(R)-Roscovitine Prolongs the Mean Open Time of Unitary N-Type Calcium Channel Currents

Abstract

(R)-roscovitine (Ros) is a cyclin-dependent kinase inhibitor that also has been shown to have direct agonist and antagonist actions on Ca_v2.1 (P/Q-type) and Ca_v 2.2 (N-type) families of voltage-gated calcium channels. These kinase-independent effects represent a novel opportunity to advance our understanding of calcium channel function and calcium-triggered neurotransmitter release. Furthermore, such actions on calcium channels may direct the development of Ros derivatives as new therapeutic agents. We used patch clamp recordings to characterize mechanisms that underlie the agonist effects of Ros on unitary N-type calcium channel gating. We found that N-type channels normally gate with either a short or long mean open time, that Ros significantly prolonged the mean open time of the long gating component and increased the probability of observing channels that gated with a long open time, but had no effect on single channel conductance. Using Monte Carlo simulations of a single channel kinetic model and Ros interactions, we were able to reproduce our experimental results and investigate the model’s microscopic dynamics. In particular, our simulations predicted that the longer open times generated by Ros were due to the appearance of a long open state combined with an increased amount of time spent in transitions between open states. Our results suggest a mechanism for agonist effects of Ros at the level of single channels, and provide a mechanistic explanation for previously reported agonist effects on whole cell calcium currents.

(R)- and (S)-Roscovitine, together with a structurally similar compound olomoucine, inhibit cyclin-dependent kinases (cdks). Of these compounds, (R)-Roscovitine (Ros) in particular also has been shown to have cdk-independent effects on calcium (Ca²⁺) channels (Yan et al., 2002; Buraei et al., 2005; 2007; Cho and Meriney, 2006). The effect of Ros on P/Q- and N-type Ca²⁺ channels (Ca_v 2.1 and Ca_v2.2) manifests itself at the population level by slowing deactivation kinetics (Yan et al., 2002; Tomizawa et al., 2002; Buraei et al., 2005; 2007). This action prolongs Ca²⁺ tail currents and has been reported to increase transmitter release at central nervous system synapses (Yan et al., 2002; Tomizawa et al., 2002) and the frog neuromuscular junction (Cho and Meriney, 2006). With increasing concentrations, Ros also displays Ca²⁺ current antagonist activity, albeit with a slower onset than observed for agonist effects (Buraei et al., 2007). Recently Buraei and Elmslie (2008) have begun to elucidate the molecular pharmacologic interactions that might underlie differences between agonist and antagonist activities of Ros on Ca²⁺ channels.

Aside from the use of Ros derivatives to study Ca²⁺ channel gating and the regulation of transmitter release, such compounds might also be developed as potential therapeutic agents that selectively target N- and P/Q-type Ca²⁺ channels. Despite recent work documenting effects on whole cell currents, it is not yet known how Ros affects single channel gating. Thus, to characterize these effects, we performed cell-attached patch clamp recordings using a cell line that stably expresses mammalian N-type Ca²⁺ channels. We show that these channels gate with distinct short or long mean open times. Ros significantly lengthened the longer mean open time component, and increased the probability of observing the longer openings. On the other hand, we did not detect any effect of Ros on single channel conductance. These results are reminiscent of the selective effects of BayK 8644 and FPL 64176 on L-type Ca²⁺ channels (Schramm et al., 1983; Kokubun and Reuter, 1984; Hess et al., 1984; Nowycky et al., 1985; Zheng et al., 1991; Kunze and Rampe, 1992; Lauven et al., 1999; Tavalin et al., 2004). We also propose a kinetic scheme for Ros modulation of voltage-gated calcium channels (modified from Buraei et al., 2005), constrained by our new single channel data and a previous estimate of the probability that N-type Ca²⁺ channels open during an action potential (Poage and Meriney, 2002; Wachman et al., 2004; King and Meriney, 2005, Luo et al., 2009). Our results provide a mechanistic explanation for the previously reported agonist effects of Ros on whole cell calcium currents.

Experimental Procedures

tsA201 cells expressing N-type calcium channels

We used a tsA201 cell line (kindly provided by Dr. Diane Lipscombe, Brown University; see Lin et al., 2004) that stably expresses all of the subunits of the N-type Ca²⁺ channel splice variant predominantly present in mammalian brain and spinal cord: Ca_v2.2 rnα_1B-c (Ca_v 2.2 e[24a,Δ31a]), Ca_vβ₃ and Ca_vα₂δ₁. The cells were maintained in DMEM supplemented with 10% fetal bovine serum, 25 ug/ml zeocin, 5 ug/ml blasticidin, and 25 ug/ml hygromycin.

