by using RNAi and replaced with either FL cortactin or a truncated form lacking the SH3 domain (cortactin ΔSH3) (Fig. 6). Removing this domain does not interfere with cortactin ΔSH3's expression or ability to bind to Kv1.2 (Fig. 6 b and c) (1). Surface Kv1.2 expression was the same in cells expressing a FL or SH3-deleted form of cortactin (1.0 ± 0.03, n = 110 vs. 0.98 ± 0.03, n = 30, respectively; P > 0.05). To our surprise, deletion of the SH3 domain had no effect on carbachol-induced channel endocytosis, producing a significant (36%) (n = 30; P < 0.0001) loss of surface channel, which was comparable to that observed in cells expressing FL cortactin (36%) (n = 110; P < 0.0001). Therefore, cortactin's SH3 domain does not contribute to Kv1.2 trafficking.

In contrast, a form of cortactin (1-330 cortactin) lacking the C-terminal region encompassing proline- and tyrosine-rich domains significantly reduced surface Kv1.2 levels relative to cells expressing FL cortactin (0.71 ± 0.08, n = 24 vs. 1.00 ± 0.03, n = 110; P < 0.001). This effect was not significant in cells that were not cotransfected with cortactin RNAi (0.83 ± 0.08, n = 14 vs. 1.00 ± 0.10, n = 14; P > 0.05). However, 1-330 cortactin did support carbachol-induced channel endocytosis to the same degree as FL cortactin. Carbachol produced a 38% reduction in surface Kv1.2 levels relative to saline control (0.44 ± 0.04, n = 23 vs. 0.71 ± 0.08, n = 24 respectively; P < 0.01). This effect is not caused by a lack of 1-330 cortactin protein expression or a disruption in its ability to bind to Kv1.2 (Fig. 6 b and c). Thus, the C terminus of cortactin contains some components of the cortactin functionality involved in regulating Kv1.2.

Cortactin is a major target for tyrosine phosphorylation, and the C-terminal tyrosines Y421, Y466, and Y482 have been identified as functionally important substrates and binding sites for Src kinase (6). Because Kv1.2 endocytosis is a tyrosine phosphorylation-dependent phenomenon (2), we asked whether these tyrosines are involved in Kv1.2 regulation. Endogenous cortactin was replaced with either FL cortactin or cortactin containing a tyrosine-to-phenylalanine mutation at Y421, Y466, or Y482. Y421F cortactin was indistinguishable from FL in its ability to maintain unstimulated surface levels and carbachol suppression (Fig. 7). In contrast, expression of Y466F cortactin produced a 19% reduction in surface Kv1.2 (n = 20, P = 0.002). Carbachol elicited a 26% loss of surface Kv1.2 (n = 20, P < 0.01). Y482F cortactin caused a 36% reduction in surface Kv1.2 (n = 23, P < 0.0001), and carbachol further reduced surface channel levels by 23% (n = 20, P = 0.04). These data suggest that Y466 cortactin and Y482 cortactin are phosphorylated in unstimulated HEK cells. Western blot analysis using a phosphospecific antibody against the tyrosine in human cortactin cognate to mouse Y482 cortactin confirmed this finding (Fig. 7a Inset), supporting the conclusion that phosphorylation of Y482 cortactin, and likely Y466 cortactin, has an important role in maintaining Kv1.2 at the cell surface. Two-way ANOVA indicates that the degree of carbachol suppression was not significantly different between the three YF mutants and wild-type cortactin (P = 0.16). These effects are not caused by a lack of protein expression of the mutant forms of cortactin or disruption in their ability to bind to Kv1.2 (Fig. 7 b and c). Therefore, although Y466 and Y482 help maintain steady-state levels of surface Kv1.2, they do not significantly affect M1 receptor-induced endocytosis.

The Actin Regulatory Domains Within Cortactin Have a Key Role in the Trafficking of Kv1.2.

