Selective phosphorylation modulates the PIP2 sensitivity of the CaM-SK channel complex

Abstract

Phosphatidylinositol bisphosphate (PIP₂) regulates the activities of many membrane proteins including ion channels through direct interactions. However, the affinity of PIP₂ is so high for some channel proteins that its physiological role as a modulator has been questioned. Here we show that PIP₂ is an important cofactor for activation of small conductance Ca²⁺-activated potassium channels (SK) by Ca²⁺-bound calmodulin (CaM). Removal of the endogenous PIP₂ inhibits SK channels. The PIP₂-binding site resides at the interface of CaM and the SK C-terminus. We further demonstrate that the affinity of PIP₂ for its target proteins can be regulated by cellular signaling. Phosphorylation of CaM T79, located adjacent to the PIP₂-binding site, by Casein Kinase 2 reduces the affinity of PIP₂ for the CaM-SK channel complex by altering the dynamic interactions among amino acid residues surrounding the PIP₂-binding site. This effect of CaM phosphorylation promotes greater channel inhibition by G-protein-mediated hydrolysis of PIP₂.

Phosphatidylinositol phosphates (PIPs) are minor acidic phospholipids found in the inner leaflet of the cell plasma membrane (~1% of total phospholipid pool)¹. PIPs play a vital role in cellular signaling, by their direct interaction with membrane proteins^1–8. In addition, PIP metabolites, such as IP₃ and diacylglycerol (DAG), are intracellular second messengers, which activate additional signaling cascades and regulate cellular activities¹. Through direct interactions, PIP lipids, particularly PI(4,5)P₂, regulate functions of many plasma membrane proteins, including the activities of different ion channels, such as Kv, Kir, KCNQ, and Cav^1,3. However, in the field of PIP₂ cell biology, an unsettled issue is that the affinity of PIP₂ can be so high for some of its target proteins, given the typical levels of PIP₂ in the cell plasma membrane, that the physiological role of PIP₂ as a modulator has been questioned^1,4,9. In such cases, it remains unclear whether cellular signaling can regulate the affinity of PIP₂ for its target proteins.

Small- and intermediate-conductance Ca²⁺-activated K⁺ channels (SK and IK) are widely expressed in excitable tissues, including the central nervous system (CNS) and the cardiovascular system^10–14. They play pivotal roles in regulating membrane excitability by Ca²⁺. In CNS, activation of SK channels generates the afterhyperpolarization (AHP) and dampens firing of action potentials, and thus contributes to Ca²⁺ regulation of neuronal excitability, dendritic integration, synaptic transmission and plasticity, and learning and memory formation^{10,12,14–22}. The functional importance of SK/IK channels is further demonstrated by their potential involvement in certain diseases^23–27. In the SK channel family, four genes have been identified, KCNN1 for KCa2.1 channels (SK1), KCNN2 for KCa2.2 (SK2), KCNN3 for KCa2.3 (SK3) and KCNN4 for KCa3.1 (IK)^12,14. Spliced variants exist for each of the four SK genes, such as SK2-b, which is less sensitive to Ca²⁺ for its activation²⁸.

Structurally, SK channels look similar to voltage-gated K⁺ channels (Kv), with four subunits forming a tetramer and each subunit having six transmembrane segments (S1 – S6). However, SK channels are activated exclusively by Ca²⁺-bound calmodulin (CaM)²⁹. CaM, constitutively tethered to the CaM binding domain (CaMBD) at the channel C-terminus, serves as the high-affinity Ca²⁺ sensor. Our recent structural data show that the channel segment (R396 – M412), which connects S6 to the CaMBD, is an intrinsically disordered fragment (IDF) that plays a unique role in coupling binding of Ca²⁺ to CaM and opening of SK channels³⁰. SK channels are subjected to regulation by intracellular second messengers. Phosphorylation of CaM, when complexed with the channel, at T79 by Casein Kinase 2 (CK2) inhibits SK channels^16,31–33. Protein phosphatase 2A (PP2A) reverses the effect of CK2. Both CK2 and PP2A interact directly with SK channels and the combined activities of CK2 and PP2A determine the phosphorylation status at T79, and thus the inhibition of SK channels³¹. It is not clear how phosphorylation of CaM at T79 inhibits SK channels^14,16,31,32. Furthermore, it remains unknown whether SK channel activity is regulated directly by PIP₂.

