“Slow” Voltage-Dependent Inactivation of CaV2.2 Calcium Channels Is Modulated by the PKC Activator Phorbol 12-Myristate 13-Acetate (PMA)

CaV2.2 (N-type) voltage-gated calcium channels (Ca2+ channels) play key roles in neurons and neuroendocrine cells including the control of cellular excitability, neurotransmitter / hormone secretion, and gene expression. Calcium entry is precisely controlled by channel gating properties including multiple forms of inactivation. “Fast” voltage-dependent inactivation is relatively well-characterized and occurs over the tens-to- hundreds of milliseconds timeframe. Superimposed on this is the molecularly distinct, but poorly understood process of “slow” voltage-dependent inactivation, which develops / recovers over seconds-to-minutes. Protein kinases can modulate “slow” inactivation of sodium channels, but little is known about if/how second messengers control “slow” inactivation of Ca2+ channels. We investigated this using recombinant CaV2.2 channels expressed in HEK293 cells and native CaV2 channels endogenously expressed in adrenal chromaffin cells. The PKC activator phorbol 12-myristate 13-acetate (PMA) dramatically prolonged recovery from “slow” inactivation, but an inactive control (4α-PMA) had no effect. This effect of PMA was prevented by calphostin C, which targets the C1-domain on PKC, but only partially reduced by inhibitors that target the catalytic domain of PKC. The subtype of the channel β-subunit altered the kinetics of inactivation but not the magnitude of slowing produced by PMA. Intracellular GDP-β-S reduced the effect of PMA suggesting a role for G proteins in modulating “slow” inactivation. We postulate that the kinetics of recovery from “slow” inactivation could provide a molecular memory of recent cellular activity and help control CaV2 channel availability, electrical excitability, and neurotransmission in the seconds-to-minutes timeframe.


Introduction
Ca V 2.2 (N-type) voltage-gated calcium channels (Ca 2+ channels) are widely expressed in neurons and neuroendocrine cells where they control neurotransmitter / hormone secretion, gene expression, activation of Ca 2+ -dependent enzymes / ion channels, and a variety of other cellular functions. Calcium entry is precisely regulated by second messengers including G proteins, kinases, and lipid signaling molecules that converge to fine tune Ca V 2 function [1][2][3][4][5][6][7][8]. Ca 2+channel inactivation also controls Ca 2+ entry and thus cellular excitability and short term synaptic plasticity [9][10][11][12]. Ca V 2 inactivation is mediated by distinct calcium or voltage-dependent mechanisms: calcium-dependent inactivation is triggered by "global" elevations of cytosolic [Ca 2+ ] and transduced via calmodulin tethered to the C-terminal tail of the channel [13][14][15][16][17]. Voltage-dependent inactivation is complex, can occur from both the open and closed states of the channel, and exhibits multiple kinetic components in response to sustained or repetitive membrane depolarization. "Fast" inactivation (onset / recovery from tens-hundreds of milliseconds) is thought to involve a "hinged-lid" type pore occlusion by the cytoplasmic loop linking the first and second domains of the α1 subunit (the I-II linker) [11,18,19]. The auxiliary β subunit of the channel binds this I-II linker and modulates the kinetics of "fast" inactivation [20,21], as do heterotrimeric G protein βγ subunits (Gβγ) [22]. An additional inactivated state, revealed by sustained membrane depolarization, displays much slower onset and recovery kinetics (seconds-to-minutes range) [23][24][25]. "Slow" inactivation is also found in potassium and sodium channels and might involve changes in the voltage-sensor domain and/or constriction of the channel pore [26][27][28][29].
Interestingly, protein kinases modulate "slow" inactivation of sodium channels and thereby control neuronal excitability [30,31]. Much less is known about how "slow" inactivation of Ca V 2 channels is regulated. The Ca V β subunit might play a role as an indirect consequence of altered "fast" inactivation [24], and syntaxin has been reported to promote "slow" inactivation of Ca V 2.2 [25,32]. In this study we show for the first time that phorbol ester (PMA) dramatically prolongs recovery of Ca V 2 channels from "slow" inactivation. We postulate this novel regulation could provide a basis for molecular memory of recent cellular activity and help control Ca 2+ channel availability, electrical excitability, and neurotransmission in the seconds-to-minutes timeframe.
