Modulation of cardiac Ca_V1.2 channels by dihydropyridine and phosphatase inhibitor requires Ser-1142 in the domain III pore loop

Abstract

Dihydropyridine-sensitive, voltage-activated calcium channels respond to membrane depolarization with two distinct modes of activity: short bursts of very short openings (mode 1) or repetitive openings of much longer duration (mode 2). Here we show that both the dihydropyridine, BayK8644 (BayK), and the inhibitor of Ser/Thr protein phosphatases, okadaic acid, have identical effects on the gating of the recombinant cardiac calcium channel, Ca_V1.2 (α₁C). Each produced identical mode 2 gating in cell-attached patches, and each prevented rundown of channel activity when the membrane patch was excised into ATP-free solutions. These effects required Ser or Thr at position 1142 in the domain III pore loop between transmembrane segments S5 and S6, where dihydropyridines bind to the channel. Mutation of Ser-1142 to Ala or Cys produced channels with very low activity that could not be modulated by either BayK or okadaic acid. A molecular model of Ca_V1.2 indicates that Ser-1142 is unlikely to be phosphorylated, and thus we conclude that BayK binding stabilizes mode 2 gating allosterically by either protecting a phospho Ser/Thr on the α₁C subunit or mimicking phosphorylation at that site.

Voltage-activated Ca_V1.2 calcium channels (α₁C) in cardiac myocytes are one of the primary targets for clinical treatments of human heart disease (1), and dihydropyridines (DHPs) have become one of the most widely studied class of calcium channel blockers (2). However, DHP-sensitive calcium channels are expressed throughout the brain, smooth muscle, and endocrine tissues with roles in neuronal plasticity, blood pressure, and hormone secretion (3–5). Therefore, understanding the molecular mechanism of DHP action could have clinical benefits for selective treatment of human cardiac disorders.

DHPs modulate the Ca_V1.2 channels by stabilizing one of three recognizable patterns of response to membrane depolarization labeled “modes” 0–2 (6). Normally when the membrane is depolarized, the channels exhibit repeated short bursts of brief (<1-ms) channel openings (mode 1). More rarely channels respond with continuous openings of much longer duration (mode 2). Interspersed among the periods of activity are periods of inactivity when individual channels show no response to physiological depolarization for several seconds (mode 0). DHP antagonists such as nimodipine increase the frequency of mode 0 behavior, whereas agonists such as BayK8644 (BayK) increase mode 2 activity (7). Although they produce opposite effects on activity, the DHPs bind competitively to the same site in the pore-forming α₁C subunit. The DHP-binding site has been confirmed experimentally by the transfer of DHP binding to other neuronal calcium channels through mutation of nine amino acids in the S5 and S6 transmembrane helices of domain III and the S6 helix of domain IV (2, 8). However, the structural basis for the three modes of gating and their modulation by DHPs remain to be elucidated.

Reversible protein phosphorylation also regulates the same three modes of gating as DHPs (9, 10). Ca_V1.2 channels are complexed in the membrane with both protein kinases and phosphatases through scaffolding adaptor proteins (11). Single-channel studies of native channels in cell-free membranes (12–14) have established that endogenous phosphatases shift channels to mode 0 behavior in the absence of ATP, and endogenous cAMP-dependent protein kinase (PKA) shifts the channels back into mode 1 activity when ATP is added back. In contrast, calmodulin-dependent protein kinase activity increases mode 2 activity (9, 15). However, the specific substrates of the kinases that produce changes in activity have not been identified conclusively (10, 11). The stimulation of mode 1 activity by PKA is reversed by several Ser/Thr protein phosphatases including PP1, PP2A, and PP2B (16–19). Inhibition of protein phosphatases also increases mode 2 activity of native channels in intact cells (20–22), but this has not been confirmed with recombinant channels. Here we show that BayK and okadaic acid (OA), an inhibitor of Ser/Thr protein phosphatases (17, 23), have identical effects on the single-channel activity. Both drugs stimulate mode 2 gating of recombinant cardiac Ca_V1.2 channels and prevent rundown in cell-free patches through a mechanism that requires Ser-1142.

