The Phosphatidylinositol 3-Phosphate Phosphatase Myotubularin- Related Protein 6 (MTMR6) Is a Negative Regulator of the Ca²⁺-Activated K⁺ Channel K_Ca3.1

Abstract

Myotubularins (MTMs) belong to a large subfamily of phosphatases that dephosphorylate the 3′ position of phosphatidylinositol 3-phosphate [PI(3)P] and PI(3,5)P₂. MTM1 is mutated in X-linked myotubular myopathy, and MTMR2 and MTMR13 are mutated in Charcot-Marie-Tooth syndrome. However, little is known about the general mechanism(s) whereby MTMs are regulated or the specific biological processes regulated by the different MTMs. We identified a Ca²⁺-activated K channel, K_Ca3.1 (also known as KCa4, IKCa1, hIK1, or SK4), that specifically interacts with the MTMR6 subfamily of MTMs via coiled coil (CC) domains on both proteins. Overexpression of MTMR6 inhibited K_Ca3.1 channel activity, and this inhibition required MTMR6's CC and phosphatase domains. This inhibition is specific; MTM1, a closely related MTM, did not inhibit K_Ca3.1. However, a chimeric MTM1 in which the MTM1 CC domain was swapped for the MTMR6 CC domain inhibited K_Ca3.1, indicating that MTM CC domains are sufficient to confer target specificity. K_Ca3.1 was also inhibited by the PI(3) kinase inhibitors LY294002 and wortmannin, and this inhibition was rescued by the addition of PI(3)P, but not other phosphoinositides, to the patch pipette solution. PI(3)P also rescued the inhibition of K_Ca3.1 by MTMR6 overexpression. These data, when taken together, indicate that K_Ca3.1 is regulated by PI(3)P and that MTMR6 inhibits K_Ca3.1 by dephosphorylating the 3′ position of PI(3)P, possibly leading to decreased PI(3)P in lipid microdomains adjacent to K_Ca3.1. K_Ca3.1 plays important roles in controlling proliferation by T cells, vascular smooth muscle cells, and some cancer cell lines. Thus, our findings not only provide unique insights into the regulation of K_Ca3.1 channel activity but also raise the possibility that MTMs play important roles in the negative regulation of T cells and in conditions associated with pathological cell proliferation, such as cancer and atherosclerosis.

Myotubularins (MTM) are a large family of evolutionarily conserved lipid phosphatases (PT) that specifically dephosphorylate the 3′ position of phosphatidylinositol 3-phosphate [PI(3)P] and PI(3,5)P₂ (28, 39). Fourteen MTMs in mammalian cells have been identified, and they can be divided into six subgroups based on sequence alignment and phylogenetic comparison. Members of one of these subgroups lack phosphatase activity due to a mutation in a critical residue within the phosphatase domain. MTM1, the founding member of this gene family, is mutated in X-linked myotubular myopathy, and MTMR2 and MTMR13 are mutated in Charcot-Marie-Tooth (CMT) syndrome type 4B (3, 30, 37). In addition to containing a phosphatase domain, most MTMs are composed of a GRAM domain which may mediate association of MTMs with membranes, a Rac-induced localization domain which mediates the association with Rac-induced membrane ruffles, and a C-terminal coiled coil (CC) domain (10, 28, 29, 39). Recent data have indicated that the MTM CC domains mediate specific heterodimerization between a MTM (PT active) and a MTM mutant for PT activity (PT inactive) and that this interaction is critical for MTM regulation, at least in part, by up-regulating PT activity (24, 33, 34).

The genetic finding that loss of MTM function leads to specific phenotypes in mammals and Caenorhabditis elegans indicates that different MTM members, even within the same subgroup, are not functionally redundant with one another. This has led to the suggestion that different MTMs regulate specific subcellular pools of PI(3)P and PI(3,5)P₂ whose function is to couple to specific downstream signaling pathways. However, as of yet, the specific downstream events regulated by different MTMs are not known. To gain insight into the specific functions and targets of MTMs, we screened a C. elegans yeast two-hybrid cDNA library for proteins that interacted with the C. elegans MTMR6 (ceMTMR6) CC domain.

MATERIALS AND METHODS

Yeast two-hybrid system.