Adrenal chromaffin cells: Male bovine adrenal glands were obtained from a local slaughterhouse (C & F Meat Co. Inc., College Grove, TN), and chromaffin cells prepared by digestion with collagenase followed by density gradient centrifugation as described previously [35]. The cells were plated onto collagen-coated coverslips at a density of~0.2 x 10 6 cell /mL. Fibroblasts and other proliferating cells were effectively suppressed with cytosine arabinoside (10 μM) (Sigma-Aldrich; St Louis MO), leaving relatively pure chromaffin cell cultures. The culture medium for chromaffin cells consisted of Dulbecco's modified Eagle medium \ F12 (1:1) supplemented with fetal bovine serum (10%), glutamine (2 mM), penicillin/streptomycin (100 unit mL -1 /100 μg mL -1 ), cytosine arabinoside (10 μM) and 5-fluorodeoxyuridine (10 μM). The culture medium was replaced the day after isolation and experiments were performed 2-5 days following cell isolation. All tissue culture reagents were from Life Technologies (Grand Island, NY).

Patch-clamp electrophysiology
Cells were placed in a recording bath (volume~300 μL) which was continually perfused with fresh solution at a flow rate of~3-4 ml/min from gravity-fed reservoirs, and viewed using a Nikon TE2000 inverted microscope. Patch pipette electrodes were pulled from borosilicate glass capillary tubes (World Precision Instruments, Sarasota, FL) using a Sutter P-97 pipette puller (Sutter Instruments, Novato, CA), coated with dental wax (Electron Microscopy Sciences, Hatfield, PA) and fire-polished to a final resistance of~2 MO when filled with a CsClbased internal solution. Cells were voltage-clamped in the conventional whole-cell configuration using an Axopatch 200B amplifier, Digidata 1400A interface, and PClamp10 (Clampex) acquisition software (Molecular Devices, Sunnyvale, CA). Analog data were filtered at 2-3 kHz and digitized at 50 kHz. Series resistance was partially compensated using the Axopatch circuitry (~60-80%). Linear capacitance and leak subtraction (performed offline) used P/N protocols (P/-4 or P/-8) with the leak pulses applied following the test pulses. Some of the voltageprotocols involved very long stimulus steps or trains. For these experiments leak subtraction was not applied, in part because the protocols were designed to monitor recovery from inactivation which could be altered in a voltage-dependent manner. In these experiments only cells with high resistance seals (> 1GO) and low holding current (< 50 pA) that remained stable for the duration of the experiment were used. Raw data were analyzed using PClamp10 (Clampfit) and graphing / statistical analyses were performed using OriginPro software (Originlab Corporation, Northampton, MA) or Prism5 software (GraphPad Software Inc., La Jolla, CA). All experiments were performed at room temperature (~23-25°C).
The extracellular NaCl-based solution used to bathe cells before and during seal formation consisted of (in mM): NaCl 145, KCl 2, MgCl 2 1, glucose 10, HEPES 10, CaCl 2 2, pH 7.3, osmolarity approx 315 mOsm. After entering the whole-cell recording configuration the bath solution was switched. For HEK293 and G1A1 cell recordings the extracellular solution contained (in mM): tetraethylammonium Cl 155, glucose 10, HEPES 10, BaCl 2 5, pH 7.3, 320-330mOsm. For chromaffin cell recording it contained (in mM): NaCl 150, KCl 2, MgCl 2 2, glucose 10, HEPES 10, CaCl 2 5, TTX 0.05-0.1, pH 7.3, osmolarity approximately 315 mOsm. Note barium and TEA were not used in chromaffin cells recordings because they both block potassium channels which results in depolarization of all the non-voltage-clamped cells in the recording chamber. These cells release a variety of neurotransmitters and hormones which can alter the cell being recorded from, for example via G protein coupled receptors [36].

Statistical analyses
Statistical analyses were performed using OriginPro software (Originlab Corporation, Northampton, MA) or Prism5 software (GraphPad Software Inc., La Jolla, CA). Recovery from inactivation was fit with a single or double exponential association function of the form: Y = Y 0 + A (1-e -X/t ) or Y = Y 0 + A 1 (1-e -X/t1 ) + A 2 (1-e -X/t2 ) in which Y 0 is the offset / initial amplitude at recovery time zero, A is the amplitude or span, X is the recovery time and t is the recovery time constant. For representation in figures, the mean recovery data were plotted and then fit with the above equations. The parameters from the fits to the mean data are reported in the figure legends. For statistical comparison of specific fit parameters (i.e. recovery time constants), each cell was individually fit with the exponential function to yield the relevant parameters and these were then pooled for statistical comparison. Statistical significance between two datasets was determined using paired or independent Student's t-test, and ANOVA was used to compare multiple datasets.