Materials and Methods

Cell Line and Transfection.

Wild-type (wt) or mutated cardiac Ca_V1.2 α₁ subunits (α₁C) were transiently expressed in a hamster kidney cell line (BHK6), which has been engineered to stably express rabbit β_1a and α₂/δ subunits (24). BHK6 cells were grown in DMEM with 10% FBS/50 mg/liter streptomycin/31 mg/liter penicillin/640 mg/liter G418 for selection. Cells were plated on 12-mm Ø glass cover slips (Deutsche Spielglas, Carolina Biological Supply) and transfected with the channel constructs and a plasmid-encoding GFP at a DNA ratio of 6:1 by using Lipofectamine 2000 (Invitrogen). Fluorescent cells were used for recordings between 36 and 72 h after transfection.

Site-Directed Mutations in α₁C Subunit.

Mutations in rabbit α₁C (GenBank accession no. X15539) were introduced by PCR with the QuikChange XL site-directed mutagenesis kit (Stratagene). Mutations at Ser-1142 (S1142A, S1142C, and S1142T) and Thr-1066 (T1066Y) were performed on a 443-bp BglII (3,074)–BglII (3,517) fragment containing the codon to be mutated. The mutated fragment then was placed into a BglII-digested 3.4-kb SacI (952)–EcoRV (4,353) fragment of α₁C [in pBluescript SK(+)], and the mutated SacI–EcoRV α₁C fragment was inserted into SacI–EcoRV-digested full-length α₁C cDNA in the expression vector pKNH (25). The mutation of Ser-1928 to Ala was constructed by the same strategy, replacing a 2.4-kb EcoRV (4,353)–HindIII (6,727) fragment of the full-length α₁C cDNA in pKNH with the suitably mutated EcoRV–HindIII fragment. Each clone was sequenced in both directions to confirm mutation and integrity.

Electrophysiological Recordings and Analysis.

Calcium channels were recorded in the cell-attached and inside-out configuration by using an EPC9 patch-clamp amplifier/interface (HEKA Electronics, Lambrecht/Pfalz, Germany) and a Power Macintosh (Apple) computer running PULSE software (HEKA Electronics) for generation of pulses and data acquisition. All experiments were performed at room temperature (20–24°C). Patch pipettes (1.5–2.5 MΩ resistance) were made from type 7052 glass (Garner Glass, Claremont, CA) and coated with Sylgard 184 (Dow-Corning) to improve the signal-to-noise ratio. Single-channel barium currents were recorded with 100 mM BaCl₂/10 mM Hepes with the pH adjusted to 7.4 with tetraethylammonium hydroxide in the patch pipette. The bath solution [high K⁺ to achieve a defined membrane potential (V_m) close to 0 mV] initially consisted of 145 mM KCl, 2 mM MgCl₂, and 10 mM Hepes with the pH adjusted to 7.4 with KOH. Before excision (inside-out), we switched to a CsCl solution with 140 mM CsCl, 1 mM MgCl₂, 1 mM 1,2-bis(2-aminophenoxy)ethane-N,N,N′, N′-tetraacetic acid (BAPTA) and 10 mM Hepes with the pH adjusted to 7.4 with CsOH. OA (LC Laboratories) and (±)-BayK (Calbiochem) were added from ≥1,000-fold concentrated DMSO solutions. ATP was added as Mg salt and cAMP as Na salt (both from Calbiochem).

