Ca²⁺- and Voltage-Dependent Inactivation of Ca²⁺ Channels in Nerve Terminals of the Neurohypophysis

Calcium current inactivation

Calcium currents were measured in response to single 500 msec pulses or to trains of 2 msec pulses. Currents elicited by a 500 msec pulse from −100 to 10 mV appeared to inactivate in two phases (Fig.1A). Double exponential fits to the decay of these Ca²⁺ currents yielded the following parameters: τ₁ = 33.7 ± 5.5 msec, τ₂= 365 ± 12 msec, andI_ss/I_peak = 0.023 ± 0.005 (n = 16). Ca²⁺currents also were elicited by physiological trains of 2 msec pulses from −100 to 50 mV designed to mimic bursts of action potentials (Jackson et al., 1991). Trains were applied at frequencies of 14 and 20 Hz, because these frequencies are relevant to neuropeptide secretion from the neurohypophysis (Poulain and Wakerley, 1982). As a 14 Hz train progressed, peak current activated by each 2 msec pulse was less than that activated by the preceding pulse. Plotting peak current versus time showed a monoexponential decay (Fig. 1B; τ₁ = 6.4 ± 0.5 sec; n = 6). After 15 sec the current was 15% below the current activated by the first pulse of the train. Ca²⁺ channel inactivation was stronger with 20 Hz trains, declining with a biexponential time course (τ₁ = 0.96 ± 0.42 sec and τ₂ = 5.8 ± 1.3 sec; n = 11) to 40% of the initial current after 15 sec. Thus, at the higher frequency a second, rapid component of inactivation became evident. The greater Ca²⁺ channel inactivation seen with 20 Hz trains may play a role in determining the optimal frequency for neuropeptide secretion. Further, the increase in Ca²⁺ channel inactivation with train frequency may be relevant to the fatigue of secretion, which is more pronounced at higher stimulation frequencies (Bicknell et al., 1984; Gainer et al., 1986; Hobbach et al., 1988).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Calcium current inactivation. A, Calcium current recorded in response to a 500 msec pulse from −100 to 10 mV. The decay shown was fit with a biexponential function with the following parameters: τ_fast = 35 msec, τ_slow = 126 msec, A_fast = −400 pA, A_slow = −336 pA, andA₀ = −32 pA (see Materials and Methods, Eq. 2). B, Ca²⁺ currents were evoked by trains of 2 msec pulses from −100 to 50 mV. Peak currents measured during trains were normalized to the first current response, averaged, and plotted versus time. Error bars are plotted at 1 sec intervals to avoid obscuring the data. Inactivation was frequency-dependent and decayed monoexponentially at 14 Hz (τ₁ = 6.45 sec;n = 6) and biexponentially at 20 Hz (τ₁ = 0.96 and τ₂ = 5.79 sec;n = 11). The fitted exponential functions were drawn in both A and B, but they are nearly concealed by the data.

Voltage- and Ca²⁺-dependent inactivation

To evaluate the contribution made by Ca²⁺ to the inactivation of Ca²⁺ channels, we replaced Ca²⁺ in the bathing solution by Ba²⁺, or we added 10 mm BAPTA to the patch pipette solution (see Materials and Methods). Normalized currents elicited by pulses from −100 to 10 mV for 500 msec are shown for each of these conditions in Figure 2A, with the Ca²⁺ current trace from Figure1A included for comparison. Figure 2Cshows representative normalized current–voltage (I–V) plots for control Ca²⁺currents, Ba²⁺ currents, and Ca²⁺currents with intracellular BAPTA. With Ba²⁺ as the charge carrier, the Ca²⁺ channel I–Vrelationship was shifted 20 to 30 mV to the left (more negative), as described previously in dissociated neurohypophysial nerve terminals (X. Wang et al., 1992).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Current through Ca²⁺ channels with Ba²⁺ substitution and intracellular BAPTA.A, Extracellular Ca²⁺ was replaced by Ba²⁺ (left), or 10 mmBAPTA was included in the patch pipette filling solution (right). Current was activated by 500 msec pulses from −100 to 10 mV, as in Figure 1. Current traces were normalized to their peak values and displayed together with normalized control Ca²⁺ current from Figure 1. Inactivation was quantified by fitting the decay of the current to a sum of exponentials. This yielded the following parameters for the traces shown: τ_fast = 26 msec, τ_slow = 144 msec,A_fast = −192 pA,A_slow = −247 pA, andA₀ = −91 pA for Ba²⁺substitution; τ_fast = 38 msec, τ_slow = 152 msec, A_fast = −397 pA,A_slow = −171 pA, andA₀ = −29 pA for intracellular BAPTA (see Materials and Methods, Eq. 2). B, Time constants for inactivation of Ca²⁺ channels are shown for pulses to 0 mV. The values for Ba²⁺ substitution and intracellular BAPTA differ significantly from controls (p < 0.01 for each pairwise comparison). For controls, n = 16; for Ba²⁺,n = 12; for BAPTA, n = 11.C, Plots of peak current versus voltage for control Ca²⁺ current (▪), Ba²⁺ current through Ca²⁺ channels (•), and Ca²⁺ current with intracellular BAPTA (▴).