PMA selectively targets recovery from "slow inactivation"
We set out to investigate voltage-dependent inactivation of Ca V 2.2 Ca 2+ channels and how it might be regulated by the PKC-activator PMA. To isolate voltage-dependent inactivation, we used barium rather than calcium as the extracellular divalent cation and included BAPTA (10 mM) in the intracellular patch-pipette solution to prevent calcium-dependent inactivation. G1A1 cells stably express Ca V 2.2, α 2 δ and β 1b calcium channel subunits. Under our recording conditions, acute application of PMA (200 nM for 5-minutes) had no effect on the peak barium current (I Ba ) amplitude (Fig 1), and did not shift the current-voltage-relationship (Fig 1B), the voltage-dependence of inactivation (Fig 1C), or the rate/extent of inactivation during a 1s step depolarization ( Fig 1D). To assess recovery of I Ba from "fast" inactivation, we used a standard protocol in which a 1s prepulse (used to produce inactivation) was followed by a short test pulse to determine the extent of recovery (Fig 1E). This double pulse protocol was repeated every 60s and the recovery interval following the prepulse increased with each stimulus. Recovery following the 1s step was relatively fast (~50% recovery within 1s). Due to concerns with rundown and the possibility of a much slower component to the recovery, we limited this (D) Inactivation of I Ba during a 1s step depolarization was not altered by PMA. Control cells or cells treated with PMA for 5-10 minutes were stimulated with a 1s step to +10mV and the evoked currents normalized to the peak amplitude to enable better comparison of the inactivation time-course. Traces show the means but error bars are omitted for clarity. (E) Recovery from "fast" inactivation was not significantly different in control cells or PMA treated cells (200nM for 5-10 minutes). Inactivation of I Ba was produced by 1s prepulse and recovery determined by a brief test pulse after the indicated interval. This was repeated once every 60s (see inset above graph for voltage-protocol). Current amplitude during the recovery test pulse was normalized to peak I Ba evoked by the prepulse. The solid lines show double exponential fits to the data. (F) PMA prolonged recovery from "slow inactivation". Inactivation of I Ba was produced by 10s prepulse and recovery was tracked using a series of brief test pulses applied at the indicated time points following the 10s prepulse (see inset). This was repeated twice in the same cell, once before application of PMA ((Ctl) and once in the presence of PMA (after 5-min exposure). Solid lines show double exponential fits to the mean data (control A 1 = 0.23, A 2 = 0.73, t 1 = 1.21 s, t 2 = 18.96 s; PMA A 1 = 0.12, A 2 = 0.76, t 1 = 1.06 s, t 2 = 49.2 s, fit comparison F = 27.6 p < 0.0001). The inset bar graph shows the mean time-constant for the slow phase of recovery calculated from fits to the individual cells (*** p = 0.001, paired t-test). protocol to five repeats, covering the first 6-seconds of recovery. These data were fit well with a double exponential ( Fig 1E) and neither of the fitted time constants were significantly changed in PMA treated cells (tau fast = 167 ± 12 ms in controls Vs. 161 ± 6 ms in PMA treated cells, p = 0.67 unpaired t-test: tau slow = 2.46 ± 0.49 s in controls Vs 3.1 ± 0.21 s in PMA treated cells, p = 0.23 unpaired t test).
To investigate recovery from "slow" inactivation we used a protocol consisting of a prepulse lasting 10 s that was followed by a series of 12 brief steps (8ms duration) applied at increasing intervals following the prepulse (Fig 1F). It was immediately apparent that recovery was slower following the 10s prepulse compared to that following the 1s prepulse. Recovery was best fit with a double exponential with a smaller fast and larger slow component (fast component = 20 ± 2% in control conditions and 13 ± 2% in PMA, n = 9, p = 0.02, paired t-test). The fast time constant was not significantly altered by PMA (890 ± 146 ms in control and 914 ± 108 ms in PMA, p = 0.84 paired t-test), but the slower time constant was dramatically prolonged from 18.3 ± 2.3 s in control conditions to 53.4 ± 6.1 s in the presence of PMA (n = 9, p = 0.0001, paired t-test). These data show that, under our recording conditions, PMA selectively targets recovery from "slow" inactivation with little or no effect on other parameters of I Ba .
Recovery from "slow inactivation" is influenced by both the Ca V β subunit and by PMA Given that PMA appeared to target recovery from "slow" but not "fast" inactivation, we investigated the effects of PMA on channels containing the β 2a subunit which dramatically reduces fast inactivation [20,21]. HEK293 cells were transiently transfected with Ca V 2.2, α 2 δ, and either β 2a or β 1b subunits (with EGFP as a marker). As expected, inactivation of I Ba during a step depolarization was much slower in β 2a than β 1b containing channels (Fig 2A). Recovery of β 2a channels was fit with a single exponential with a time constant (at a holding potential of -100 mV) of 48.6 ± 4.7 s (n = 16), which was significantly longer compared β 1b containing channels (19.5 ± 1.8 s; n = 4; p = 0.0007 unpaired t-test). As expected for voltage-dependent inactivation, the recovery kinetics of I Ba showed no correlation with the amount of barium entry ( Fig 2B). The overall amount of barium entry (charge density) was much greater in cells expressing the β 2a subunit than in G1A1 cells (β 1b expressing cells). However, when the recovery time constant was plotted against charge density, the slope of a linear fit using Deming regression was not significantly different from zero (0 = 0.75 for β 1b expressing cells and p = 0.76 for β 2a expressing cells). This was confirmed using Pearson's correlation coefficient which again showed no statistically significant correlation between recovery rate and charge density: for β 1b cells r = -0.13, p = 0.74; for β 2a cells r = -0.08, p = 0.76. We also found that, as predicted for voltage-dependent inactivation, recovery was significantly accelerated at more hyperpolarized holding potentials ( Fig 2C). Thus, recovery following the conditioning prepulse demonstrated the expected features for voltage-dependent inactivation, and the recovery kinetics were influenced by the subtype of the Ca V β subunit.