Whole-cell currents (Fig. 1A) were recorded in nystatin-perforated patches with the CsCl solution described above in the pipette and a bath solution with 106 mM tetraethylammonium-Cl, 1 mM MgCl₂, 10 mM CaCl₂, 20 mM glucose, and 10 mM Hepes with the pH adjusted to 7.4 with tetraethylammonium-OH. Single-channel currents were recorded at a sampling frequency of 20 KHz and filtered at 1.5 or 2 KHz. TAC and TACFIT software (Bruxton, Seattle) were used to idealize the single-channel data and construct and fit open-time histograms. The data for dwell-time histograms were collected during voltage steps from −50 to 0 mV of 200- to 1,000-ms duration or during 1-min periods of constant depolarization. Although channels showed some steady-state inactivation during longer depolarization, the open-time constants and, more importantly, the proportion of long to short openings was the same as in histograms from the shorter voltage jumps. Differences between data sets (paired or unpaired, as appropriate) were evaluated with a Student's t test, and values of P ≤ 0.05 were considered significant.

Figure 1.

Ca_v1.2 α₁ subunits expressed in BHK6 cells retain properties of native L-type calcium channels. (A) Whole-cell calcium current recorded in a nystatin-perforated patch is potentiated by forskolin, and the current is blocked by the DHP antagonist nimodipine. (B) Single-channel barium currents from a cell-attached patch shows mainly brief (<1-ms) openings and some long openings (trace 4). The voltage protocols for A and B are shown at the bottom of each panel. (C) Channel open-time histograms at 0 mV are described best by the sum of two open times.

Calcium Channel-Structure Modeling.

The alignment of transmembrane helices M1 and M2 from the KcsA potassium channel structure (26) with the corresponding transmembrane segments S5 and S6 from the four homologous rabbit α₁C repeats was taken from Huber et al. (8). Side-chain positions for the model were subsequently generated from the alignment by using the SCWRL program (27). The P loops forming the selectivity filter of the KcsA channel were aligned with the P-loop regions of repeats I–IV by using the CLUSTALW program (28). Because the backbone coordinates of the KcsA selectivity filter are incompatible with the larger apparent pore size in L-type calcium channels (29), the P-loop coordinates were generated separately for each repeat by using the KcsA P-loop backbone and substituting the side chains as above. Then the four P loops were docked rigidly into their respective positions in the KcsA tetramer such that the glycines adjacent to the conserved glutamates that provide calcium coordination in the pore were arranged symmetrically around the pore. This required minor adjustments to the backbone N-terminal to the glutamates to avoid steric clashes between the gating helices. The resulting P-loop tetramer is asymmetric, with glutamates I and III binding the calcium from below and those from repeats II and IV binding the calcium from above. Docking and concomitant side-chain adjustment was performed by using the Swiss PDB VIEWER software (30). The R-(+) antagonist and S-(−) agonist enantiomers of BayK were generated and optimized by using the GAUSSIAN 98 program (Gaussian, Pittsburgh). The optimization was carried out at the Hartree Fock 3-21G basis-set level.

Results

The pore-forming α₁C subunit of the rabbit cardiac Ca_V1.2 calcium channel was transiently expressed in a hamster kidney cell line, BHK6, which has been engineered to stably express the auxiliary β₁, and α₂/δ subunits of L-type cardiac calcium channels (24). The resulting channels have all the major characteristics of native cardiac calcium channels including modulation by DHPs and PKA (31). In intact cells the pharmacologically isolated calcium current is increased rapidly by forskolin, which stimulates adenylyl cyclase, and the resulting current is inhibited completely by 1 μM nimodipine (Fig. 1A). At the single-channel level, unitary barium currents recorded from on-cell membrane patches spontaneously exhibit all three modes of L-type calcium channel gating (Fig. 1B) including two distinct populations of open times with mean durations of <0.2 and >2 ms (Fig. 1C). Application of 200 nM BayK significantly increased the mean duration of the shorter (mode 1) events from 0.2 to 0.7 ms and increased the proportion of longer duration (mode 2) events from 7% to 33% (Table 1). Less than 24 h after transfection, individual channels exhibited considerably more variability in their biophysical properties and susceptibility to regulation, and thus we confined our observations to a period between 36 and 72 h after transfection.

Table 1.