Here we report that PIP₂ is an important cofactor for Ca²⁺-dependent activation of SK channels. Removal of the endogenous PIP₂ results in inhibition of SK channels, and such inhibition can be reversed by application of synthetic PIP₂ derivatives. Using computational and experimental tools, we have identified the PIP₂-binding site, which includes amino acids from both CaM and the IDF of SK channels. We have further established that CK2 phosphorylation of CaM T79, which is located in the vicinity of the PIP₂-binding site, will weaken the interaction between PIP₂ and the CaM-SK channel complex by altering the dynamic interactions among amino acid residues around the PIP₂ binding site. The reduced affinity of PIP₂ for the CaM-SK channel complex greatly facilitates inhibition of SK channels by Gq-mediated hydrolysis of PIP₂.

RESULTS

PIP₂ is a cofactor for activation of SK channels by Ca²⁺

To test whether SK channels are regulated by PIP₂, SK2-a channels (WT) together with CaM (WT) were expressed in TsA cells and their currents were recorded using inside-out membrane patches^30,34. The patch membrane was exposed to Ca²⁺ and other reagents sequentially, as depicted in Supplementary Fig. 1 (Supplementary Results). SK2-a channels were activated by a bath solution containing 2 μM Ca²⁺, which should produce near-maximal channel open probability^29,34. In the presence of 2 μM Ca²⁺, application of poly-lysine (poly-K), at 900 μg/ml, inhibited the SK2 currents by sequestering the endogenous PIP₂ (Supplementary Fig. 1)³⁵. The current amplitude decreased rapidly (Figs. 1a and 1b). Once the endogenous PIP₂ was scavenged by poly-K, SK2 currents could not be recovered even when higher Ca²⁺ concentrations were applied (not shown).

Figure 1. PIP₂ is an essential cofactor for activation of SK2 channels by Ca²⁺.

a, Representative time course of inhibition of the SK2 channel activity by depletion of the endogenous PIP₂ upon application of poly-lysine (poly-K, 900 μg/ml). b, Raw current ramp traces before and after application of poly-lysine. Both traces are from the same patch in panel a (indicated by arrows). c, Restoration of the SK2 channel activity after application of exogenous diC₈-PIP₂ at the concentrations indicated once the endogenous PIP₂ was depleted by poly-K. d, Dose-response curve of channel reactivation by exogenous diC₈-PIP₂ (n = 6). The bath solutions contained 2 μM Ca²⁺. Data in panel d represent mean values ± s.e.m.

We next tested whether inhibition of the SK currents by poly-K could be reversed by exogenous PIP₂, such as the water soluble synthetic PIP₂ derivative, diC₈-PIP₂36. After inhibition of SK2-a channels by poly-K reached the steady state (Supplementary Fig. 1), increasing concentrations of diC₈-PIP₂ were added to the bath solution (Fig. 1c). Maximal currents were obtained at diC₈-PIP₂ concentrations between 30 to 100 μM (Fig. 1c). Fitting of the data to a dose-response equation yielded an EC₅₀ of 1.9 ± 0.22 μM (n = 6, Ca²⁺ = 2 μM, Fig. 1d). The EC₅₀ for activation of SK2 channels by diC₈-PIP₂ is several-fold higher than that for activation of GIRK channels³⁷. Collectively, results of inhibition of SK2 by poly-K depletion of the endogenous PIP₂ and subsequent restoration of the channel activity by application of the exogenous diC₈-PIP₂ suggested that PIP₂ was required for the normal functions of SK channels.