PMA significantly prolonged the recovery time constant in β 2a containing channels from 46.7 ± 9.4 s to 108.6 ± 9.9 s (n = 6, p = 0.00001, paired t-test) (Fig 3). We also tested the closely related Ca V 2.1 (P/Q-type) channel (Ca V 2.1, β 2a and α 2 δ). Recovery from inactivation in Ca V 2.1 channels was faster than in Ca V 2.2 channels, but was still significantly prolonged by PMA from 14.8 ± 3.2 s to 35.5 ± 9.7 s (n = 5, p = 0.04, paired t-test). To compare the effects of PMA across these channels with different subunit combinations that had inherently different rates of recovery, we calculated the change in recovery time constant (i.e. tau in the presence of PMA / tau before application of PMA). This tau ratio showed that PMA prolonged recovery from inactivation to a similar extent in all cases: the tau ratio was 2.66 ± 0.35 for Ca V 2.2 + β 2a (n = 6), 3.09 ± 0.3 for Ca V 2.2 + β 1b (n = 9), and 2.38 ± 0.38 for Ca V 2.1 + β 2a (n = 5) (p = 0.34; F = 1.15, One-way ANOVA).

PMA prolonged recovery from inactivation following trains of brief stimuli applied at physiologically relevant frequencies
To more closely mimic physiologically relevant electrical activity, we stimulated cells with trains of brief depolarizations applied at 50Hz for 1s, or 5Hz for 10s (Fig 4). The extent of inactivation at the end of the 50Hz train was not altered by PMA (73% ± 3.6% in control conditions Vs. 69 ± 3.6% in PMA; n = 4; p = 0.21, paired t test) (Fig 4A). The recovery was biphasic, and PMA slightly reduced the magnitude of the initial fast component: in control conditions 52 ± 3% of the inactivation was recovered by 1s (the first recovery time point) and this was reduced to 40 ± 4% by PMA (n = 4, p = 0.021, paired t-test). PMA significantly prolonged the time constant of the slow recovery phase from 29.7 ± 4.6 s in control conditions to 112.9 ± 23.9 s in the presence of PMA (n = 4; p = 0.045 paired t-test). As expected, the 5Hz train produced less inactivation than the 50Hz train ( Fig 4B). The extent of inactivation at the end of the 5Hz train was significantly increased by PMA (33 ± 2% Vs. 41 ± 3%, n = 6; p = 0.015, paired t-test). Recovery following the 5 Hz train was well fit with a single exponential and lacked the initial fast component seen after the 50 Hz train. The time constant of recovery was significantly slowed from 35.1 ± 6.7 s in control conditions to 118.5 ± 13.8 s in the presence of PMA (n = 6, p = 0.0042, paired t-test). These data show that PMA prolonged recovery from "slow" inactivation following physiologically relevant stimulus trains as well as following sustained step depolarization.