Values for the open-time constants (τ_o1 and τ_o2) and their relative proportion (%τ_o1 and %τ_o2), which were determined by fitting channel open-time histograms at 0 mV as shown in Fig. 1C

Mutation	τo1, μs	τo2, ms	% τo1	% τo2
wt	171  ± 15 (12)	5.6  ± 1.7 (7)	93  ± 2.6 (12)	7.4  ± 2.8 (12)
wt + BayK	664  ± 96 (9)*	6.3  ± 0.5 (9)	65.6  ± 5.8 (9)*	33.2  ± 6.1 (9)*
wt + OA	269  ± 30 (4)*	6.6  ± 2.6 (4)	66.9  ± 9.7 (4)*	33.1  ± 9.6 (4)*
S1142A	204  ± 24 (8)	1.5  ± 0.5 (7)	95.6  ± 1.4 (8)	4.3  ± 1.4 (8)
S1142A + BayK	191  ± 49 (3)	3.0  ± 1.7 (3)	91.9  ± 4.7 (3)	8.1  ± 4.7 (3)
S1142A + OA	208  ± 13 (2)	1.1  ± 0.4 (2)	96.0  ± 1.0 (2)	4.0  ± 0.3 (2)
S1142C	143  ± 9 (2)	0.6  ± 0.1 (2)	94.6  ± 0.9 (2)	5.3  ± 0.9 (2)
S1142C + BayK	137  ± 8 (3)	0.3  ± 0.3 (3)	96.0  ± 1.3 (3)	4.0  ± 1.3 (3)
S1142T	164  ± 22 (9)	2.8 (1)	99.8  ± 0.2 (9)	0.2  ± 0.2 (9)
S1142T + BayK	157  ± 45 (4)	3.2  ± 1.1 (4)	77.1  ± 7.4 (4)*	22.9  ± 7.4 (4)*

Discrepancies in the number of patches (in parentheses) between τ_o1 and % τ_o1 are due to the fact that not all open-time histograms had a second exponential component (see Materials and Methods). The values marked by asterisks indicate statistically significant differences (P ≤ 0.05) between the untreated and BayK or OA (both 0.5 or 1 μM)-treated patches. Values are given as mean ± SE. 

In untreated cells, the activity of the recombinant channels declined rapidly when the patches were excised into an ATP-free saline solution (Fig. 2A, i-o), which does not support protein phosphorylation. Within 1–5 min the channels entered mode 0 and, except for the occasional isolated opening of very brief duration, stopped responding to membrane depolarization. Channel activity was partially restored in a fraction of the patches (3 of 5) by adding cAMP and ATP (Fig. 2A), but even with the addition of a purified preparation of PKA catalytic subunit, activity never reached >10% of its original level before excision, and longer duration openings (mode 2 activity) were not observed (data not shown). This inability to reconstitute channel activity fully by exogenous PKA could reflect the absence of other essential cytoplasmic factors (32, 33) or a hyperactive protein phosphatase that remains closely associated with the channel in cell-free patches. To test the latter possibility we bathed the cells in OA for 5–10 min before excision. At concentrations ≥500 nM, OA not only prevented the loss of activity in ATP-free solutions (Fig. 2B), but it also shifted the channels to mode 2 gating as effectively as BayK (Fig. 2 C and D). Concentrations of OA ≤100 nM did not have this effect (data not shown). Although similar results have been reported for native channels (20–22), this shows that mode 2 gating in response to protein phosphatase inhibition can also be recapitulated with recombinant channel proteins in intact cells.

Figure 2.

Ca_v1.2 channel activity quickly subsides after excision of the patch into solution without ATP/cAMP, but activity is maintained in the presence of OA or BayK. (A) Rundown of activity is reversed only partially after the addition of ATP/cAMP (0.5 mM/0.1 mM) and 0.2 μM BayK. Activity is maintained if the cell is pretreated (5′) with OA. (B) Diary plots of single-channel activity during 200-ms voltage steps to 0 mV. Recordings were started cell-attached in high-potassium saline, and patches were subsequently excised into CsCl solution. i/o denotes time of excision into the inside-out configuration. (A Right and B Right) Representative channel openings observed cell-attached (c-a) and after excision (i-o). (C) Amplitude histograms of the absolute value of channel open times determined from activity plots such as the one shown in Fig. 1C. BayK and OA increase the mean duration of the short openings but not of the long openings. (D) Frequency histograms of the relative percentage of short and long openings. BayK and OA increase the frequency of long openings to the same extent. (C and D) The first bar in each group is the control value in the absence of BayK or OA. The error bars are SEMs. *, P ≤ 0.05 and **, P < 0.01. The collated data are shown in Table 1.