The ProQuest two-hybrid system (Invitrogen) was used according to the manufacturer's protocol to screen a C. elegans cDNA library (Invitrogen) for proteins that bound the CC domain (amino acids [AA] 508 to 570) of CeMTMR6 (F53A2.8). The CC domains of CeMTMR6, CeMTMR1 (CeY110ATA.5, amino acids 580 to 613), and CeMTMR3 (CeT24A11.1, amino acids 700 to 750) were amplified by PCR and cloned into the vector pDBLeu (Invitrogen). These constructs were then cotransfected with the original MTMR6 CC domain-interacting clone (the C-terminal 126 amino acids of F08A10.1 [AA 781 to 907]), and interaction was tested according to the manufacturer's protocol. The CC domains were determined using a program developed by Lupas et al. (31).

Constructs, cell lines, and cell culture.

Flg-tagged Human K_Ca3.1 (SK4) and Rat K_Ca2.2 (SK2) were cloned into pcDNA3 and transfected into CHO cells using Lipofectamine (Clontech). Geneticin-resistant colonies were isolated, and homogeneous expression of K_Ca3.1 (CHO-K_Ca3.1) and K_Ca2.2 (CHO-K_Ca2.2) was confirmed by immunofluorescence using anti-Flg antibodies. Green fluorescent protein (GFP)-tagged human MTMR6, MTMR8, and MTMR1 were generated by amplifying the various cDNAs by PCR and then cloning in frame into the vector pEGFP-C1 to generate an amino-terminal tagged GFP fusion protein. MTMR6 mutant for PT activity [MTMR6-PT(dead)] in which the catalytic cysteine at position 336 was mutated to serine was generated using a Transformer mutagenesis kit (Clontech) and MTMR6 deleted of its CC domain was generated by PCR using a 5′ sense primer containing the amino-terminal ATG and a 3′ antisense primer 5′ to amino acid 514. MTM1/6CC was generated by overlapping PCR to generate a chimeric molecule that consisted of AA 1 to 546 of MTM1 and AA 512 to 560 of MTMR6. Accession numbers for the various constructs are as follows: AAP3609 for human SK4, NP_062187 for rat SK2, CAI39897 for human MTMR6, NP_000243 for human MTM1, and CAI43025 for human MTMR8.

Immunoprecipitation, Western blotting, and immunofluorescence.

HEK293 cells were transfected with GFP-tagged MTMR6, MTMR8, or MTMR6 mutants as described above together with Flg-tagged K_Ca3.1 or Flg-tagged K_Ca2.2. Cell lysis, immunoprecipitation, and Western blotting were performed as previously described (1).

To determine whether overexpression of MTMR6 affects the amount of K_Ca3.1 at the plasma membrane (PM), MTMR6 was transfected with K_Ca3.1 containing an hemagglutinin (HA) epitope tag inserted into the extracellular loops between S3 and S4. Previous studies have demonstrated that K_Ca3.1 with this extracellular tag functions normally and can be specifically detected at the PM by immunofluorescence on nonpermeabilized cells (38). CHO cells were transfected with HA-K_Ca3.1 together with the GFP control or the GFP-tagged MTMR6 wild type (GFP-MTMR6-WT), and cell surface expression was performed on nonpermeabilized cells using antibodies to HA as described previously (38).

Patch clamping.