Probing the involvement of PKC on recovery from "slow" inactivation 4α-PMA is an inactive control for PMA that does not activate PKC. We found that 4α-PMA did not significantly alter recovery from inactivation following a 5Hz/10s stimulus train in HEK cells expressing Ca V 2.2, β 1b and α 2 δ, (tau control = 19.8 ± 3.1 s Vs tau 4α-phorbol = 22.5 ± 4.7 s; n = 4; p = 0.28, paired t-test). This is consistent with the involvement of PKC, so we tested several PKC inhibitors to see if they blocked the effect of PMA on recovery from inactivation. Cells were pre-incubated for 20-30 minutes with the inhibitors and then stimulated first in the presence of the inhibitor alone and then in the presence of the inhibitor + PMA. The inhibitors used were PKC(19-36), a pseudosubstrate peptide inhibitor of PKC that was added to the intracellular patch-pipette solution (2 μM); a combination of bisindolylmaleimide-1 (500nM) + Go6983 (100nM); calphostin C (200 nM). Recovery from inactivation following a 5Hz/10s train was fit with an exponential to determine the recovery time constant (Fig 5A and 5B). None of the drug pretreatments significantly altered the baseline recovery rate prior to application of PMA (Fig 5B) (p = 0.09, F = 2.28; one-way ANOVA followed by Dunnett's post-test). However, the recovery time constant in the presence of PMA was significantly reduced by pretreatment with bisindolylmaleimide-1 + Go6983 or calphostin C (p < 0.0001, F = 11.64; one-way ANOVA followed by Dunnett's post-test for multiple pairwise comparisons) (Fig 5B). To compare the various treatment groups, the change in recovery time constant was expressed as a ratio (tau in the presence of PMA / tau before PMA) (Fig 5C). The slowing  . Also shown is data for cells treated with 4α-PMA (n = 4). Pretreatment with the various drugs did not significantly alter the recovery time constant before application of PMA (left panel). The effect of PMA was significantly reduced by pretreatment with Bis + Go or calphostin C (right panel) (ns, not significantly different, * p < 0.05, *** P < 0.001 compared to the control PMA cells (red bar) determined using one-way ANOVA and Dunnett's post-test). (C) To quantify the change in recovery rate produced by PMA, a ratio of the recovery time constants was calculated in each cell (tau in the presence of PMA / tau before application of PMA). Statistical significance compared to the control PMA cells (red bar) was determined using one-way ANOVA and Dunnett's post-test (ns, not significantly different, ** P < 0.01). (D) Another index to compare the various drug treatments is the percent inhibition of recovery at the 40 s time point (R 40 -see panel A and results section for more detail). Statistical significance compared to the control PMA cells (red bar) was determined using oneway ANOVA and Dunnett's post-test (ns, not significantly different, * p < 0.05, *** P < 0.001). of recovery produced by PMA (tau ratio) was not seen with the control 4α-PMA, was abolished by pretreatment with calphostin C, and partially, but not significantly reduced in cells pretreated with PKC(19-36), or bisindolylmaleimide-I + Go6983 (p = 0.005, F = 4.885, one-way ANOVA, followed by Dunnett's post-test). It has been reported previously that the extent, but not the rate, of Ca V 2.2 recovery from inactivation can be modulated by alternative splicing [37]. Therefore, we calculated the percent change in recovery produced by PMA at the 40 s time point (R 40 ) (Fig 4A and 4C). In control cells (before application of PMA) the fractional recovery at this time was 0.67 ± 0.07, and in the presence of PMA this was reduced to 0.23 ±0.02 (n = 6, p = 0.0026, paired t-test) resulting in a calculated R 40 of 61 ± 6.9%. We calculated R 40 for the various treatment groups (different PKC inhibitors) and found it was significantly reduced by bisindolylmaleimeide-1 + Go6983 and abolished by calphostin C (Fig 5D) (p = 0.0001, F = 29.3, one-way ANOVA followed by Dunnett's post-test). Thus, our data show that the effect of PMA on recovery from inactivation was abolished by calphostin C and partially blocked by bisindolylmaleimeide-1 + Go6983.
Given the incomplete block of PMA by bisindolylmaleimide-I and especially PKC(19-36) (see above and Fig 5), we wanted to test their ability to antagonize PKC under our recording conditions. To do so, we investigated the inhibition of Ca V 2.2 channels by G protein βγ subunits (Gβγ) [1,2]. Gβγ inhibits I Ba by binding directly to the Ca V 2.2 subunit, and activation of PKC by phorbol ester has been shown to partially reverse this inhibition [38][39][40][41][42][43]. We transiently transfected G1A1 cells with GFP-tagged Gβγ subunits (Fig 6) which results in tonic inhibition of I Ba . A trademark characteristic of Gβγ-mediated inhibition is that it can be reversed by strong membrane depolarization due to transient voltage-dependent dissociation of Gβγ from the channel [1]. Thus, the extent of Gβγ-mediated inhibition can be revealed using a prepulse facilitation protocol as shown in Fig 6A. Treating cells with PMA for 5-minutes significantly reduced the extent of Gβγ-mediated inhibition (Fig 6B). Pretreating cells with bisindolylmaleimide-I (Fig 6C) or PKC (19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35)(36) (Fig 6D) blocked the ability of PMA to reduce Gβγ-mediated inhibition. Thus the drugs were effective under our recording conditions, suggesting the partial inhibition of PMA's actions on slow inactivation is not simply due to inactive drugs or other artefacts.