Whole-cell patch clamp recordings

Whole-cell currents through Ca²⁺ channels were recorded as previously described (White et al., 1997; Yazejian et al., 1997; Pattillo et al., 1999). Briefly, the pipette solution consisted of (mM): 135 CsCl, 4 MgCl₂, 10 HEPES, 1 EGTA, 1 EDTA, pH 7.4. The culture was bathed in a solution consisting of (mM): 130 ChCl, 10 TEA-Cl, 2 CaCl₂, 1 MgCl₂, 10 HEPES, pH 7.4. In some experiments the cells were bathed in a high barium solution (100 mM BaCl₂, 10 mM HEPES, 10 mM TEA-Cl, pH 7.4). Patch pipettes were fabricated from borosilicate glass and had an average access resistance of 4.3 ± 2.9 MΩ (mean ± SD, n=37), compensated for by 85% (lag setting = 10 μsec). Capacitive currents and passive membrane responses to voltage commands were subtracted using four waveforms of reverse polarity, each 1/4 the size of the full waveform. Ca²⁺ currents were amplified by an Axopatch 200B amplifier, filtered at 5 KHz, and digitized at 10 KHz for subsequent analysis using pClamp software (Axon Instruments / Molecular Devices; Sunnyvale, CA). Liquid junction potentials of −7 and −9 mV (for calcium and barium containing solutions respectively) were accounted for in the plotting of data. Fitting of tail current deactivation was begun 50 μsec after the peak of the tail current. All experiments were carried out at room temperature (22° C).

Cell-attached patch clamp recordings

Cell-attached recordings of single Ca²⁺ channel currents were performed using a bath saline that collapsed the membrane potential (mM): 140 K-aspartate, 10 HEPES, 5 EGTA, 5 MgCl₂, pH 7.4 with KOH. The pipette solution included barium as the major charge carrier to increase the size of current through Ca²⁺ channels (mM): 100 BaCl₂, 10 HEPES, 10 TEA-Cl, pH 7.4. For data acquisition and analysis we used the pClamp software package (Axon Instruments / Molecular Devices; Sunnyvale, CA). Single channel currents were amplified by an Axopatch 200B amplifier with an actively cooled capacitive head stage, filtered at 5 KHz, and digitized at 10 or 20 KHz for subsequent analysis. Before analysis, unitary currents were digitally filtered offline at 2 KHz. Patch pipettes were fabricated from quartz (1.5 mm outside diameter; 0.5 mm inside diameter; Sutter Instrument Co., Novato, CA) using a laser-based pipette puller (P-2000, Sutter Instrument Co.; Novato, Ca) to provide low noise recordings. Capacitive and leak currents were subtracted off-line using null sweeps. A liquid junction potential of −20 mV was accounted for in the plotting of data. Unitary event transitions were recognized using a 50% threshold method, and transitions shorter than 0.2 msec were ignored (consistent with our filtering conditions).

Reagents

(R)-roscovitine (a gift of Dr. Laurent Meijer, CNRS, Roscoff, France) was dissolved in DMSO as a 100 mM stock and stored at −20°C. For whole-cell recordings, aliquots were diluted on the day of use into extracellular saline at a final concentration of 100 μM, and bath applied via a delivery pipette in a ~0.3 ml chamber using a rapid perfusion system running at ~1 ml/min. Control recordings performed with 0.1% DMSO alone added to the drug delivery pipette solution revealed no significant effects on whole cell Ca²⁺ currents. For cell-attached recordings, aliquots were diluted into the patch pipette solution at the same final concentration. All other salts and chemicals were obtained from Sigma-Aldrich chemical company (St. Louis, MO).

Kinetic modeling of channel gating

Calcium channel kinetics were simulated with MCell (Monte Carlo Cell; Stiles et al., 1996; Stiles and Bartol, 2001), which in its latest revision (MCell version 3) allows continuous-time stochastic simulation of voltage-dependent state transitions and ion flux (Kerr et al., 2008). Our model consisted of one or more Ca²⁺ channels embedded in a surface that represented a cell membrane. During simulations, the channel(s) underwent stochastic transitions between closed and open states according to the kinetic scheme shown in Fig. 8, with voltage-dependent rate constants assigned to each transition except Ros binding and unbinding. Voltage-dependent rate constants were calculated from K_x = A_x exp((V-V_c) z_x F/RT), where A_x is the rate amplitude at the characteristic voltage V_c = 10 mV, z_x is the charge moved, and F, R, and T are the Faraday constant, ideal gas constant, and temperature, respectively. For the sake of computational efficiency, Ros binding at a constant concentration (100 μM) was simulated using a pseudo-first order rate constant for each binding transition (the pseudo-first order rate constants were obtained from the product of a bimolecular binding rate constant and the constant Ros concentration).