For electrophysiological studies, CHO-K_Ca3.1 cells were transfected with GFP-tagged constructs as indicated above and plated on 12-mm-diameter coverslips. After 24 to 48 h, a coverslip was positioned in a recording chamber, and patch clamping was performed under one of two experimental conditions that yielded similar results. One set of solutions consisted of a pipette solution containing K⁺-aspartate (110 mM), KCl (30 mM), HEPES (5 mM), MgCl₂ (1.13 mM), EGTA (1 mM), CaCl₂ (0.985 mM; pH 7.2, adjusted with KOH; calculated free Ca²⁺, 10 μM) and a high-K⁺ bath solution (KCl [140 mM], CaCl₂ [2 mM], MgCl₂ [1 mM], HEPES [10 mM], glucose [5 mM; pH 7.4, adjusted with KOH]). The second condition utilized a pipette solution containing: K⁺-gluconate (100 mM), KCl (30 mM), HEPES (10 mM), MgCl₂ (1.15 mM), EGTA (5 mM), CaCl₂ (4.27 mM; pH 7.2 with 1 N KOH; calculated free Ca²⁺, 1 μM), and a bath solution containing NaCl (140 mM), KCl (5 mM), CaCl₂ (1 mM), MgCl₂ (1 mM), HEPES (10 mM), and glucose (10 mM). Patch clamp pipettes had resistances ranging between 2 and 4 MΩ. We performed whole-cell patch clamp recordings at room temperature. Current-voltage (IV) relationships were measured using ramp voltage clamp protocols (at 15-s intervals) from a holding potential of −70 mV to −120 mV, followed by ramp depolarization to +60mV (symmetrical ramp rate of 0.18 mV · ms⁻¹). The current-voltage relationship was obtained by plotting the current during the depolarizing ramp phase as a function of the corresponding voltage. In some experiments, a square step voltage clamp protocol was used (200-ms duration) to assess alterations in kinetics. Membrane currents filtered (−3 dB at 1 kHz) and digitized at 10 kHz (pClamp 9.2 with Digidata 1200 ADC interface; Axon Instruments). Cell capacitance and pipette series resistances were compensated (usually >80%) and were obtained using the “membrane test” function of Clampex. Whole-cell current density was expressed as pA/pF.

Currents recorded from stably transfected cells were verified to be Ca²⁺-activated K⁺ currents by (i) their sensitivity to cytosolic Ca²⁺, (ii) the dependence of the reversal potential on the extracellular K⁺ concentration, (iii) their absence in parental CHO cells, (iv) voltage-independent gating, and (v) pharmacological sensitivities (K_Ca3.1 was inhibited by Tram 34, but not apamin, whereas K_Ca2.2 was inhibited by apamin but not by Tram 34).

To determine whether PI(3)-kinase inhibition affects channel activity, stably transfected cells were treated with the PI(3) kinase [PI(3)K] inhibitor LY294002 (10 μM) or wortmannin (100 nM) 15 min prior to patch clamping, and the IV relationship was performed as described above. Phosphatidylinositols (C₈) were purchased from Echelon Biosciences and used according to the manufacturer's specifications. Briefly, lipids were resuspended in water and sonicated for 20 min and used at concentrations of 100 nM in the pipette solution.

RESULTS

Identification of a calcium-activated K⁺ channel (F08A10.1) that specifically binds the CC domain of ceMTMR6.

One of the C. elegans ceMTMR6 CC domain-interacting proteins we identified in the yeast two-hybrid assay was a calcium-activated K⁺ channel (F08A10.1) that is most similar to the mammalian intermediate K⁺ channel, K_Ca3.1 (also known as KCa4, IKCa1, hIK1, or SK4), but also has significant similarity to other small-conductance (SK) channels (4, 6, 7). The original interacting clone was only a partial cDNA of F08A10.1 and encoded only the C-terminal 126 amino acids of F08A10.1 (AA 781 to 907), of which AA 811 to 895 is strongly predicted to be a CC domain. Thus, the interaction between CeMTMR6 and F08A10.1 is mediated by CC domains on both proteins. In addition, the interaction was specific; the same clone failed to bind the CC domain of both CeMTMR3 and CeMTMR1 (Fig. 1A).

FIG. 1.