PMA can also activate some additional signaling pathways, for example by binding to the C1-domain of RasGRP (Ras Guanine nucleotide Releasing Protein) which in turn leads to activation of the monomeric G protein Ras [44,45]. To investigate a potential role for Ras signaling we transfected G1A1 cells with Ras S17N (a dominant negative mutant) or Ras G12V (a constitutively active mutant) and compared recovery to control cells transfected with GFP alone. Neither of the Ras mutants altered the baseline recovery rate before application of PMA, nor did they alter the slowing of recovery produced by PMA. The change in recovery rate (tau ratio) produced by PMA was 3.09 ± 0.3 in GFP controls (n = 9), 2.42 ± 0.34 in S17N (n = 7) and 2.55 ± 0.26 in G12V (n = 7) (p = 0.25, F = 1.483, one-way ANOVA). Similarly, R 40 (% inhibition of recovery at 40s by PMA) was not significantly altered by the Ras mutants: 36 ± 4% in controls, 27 ± 7% in S17N and 31 ± 6% in G12V (p = 0.53, F = 0.664, one-way ANOVA). Phorbol esters / PKC have also been reported to modulate surface expression of some ion channels and transporters through dynamin dependent endocytosis [46][47][48][49] and Ca V 2 channels are known to associate with endocytic complexes and are subject to internalization by various stimuli [50][51][52][53]. However, when a dynamin inhibitory peptide (50 μM) was included in the patch-pipette solution, PMA still significantly slowed the recovery time constant in HEK cells expressing Ca V 2.2, β 1b and α 2 δ, from 28.9 ± 3 s to 65.5 ± 14.1 s (n = 5; p = 0.033 paired t-test) suggesting endocytosis is not playing a significant role under our recording conditions.

The endogenous I Ca in neuroendocrine chromaffin cells is regulated in a similar manner by PMA
It is possible that channels in heterologous expression systems are regulated in a different manner than would be the case in a native cellular environment. Therefore, we investigated the ability of PMA to slow recovery of I Ca from inactivation in adrenal chromaffin cells, an important neuroendocrine component of the sympathetic nervous system [54]. In bovine chromaffin cells Ca V 2 channels account for~85-90% of the whole cell calcium current in a roughly 1:1 ratio of N-type and P/Q-type I Ca , with the remainder due primarily to a small (~10%) Ca V 1 (Ltype) component [36,55,56]. As before, we included 10mM BAPTA in the patch pipette solution which effectively blocked calcium dependent inactivation of I Ca to enable investigation of voltage-dependent inactivation [22,57]. However, we used Ca 2+ as the extracellular divalent charge carrier because Ba 2+ depolarizes the surrounding cells in the recording chamber leading to exocytosis of catecholamines, ATP, and opioids that can act in a paracrine manner to modulate the I Ca in the cell being recorded from [36].
"Fast" voltage-dependent inactivation is minimal in chromaffin cells, possibly due to expression of the β 2a subunit [58,59]. However, the 10s step depolarization paradigm produced robust inactivation of I Ca both in control conditions (95 ± 1%) and in the presence of PMA (97 ± 0.5%, n = 7; p = 0.132, paired t-test). Recovery from inactivation exhibited at least two kinetic components (Fig 7). PMA clearly prolonged recovery from inactivation, with a particularly prominent effect on the slower time constant which was increased from 26.6 ± 1.4 s to 90.9 ± 8.8 s (n = 7, p = 0.0002, paired t-test). The R 40 (% change in recovery at 40s) showed that PMA inhibited recovery by 35.5 ± 1.6% (n = 7). Pretreating cells with bisindolylmaleimide-1 modestly reduced R 40 to 21.6 ± 7% (n = 5) although this was not statistically significant, and calphostin C significantly reduced R 40 to 12.8 ± 4% (n = 6; p < 0.01 one-way ANOVA followed by Dunnetts post-test) (Fig 7D). These data mirror those obtained in recombinant channels and confirmed that PMA prolongs recovery from "slow" inactivation for endogenously expressed I Ca in neuroendocrine cells.

GDP-β-S reduces the slowing of recovery produced by PMA
To investigate any potential involvement of G protein signaling in the effects of PMA we replaced GTP with GDP-β-S (0.5 mM) in the intracellular patch pipette solution to prevent activation of heterotrimeric and monomeric G proteins (Fig 8). First we used this solution to record from HEK cells expressing Ca V 2.2, β1b and α2δ subunits stimulated with a 5Hz/10s train. In the absence of PMA, GDP-β-S increased the extent of inactivation at the end of the 5Hz train compared to control cells (50 ± 4%, n = 8 Vs 34 ± 2%, n = 6; p = 0.011, unpaired ttest). The rate of recovery from inactivation under control conditions (before application of PMA) was not altered by GDP-β-S, but the slowing of recovery produced by PMA was significantly reduced, as reflected in the tau ratio (Fig 8A and 8B). We repeated a similar experiment but with GTP-γ-S added to the patch-pipette solution. Again, the prolonged recovery from inactivation produced by PMA was blocked in these cells (Fig 8B). Similarly, in adrenal Control of CaV2.2 "Slow" Inactivation chromaffin cells GDP-β-S had no effect on recovery from inactivation under control conditions, but significantly reduced the slowing of recovery produced by PMA (Fig 8C and 8D). Together these data suggest that G protein signaling pathways can influence the kinetics of recovery from inactivation, opposing the actions of PMA.