Fig. 8.

Kinetic model and simulations of Ros effects on channel gating. (A) Kinetic scheme (see Experimental Procedures and Table 1 for rate constant values). (B) Open time distribution for simulated control conditions was best fit by the sum of two exponentials, similar to experimental results (Fig. 4B and Table 2). (C) Open time distribution for simulated Ros-treatment was also best fit by the sum of two exponentials, but the longer component was prolonged and larger, again similar to experimental results (Fig. 4D and Table 2). (D) Time course of fractional occupancy under control conditions for the terminal closed state and the two open states (simulated step to −20 mV averaged over 10,000 simulations). (F) Time course of fractional occupancy as in (E), but after Ros treatment, showing the terminal closed state, the two unbound open states, and the two Ros-bound open states.

Each time a channel entered a new state, or at times when the membrane potential changed, a random number was used to choose a lifetime from the exponential distribution of possible lifetimes for the new state and/or voltage. Whenever the current lifetime elapsed, another random number was used to decide which transition would occur and a subsequent lifetime was generated for the destination state. Production of Ca²⁺ ions from open states was handled in a similar stochastic fashion, using rate constants calculated from the membrane voltage, reversal potential, and single channel conductance (2.6 pS in 2 mM extracelluluar calcium; Church and Stanley, 1996). Whenever an event was scheduled to occur from an open state, the channel could either make a transition to an adjacent state, or produce a calcium ion and remain in the original open state until the next event occurred. Since the next event also might produce a calcium ion (and might even occur during a given simulation timestep), it was possible for multiple ions to be produced prior to exit from an open state.

Simulation output included the times of all transitions induced by a step depolarization, but to be consistent with experimental (steady-state) patch clamp data, we analyzed simulated lifetime distributions only after the channel had equilibrated at the new holding potential (the last 20 ms of each simulation; see Figs. 8D and E). Mean open times were calculated from 10000 single channel simulations, each of which used a different random number seed and a 50 ms step depolarization from the resting potential (−60 mV) to −20 mV (the same step size used in experimental recordings). As with our experimental data, we omitted open times < 0.2 ms from analysis (including such short events changed calculated time constants by less than 0.1 ms). Final model parameters were determined by an extensive parameter sweep, adjusting inputs until the model yielded good agreement with the experimental open time distribution while maintaining microscopic reversibility for the cyclic part of the kinetic scheme. The same final model then was used to simulate whole cell currents using a population of 10,000 channels.

Results

Roscovitine effects on whole cell currents

Whole cell currents (Fig. 1A) were recorded from tsA201 cells expressing N-type Ca²⁺ channels Ca_v2.2e[24a, Δ31a] together with Ca_vβ₃ and Ca_vα₂δ₁. Using 2 mM Ca as the charge carrier, the average peak current was 1.334 ± 0.133 nA (mean ± SEM, n = 6 cells), similar to results reported by Lin et al. (2004). In 100 mM Ba²⁺ (used below for single channel measurements), currents were larger (average peak of 2.714 ± 0.648 nA, n = 7) and the current-voltage relationship (Fig. 1B) was shifted to the right by about 20 mV due to increased surface charge on the membrane (see Zhou and Jones, 1995). As in previous reports (Buraei et al., 2005; 2007; Cho and Meriney, 2006), Ros slowed deactivation of tail currents, which were best fit by a sum of two exponentials (Fig. 2). After an acute (15-30 second) application of 100 μM Ros, the long decay time constant (long tau) was significantly increased while the short component was unchanged (Fig. 2A). In addition, the fitting weight of the long component was significantly increased (Fig. 2B). Similar effects of roscovitine on whole cell current were observed when using either 2 mM calcium or 100 mM barium as the charge carrier (data not shown). Ros is a use-dependent Ca²⁺ channel agonist (Buraei et al., 2005; 2007) with effects that are more pronounced during longer and stronger depolarizations (Fig. 2C). Following prolonged application, 100 μM Ros also decreased the amplitude of the whole cell current (data not shown; see Buraei and Elmslie, 2008 for description of antagonist effects). For this report, we have focused on the agonist effects.

Fig. 1.