In contrast to PKA, BayK by itself was completely ineffective at restoring any activity after rundown (data not shown). Even after activity was restored partially by the addition of ATP and cAMP, BayK did not elicit mode 2 gating (Fig. 2A). However, when cells were pretreated with BayK before excision, intermittent mode 2 activity persisted for the lifetime of the patch in the absence of exogenous ATP (Fig. 3A). This effect of BayK on rundown also required direct interaction with the pore-forming subunit, because the effectiveness of BayK at preventing rundown was eliminated by mutation of Thr-1066 to Tyr (Fig. 3B). The T1066Y mutation was shown previously to prevent agonist binding (34). Thus, BayK is just as effective as OA at preventing the loss of activity, which raises the question of whether BayK produces its effects on both rundown and mode 2 gating by inhibiting dephosphorylation of the channel. Although the calcium/calmodulin-dependent protein kinase II (CAMKII) is the only kinase that has been reported to produce mode 2 gating of the cardiac calcium channel (9, 15), the substrate for CAMKII stimulation of mode 2 activity has not been identified. In contrast, phosphorylation of Ser-1928 in the cytoplasmic carboxyl terminus of the α₁C subunit is required for stimulation of the channel by PKA in other systems (35, 36). However, neither of the effects of BayK in our expression system were blocked by mutating Ser-1928 to Ala (Fig. 3C). This result is consistent with earlier reports that truncation of the carboxyl terminus of the cardiac a1C subunit did not alter responsiveness to BayK or PKA (37, 38). Therefore, we considered Ser-1142 in the domain III pore loop between transmembrane segments S5 and S6 of the α₁ subunit.

Figure 3.

The effect of BayK on rundown of channel activity when in cell-free patches without ATP. (A) BayK application before patch excision maintains channel activity after excision into ATP-free bath solution. (B) T1066Y, which prevents BayK binding to the α₁C subunit, also prevents BayK from maintaining activity. (C) S1928A, which prevents stimulation of Ca_V1.2 by PKA in other systems, does not prevent the effects of BayK on rundown or mode 2 gating.

Ser-1142 was unexpectedly identified as a critical determinant of DHP action on mammalian channels in a recent study of the ascidian orthologue of Ca_V1.2, in which Ser-1142 is replaced by Ala (39). Although the ascidian orthologue contained all the other residues that were believed to be sufficient for DHP sensitivity (2), it was completely insensitive to both agonists and antagonists (39). We found that S1142A reduces overall activity to <5% of wt even in intact cells, indicative of predominantly mode 0 gating (Fig. 4A). In addition it blocks the effect of both OA and BayK to stabilize mode 2 gating (Table 1), providing further support for the hypothesis that BayK and OA produce mode 2 gating through a common mechanism. Cys was also ineffective at substituting for Ser at position 1142 (S1142C, Table 1). Only Thr could substitute for Ser (S1142T) in supporting mode 2 activity (Table 1), even though higher doses of BayK (≥500 nM) were required to produce an equivalent change in gating (Fig. 4B).

Figure 4.

BayK promotes mode 2 gating in the S1142T but not the S1142A mutant. Open-time histograms (Upper) and representative single-channel activity during continuous recording at 0 mV (Lower) of S1142A (A) and S1142T (B) in 1 μM BayK.