The kinetics of inactivation of the Ca²⁺ currents shown in Figure 2A appear to be altered both by Ba²⁺ substitution and by intracellular BAPTA. The current decays were fit to a double exponential function (see Materials and Methods), and the results of these fits showed that the change in inactivation kinetics results from a shortening of the time constants of both the fast and slow components of inactivation (Fig.2B). The relative proportions of the two components of inactivation showed no significant differences among controls, Ba²⁺ substitution, and intracellular BAPTA (data not shown). Figure 2B shows the time constants for decay at 0 mV, a voltage at which Ca²⁺ current was near maximal. Inactivation kinetics also was analyzed over five voltage points in the range −20 to 20 mV, and a two-way ANOVA showed a statistically significant interaction between solution and voltage for τ_slow (p = 0.002) and for the ratio of final to peak current (p = 0.024; for control, n = 16; for Ba²⁺,n = 12; for BAPTA, n = 11). More data will be presented below showing a decrease in steady-state inactivation of Ca²⁺ channels resulting from Ba²⁺ substitution and intracellular BAPTA.

It is interesting that the time constants for current inactivation were faster with Ba²⁺ substitution and intracellular BAPTA when compared with Ca²⁺ (Fig.2B). Ca²⁺ channels in other preparations inactivate more slowly under these conditions (Chad and Eckert, 1984; Chad, 1989), and the increased speed of inactivation that we saw was the opposite of what one would expect for inactivation caused by intracellular Ca²⁺. Because additional results are presented below that are consistent with Ca²⁺-dependent inactivation, it is important to address this matter before proceeding. The situation can be clarified with the aid of a simple two-state model in which Ca²⁺ channels interconvert between an open (O) and inactivated (I) state with a forward rate constant, α, and a reverse rate constant, β. Equation 4For this model the time constant of inactivation and the steady-state fraction of inactivated channels are given by the expressions 1/(α + β) and α/(α + β), respectively. The more rapid time constant and lower level of steady-state inactivation observed here can be explained by Ca²⁺ inhibition of both the forward and reverse rates, with a greater effect on the reverse rate. This interpretation is supported by the results presented below showing a strong inhibitory effect of Ca²⁺ on recovery from inactivation (β) at negative potentials (see Figs. 6,7). Inhibition of recovery from inactivation by Ca²⁺has been seen in a number of other preparations (Brehm et al., 1980;Yatani et al., 1983; Gutnick et al., 1989). More detailed kinetic studies of Ca²⁺ channel inactivation have been undertaken by others to explain a large body of data (Gutnick et al., 1989; Mazzanti et al., 1991; Fryer and Zucker, 1993; Imredy and Yue, 1994). The comparatively simple two-state analysis used here is presented solely to clarify two apparently conflicting results of increased rate of inactivation and reduced steady-state level of inactivation.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

Recovery of Ca²⁺ current from inactivation. Ca²⁺ current was inactivated with 500 msec pulses from −100 to 10 mV. Then recovery was examined with 100 msec test pulses applied at various time intervals after the end of the inactivating pulse. A, Ba²⁺currents are shown, and in these traces the noise appears greater than it actually is because the four traces selected do not superimpose perfectly. B, Normalized peak current from traces such as those in A was plotted versus the interpulse interval to show the time course of recovery. The current at the end of the first inactivating pulse was subtracted from the peak current, and this difference was plotted versus interpulse interval. The best-fitting double-exponential functions were drawn through the data (see text for parameter values and n values). In these fits we imposed the constraint A_fast +A_slow = −A₀ and set t₀ = 0 (see Materials and Methods, Eq. 2). C, Data in B are replotted with an expanded time scale to show the rapid component of recovery more clearly.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.