The PIP₂-binding site in the CaM-SK2 complex

Typically, PIP₂-binding sites consist of positively-charged amino acid residues and are located near the channel gating machinery^1,38. Analysis of our structural model, which includes both CaM and the SK2 channel fragment³⁰, showed the surface electrostatic potential distribution (Supplementary Fig. 2). While the protein complex surface bore primarily negative charges (red), there were two loci, near the IDF, which showed clusters of positive charges (blue). Molecular docking revealed that the PIP₂ head group could fit well into these two positively charged loci, interacting with residues from both proteins, most notably the positively charged residues that interacted with the phosphate groups of PIP₂, including K402 and K405 of SK2 and R74 and K77 of CaM (Fig. 2a). K402 and K405 are highly conserved among different members of the SK channel family³⁰. To test whether the putative PIP₂-binding site represented the functional PIP₂-binding site, we mutated the positively charged residues predicted to interact with the PIP₂ head group and examined the effects of mutations on modulation of SK channels by PIP₂. K402N, K405N, R74N and K77N, all tested at 2 μM Ca²⁺, significantly increased the EC₅₀ for activation of SK2 channels by diC₈-PIP₂ to 14.2 ± 1.93 μM (n = 4, p < 0.001), 15.2 ± 2.14 μM (n = 4, p < 0.001), 9.6 ± 0.66 μM (n = 5, p < 0.001) and 5.7 ± 0.75 μM (n = 5, p < 0.001), respectively (a 3.0 – 8.0-fold increase, Fig. 2b). In contrast, K397N, located outside of the putative PIP₂-binding site (Fig. 2a), had little effect (EC₅₀ = 2.0 ± 0.52 μM, n = 6, p = 0.823).

Figure 2. The PIP₂-binding site resides at the interface of the CaM-SK complex.

a, Molecular docking reveals that the putative PIP₂-binding site includes the CaM linker and the IDF of SK2, with the following positively charged residues interacting with the PIP₂ head group: R74 and K77 of CaM and K402 and K405 of the channel. K397 is located outside of the putative PIP₂-binding site, ~17 Å from P5 of the PIP₂ head group. b, Mutations of the four positively charged residues decrease the sensitivity of SK2 activation by diC₈-PIP₂ (a 3.0 – 8.0-fold increase in the EC₅₀ for diC₈-PIP₂). The effects of mutations were measured in the bath solution containing 2 μM Ca²⁺. The numbers of experiments are shown in the parentheses of the bar graph. c, Mutations of the four positively charged residues have much less direct impact on Ca²⁺-dependent activation of the SK2 channels (n = 5 – 8, a 1.4 – 2.0-fold increase in the EC₅₀ for Ca²⁺). At 2 μM Ca²⁺, the channel open probability is > 0.98 for all mutants (dashed vertical line). d, There exists a clear correlation between the Ca²⁺ dependent channel activation of SK channels and the apparent PIP₂ affinity for the CaM-SK2 complex. Data in panels b, c and d represent mean values ± s.e.m.

While the results indicated that these selective mutants might disrupt the interaction of PIP₂ with the CaM-SK2 complex at the putative PIP₂-binding site, mutations could inhibit SK channels via allosteric effects due to altered energetics of channel opening. To address the potential allosteric effects by mutations, we measured the Ca²⁺-depedent channel activation for each mutant. The EC₅₀ for Ca²⁺ dependent channel activation was increased from 0.32 ± 0.03 μM (WT, n = 8) to 0.63 ± 0.04 μM (K402N, n = 5, p < 0.001), 0.62 ± 0.04 μM (K405N, n = 6, p < 0.001), 0.56 ± 0.03 μM (R74N, n = 5, p < 0.001), and 0.44 ± 0.03 μM (K77N, n = 5, p = 0.019), while the EC₅₀ for K397N remained unchanged (0.34 ± 0.02 μM, n = 5, p = 0.626, Fig. 2c). However, at 2 μM Ca²⁺, the Ca²⁺ concentration used to measure the PIP₂ effects, the channel open probability was 0.981 – 0.997 for WT as well as the mutants (dashed vertical line, Fig. 2c), demonstrating that both WT and mutants had achieved the maximal channel open probability at 2 μM Ca²⁺, even though mutations caused a minor shift of the Ca²⁺-dependent channel activation to the right (a 1.4 – 2.0-fold increase in the EC₅₀, Fig. 2c). Thus, K402N, K405N, R74N and K77N primarily decreased the affinity of PIP₂ for the CaM-SK2 complex at 2 μM Ca²⁺, establishing that K402N, K405N, R74N and K77N were part of the functional PIP₂-binding site. Furthermore, there was a strong correlation between Ca²⁺ dependent channel activation and the apparent PIP₂ affinity for the CaM-SK2 complex (Fig. 2d). Collectively, these results demonstrated that the putative PIP₂-binding site, identified by our docking simulations, controled the PIP₂ activation of SK2 channels and suggested that these interactions serve to modulate the Ca²⁺-dependence of channel activation. However, at the EC₅₀ of 1.9 μM for diC₈-PIP₂, the affinity of the endogenous PIP₂ for the SK channel complex is likely to be much higher, casting doubts of whether decreases in PIP₂ levels by hydrolysis could produce any meaningful modulation of SK channels under physiological conditions^1,4,9. In the next series of experiments, we attempted to address whether the affinity of PIP₂ for the SK channel complex might be subjected to modulation by cellular signaling.