Voltage-gated currents through N-type Ca²⁺ channels expressed in tsA201 cells recorded using whole cell patch clamp methods. (A) Representative family of currents recorded using 2 mM Ca²⁺ (left) or 100 mM Ba²⁺ (right) as the charge carrier. Top: voltage waveforms used to activate current start from a holding potential of −107 mV (Ca²⁺), or −89 mV (Ba²⁺), and stepped to the depolarized potentials shown in B. Bottom: family of currents generated by the voltage waveforms. (B) Current voltage relationship for currents recorded in 2 mM Ca²⁺ (filled circles, n = 6), or 100 mM Ba²⁺ (open circles, n = 7). Data are mean ± SEM.

Fig. 2.

Ros effects on whole cell Ca²⁺ current deactivation kinetics. (A) Tail currents (see Inset) were best fit by a sum of two exponentials, and 100 μM Ros (filled bars) selectively and significantly increased the time constant of the longer component. Inset: representative currents (bottom traces) elicited in the presence (Ros) and absence (cont) of 100 μM Ros using a depolarizing voltage step to +53 mV followed by a repolarization to −43 mV (top trace). (B) Ros treatment also increased the fitting weight for the long component of deactivation, subsequently explained by increased frequency of prolonged unitary currents (see Fig. 4). *significantly different, Student’s t-test, p<0.05, n = 4. (C) Ros treatment effects show dependence on channel use. Top traces: A depolarizing voltage step from −107 to +53 mV was maintained for 2 or 14 ms (squares or circles, respectively) prior to repolarization to −43 mV. In the absence of Ros (open symbols), the duration of depolarization had no effect on the subsequent rate of current deactivation. In contrast, Ros-dependent slowing of deactivation (filled symbols) was more pronounced after the longer depolarizing step. Data shown with open and filled symbols were recorded from different cells.

Roscovitine effects on single channel currents

In cell attached recordings we found that tsA201 cells expressed N-type Ca²⁺ channels in clusters of variable size. Figure 3 shows representative recordings from two different patches chosen to demonstrate the variability in recorded currents. Using a voltage ramp protocol (Fig. 3A) to activate channel openings, some patches showed few or no channel openings (Fig. 3B), while others had very large currents generated by nearly simultaneous opening of many channels (Fig. 3C). Because we did not observe this degree of heterogeneity in our recordings of whole cell currents, we concluded that N-type Ca²⁺ channels are clustered on the cell surface.

Fig. 3.

N-type Ca²⁺ channels are clustered on tsA201 cell membranes. (A) Ramp voltage protocol used to evoke channel openings. (B) Representative recording of openings in a patch that contained few channels. (C) Representative recording for a patch that contained many channels.

Since channels were clustered, we measured unitary open time distributions by using small depolarizing steps to voltages (usually −30 or −20 mV) that resulted in rare openings of single channels (Fig. 4A). In control patches, most open time distributions were best fit using a sum of two exponentials, with short openings predominant (Figs. 4B and 5). Occasionally, an equally good fit was obtained using a single exponential, as the longer open time component was essentially absent.

Fig. 4.

Ros effects on single channel open times. (A) Openings were evoked by voltage steps from −120 mV to −30 mV. Representative sweeps from an untreated cell mostly show openings of short duration. (B) Open time distribution for a representative untreated patch is best fit by the sum of two exponentials, with short openings predominant. (C) Representative sweeps from a Ros-treated patch; under these conditions we observed a mix of short and longer openings. (D) Open time distribution for the Ros-treated patch shown in (C) is best fit by the sum of two exponentials, but the longer component is prolonged and more evident than in untreated cells (B). Representative records in (A) and (C) were taken from full sweeps that were 500 ms in duration and are not consecutive; a solid line is drawn through the zero current level, and a dashed line is drawn through the unitary open current level.

Fig. 5.

Summary data for Ros effects on open time distributions. (A) Effect on open time constants. Average values are shown for short (circles) and long (squares) components in control (open symbols) and Ros-treated patches (filled symbols) recorded using voltage steps to either −30 or −20 mV. In control patches open times (in ms) at −30 mV (τ1 = 0.35 ± 0.02, τ2 = 1.24 ± 0.20) or at −20 mV (τ1 = 0.44 ± 0.03, τ2 = 1.36 ± 0.18) were similar. Ros did not affect the short open time at either voltage (τ₁ = 0.38 ± 0.02 at −30mV; 0.40 ± 0.03 at −20mV), but significantly increased the long open time (τ₂ = 3.42 ± 0.20 at −30 mV; 3.03 ± 0.20 at −20 mV). (B) Effect on open time weights for steps to −30 mV. Ros significantly increased the probability of long openings. *significantly different, student’s t-test, p<0.05; n = 8-10 patches at each voltage.