Inverse correlation between recovery and [Ca²⁺]_i. The time course of recovery of Ca²⁺ current from inactivation (τ = 1.25 sec) was similar to that for recovery of [Ca²⁺]_i (τ = 2.20 sec). Data were pooled from five nerve terminals. Inset, [Ca²⁺]_i measured during a 300 msec pulse from −100 to 10 mV. Arrows indicate the times at which the test pulses were applied to produce points in the plot.

Steady-state inactivation

The contributions of voltage and Ca²⁺ to steady-state inactivation also were investigated by using a paired-pulse protocol to generate steady-state inactivation curves (Fig. 3). Long prepulses (500 msec) to different voltages were applied to allow inactivation to reach a steady state. Then the voltage was returned to a holding potential of −100 mV for 5 msec to close channels, after which a test pulse to 10 mV for 100 msec was applied to assess the state of inactivation of the channels. The peak current measured in response to this test pulse was normalized to the peak current of an identical control pulse given at least 1 min before the paired pulse protocol (Fig. 3A). Inactivation increased sharply as the prepotential increased from −60 to −20 mV, reaching a maximum of 93% at +10 mV (Fig. 3B). The data from −120 to 10 mV were fit to a Boltzmann function (see legend, Fig.3B), and this curve was extended to 70 mV to emphasize the upward slope of the data at more positive voltages. For Ca²⁺-dependent inactivation, an approximately U-shaped inactivation curve is expected, because as the prepulse voltage becomes more positive, Ca²⁺ influx first increases and then decreases (Fig. 2C). The data in Figure3B indicate a relatively weak component of Ca²⁺-dependent inactivation. Inactivation decreased from 93% at 10 mV to ∼65% at 70 mV. The data above 10 mV were fit to a line with a slope of 0.0040 ± 0.0003 mV⁻¹.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Steady-state inactivation. A, The current traces shown were recorded by using the steady-state inactivation protocol indicated schematically by theinset. The prepulses were −50 mV (a), −10 mV (b), and 50 mV (c). Control pulses were similar each time and therefore are not labeled. Note the decrease in inactivation with a prepulse of 50 mV (c) when compared with a prepulse of −10 mV (b).B, Peak test pulse currents were normalized to peak control pulse currents and plotted against the prepulse voltage (n = 21). Points from −120 to 10 mV were fit to a Boltzmann equation (I₁ = 0.98,I₂ = 0.07,V_½ = −30.5 mV, andk = 12.1 mV; see Materials and Methods, Eq. 3), and points from 10 to 70 mV were fit to a line (slope = 0.0040 ± 0.0003 mV⁻¹). Inactivation decreased at prepulse potentials more positive than 10 mV.

The possibility that positive prepulses may facilitate Ca²⁺ current (Hoshi et al., 1984) also was examined. When we used short prepulses of 25 msec to minimize inactivation, varied the prepulse potential from −40 to 100 mV, and varied the interpulse interval from 20 to 100 msec, we saw no evidence for facilitation of Ca²⁺ current in these nerve terminals (data not shown). Thus, the modest reduction in inactivation with increasing prepulse potential above 10 mV cannot be attributed to facilitation counteracting the effect of inactivation.