Phosphorylation boosts inhibition of SK2 by PIP₂ removal

Our structural model showed that a unique phosphorylation site, T79 of CaM, known to be phosphorylated by CK2 to inhibit SK channels, was located close to the PIP₂-binding site (Fig. 2a). Studies have shown that the phosphorylation status at CaM T79 is determined by the combined activities of CK2 and PP2A determine^16,31–33. Accordingly, two channel mutants were created, SK2(K121A), known to prevent activation of CK2, and SK2(AQAA), known to disrupt the interaction between the SK channel and PP2A, thus stopping dephosphorylation by PP2A at T79³¹. We used these two mutants to ask whether phosphorylation of the CaM-SK complex by CK2 might affect modulation of SK channels by PIP₂. Membrane patches of SK(K121A) were exposed to 10 μM TBB, which compromises activation of the endogenous CK2, to render the CaM-SK2 complex in a dephosphorylated state at T79 (dephospho-, Fig. 3), while membrane patches of SK(AQAA) were exposed to the active catalytic domain of CK2 (CK2α) to create a permanently phosphorylated CaM-SK2 complex at T79 (phospho-, Fig. 3). Both mutants then underwent PIP₂ depletion by poly-K, followed by application of diC₈-PIP₂.

Figure 3. Phosphorylation of CaM T79 by CK2 enhances inhibition of SK channels by PIP₂.

a, After phosphorylation by the active catalytic domain of CK2 (CK2α), SK2 (AQAA) channels (phospho-) become less responsive to diC₈-PIP₂, resulting in a shift of the dose-response curve for diC₈-PIP₂ to the right (n = 6), compared with the SK2 (K121A) channels (dephospho-, n = 5). b, Phosphorylation by CK2α increases the off-rate for dissociation of the endogenous PIP₂ from the CaM-SK2 complex. The apparent off-rate was measured from the time course of exposure of SK2 channels to application of poly-K. The numbers of experiments are shown in the parentheses of the bar graph. c and d, In vitro phosphorylation assays, detected by Pro-Q (c) or western blotting (WB) using an antibody against CK2 specific substrate (CK2S-Ab, d), show phosphorylation of CaM by CK2α in different conditions, as indicated. While 200 μM diC₈-PIP₂ produces moderate inhibition of phosphorylation by CK2α, the presence of Ca²⁺ virtually inhibits the CK2 activity. Coomassie brilliant blue (CBB) staining showed that the same amount of CaM and the SK2 fragment (SK2 Fx) were present in all lanes (panel c). Likewise, CaM Ab blotting showed the same amount of CaM was present in all lanes (panel d). e and f. Quantification of the in vitro phosphorylation assays as shown in c and d. Gel images in c and d were quantified by normalizing the fluorescence intensity to that obtained at 0 Ca²⁺ without diC₈-PIP₂. Data in panels a, b, e, and f represent mean values ± s.e.m.