To facilitate thinking about the role of Ser-1142 in DHP action, the recent high-resolution crystal structure (26) of the KcsA potassium channel was used as a structural template for homology modeling of the BayK-binding pocket in rabbit cardiac calcium channels (Fig. 5). The alignment of KcsA transmembrane helices M1 and M2 with the corresponding transmembrane segments S5 and S6 from the four homologous rabbit α₁C repeats was taken from Huber et al. (8), but the pore loop placement was modified as described in Materials and Methods. This allowed us to predict the orientation of Ser-1142 relative to the binding site formed by the transmembrane helices IIIS5, IIIS6, and IVS6. Despite the difference in selectivity and regulation of gating, the sequence of the pore loops is remarkably conserved between KcsA and Ca_V1.2. Based on this alignment, Ser-1142 is located near the end of the gating helix that preceeds the selectivity filter (Fig. 5).

Figure 5.

Model of the DHP BayK docked to the putative DHP-binding site of Ca_V1.2. Transmembrane helices S5 and S6 of domains I (pink), II (cyan), III (red), and IV (white) are shown. Calcium ion (blue) is held by a ring of conserved glutamic acid carboxylates of 6 Å in diameter. BayK (green) shown between helices IIIS5, IIIS6, and IVS6. Ser-1142 of helix IIIS5 is highlighted in yellow, and Thr-1066 of IIIS5 is highlighted in light blue.

Substitution of either Ala or Cys for Ser-1142 produced channels with very low open probability that could not be shifted to mode 2 by either BayK or OA. Only Thr substituted for Ser at this position (Fig. 4B and Table 1), which indicates the importance of a hydroxyl group at this position. Therefore, either Ser-1142 is essential for DHP action because it hydrogen-bonds to the DHP or to an adjacent structure that participates in DHP binding such as the IIIS6 helix, or Ser-1142 is phosphorylated. Although phosphorylation of Ser-1142 seems unlikely, if it were phosphorylated by a closely associated protein kinase, the model predicts that DHP antagonists would not fit in the binding pocket. This prediction also applies to the enantiomer of BayK, which acts as a weak antagonist (40). Only the agonist enantiomer of BayK, could be docked successfully into the site when the phosphorylated form of Ser-1142 was incorporated into the model (not shown). Furthermore, once the DHP was bound, it would occlude any further access to Ser-1142 by protein phosphatases. Therefore, if BayK physically blocked phosphatase access to the phosphorylation site controlling mode 2 gating, it would provide a simple, direct, testable explanation for the effects of BayK on rundown and the effects of OA on mode 2 gating.

Discussion

We have studied the modulation of the recombinant, DHP-sensitive, voltage-activated calcium channel, Ca_V1.2, from rabbit cardiac myocytes (GenBank accession no. X15539) in a cell line (BHK6) where Ca_V1.2 retains all the major properties of native cardiac channels (31), including spontaneous transitions between all three modes of gating, stimulation by PKA, and modulation by DHPs (Fig. 1). We have used this system to test the interrelationship between modulation by DHPs and phosphorylation. Surprisingly BayK, the DHP agonist (7), and OA, an inhibitor of Ser/Thr protein phosphatase activity (23), had exactly the same effects on channel activity. Both drugs promote mode 2 activity in cell-attached patches. Both drugs maintain activity in cell-free patches if they are applied before excision into ATP-free solutions. The effects of both drugs were lost in S1142A or S1142C mutant channels, and neither drug could produce mode 2 gating on wt channels in cell-free patches that had been resurrected after rundown by the readdition of ATP.

The simplest interpretation of these results would be that BayK, similar to OA, inhibits dephosphorylation of a site that is phosphorylated by a cytoplasmic kinase, which does not remain closely associated with the channel protein in cell-free patches. However, unlike OA, which binds directly to Ser/Thr phosphatases (23), BayK binds directly to the α₁ subunit of Ca_V1.2 (2), and blocking DHP binding by mutating T1066 (34) prevents the effect of BayK on channel lifetime and on rundown (Fig. 3B). Therefore, BayK does not prevent rundown by binding to some other site on the channel or to another protein. An alternative view is that BayK, instead of blocking dephosphorylation, allosterically mimics the effect of phosphorylation.