To determine whether the decline in steady-state inactivation at positive potentials in Figure 3B was accompanied by reduced Ca²⁺ entry, we made simultaneous measurements of Ca²⁺ current and [Ca²⁺]_i, using the Ca²⁺-sensitive fluorescent dye fura-2. Figure4A shows representative changes in [Ca²⁺]_i evoked by 500 msec prepulses to −50 mV (a), 10 mV (b), and 50 mV (c), each of which was followed by a 100 msec test pulse to 10 mV to assess the state of the Ca²⁺ channels (as in Fig. 3). Note that in these experiments [Ca²⁺]_i rose as high as 1.5 μm, which is considerably higher than that seen in previous studies (Stuenkel, 1990, 1994; Jackson et al., 1991). This difference reflects the lower amount of exogenous chelator used in the present experiments.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Steady-state inactivation is correlated with prepulse [Ca²⁺]_i. A, [Ca²⁺]_i was measured during the steady-state inactivation protocol illustrated in Figure 3. Prepulse steps were given at t = 0.5 sec for 500 msec to the following voltages: −50 mV (a), 10 mV (b), and 50 mV (c). Thearrow at the top indicates the [Ca²⁺]_i level at the end of the prepulse that is plotted in B. With no added Ca²⁺ buffer (other than 0.1 mm fura-2), resting [Ca²⁺]_i was 0.35 ± 0.04 μm (n = 7). B, Steady-state inactivation and [Ca²⁺]_iat the end of the prepulse (shown in A) are plotted as a function of prepulse voltage (n = 7). The fraction of noninactivated current and [Ca²⁺]_iis inversely related.

If intracellular Ca²⁺ contributes to Ca²⁺ channel inactivation, an inverse relationship would be expected between [Ca²⁺]_i at the start of the test pulse and the noninactivated Ca²⁺ current evoked by the test pulse. Pooled data from seven nerve terminals show a correlation (Fig.4B), but the decrease in [Ca²⁺]_i at voltages more positive than 10 mV is much greater than the decrease in inactivation. To explain this behavior entirely in terms of Ca²⁺-dependent inactivation would require a very steep functional dependence. Further support for inactivation attributable primarily to voltage rather than to Ca²⁺ is provided by the result with a prepulse voltage of −30 mV, for which there is strong inactivation (∼50%) with almost no increase in [Ca²⁺]_i. Thus, Figure 4B suggests that positive voltages can inactivate Ca²⁺ channels without raising [Ca²⁺]_i.

As an additional test of the Ca²⁺ dependence of steady-state inactivation of Ca²⁺ channels, the paired-pulse protocol was used with Ba²⁺substitution and with intracellular BAPTA. The results are presented in Figure 5, with control data from Figure 3Breproduced for comparison. These experiments showed that the Ca²⁺-dependent component of steady-state inactivation, represented by the upward slope above 10 mV, was reduced but not eliminated by either Ba²⁺ substitution (Fig.5A) or intracellular BAPTA (Fig. 5B). The slopes between 10 and 70 mV of 0.0030 ± 0.0011 mV⁻¹for Ba²⁺ and 0.0010 ± 0.0004 mV⁻¹ for BAPTA were significantly different from zero. It is unclear whether Ba²⁺ influences this slope, but the slope in BAPTA is significantly less than the control value of 0.004 mV⁻¹. Although these results are consistent with our interpretation of the upward slope in steady-state inactivation in terms of Ca²⁺ (Fig. 3B), they suggest that neither Ba²⁺ nor BAPTA is completely effective in blocking this effect. Intracellular BAPTA had little effect on steady-state inactivation at prepulse potentials below 10 mV (Fig. 5B). However, substitution of Ca²⁺ by Ba²⁺ reduced the maximum level of inactivation at 10 mV from 93 to 81% and shiftedV_1/2 by −11 mV (see legend for parameter values from Boltzmann fit, Fig. 5A). The decrease in maximum inactivation is consistent with a reduction in the Ca²⁺-dependent component of inactivation. The leftward shift in V_1/2 may be related to the shift in the I–V relationship produced by Ba²⁺ (Fig. 2C).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Dependence of steady-state inactivation on [Ca²⁺]_i. Steady-state inactivation of Ca²⁺ current was studied with the paired-pulse protocol shown in Figure 3, using Ba²⁺ substitution (A) and intracellular BAPTA (B). Points from −120 to −10 mV for Ba²⁺ and −110 to 10 mV for BAPTA were fit to a Boltzmann equation (see Materials and Methods, Eq. 3) to obtain the following parameters: for Ba²⁺,I₁ = 1.01, I₂ = 0.19, V_½ = −41.9 mV, andk = 10.0 mV; for BAPTA,I₁ = 0.98, I₂ = 0.11, V_½ = −27.4 mV, andk = 10.8 mV (see Materials and Methods, Eq. 3). The curves based on these fits were extended to 70 mV to emphasize the upward slope at positive potentials. Points from 10 to 70 mV were fit to a line (see text for slopes). Control data from Figure3B with 5 mm Ca²⁺ as the charge carrier are reproduced for comparison.