(A) Interaction of CeMTMs with F08A10.1 (AA 781 to 907) by two-hybrid analysis. The CC domains of CeMTM6-CC (AA 508 to 570), CeMTMR1 (AA 580 to 613), and CeMTMR3 (AA 700 to 750) were cotransfected with the original MTMR6CC domain-interacting clone F08A10.1. Transfection efficiency was assessed by plating on medium lacking tryptophan and leucine (+His, −3-aminotriazole [−3AT]), and interaction was assessed by growth on plates also lacking histidine and containing 50 mM 3AT (−His, +50 mM 3-aminotriazole). a, bait vector alone; b, bait vector containing ceMTM6CC domain; c, bait vector containing MTM1CC domain; and d, bait vector containing ceMTM3CC domain. (B) Phylogenetic tree of mammalian and C. elegans K⁺ channels related to F08A10.1. (C) Coimmunoprecipitation of hMTMR6 and hMTMR8 with K_Ca3.1 and K_Ca2.2. HEK-293 cells were cotransfected with Flg-tagged K_Ca3.1 (lanes 1 to 10) together with GFP-tagged MTMR8 (lanes 1 and 6), MTMR6-WT (lanes 5 and 9), MTMR6 mutant for PT activity [MTMR6-PT(dead)] (lanes 3 and 7), MTMR6 lacking its CC domain (MTMR6ΔCC) (lanes 4 and 9), or GFP alone (lanes 2 and 10). Lysates were immunoprecipitated with α-Flg antibodies, separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (10%) and then Western blotted with α-GFP (upper) or α-Flg (lower) antibodies. Lanes 1 to 5, lysates; lanes 6 to 10, α-Flg immunoprecipitation. HEK-293 cells were also transfected with GFP-MTMR6, either alone (lane 11) or together with Flg-tagged K_Ca2.2 (lane 12). Immunoprecipitation and Western blotting were performed as described above.

Mammalian MTMR6 family members interact with K_Ca3.1 in cells.

Mammalian cells contain four genes encoding small or intermediate K⁺ channels. Three of these belong to the same subfamily (K_Ca2.1, K_Ca2.2, and K_Ca2.3, also named SK1-3) (4, 6, 7). Within the SK gene family, K_Ca3.1 (also named SK4) diverges from other SK family members, suggesting that SK4 represent a member of a new emerging subfamily of SK genes. Both the small conductance and intermediate conductance K⁺ channels are activated by cytosolic Ca²⁺ and are weakly rectifying, and their gating is voltage independent (4, 19, 22). The C. elegans channel that bound CeMTMR6 (F08A10.1) is the only member of this family found in C. elegans (Fig. 1B) and has the highest homology to K_Ca3.1. While C. elegans has only one MTMR6 subgroup member, mammals have three, MTMR6, 7, and 8 (28, 39, 42).

To determine whether mammalian MTMR6 family members interact with K_Ca3.1 in cells, we tested whether human MTMR6 (hMTMR6) or a related subfamily member, hMTMR8, coimmunoprecipitate with K_Ca3.1. HEK-293 cells were transfected with GFP-tagged MTMR6 or GFP-tagged MTMR8 together with Flg-K_Ca3.1. We found that both GFP-tagged MTMTR6 and GFP-tagged MTMR8 coimmunoprecipitate with Flg antibodies (Fig. 1C, lanes 6 and 9). The interaction between MTMR6 and K_Ca3.1 did not require phosphatase activity since MTMR6-PT(dead), in which the catalytic cysteine at position 336 was changed to serine, also coimmunoprecipitated to a similar degree as the MTMR6-WT (Fig. 1C, compare lanes 7 and 9). In contrast, mutation of the MTMR6 CC domain by truncating MTMR6 at amino acid 514, which is predicted to delete the CC domain (AA 514 to 546), plus the C-terminal AA 547 to 621, markedly diminished binding (lane 8) despite similar levels of expression (Fig. 1C, compare lanes 3, 4, and 5) and similar amounts of immunoprecipitated K_Ca3.1 (lower part of lanes 6 to 10). In similar experiments, we found that K_Ca2.2 (SK2) also coimmunoprecipitated with MTMR6, although to a lesser extent than K_Ca3.1 (Fig. 1C, lanes 11 and 12).

MTMR6 inhibits K_Ca3.1 channel activity.