After depletion of the endogenous PIP₂, application of diC₈-PIP₂ could restore the channel activities for both the dephosphorylated and phosphorylated CaM-SK2 complex (Supplementary Fig. 3). However, the phosphorylated channel complex required much higher diC₈-PIP₂ concentrations to achieve maximal activation. The EC₅₀ for channel activation by diC₈-PIP₂ was increased from 1.24 ± 0.30 μM for the dephosphorylated CaM-SK2 (n = 5) to 18.04 ± 0.91 μM for the phosphorylated CaM-SK2 (n = 6), a 14.5-fold increase (p < 0.001, Fig. 3a). Furthermore, phosphorylation by CK2 significantly increased the off-rate for dissociation of the endogenous PIP₂ from the CaM-SK2 complex, from 0.11 ± 0.02 s⁻¹ for the dephosphorylated (n = 5) to 0.27 ± 0.03 s⁻¹ for the phosphorylated (n = 7, p = 0.002, Fig. 3b). Such an increase in the off-rate greatly facilitated the inhibition of the phosphorylated SK channels by sequestration of endogenous PIP₂ (Supplementary Fig. 4). In vitro phosphorylation assays were performed to confirm that use of CK2α had resulted phosphorylation of the CaM at T79, using purified CaM and the SK2 fragment. The phosphorylated CaM was readily detectable by both Pro-Q (Fig. 3c and Supplementary Fig. 5a) and an antibody against the phosphorylated CK2 substrates (CK2S-Ab, Fig. 3d and Supplementary Fig. 5b). We further tested whether phosphorylation of the CaM-SK2 fragment by CK2α could be affected by Ca²⁺ or PIP₂ which activate SK channels. Without Ca²⁺, the presence of 200 μM diC₈-PIP₂ produced a moderate inhibition of phosphorylation by CK2α (down to 59.0 ± 3.32% of the control, n = 6, by Pro-Q, Fig. 3e; or 77.1 ± 3.30% of the control, n = 6, by CK2S-Ab, Fig. 3f). Ca²⁺, on the other hand, dramatically reduced phosphorylation at T79, down to 11.7 ± 1.66% of the control (n = 6, by Pro-Q) or 18.7 ± 3.47% of the control (n = 6, by CK2S-Ab). In the presence of Ca²⁺, 200 μM diC₈-PIP₂ did not cause any further reduction of phosphorylation at CaM T79. The effects by Ca²⁺ are consistent with previous reports³¹.

Previous studies have shown that inhibition of SK channels by CK2 can be mimicked by the phosphomimetic CaM mutant, T79D, and blocked by the CaM mutant, T79A, in mammalian cells including neurons^16,31,32. The effects of CK2 phosphorylation on modulation of SK channels by PIP₂ (Fig. 3) led us to test whether the phosphomimetic CaM mutant T79D could achieve the same effect on modulation of SK channels by PIP₂. Coexpressed with CaM(T79D), SK2 became less responsive to diC₈-PIP₂ for its activation (Supplementary Fig. 6). The dose-response curve for activation of SK2 by diC₈-PIP₂ was significantly shifted to the right, with an 11.6-fold increase in the EC₅₀ for diC₈-PIP₂, from 1.9 ± 0.22 μM (n = 6) to 22.1 ± 2.22 μM (n = 4, p < 0.001, Fig. 4a) at 2 μM Ca²⁺. Consistent with this observation, the apparent off-rate for dissociation of the native PIP₂ from the CaM-SK2 complex was significantly increased, from 0.086 ± 0.020 s⁻¹ (n = 6) to 0.33 ± 0.028 s⁻¹ (n = 6, p < 0.001, Fig. 4b). The EC₅₀ for Ca²⁺-dependent activation was increased as well, from 0.32 ± 0.03 μM (WT, n = 8) to 1.2 ± 0.19 μM (T79D, n = 4, 3.8-fold, p < 0.001)³⁰, with a channel open probability of 0.88 at 2 μM Ca²⁺. In addition to confirming the effects of CK2 phosphorylation on modulation of SK channels by PIP₂, the CaM T79D mutant would be used as an effective tool to mimic phosphorylation for later experiments.

Figure 4. The phosphomimetic CaM mutant T79D mimics the effect of phosphorylation by CK2.

a, The T79D mutant, which mimicks CK2 phosphorylation of the CaM-SK2 complex, increases the EC₅₀ for dose-dependent activation of SK2 by diC₈-PIP₂ (n = 6 for WT, and n = 4 for T79D). b, T79D significantly accelerates the dissociation of PIP₂ from the CaM-SK2 complex. The off-rate, obtained from fitting the time course of depletion of the endogenous PIP₂ by poly-K to an exponential function, is significantly increased. The numbers of experiments are shown in the parentheses of the bar graph. Data represent mean values ± s.e.m.