Ca²⁺ inhibition of recovery from inactivation

The results described above suggest relatively weak effects of intracellular Ca²⁺ on the rate and extent of inactivation at positive potentials. Recovery from inactivation is another process that can be influenced by Ca²⁺(Brehm et al., 1980; Yatani et al., 1983; Gutnick et al., 1989). To study recovery, we inactivated most of the Ca²⁺current with a 300 msec pulse to 10 mV, and after a variable recovery period at the holding potential of −100 mV, we applied a second 100 msec pulse to 10 mV to assess recovery. Results with control solutions, Ba²⁺ substitution, and intracellular BAPTA show that recovery from inactivation is faster in the Ba²⁺ and BAPTA solutions (Fig. 6), implying that recovery is inhibited by intracellular Ca²⁺. These experiments also resolved recovery into at least three distinct kinetic phases, the first two of which were analyzed by fitting to a double exponential function. The third component was too slow to quantify easily but appeared to require >10 sec.

The data in Figure 6C suggest that the rapid component of recovery is strongly inhibited by intracellular Ca²⁺and that intracellular BAPTA is more effective at removing this inhibition than Ba²⁺ substitution. The time constant of the rapid component decreased from 0.17 ± 0.03 sec in Ca²⁺ (n = 18) to 0.10 ± 0.01 sec with Ba²⁺ substitution (n = 12) and 0.073 ± 0.018 sec with intracellular BAPTA (n= 9). The weight of the rapid component increased from 0.42 ± 0.07 with Ca²⁺ to 0.67 ± 0.05 with Ba²⁺ substitution and 0.86 ± 0.06 with intracellular BAPTA. The intermediate time constant of recovery was less affected by Ca²⁺ removal: τ_intermediate = 1.58 ± 0.36 sec in Ca²⁺ and 1.82 ± 0.69 sec with Ba²⁺ substitution. With intracellular BAPTA the value seemed to be similar, but the weight of this component was too small to permit a quantitative determination of the time constant. In summary, Ba²⁺ substitution and intracellular BAPTA accelerated recovery by increasing both the speed and the weight of the fast component. Intracellular Ca²⁺ thus seems to inhibit the recovery of Ca²⁺ channels from inactivation at negative potentials. Recall that Ca²⁺ inhibition of the recovery rate, β, was invoked earlier in our analysis of inactivation at positive potentials in terms of a two-state model (Eq. 4). Thus, it seems that intracellular Ca²⁺ inhibits the reverse process at both positive and negative potentials.

Insight into the relationship between intracellular Ca²⁺ and the recovery of Ca²⁺current from inactivation can be gained by inspection of our fluorometric [Ca²⁺]_i measurements (Fig. 4). The return of [Ca²⁺]_i to baseline requires seconds and is clearly too slow to account for the fastest component of recovery of Ca²⁺ current. However, the intermediate component of recovery from inactivation appeared to have a similar rate. To compare these two processes further, we made simultaneous measurements of [Ca²⁺]_i and Ca²⁺current, using pairs of pulses as in Figure 6. In this experiment interpulse intervals of 1–10 sec were used, so the rapid component of Ca²⁺ current recovery was complete at the start of the test pulse. Current recovery approximately mirrored the return of [Ca²⁺]_i to baseline (Fig.7). The recovery of current and [Ca²⁺]_i each was fit to single exponentials, yielding comparable time constants of τ = 1.25 ± 0.77 and τ = 2.20 ± 0.31 sec, respectively (p = 0.14). (Note that the time constants here are similar to the intermediate time constant of recovery of Ca²⁺ current of 1.58 sec from Fig. 6.) Thus, the intermediate component of recovery of Ca²⁺ current from inactivation could reflect Ca²⁺ removal from the nerve terminal. This is consistent with the finding that, when Ca²⁺ rises were prevented by intracellular BAPTA (Fig. 6), the weight of the intermediate component was reduced from 0.58 to only 0.14.