We next included 100 μM Ros in the patch pipette solution to determine resulting changes in the open time distributions. In the presence of the drug, longer openings were more frequent (Fig. 4C) although distributions remained best fit by a sum of two exponentials (Fig. 4D). While there was no significant effect on the shorter time constant, the longer time constant was significantly increased (Figs. 4D and 5A). Therefore, Ros selectively prolonged the open time for the longer gating component of the N-type Ca²⁺ channel. In addition, we showed that Ros exposure significantly increased the probability of the longer openings (Fig. 5B).

We also determined whether single channel conductance was affected by Ros (Fig. 6). In control patches, conductance was 17.4 pS (Fig. 6B, open circles), and no significant difference was observed after Ros treatment (Fig. 6B, filled circles). To further ensure that drug treatment had no effect, we examined a subpopulation of openings consisting only of events longer than 1 msec (>3 SD longer than the average short time constant). Under these conditions, virtually all of the events should reflect the behavior of Ros-modified channels. As shown by the filled squares in Fig. 6B, we still observed no significant change in single channel conductance. Thus, Ros does not alter the conductance of this splice variant of the N-type Ca²⁺ channel.

Fig. 6.

Single channel conductance is unaffected by Ros treatment. (A) Representative sweeps from Ros-treated patches recorded at the indicated voltages (mV). Solid and dashed lines indicate zero and unitary current levels (as in Fig. 4). (B) Current-voltage relationships for control patches (open circles; n = 4-11 patches at each voltage), Ros-treated patches (filled circles; n = 4-10), and for selected long openings (see text) from Ros-treated patches (filled squares; n = 4-10). Regression line shown for control data has a slope of 17.4 pS; fitting Ros-treated data produced a similar result.

To increase the frequency of observing channels gating with the drug-prolonged open time, we applied a 20 msec depolarizing prepulse to a voltage (+60 mV) near the Ca²⁺ channel reversal potential (see Fig. 1B) before stepping back to the test voltage (−30 mV) to measure currents. The prepulse had no significant effect on short and long open time values (data not shown), but, as expected, did increase the probability of observing long openings. We analyzed openings that occurred at the beginning of the test pulse (10-50 msec after the prepulse) separately from those that occurred at the end of the test pulse (460-500 msec after the prepulse). The probability (weight) obtained for long openings was much higher for early as opposed to late openings (29% vs. 11%, respectively). These data are consistent with the known use-dependent nature of (R)-roscovitine modulation (Fig. 2C) previously characterized using only whole cell recordings (Buraei et al., 2005; 2007).

Lastly, we examined unitary and multi-channel currents in the presence of 100 μM Ros. As illustrated by Fig. 7A, we used a prepulse to +60 mV (as above) and then stepped to −60 mV to measure the currents. Figure 7B shows a series of representative sweeps that illustrate how channels opened by the step to +60 mV can remain open for variable lengths of time after repolarization, occasionally resulting in two openings superimposed on one another. On the other hand, we never detected consecutive unitary openings that could be interpreted as single channel re-opening, as has been suggested for one type of dihydropyridine-sensitive L-type calcium channel (Kavalali and Plummer, 1994; 1999; Hivert et al., 1999). Figure 7C shows an ensemble current generated by averaging 100 individual sweeps like those of Fig. 7B. As expected from the appearance of the single sweeps, the ensemble average could be fit by a sum of two exponentials, and, in addition, was similar in time course to the whole cell tail current shown previously in Fig. 2A. Therefore, we conclude that the prolonged single channel open time resulting from Ros exposure can account for the slowed current deactivation observed in whole cell recordings. As noted above, some L-type calcium channels show dihydropyridine-dependent re-openings near resting membrane potential and this may contribute to prolongation of corresponding tail currents, but based on our data, this mechanism does not apply to Ros-mediated effects on N-type calcium channels.

Fig. 7.

Unitary and composite tail currents after Ros treatment. (A) Voltage protocol used to evoke channel openings. (B) Representative sweeps from a Ros-treated patch (sweeps are not consecutive; solid and dashed lines as in Figs. 4 and 6). (C) Ensemble average current from all sweeps recorded from this patch, including nulls (n = 100). Deactivation was fit with the sum of two exponentials (τ₁ = 0.6; τ₂ = 3.6 ms) and was similar to the Ros-modulated whole cell current deactivation shown in Fig. 2A.