Phosphorylation reduces the PIP₂ affinity for CaM-SK2

In the next set of experiments, we attempted to address how phosphorylation at CaM T79, located in the vicinity of the PIP₂-binding site (Fig. 2a), could facilitate modulation of SK channels by PIP₂. First, we examined whether phosphorylation of CaM T79 might have changed the global conformation of the complex of CaM(T79D)-SK2 fragment by X-ray crystallography. CaM(T79D) was co-crystallized with the SK2 channel fragment in the presence of Ca²⁺ (Supplementary Table 1). Structure determination showed no significant changes in the global structure of the CaM(T79D)-SK2 fragment complex (blue, Fig. 5a) compared to that of the WT complex (salmon), with rmsd of 0.37 Å and 0.30 Å for CaM and the SK2 fragment, respectively. Since the crystallographic structure represents the energetically most stable conformation, we next performed molecular dynamic (MD) simulations, using the PIP₂-docked structural model (Fig. 2a), to further explore any potential changes in the dynamic interactions among amino acid residues locally.

Figure 5. Phosphorylation of T79 of CaM disrupts the interaction between PIP₂ and its binding site and reduces the apparent affinity of PIP₂ for the CaM-SK2 complex.

a, T79D mutation does not alter the global conformation of the CaM-SK channel fragment protein complex. The crystal structure of CaM(T79D)-SK channel fragment complex (blue) was determined in the presence of Ca²⁺, and compared to that of the WT (salmon). b, In MD simulations, CaM(T79D) increases the distance between K77 of the PIP₂-binding site and P5 of the PIP₂ head group. The distance distribution histogram shows that T79D significantly increases the mean K77 – P5 distance, from 3.6 Å (WT) to 5.8 and 8.6 Å (T79D). c, T79D increases the distance between K402 of the PIP₂-binding site and the P4 of the PIP₂ head group. The distance distribution histogram shows that T79D significantly increases the mean K402 – P4 distance, from 3.5 Å (WT) to 8.0 Å (T79D). d and e, Snapshots from the MD simulations showing the interactions of the PIP₂ phosphates to K77 of CaM, and K402 and K405 of SK2 (WT, d) and the effect of CaM(T79D) on these interactions (e).

The distance distribution histogram of the MD simulations showed that the T79D mutation reduced the mean K77 – T79 (or D79) distances from 8.5 and 11.5 Å to 2.8 and 8.0 Å, suggesting formation of a new intramolecular salt bridge between D79 and K77, a key residue of the PIP₂-binding site (Fig. 2a and Supplementary Fig. 7). Further analysis showed that the T79D mutation changed how the PIP₂ head group interacted with the PIP₂-binding site. For instance, there was an increase in the distance between K77 and P5 of the PIP₂ head group, with the mean P5 – K77 distance increased from 3.6 Å (WT) to 5.8 Å and 8.6 Å (T79D, Fig. 5b). Similarly, the distance between K402 and P4 of the PIP₂ head group was also increased significantly, from 3.5 Å (WT) to 8.0 Å (T79D, Fig. 5c). Such changes were demonstrated by individual frames of MD simulations which showed that the distances for P5 – K77 and P4 – K402 were increased from 3.8 Å and 3.8 Å (WT, Fig. 5d) to 9.6 Å and 8.3 Å (T79D, Fig. 5e) respectively. Furthermore, the T79D mutation reduced the calculated interaction energy between PIP₂ and its binding site in the CaM-SK2 complex from -19.20 kcal/mol (WT) to -14.99 kcal/mol (T79D), consistent with the experimental data (Figs. 3a, 3b, 4a and 4b). Thus, the MD simulation results suggested that phosphorylation of CaM T79 by CK2 could effectively weaken the interaction between PIP₂ and the CaM-SK2 complex, a plausible mechanism to explain the experimental data (Figs. 3a, 3b, 4a and 4b).