Ser-1142 on the gating helix of the domain III pore loop between transmembrane helices S5 and S6 (39) is required for the channel to show mode 2 activity. BayK binding could prevent dephosphorylation of the α₁C subunit directly if the substrate of the kinase were Ser-1142. However, Ser-1142 is not in a recognized consensus sequence for PKA or CAMKII, nor is it easy to imagine how a cytoplasmic kinase would get access to this site. It is also unlikely that the only function of Ser-1142 is contributing directly to the DHP-binding site because S1142A reduces spontaneous mode 2 activity in the absence of DHP and prevents OA from stimulating mode 2 activity (Table 1). Thus, Ser-1142 seems most likely to play a structural role in allosterically transducing both DHP binding and channel phosphorylation into mode 2 gating. However, the latter possibility requires a structure in the protein that can link both BayK binding and phosphorylation to Ser-1142. The S6 helices of domain III and IV would be obvious candidates for such a structure, so it is interesting to ask whether there are any more plausible potential phosphorylation sites adjacent to either of these helices. Ser-1173 at the beginning of IIIS6 might be a candidate, but it is predicted to be extracellular. On the other hand, Ser-1627 and Ser-1700 in the cytoplasmic carboxyl terminus following IVS6 have both been reported to be phosphorylated in bovine cardiac tissue (41), and they are located in a region of the channel protein that was shown to be required for modulation by BayK (37).

In contrast, Ser-1928, which is required for stimulation by PKA in other heterologous expression systems (10, 11), is not essential for the effects we see. There have been reports of interactions between cAMP-dependent phosphorylation and DHP action (42, 43), but these might reflect separate effects on mode 1 and mode 2 gating. The calcium/calmodulin-dependent protein kinase, CAMKII, has been implicated by many investigators in cardiac calcium channel regulation (10, 11). If CAMKII were required for mode 2 gating, it might explain why neither BayK nor OA restore mode 2 gating after rundown in ATP-free solutions. Drug-induced disruption of the cytoskeleton in intact cells eliminated the stimulation of mode 2 gating by CAMKII (44), an effect that could be prevented by inhibition of protein phosphatases (33). Patch excision into ATP-free solutions could have similar effects on cytoskeletal integrity, and thus CAMKII remains the best candidate for stimulating mode 2 activity in intact cells when phosphatases are inhibited.

The endogenous phosphatase that reverses mode 2 activity also needs to be identified. Stimulation of channel activity by PKA is reversed by several Ser/Thr protein phosphatases including PP1, PP2A, and PP2B (16–19), and purified PP2A was reported to reverse mode 2 activity in cell-free patches from smooth muscle (22). In our study of recombinant channels and in previous studies of native channels (20–22) concentrations of OA ≥500 nM were required to produce mode 2 gating. PP2B (calcineurin) is the only Ser/Thr protein phosphatase with an inhibition constant for OA in this range (23). Nevertheless, inhibition constants vary widely depending on substrates, conditions, and the source of the enzyme that might be associated with different regulatory subunits in different tissues. Thus, PP1 from cardiac myocytes is also relatively insensitive to OA (17), but other potent PP1 inhibitors do not promote mode 2 activity (21), and more specific inhibitors of PP2B remain to be tested on intact cells.

Regardless of the identity of the kinase and phosphatase that regulate mode 2 transitions or the site in the channel complex at which they act, we have demonstrated a link between modulation by DHPs and modulation by phosphorylation. In view of the increasing number of channels and signaling proteins of many varieties, which are complexed in the cell with the kinases and phosphatases that regulate their activity, altering susceptibility of the drug target to phosphorylation and dephosphorylation by closely associated kinases or phosphatases might be a more general mechanism for drug action (45–47) than previously anticipated.

Abbreviations

DHP

dihydropyridine

BayK

BayK8644

PKA

cAMP-dependent protein kinase

OA

okadaic acid

wt

wild-type Ca_v1.2 α₁

CAMKII

calcium/calmodulin-dependent protein kinase II