A kinetic scheme for N-type calcium channel gating and modulation by roscovitine

To further analyze our experimental data, we developed a model of N-type calcium channel kinetics. Buraei et al. (2005) previously have suggested several possible kinetic schemes based on their recordings of Ros effects on whole cell calcium currents. Our channel state diagram (Fig. 8A) is a hybrid between schemes 2 and 3 in Buraei et al. (2005), and was parameterized to reproduce previously determined experimental estimates of channel opening during an action potential (p = 0.2 – 0.3; Poage and Meriney, 2002; Wachman et al., 2004; King and Meriney, 2005, Luo et al., 2009), as well as the open time distributions obtained under control conditions and after Ros treatment in the present study. We first tried schemes 2 and 3 exactly as reported previously (Buraei et al., 2005), but under those conditions the simulated probability of opening during an action potential waveform (p = 0.84 and 0.77 respectively for schemes 2 and 3) and open time distributions were inconsistent with our experimental observations. Therefore, we introduced an additional closed state (C4 in Fig. 8A) to avoid an overly large opening probability during an action potential and to reproduce our experimental single channel measurements (see below). Our need for an additional closed state thus was very different from the motivation behind two additional closed states connected by voltage independent rates as used by Yarotskyy et al. (2009), who modeled gating charge movements and hence focused on data quite different from those reported here.

As introduced above, our kinetic scheme (Fig. 8A and Table 1) satisfies the relatively low opening probability of N-type calcium channels during an action potential (p = 0.3) and also reproduced our experimental open time distributions and fitting weights for control and Ros-treated preparations (compare Figs. 8B and C with Figs. 4B and D; Table 2). To determine simulated open times, we used a step depolarization to −20 mV as in our experimental protocol (see Experimental Procedures). Our simulations also reproduced the experimental effects of Ros on whole cell currents (compare Fig. 9A to Fig. 2C). In particular, we simulated our experimental prepulse protocol and were able to reproduce the use dependence of Ros-mediated modulation of calcium channel gating. Our simulations showed that slowed deactivation was more pronounced when the prepulse was increased from 2 to 14 ms, whereas control currents simulated without Ros-modulated states were unaffected by the length of the prepulse.

Table 1.

As described in Buraei et al. (2005), voltage-dependent rate constants (kxx) in the kinetic model (see Fig. 8) are computed using: k_x = A_x exp ((V − C_x)z_xF/RT). For all simulations, [R] = 100 μM. The values shown here are consistent with microscopic reversibility for the cyclical transitions introduced by Ros binding

	A	z
k12	5000	0.8
k21	500	−0.8
k23	4000	0.9
k32	500	−0.9
k34	1200	0.3
k43	8000	−0.3
k45	10000	0.3
k54	1600	−0.3
k56	380	0.3
k65	595	−0.3
k57	4[R]	--
k75	280	--
k68	5[R]	--
k86	400	--
k78	152	0.3
k87	272	−0.3

Table 2.

Comparison of fitting results for experimental and simulated open time distributions (based on a depolarizing step to −20 mV)

Condition	tau 1 (ms)	tau 1 weight (%)	tau 2 (ms)	tau 2 weight (%)
Control, experimental	0.44	90	1.36	10
Control, simulated	0.37	80	1.35	20
Ros-treated, experimental	0.40	84	3.03	16
Ros-treated, simulated	0.34	71	3.68	29

Fig. 9.

Simulations of whole cell currents. (A) Currents were generated using voltage protocols indicated by symbols as in Fig. 2C. (B-E) Time courses of fractional occupancy (open states) under control and Ros-treated conditions. Each panel corresponds to the currents identified by the matching symbols in (A). Occupancy plots were generated by averaging over 10,000 simulations, each of which contained a single channel.

Having fit the kinetic model to our observed experimental data, we used additional simulations to trace transitions of the channel through the different closed and open states, and to quantify average residence time in each state. This analysis then allowed us to correlate transitions and open times in the simulations with observed open times obtained with experimental recordings. Once again we used a simulated step depolarization to −20 mV to match our experimental conditions. As shown in Fig. 8D, under control conditions the first open state (O₅) was populated rapidly and predominated as the channel transitioned between closed states and open states O₅ and O₆. In addition, the occupancy plots for C₄ and O₅ exhibited significantly faster fluctuations than seen for O₆, thus suggesting rapid transitions between the two proximal states. By directly counting transitions between C₄ and O₅, we confirmed that the channel moved between these two states much more frequently than between O₅ and O₆ (data not shown). The mean residence times in the O₅ and O₆ states were 0.59 and 1.35 ms, respectively. The values, together with frequent fluctuations between C₄ and O₅, indicate that the predominant short openings seen in experimental recordings correspond to channel flickering between C₄ and O₅, while the less frequent long openings seen experimentally occur when the channel also enters the O₆ state.