Phosphorylation enhances Gq-mediated inhibition of SK2

The high apparent affinity of the SK2 channel-CaM complex for PIP₂ suggests that it might serve to protect against inhibition from signals that decrease PIP₂, such as PIP₂ hydrolysis. If this were true, we wondered whether phosphorylation of T79 of CaM would facilitate modulation of SK2 channels by PIP₂ hydrolysis stimulated by Gq protein-coupled receptors. We reconstituted the classical phospholipase C (PLC) mediated PIP₂ hydrolysis using a cell line stably expressing the human muscarinic receptor (hM1R)³⁹. Cell-attached patches were used, and the intracellular Ca²⁺ was elevated by application of a Ca²⁺ ionophore, A23187 (15 μM) in the presence of 5 μM Ca²⁺ in the bath solution. To eliminate potential phosphorylation of the WT CaM by endogenous CK2³¹, we coexpressed SK2 with CaM(T79C) as the control. We have previously shown that CaM(T79C) does not have any effect on Ca²⁺ dependent activation of SK2 channels³⁰. Application of acetylcholine (ACh, 10 μM) led to a moderate inhibition of the SK2 current (Fig. 6a, filled circles, and Supplementary Fig. 8a). In contrast, coexpression of CaM(T79D) greatly increased the effectiveness of ACh in its inhibition of the SK2 current under identical conditions used with CaM(T79C) (Fig. 6a, open circles, and Supplementary Fig. 8b). Likewise, coexpression of T79D increased significantly the off-rate for dissociation of the native PIP₂ from the CaM-SK2 complex, from 0.019 ± 0.0019 s⁻¹ (n = 5, T79C) to 0.029 ± 0.0038 s⁻¹ (n = 4, T79D, p = 0.002, Fig. 6b). Consequently, the remaining current, after application of ACh, is significantly reduced, from 70.3 ± 5.2% (T79C, n = 5), compared to the current prior to ACh application, to 25.1 ± 2.84% (T79D, n = 4, p < 0.001, Fig. 6c). It is interesting to note that NS309, a positive SK channel modulator⁴⁰, was able to overcome the inhibitory effect by PIP₂ hydrolysis (Fig. 6a).

Figure 6. CaM T79 phosphorylation facilitates modulation of SK2 by PIP₂ hydrolysis triggered by stimulation of a G-protein coupled receptor, hMR1.

a, Activation of hMR1 inhibits the SK2 channel and such inhibition is significantly enhanced by the phosphomimetic mutation, T79D. Activation of hMR1 by ACh triggers Gq-PLC-mediated hydrolysis of PIP₂. T79C was used instead of WT to eliminate potential phosphorylation of WT CaM at T79. NS309, a positive SK channel modulator, or a 0 Ca²⁺ solution (no activity) were used to ensure that the currents recorded were indeed due to activation of SK channels by Ca²⁺. b, T79D significantly increases the off-rate, obtained from fitting the time course of reduction of the current amplitudes after application of ACh. c, T79D promotes inhibition of the SK2 channel activity by ACh-induced PIP₂-hydrolysis. The remaining current due to ACh-induced inhibition is reduced from 70.3 ± 5.2% (n = 5, T79C) to 25.1 ± 2.84% (n = 4, T79D, p < 0.001). d, T79D facilitates inhibition of SK2 by Ci-VSP-triggered dephosphorylation of PIP₂. Activation of Ci-VSP by membrane depolarization (pre-pulse) dephosphorylates PIP₂ and inhibits the SK2 channel activity. This inhibitory effect is significantly increased by the T79D mutation, and the remaining current after membrane depolarization is reduced from 88.7 ± 1.6% (n = 4, no VSP) to 68.6 ± 4.7% (n = 10, T79C + VSP) and 23.1 ± 5.0% (n = 5, T79D + VSP). Data in panels b, c and d represent mean values ± s.e.m. and the numbers of experiments are shown in the parentheses of the bar graph.