To determine whether MTMs regulate K_Ca3.1 function, a CHO cell line stably expressing Flg-tagged K_Ca3.1 (CHO-K_Ca3.1) was generated and the effect of overexpressed GFP-tagged MTMR6 on channel activity was determined. The IV relationships from CHO-K_Ca3.1 cells transiently transfected with GFP (as a control) are shown in Fig. 2A. Also shown is the IV curve following application of Tram 34. In contrast to these control cells, K_Ca3.1 currents were markedly reduced when CHO-K_Ca3.1 cells were cotransfected with GFP-MTMR6 (Fig. 2B), without affecting the reversal potential (Fig. 2B) or time dependence of the currents (not shown). The inhibition of K_Ca3.1 by MTMR6 required both MTMR6 CC and PT domains, because mutation of either domain abrogated the inhibitory effects of MTMR6 (Fig. 3). In fact, current activity was reproducibly greater in CHO-K_Ca3.1 cells transfected with MTMR6-PT(dead) than in GFP-transfected cells, suggesting that this mutant may function as dominant negatives to inhibit MTMR6 family members endogenously expressed in these cells. Immunofluorescence studies demonstrated that expression of GFP-MTMR6 or treatment of cells with the PI(3) kinase inhibitor wortmannin (which also inhibits K_Ca3.1; see Fig. 6) did not affect cell surface expression of K_Ca3.1, indicating that altered trafficking of K_Ca3.1 to the plasma membrane does not account for the decrease in current in GFP-MTMR6-transfected cells (Fig. 4).

FIG. 2.

Expression of MTMR6 inhibits K_Ca3.1 but not K_Ca2.2 channel activity. CHO-K_Ca3.1 or CHO-K_Ca2.2 stably expressing cells were transfected with GFP or GFP-MTMR6, and channel activity was recorded using patch clamping. (A) The IV plot of control CHO-K_Ca3.1 cells transfected with GFP as recorded before and after treatment with Tram 34 (1 μM for 3 to 5 min), demonstrating the identity of the channels as K_Ca3.1. The K⁺ current displays the reversal potential of −73 mV (not corrected for the liquid junction potential, calculated to be about 13 mV), which is similar to the calculated reversal potential of −84 mV at an extracellular K⁺ of 5 mM. (B) Currents recorded from a cell transfected with GFP-MTMR6 (before and after Tram 34). (C) IV plots from CHO-K_Ca2.2 stably transfected cells that have been transfected with GFP (as a control) before and after treatment with apamin demonstrating that the current measured is K_Ca2.2. (D) Currents from a CHO-K_Ca2.2 cell transfected with GFP-MTM6 demonstrates that MTM6 does not inhibit K_Ca2.2.

FIG. 3.

MTMR6 inhibition of K_Ca3.1 channel activity requires MTMR6's phosphatase activity and CC domain. K_Ca3.1-stable cells were transfected with GFP as a control, GFP-MTM6-WT, GFP-MTM6-PT(dead), GFP-MTM6ΔCC (MTMR6 lacking the CC domain, AA 1 to 514), GFP-MTM1, or a MTM1/MTMR6CC chimera consisting of MTM1 (AA 1 to 546, lacking the MTM1 CC domain) fused to the C-terminal CC domain of MTMR6 (AA 512 to 560). K_Ca3.1 channel density was calculated by measuring current at +60 mV and dividing by the cell capacitance as a measure of the electrically active membrane surface. Relative K_Ca3.1 channel density is plotted as a percent of GFP control-transfected cells.

FIG. 6.

The PI(3) kinase inhibitors LY294002 and wortmannin inhibit K_Ca3.1, but not K_Ca2.2, channel activity. IV plots of current densities recorded in CHO-K_Ca3.1 cells after treatment with the PI(3)K inhibitor (A) LY294002 or (B) wortmannin for 15 min at 37°C prior to patch clamping. (C) Bar graph summary showing Tram 34-sensitive currents at −120 and +60 mV from control, LY294002-treated, or wortmannin-treated cells. IV plot of currents measured for (D) control CHO-K_Ca2.2 cells or (E) CHO-K_Ca2.2 cells after treatment with wortmannin. (F) Bar graph summary showing apamin-sensitive currents at −120 and +60 mV.

FIG. 4.

MTMR6 does not affect trafficking of K_Ca3.1 to the PM. CHO cells were transfected with exofacial-tagged HA-K_Ca3.1 together with GFP or GFP-MTMR6. Nonpermeabilized, fixed cells were stained with anti-HA antibodies and visualized with Texas Red anti-mouse immunoglobulin G. Similar experiments were performed on GFP-transfected cells treated with wortmannin (GFP + Wort).