Actin dynamics can be regulated by phosphorylation of cortactin tyrosines Y421, Y466, or Y482 (6). Therefore, we hypothesized that the mutation-induced effects shown in Fig. 7 may indicate a role for cortactin-mediated actin dynamics in regulating Kv1.2. Replacement of FL cortactin with a mutant version (Δ4 cortactin), which cannot bind F-actin, caused a 33% reduction in surface Kv1.2 relative to FL cortactin (n = 38, P < 0.001) (Fig. 8). Furthermore, carbachol produced a 28% decrease in surface Kv1.2 in Δ4 cortactin-expressing cells (n = 34, P = 0.006). Analysis by two-way ANOVA indicates that the percent suppression induced by carbachol in Δ4 cortactin-expressing cells is not significantly different from that produced in wild-type cortactin-expressing cells (P = 0.34). These effects are not caused by a lack of Δ4 cortactin protein expression or a disruption of its ability to bind to Kv1.2 (Fig. 8 b and c). Therefore, cortactin's F-actin-binding functionality has a role in regulating Kv1.2 surface levels, but is not directly involved in receptor-mediated suppression. Cortactin also is a key activator of the Arp2/3 complex (14), recently hypothesized to contribute to early stages of endocytosis (15). The W22A point mutation completely ablates cortactin's ability to activate Arp2/3 (14). Replacement of FL cortactin with W22A cortactin had no significant effect on surface Kv1.2 levels relative to FL cortactin (n = 36, P = 0.16). Strikingly, surface Kv1.2 levels between saline- and carbachol-treated W22A cortactin-expressing cells were not significantly different (n = 34, P = 0.1). This effect was lost in cells not cotransfected with cortactin RNAi (1.00 ± 0.08, n = 12 vs. 0.67 ± 0.05, n = 13; P < 0.001). Further, these effects are not caused by a lack of W22A cortactin protein expression or a disruption of its ability to bind to Kv1.2 (Fig. 8 b and c). Therefore, cortactin's ability to activate the Arp2/3 complex is required for M1 receptor-mediated Kv1.2 endocytosis, but has no role in maintaining surface Kv1.2 levels.

Tissue Culture.

HEK293 cells stably expressing Kv1.2, Kvβ2, and the M1 muscarinic acetylcholine receptor under Zeocin and G418 selection, respectively, were cultured as reported (2).

Flow Cytometery.

Surface Kv1.2 levels were measured by flow cytometry as described (2).

Electrophysiology.

Xenopus oocytes were injected with 1 ng of cRNA 48 h before recording. When indicated, oocytes were injected with an HPLC-purified antisense DNA oligonucleotide directed against Xenopus cortactin: C*A*G*T*CGTCTGTCTCA*T*C*T*G (asterisks indicate phosphorothioated bases).

FRET.

CFP cortactin, YFP-Kv1.2, and Kvβ2 were transiently expressed in HEK cells stably expressing the M1 muscarinic acetylcholine receptor. Where indicated, cells were fixed in 4% formaldehyde as described for standard epiflouresence microscopy (2).

Acceptor Photobleach.

Fixed cells were analyzed on a LSM 510 meta-scanning laser confocal microscope (Carl Zeiss, Thornwood, NY). Excitation of CFP cortactin was with a 405-nm laser diode. CFP emission was detected with a 470- to 500-nm band-pass filter. Excitation of YFP-Kv1.2 was with an argon laser with a 488-nm line. YFP emission intensities were detected with a 505- to 550-nm band-pass filter. FRET efficiency was quantified following Karpova et al. (17).

Donor Photobleach.

Fixed cells were imaged on the Zeiss LSM 510 meta-scanning laser confocal microscope. Excitation and emission parameters for YFP and CFP were as described for acceptor photobleaching.

MicroFRET.

Cells were fixed as before or imaged live in an environmental control chamber heated to 37°C and imaged on the DeltaVision deconvolution-reconstruction microscopy system (Applied Precision). Spectral bleed-through in FRET images was corrected by using the method of Gordon et al. (12).

GST Pull-Down Assays.

GST-fusion protein pull-down experiments for Fig. 4 were performed by using purified recombinant cortactin as described (1). Pull-downs were done at minimum in triplicate.

Statistics.

Detection of statistical difference between two independent measurements was by one-way Student's t test. To avoid inflated type I error, comparisons between three or more groups were made with the Kruskal–Wallis test. Comparison of percent changes between pairs of independent measurements was by two-way ANOVA. Sample populations were considered to be significantly different at P ≤ 0.05.