ACh-induced PIP₂ hydrolysis is known to activate the intracellular signaling cascades, such as generation of DAG which activates protein kinase C (PKC)¹. To exclude the potential involvement of the PIP₂ metabolites in inhibition of SK2 channels in our reconstituted system, we turned to the voltage-sensitive phosphatase (Ci-VSP), which dephosphorylates PI(4,5)P₂ to PI(4)P^39,41,42. Ci-VSP was coexpressed with CaM and SK2-a, and its activity was elicited by membrane depolarization. Activation of Ci-VSP reduced the remaining current of the T79D containing complex significantly more than the T79C containing complex (from 68.6 ± 4.67%, T79C, n = 10 to 23.1 ± 5.02%, n = 5, p < 0.001, Fig. 6d). The results suggested that the phosphomimetic T79D mutation directly facilitated the inhibition of the SK channel activity by PLC-induced PIP₂ depletion. The results also lent further support of our conclusion that PIP₂ was required for Ca²⁺-dependent activation of SK channels.

DISCUSSION

Like many other ion channels^1,3, PIP₂ is a necessary cofactor for the normal functions of SK channels, such as Ca²⁺-dependent activation of SK. Removal of PIP₂, e.g. by Gq-mediated PIP₂ hydrolysis, has led to inhibition of the channel activities (Figs. 1 and 6). Such inhibition is not caused by PI metabolites (Figs. 1 and 6d). Unique to this case is that the PIP₂-binding site is located at the interface of the CaM-SK2 complex (Fig. 2a), including R74 and K77 of the CaM linker region and K402 and K405 within the IDF of the SK channel C-terminus. Sequence comparison shows that the amino acid residues that contribute to the PIP₂-binding site are highly conserved among all SK channel family members³⁰. IDF, a channel fragment before the CaMBD, is known to couple Ca²⁺ binding to CaM and channel opening³⁰. Thus, like many other ion channels, the location of the PIP₂-binding site in the CaM-SK complex is close to the channel gating machinery^1,38.

Our results further show that inhibition of SK channels by PIP₂ hydrolysis can be enhanced via phosphorylation of CaM T79 by CK2. CaM T79 is located at the vicinity of the PIP₂-binding site (Fig. 2a). Its phosphorylation by CK2, mimicked by the phosphomimetic mutant T79D, can significantly reduce the affinity of PIP₂ for its binding site and therefore enhance inhibitory effects via PIP₂ hydrolysis (Figs. 3 and 4). While phosphorylation of T79 by CK2 does not produce significant global conformational changes (Fig. 5a), it significantly alters the dynamic interactions among amino acid residues surrounding the PIP₂-binding site, such as formation of a new salt bridge, and ultimately weakens the interaction between the PIP₂ head group and its binding site (Fig. 5). Proteins are known to exhibit dynamic behaviors, such as small-scale movements at the level of individual amino acid residues and larger-scale movements between domains at different time scales⁴³. Changes in the dynamic interactions among amino acid residues and formation of new salt bridges are the molecular mechanisms for activation of G-protein coupled receptors (GPCRs) upon binding of their ligands^44–46.

CK2 and PP2A are known to form a signaling complex with SK channels and regulate the channel activity through phosphorylation and/or dephosphorylation³¹. Before this study it was unknown how phosphorylation of CaM at T79 inhibited SK channels^16,31,32, although our recent work demonstrated that the phosphomimetic T79D mutation did not affect the Ca²⁺ sensitivity for formation of the CaM-CaMBD complex²⁹. Based on the results of this study, we propose that the primary consequence of phosphorylation of CaM T79 by CK2 is to reduce the affinity of PIP₂ for the CaM-SK channel complex and promote the inhibitory effects by PIP₂ hydrolysis (Fig. 6). SK channels are known to be inhibited by activation of Gq-coupled GPCPs, such as cholinergic and adrenergic receptors^15–17,33. Inhibition of SK channels in neurons by such a mechanism contributes to regulation of synaptic plasticity and memory formation^{15–18,47,48,49} . The results of our study provide a prime example of how physiological stimuli, such as protein phosphorylation, can directly modulate the affinity of PIP₂ for its target proteins, such as ion channels, and make them more susceptible to modulation by stimuli that deplete PIP₂.

METHODS

Methods and any associated references are available in the online version of this paper.

Supplementary information is available in the online version of the paper.