We next assessed whether the overexpression of GFP-MTM6 inhibits K_Ca2.2 channel activity (Fig. 2C and D). Using a similar approach to that described for K_Ca3.1, stable CHO-K_Ca2.2 cell lines were generated and the effect of GFP-MTM6 overexpression on K_Ca2.2 channel activity was assessed. A typical K_Ca2.2 current in the GFP control-transfected CHO-K_Ca2.2 cells is shown in Fig. 2C. The specificity of the current was documented using criteria similar to those described above for K_Ca3.1, with the exception that K_Ca2.2 current was inhibited by apamine (Fig. 2C) and not by Tram 34 (data not shown). In contrast to K_Ca3.1, overexpression of GFP-MTMR6 did not inhibit K_Ca2.2 channel activity (Fig. 2D). Thus, MTMR6 does not inhibit all SK channels. Future studies will address whether MTMR6 inhibits the K_Ca2.1 or K_Ca2.3 channel.

The CC domain of MTMR6 is sufficient to enable other MTMs to inhibit K_Ca3.1 channel activity.

In contrast to MTMR6, GFP-MTM1 did not inhibit K_Ca3.1 (Fig. 3). However, a chimeric MTM1 in which the MTM1 CC domain was swapped for the MTMR6 CC (MTM1/6CC)-inhibited K_Ca3.1 to a similar degree as MTMR6, indicating that the CC domain is sufficient to confer target specificity (Fig. 3).

MTMR6 inhibits K_Ca3.1 by decreasing intracellular levels of PI(3)P.

One function of MTMR6 binding to K_Ca3.1 may be to localize MTMR6 activity to the plasma membrane adjacent to K_Ca3.1, thereby enabling MTMR6 to selectively dephosphorylate a local PI(3)P pool in a lipid microdomain adjacent to K_Ca3.1. The finding that overexpression of MTMR6 turns off channel activity predicts that for this model to be correct, intracellular application of PI(3)P or PI(3,5)P₂ should reactivate K_Ca3.1. Consistent with this hypothesis, dialysis of PI(3)P into CHO-K_Ca3.1 cells that have been cotransfected with MTMR6 completely restored K_Ca3.1 channel activity (Fig. 5A to C).

FIG. 5.

PI(3)P rescues inhibition of K_Ca3.1 by MTMR6. (A) IV plots of currents recorded from stably transfected K_Ca3.1 CHO cells (before and after Tram 34). (B) A similar experiment as that shown in panel A, but the pipette contained 100 nM PI(3)P, demonstrating rescue of current by dialyzing PI(3)P intracellularly. (C) Bar graph showing Tram 34-sensitive current densities measured at −120 mV and +60 mV with CHO-K_Ca3.1 expressing GFP-MTMR6 alone or dialyzed with 100 nM PI(3)P.

Phosphoinositide PI(3)P is required for K_Ca3.1 activity.

Our data so far demonstrate a key role for PI(3)P in regulating K_Ca3.1 channel activity. We performed additional experiments with PI(3)K inhibitors, which are predicted to decrease PI(3)P levels and hence, according to our model, inhibit K_Ca3.1 channel activity. Treatment of CHO-K_Ca3.1 cells with the specific PI(3) kinase inhibitors LY294002 or wortmannin for 15 min prior to patch clamping completely abrogated channel activity (Fig. 6A to C). The inhibition of K_Ca3.1 by LY294002 or wortmannin was due to decreased levels of PI(3)P, because channel inhibition by LY294002 or wortmannin could be completely restored by dialysis of 100 nM PI(3)P into the cell through the patch pipette (Fig. 7A to C). The rescue of current activity was specific to PI(3)P, since similar experiments using PI(4)P, PI(5)P, PI(3,5)P₂, PI(3,4)P₂, or PI(3,4,5)P₃ in the pipette did not restore current activity over a time course of 5 min in cells treated with LY294002 or with wortmannin (Fig. 7C and data not shown). Neither wortmannin nor LY294002 inhibited K_Ca2.2 (Fig. 6D to F). These data, together with the finding that the overexpression of MTMR6 did not inhibit K_Ca2.2, indicates that, unlike K_Ca3.1 channels, K_Ca2.2 is not regulated by PI(3)P.

FIG. 7.