PMA prolongs recovery from "slow" inactivation
Voltage-dependent inactivation of Ca V 2.2 plays an important role in controlling the temporal response of the channels to different electrical stimuli. "Fast" inactivation is relatively wellunderstood and thought to involve pore occlusion by the cytoplasmic I-II linker [11,18,19]. As with other voltage-gated channels, there is an additional mode of inactivation with slower onset and recovery kinetics. This "slow" inactivation is molecularly distinct but poorly understood. It has been investigated somewhat more in sodium and potassium channels where GDP-β-S had no effect on recovery rate in the absence of PMA (black bars) but significantly reduced the slowing of recovery produced by PMA (red bars) (ns, not significant, * p < 0.05; ANOVA with Bonferroni's post-test for multiple pairwise comparisons). doi:10.1371/journal.pone.0134117.g008 proposed mechanism(s) involve altered movement of the voltage-sensor and/or constriction of the channel pore [26][27][28][29]. "Slow" inactivation of sodium channels is modulated by second messengers including PKA and PKC, which in turn controls neuronal excitability [30,31,60,61]. It has been reported that co-expression with syntaxin can promote "slow" inactivation of Ca V 2 channels [25] but little else was known about the how it might be controlled. Therefore, we set out to investigate "slow" inactivation of Ca V 2.2 channels using both recombinant channels expressed in HEK293 cells, and endogenous channels in adrenal chromaffin cells. In both these cell types the Ca V 2 channels undergo Ca 2+ -dependent inactivation which confounds investigation of voltage-dependent inactivation. Therefore, we included 10mM BAPTA in the patch pipette solution and, for HEK293 cells, we also used Ba 2+ rather than Ca 2+ as the charge carrier to prevent Ca 2+ -dependent inactivation.
In the literature, the effects of PMA/PKC on Ca V 2.2 channels are mixed/complex. In some cases there is a pronounced increase in current amplitude, at least in part due to reversal of tonic Gβγ-mediated inhibition [38][39][40][41][42][43]62]. PKC can also inhibit or exert bidirectional modulation of I Ca , perhaps through distinct kinase isoforms and phosphorylation sites, and its effects might be altered by the Ca V β subunit [63][64][65][66][67]. We found there was little effect of PMA on I Ba amplitude (Fig 1). This might reflect a lack of tonic inhibition by Gβγ under our conditions, or could be due to the strong Ca 2+ buffering (10mM BAPTA) which might preclude activation of classic (Ca 2+ -dependent) PKC isoforms. We also found no effect of PMA on the I-V curve, inactivation curve, or "fast" inactivation during a step depolarization (Fig 1). During a step depolarization the onset of "fast" inactivation masks that of "slow" inactivation. However, tracking the recovery from inactivation provides a convenient solution: recovery from "fast" inactivation is similarly fast, whereas recovery from "slow" inactivation is slow. Thus we focused on recovery from inactivation following different stimulus protocols, and how this was modulated by the phorbol ester PMA. Recovery following a 10s step depolarization was biphasic with the slower component accounting for >80% of the total. As expected for recovery from voltage-dependent inactivation, the kinetics of this slow component showed no correlation with amount of barium entry ( Fig 2B) and were accelerated at more hyperpolarized potentials ( Fig 2C). When cells were stimulated with a 1s step depolarization recovery was faster than following a 10s step consistent with the prediction that fewer channels enter the "slow" inactivated state during a shorter duration stimulus (Fig 1E and 1F). We show that PMA dramatically prolonged recovery from "slow" inactivation with little impact on recovery from "fast" inactivation. We found a similar effect when using a train of brief stimuli to more closely resemble physiological patterns of electrical activity (Fig 4). Importantly, we show that PMA also prolonged recovery of Ca V 2 channels from "slow" inactivation in adrenal chromaffin cells, an important neuroendocrine component of the sympathetic nervous system (Fig 7). Thus, this phenomenon is apparent in a native cellular context and is not an artefact of the heterologous expression system. Taken together, we interpret our data as reflecting the activity-dependent entry of Ca V 2.2 channels into a "slow" inactivated state during sustained periods of stimulation, and that PMA prolongs recovery from this state.