With Ros treatment, i.e., in the presence of the RO₇ and RO₈ states, the mean residence times for O₅ and O₆ changed little (to 0.53 and 0.92 ms, respectively), but the channel now resided in the RO states for a significant amount of time. Analysis of the single state occupancy plot (Fig. 8E) revealed that the channel spent most of its time in states O₅ and RO₇, with the latter having the longest mean open time (2.53 ms, vs. 1.40 ms for RO₈). As in the control case, the most frequent fluctuations were between C₄ and O₅, and thus the frequent short experimental openings that we observed again correspond to channel flickering between C₄ and O₅. Experimental Ros-dependent changes (more frequent longer openings) resulted from the addition of the long-lived RO₇ state, as well as an increase in the number of transitions between all of the neighboring open states (as explicitly confirmed by counting all transitions between the four open states, data not shown). Despite four distinct mean open times in the simulations, our simulated open time distribution was best fit by the sum of only two exponentials, as was our experimental data (Figs. 8C and 4D; Table 2). This collapse of four open states into two observed exponentials results from several relatively similar mean residence times and relatively infrequent occupancy of several open states (Fig. 8E), yielding populations of openings that resolve into only two apparent mean open times.

As shown in Fig. 9, we also used our kinetic model to simulate whole cell tail currents based on a protocol similar to that used experimentally (Fig. 2C). Simulated current amplitudes obtained with 10,000 channels approximated experimental amplitudes (Fig. 9A). As in experiments, deactivation kinetics were best fit by a sum of two exponentials (in ms, control: τ₁ = 0.34, τ₂ = 0.97; Ros-treated: τ₁ = 0.63, τ₂ = 4.35). Also consistent with experiments, a longer prepulse had no significant effect on control currents, while Ros-modulation increased with a longer prepulse (τ₂ weight increased from 70% to 84% when the prepulse increased from 2 to 14 ms). In addition, Figs. 9B-E show the time course of occupancy plots for control (Figs. 9B and D) and Ros-treated (Figs. 9C and E) conditions. In our model, use-dependent Ros modulation occurs because Ros binds only to the open states and maximal Ros occupancy is attained only after 5 to 10 ms have elapsed during the prepulse (Fig. 9E).

Discussion

We studied the agonist effects of Ros on unitary N-type Ca²⁺ channel currents to determine the effects of this novel Ca²⁺ channel modulator on single channel gating. Under control conditions, we found that N-type Ca²⁺ channels expressed in the tsA201 cell line gate with either a short (frequent) or long (infrequent) mean open time. After acute Ros exposure, the long gating component was selectively prolonged and increased in frequency (as was predicted by Buraei et al., 2005, based only on whole cell recordings). These single channel effects appear to underlie the previously reported agonist effects of Ros on calcium current and transmitter release (Buraei et al., 2005; 2007; Cho and Meriney, 2006). We showed that tsA201 cells express N-type channels in clusters and found that most patches contained multiple channels, precluding analysis of unitary closed time distributions, and thus also precluding detailed investigations of inactivation states and effects on probability of opening (which may underlie the reported antagonist effects obtained with high Ros concentrations; see Buraei and Elmslie, 2008).

Gating of calcium channels

Historically, L-type voltage-gated Ca²⁺ channel gating modes were classified as “mode 0” when the channel was closed or unavailable, “mode 1” when the channel opened with a brief mean open time (less than 1 ms), and “mode 2” when the channel opened with a long mean open time (lasting many ms; Hess et al., 1984; Nowycky et al., 1985). Interestingly, several L-type Ca²⁺ channel agonists (BayK 8644, CGP 28392, and FPL 64176) have been shown to promote prolonged “mode 2” gating (Kokubun and Reuter, 1984; Hess et al., 1984; Nowycky et al., 1985; Kunze and Rampe, 1992) in a manner that appears similar to the effects reported here for Ros on N-type Ca²⁺ channels. Similar gating “modes” also have been described in earlier studies of unitary N- or R-type Ca²⁺ channel openings (Plummer and Hess, 1991; Delcour et al., 1993; Rittenhouse and Hess, 1994; but see Elmslie et al., 1994; Elmslie, 1997). In some of these studies, N- or R-type channels were observed to gate in one of two or three modes characterized by open probability, inactivation kinetics, open time, closed time, and current amplitude. Voltage-gated Ca²⁺ channels appear to gate in these different modes presumably by adopting different protein conformations, and may be able to switch between these modes quickly (see Plummer and Hess, 1991; Delcour et al., 1993; Rittenhouse and Hess, 1994; Lee and Elmslie, 1999). Various forms of physiological modulation have been shown to favor Ca²⁺ channel gating in one particular mode or another depending on the manner in which the channel protein is altered (Delcour and Tsien, 1993; Ono and Fozzard, 1993; Patil et al., 1996; Carabelli et al., 1996; Lee and Elmslie, 2000; Liu and Rittenhouse, 2000; Colecraft et al., 2001; Zhong et al., 2001).