PI(3)P is required for K_Ca3.1 channel activity. IV plot of current densities measured for CHO-K_Ca3.1 cells (A) treated with wortmannin and (B) showing rescue of wortmannin-treated cells by dialyzing PI(3)P (100 nM). (C) Bar graph showing Tram 34-sensitive current density measured at −120 mV and +60 mV in wortmannin (100 nM)-treated CHO-K_Ca3.1 cells dialyzed with 100 nM of a number of different C8 phospholipids. Rescue was seen only with intracellular dialysis of PI(3)P.

DISCUSSION

Our finding that MTMR6 inhibits K_Ca3.1 channel activity provides the first identification of a downstream target that is specifically regulated by an MTM family member. Moreover, this is the first demonstration that K_Ca3.1 is specifically regulated by PI(3)P and that K_Ca3.1 requires PI(3)P for channel activity. Our data, when taken together, support a model whereby binding of the CC domain of MTMR6 to the CC domain on K_Ca3.1 leads to selective dephosphorylation of PI(3)P in a lipid microdomain adjacent to K_Ca3.1, resulting in a decrease in K_Ca3.1 channel activity.

An important role for CC domains in regulating specific functions of MTMs has been established. However, unlike the findings reported here, previous studies have focused on the importance of CC domains in mediating specific heterodimerization between a PT-active and PT-inactive MTM family member (24, 33, 34). The importance of this interaction is supported by both biochemical and genetic data. Genetic evidence with humans and C. elegans has shown that mutations in a PT-active and a PT-inactive MTM that heterodimerize give rise to similar phenotypes; mutations in a PT-active (MTMR2) and PT-inactive (MTMR13) MTM give rise to CMT type 4B in humans and mutations in a C. elegans PT-active (CeMTM6), and PT-inactive (CeMTM9) MTM lead to defective fluid phase endocytosis in coelomocytes in C. elegans (3, 8, 37). The finding that heterodimerization increases PT activity has led to the idea that this interaction is critical, at least in part, for optimal PT activity. We now show that MTM CC domains have an additional function, to direct MTMs to specific targets which are regulated by PI(3)P. We propose that at least part of the specificity in MTM function is mediated by the recruitment of different MTM CC domains to specific targets. This model is supported by our finding that while MTM1 does not inhibit K_Ca3.1, a chimeric MTM1 containing the MTMR6 CC domain inhibits K_Ca3.1 channel activities. Thus, once MTM1 is targeted to the channel, other domains on MTM1 that are required to inhibit K_Ca3.1, such as the PT domain, function similarly in MTMR6 and MTM1.

The idea that CC domains specifically target MTMs may also have important implications for understanding the mechanisms whereby loss of MTMs leads to disease. Although altered levels of PI(3)P or PI(3,5)P₂ likely account for clinical disease in X-linked myotubular myopathy and CMT, downstream targets or signaling pathways that are altered due to loss of MTM function are not yet known. The identification of binding partners for the CC domains on these MTMs (MTM1, MTMR2, and MTMR13) may thus provide important insights into the mechanisms whereby loss of MTMs leads to disease. One important question that needs to be addressed is whether a CC domain-interacting protein forms a trimolecular complex with a PT-live and a PT-inactive MTM. With regard to MTMR6, genetic and biochemical studies have indicated that MTMR6 family members heterodimerize specifically with MTMR9 (PT inactive) (33, 42). We have found that while C. elegans that has mutations in either MTMR6 and MTMR9 is defective in fluid phase endocytosis, RNA interference that target MTMR6, but not MTMR9, rescues lethality for vps34 in the C. elegans mutant. These data indicate that MTMR6, under some circumstances, can act independently of MTMR9. We have thus far been unable to identify a function of F53A2.8 for C. elegans or whether F53A2.8 functions in a pathway similar to CeMTMR6; C. elegans fed or injected with RNA interference to F53A2.8 to exhibit any obvious phenotype. Future studies will address whether inhibition of a specific target (K_Ca3.1) by MTMR6 in mammalian cells requires MTMR9.