Although we did not systematically compare the effects of different β subunits, we did investigate the β 2a subunit because it dramatically reduces "fast" inactivation. Recovery from inactivation took significantly longer with β 2a than in cells with β 1b . It was previously reported that the β subunit had no effect on recovery from "slow" inactivation in Ca V 2.1 channels, although it did indirectly modulate the onset kinetics via altered fast inactivation [24]. It is possible that the β subunit has a different impact on "slow" inactivation of Ca V 2.2 and CaV2.1 channels, but this will need to be confirmed by direct comparison in future investigations. The focus of our study was the impact of PMA on "slow" inactivation and we found that, despite the different baseline recovery rates, PMA produced a similar~2.5-3 fold increase in the recovery time constant for Ca V 2.2 channels containing either the β 2a or β 1b subunits. We also found that the rate of recovery Ca V 2.1 channels (with β 2a ) following a 10s step was faster than for the Ca V 2.2 channels, but still prolonged by treatment with PMA. Thus, PMA prolongs recovery from "slow" inactivation in both Ca V 2 channel types, and it will be interesting to determine if this extends to the Ca V 1 or Ca V 3 Ca 2+ channels.
Does PKC underlie the slowed recovery from inactivation produced by PMA?
Recovery from inactivation was prolonged by PMA but not by 4α-PMA, a control analogue that does not activate PKC. Also consistent with the involvement of PKC, pretreating cells with calphostin C prevented the slowing of recovery produced by PMA (Fig 5). However, a combination of bisindolylmaleimide-1 + Go6983 only partially reduced the effect of PMA while PKC (19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35)(36) had little effect. Similarly, in adrenal chromaffin cells calphostin C blocked the action of PMA while bisindolylmaleimide-1 only had a partial, statistically non-significant effect ( Fig  7D). Both PKC(19-36) and bisindolylmaleimide-1 were able to effectively antagonize the ability of PMA to reverse Gβγ-mediated inhibition of Ca V 2.2 channels (Fig 6). So why were these drugs less effective at antagonizing the effect of PMA on "slow" inactivation? One clue might come from the mechanism of action of the different antagonists: calphostin C targets the regulatory C1-domain of PKC whereas the other antagonists target the catalytic domain. It has been shown that A-kinase anchoring protein-79 (AKAP-79) scaffolds a signaling complex between PKC and KCNQ potassium channels [68]. Moreover, when PKC is in this complex it is protected from antagonists that target the ATP binding catalytic domain, but still inhibited by calphostin C [68]. Perhaps a similar situation exists for the Ca V 2 channels, which would explain the differential sensitivity to the PKC antagonists.
The effects of phorbol esters / PMA are not always recapitulated by stimulating endogenous pathways that activate PKC, such as Gq-coupled GPCRs. One possible explanation is that PMA acts at least in part through a non-PKC signaling pathway. Phorbol esters can bind to the C1-domain of other proteins, for example RasGRPs which activate the monomeric G protein Ras [44,45]. However, our data suggest this pathway is not involved, because overexpression of constitutively active or dominant negative Ras mutants had no effect on the ability of PMA to slow recovery from inactivation. Potential involvement of other C1-domain proteins will require further investigation. PMA has also been reported to promote removal of ion channels and transporters from the plasma membrane through dynamin-dependent endocytosis [46][47][48][49]. Although we cannot categorically rule out a role for channel trafficking, our data are not consistent with this playing a major role. First, a dynamin inhibitory peptide did not significantly change the effect of PMA on recovery from inactivation. Second, PMA had no effect on the amplitude of I Ba prior to the stimulus train / 10s step, and the extent of inactivation was only modestly altered. Recovery from fast inactivation was unaltered by PMA and recovery from slow inactivation was voltage-dependent ( Fig 2C). None of these features are consistent with endocytic recycling of the channels playing a major role under our recording conditions. Finally, our data also point to a role for G protein signaling in helping to control slow inactivation. Disrupting G protein signaling using intracellular GDP-β-S had no obvious effect on recovery from inactivation per se, but significantly reduced the ability of PMA to slow recovery (Fig 8B). Similarly, GTP-γ-S (which should activate rather than block G protein signaling) also reduced the effects of PMA. While it is possible that PMA recruits a second messenger pathway involving G protein signaling, it is also possible that G proteins or guanosine nucleosides exert a parallel, allosteric effect that is permissive for the actions of PMA. This possibility will require extensive future investigations.

Conclusions
"Fast" voltage-dependent inactivation of Ca 2+ channels is relatively well characterized and helps shape channel activity and short-term synaptic depression during brief trains of action potentials (tens-hundreds of milliseconds). Superimposed on this is the molecularly distinct, but poorly understood process of "slow" inactivation which develops and recovers over the course of seconds-to-minutes. The slow onset favors recruitment by sustained periods of electrical activity or membrane depolarization, while the slow recovery kinetics might confer a short-term "molecular memory" for this preceding cellular activity. In this study we report the novel finding that PMA dramatically prolongs recovery of Ca V 2.2 channels from "slow" inactivation. This effect is likely mediated at least in part through PKC and might also involve G protein / guanosine nucleoside signaling. Regardless of the precise molecular details, this could provide a novel mechanism for dynamic, activity-dependent regulation of Ca 2+ channel availability, electrical excitability, and neurotransmission in the seconds-to-minutes timeframe.