In this study, we characterized a Ros-mediated shift in the propensity of the N-type Ca²⁺ channel to gate in a low frequency, prolonged open state that is ~3 times longer than in untreated cells. Furthermore, we have developed a kinetic model that captures these findings and which has allowed us to gain microscopic insight into the dynamics of the N-type Ca²⁺ channel under both control and Ros-modulated conditions. The effects of Ros on N-type Ca²⁺ channels are reminiscent of the effects of BayK 8644 and FPL 64176 in the promotion of so-called “mode 2” gating in some L-type channels (see Hess et al., 1984; Nowycky et al., 1985; Tavalin et al., 2004). In particular, Tavalin et al. (2004) quantified mean open times in control and agonist-modified L-type channels under different divalent ion conditions. Using this approach, they showed that L-type unitary conductances recorded in high Ba²⁺ gated with both a short and long mean open time, and that BayK 8644 and FPL 64176 prolonged and increased the frequency of the long openings. Thus, our observed effects of Ros on N-type Ca²⁺ channels are similar to observations for some L-type Ca²⁺ channels modulated by agonists.

Clustered events of a particular mean open time are sometimes interpreted as indicative of modal gating (see for example, Nowycky et al., 1985). In our recordings, patches usually contained many channels, forcing us to step to relatively weak depolarizations to see unitary openings, which also precluded analysis of closed times and clusters of openings. In addition, our results suggest that Ros modulation was present for only a small percentage of channels in the patch (as would be expected at −20 mV), so infrequent prolonged openings could be interspersed with many other un-modulated brief openings. Thus, we would not expect to detect clustered long openings, even if such could occur under other conditions.

Earlier studies disagree with regard to changes in single channel current amplitude during long openings. Some report an increase (Delcour et al., 1993; Rittenhouse and Hess, 1994; Cloues et al., 1997; Tavalin et al. 2004), while others report no change (Hess et al., 1986; Fox et al., 1987). Consistent with the latter, in the present study we found no change in current amplitude after Ros treatment, and thus conclude that N-type Ca²⁺ channels of this central nervous system splice variant (Ca_v2.2e[24a, Δ31a]; see Lin et al., 2004) do not show a change in single channel conductance despite a prolongation of mean open time. It is conceivable that different reported findings for agonist effects on single channel conductance are specific to different subtypes of calcium channels.

Impact of single channel gating on calcium-triggered transmitter release

Single channel gating properties are critical determinants of the magnitude and temporal characteristics of Ca²⁺ flux into the presynaptic nerve terminal. At synapses where vesicle release can be triggered by Ca²⁺ flux through a single open Ca²⁺ channel (Stanley, 1993; Yoshikami et al., 1989; Wachman et al., 2004; Shahrezaei et al., 2006), the gating properties of these channels can have an especially significant impact on the triggering of vesicle fusion. Even small changes in the kinetics of channel gating can have dramatic effects on the magnitude and temporal characteristics of Ca²⁺ influx during an action potential (see McCobb and Beam, 1991; Sabatini and Regehr, 1999; Meinrenken et al., 2003). Given the steep non-linear dependence of transmitter release on Ca²⁺ influx (Dodge and Rahamimoff, 1967), even small changes in Ca²⁺ influx can have significant effects on transmitter release. Therefore, a Ros-mediated change in presynaptic Ca²⁺ channel gating could have a significant impact on the local Ca²⁺ dynamics around open Ros-bound channels. Such effects likely would explain reported Ros-mediated increases in transmitter release at the frog neuromuscular junction (Cho and Meriney, 2006), where vesicle fusion is triggered predominantly by Ca²⁺ ions entering through one or two open Ca²⁺ channels in the active zone (Yoshikami et al., 1989; Wachman et al., 2004; Shahrezaei et al., 2006; Luo et al., 2009).

Abbreviations

Ros

(R)-roscovitine

cdk

cyclin-dependent kinase

Ca²⁺

calcium

P_o

Probability of opening

DMEM

Dulbecco’s Modification of Eagle’s Medium

HEPES

4-(2-hydroxyethyl)-1-piperazine ethane sulfonic acid

EGTA

ethylene glycol tetraacetic acid

DMSO

Dimethyl sulfoxide

EDTA

ethylene diamine tetraacetic acid

TEA

tetraethylammonium

MCell

Monte Carlo simulator of cellular microphysiology