An important physiological role for K_Ca3.1 channels is to regulate membrane potential and calcium signaling in a variety of cell types. In response to a rise in cytosolic calcium, K_Ca3.1 channels are activated via calmodulin bound to the C terminus of K_Ca3.1, leading to cross-linking of the channel, pore opening, and subsequent K⁺ efflux (11, 36). Efflux of K⁺ maintains the membrane potential at fairly negative values, providing an electrical gradient that facilitates Ca²⁺ influx. Beyond the role of Ca²⁺, little is known about the regulation of K_Ca3.1 channel activity. Kinases, such as protein kinase A, have been proposed to play a role in K_Ca3.1 activation (13, 14), although direct phosphorylation of K_Ca3.1 has not been shown. In addition, arachidonic acid has been shown to inhibit K_Ca3.1 by binding the S5-pore-S6 region (16). We now provide evidence that K_Ca3.1 is activated by PI(3)P and that MTMs negatively regulate K_Ca3.1, possibly by decreasing the concentration of PI(3)P in the plasma membrane. The finding that K_Ca2.2 is not regulated by PI(3)P provides further evidence that while K_Ca2.2 and K_Ca3.1 share conserved amino acid sequences and structure and both require Ca²⁺ for activation, they also have distinct mechanisms for regulation. This is consistent with previous studies that demonstrated that K_Ca3.1, but not K_Ca2.2, is regulated by kinases and arachidonic acid (14, 16).

While a number of channels have been shown to be regulated by phosphoinositides, we believe that this is the first demonstration of specific regulation of a channel by PI(3)P. It is not clear whether PI(3)P activates K_Ca3.1 directly or whether PI(3)P activates K_Ca3.1 indirectly by binding and recruiting a different molecule to a membrane (such as the PM), which then activates K_Ca3.1. Direct regulation of channels, such as the Ca²⁺ transporter NCX1 and voltage-gated K channels, by PIP₂ involves the interaction of PIP₂ with cationically charged inhibitory domains in the cytoplasmic face of the channel, leading to conformational change and channel activation (17, 18). The ability of PIP₂ to activate channels under these circumstances has been attributed to its high physiologic concentrations (>1% of anionic phospholipids), which are far in excess of the concentration of PI(3)P in membranes. Thus, it is unlikely that PI(3)P regulates K_Ca3.1 by a similar mechanism. Rather, we favor the hypothesis that PI(3)P regulates K_Ca3.1 by specifically binding an as-yet-unidentified PH or FYVE domain-containing protein, which in turn regulates K_Ca3.1. In support of this model, we have preliminary evidence that PI(3)P does not activate K_Ca3.1 in isolated membrane patches (data not shown).

The finding that MTMR6 family members downregulate K_Ca3.1 suggests that this family of PI(3)P PTs plays unexpected roles in downregulating biological responses that are mediated by K_Ca3.1. K_Ca3.1 is required for optimal Ca²⁺ influx which, in turn, leads to proliferation by a variety of cells, including T lymphocytes, keratinocytes, vascular smooth muscle cells, and cancer cells (2, 5, 12, 15, 21, 25, 27, 40). In response to stimulation, K_Ca3.1 is upregulated in naïve and central memory T cells and naïve and memory B cells (12, 15, 23, 40, 41). Moreover, K_Ca3.1 is required for proliferation of preactivated naïve T cells and naive B cells, as demonstrated by the finding that the specific K_Ca3.1 inhibitor Tram 34 suppresses proliferation of these T and B cell subsets (15, 41). K_Ca3.1 is also upregulated in proliferative vascular smooth muscle cells, and the inhibition of K_Ca3.1 inhibits angioplasty-induced restenosis in animal models (27). K_Ca3.1 also contributes to the proliferation of a number of different cancer cell lines. Thus, MTM6 family members may function physiologically to negatively regulate immune cells and to inhibit mitogenic responses in conditions associated with pathological proliferation, such as cancer and restenosis after angioplasty. Finally, K_Ca3.1 also permits the passive efflux of Cl⁻, driving water and Na⁺ secretion from epithelia and red blood cells (9, 20, 32). Agonists of K_Ca3.1 have been proposed as therapy for cystic fibrosis, in which enhanced K⁺ secretion via K_Ca3.1 can be a means to promote Cl⁻ secretion. However, only poor K_Ca3.1 agonists exist, and they are ineffective because they result in downregulation of K_Ca3.1 (26, 35). Thus, the inhibition of MTM6 family members may provide a novel target strategy for upregulating K_Ca3.1 activity.