Frequency dependence of Ca²⁺-dependent inactivation

The data in Figure 1B showed that Ca²⁺ current inactivation during trains of brief pulses increased when the train frequency was increased from 14 to 20 Hz. To determine whether Ca²⁺ influences inactivation during trains, we performed experiments like those in Figure 1 by using intracellular BAPTA. Intracellular BAPTA rather than Ba²⁺ substitution was chosen because of its greater efficacy in accelerating recovery from inactivation (Fig. 6) and in removing the upward slope in steady-state inactivation above 10 mV (Fig. 5). These experiments showed that intracellular BAPTA reduced inactivation during 20 Hz trains (Fig.8A). In contrast to the control data in Figure 1, the time constants of inactivation with BAPTA in response to trains of stimulation were the same for 14 and 20 Hz. This suggests that the frequency-dependent component of inactivation is mediated by Ca²⁺ entry. The results above obtained with long-duration pulses indicated that at positive potentials intracellular BAPTA accelerated both inactivation and recovery, and at negative potentials intracellular BAPTA accelerated recovery. The reduction of Ca²⁺ channel inactivation by intracellular BAPTA during high-frequency trains, therefore, would appear primarily to be attributable to an acceleration of recovery of Ca²⁺ current from inactivation during the intervals between pulses.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 8.

Ca²⁺ dependence of inactivation during trains. A, Ca²⁺ current was evoked by a train of 2 msec pulses, as in Figure1B, but with BAPTA included in the patch pipette solution. BAPTA reduced inactivation during a 20 Hz train to that seen with a 14 Hz train. Normalized peak Ca²⁺ currents were averaged and plotted versus time with error bars at 1 sec intervals. The time constants from fits of double exponential functions (see Materials and Methods, Eq. 2) were τ₁ = 0.69 and τ₂ = 26.3 sec at 14 Hz (n = 4), and τ₁ = 0.40 and τ₂ = 24.7 sec at 20 Hz (n = 5). B, [Ca²⁺]_i rises to a plateau during 14 and 20 Hz trains (with no intracellular BAPTA). The increases and decreases were both faster at 20 Hz. Baseline [Ca²⁺]_i was 300 nm in this terminal, and the plateau was 0.7 μm for both frequencies (see text for averages). The inset shows two selected [Ca²⁺]_i traces from different terminals on an expanded time scale to illustrate the different rates of rise for 14 and 20 Hz trains. The time constants in these two traces were 3.2 and 0.5 sec, respectively (see text for average time constants for both rises and falls in [Ca²⁺]_i).

If, as suggested by the data in Figure 8A, the greater inactivation of Ca²⁺ current with 20 Hz trains in Figure 1B is attributable to a greater rise in [Ca²⁺]_i, we would expect that the [Ca²⁺]_i level reached during a 20 Hz train would be significantly higher than that reached during a 14 Hz train. However, we found no significant difference in the level of [Ca²⁺]_i reached during trains at these two frequencies (Fig. 8B). The increase in [Ca²⁺]_i was 0.46 ± 0.06 μm at 14 Hz (n = 5) and 0.55 ± 0.13 μm at 20 Hz (n = 5). Although the plateau level of [Ca²⁺]_i was not significantly different, the rate of rise and the rate of decay of [Ca²⁺]_i were faster during 20 Hz bursts. The comparison of the rates of rise can be seen in the inset of Figure 8B. At 20 Hz we found τ_rise = 0.62 ± 0.08 sec and τ_decay = 1.82 ± 0.17 sec (n = 5). At 14 Hz we found τ_rise = 1.92 ± 0.41 sec and τ_decay = 3.18 ± 0.61 sec (n = 5). The threefold increase in rate of rise in [Ca²⁺]_i during 20 Hz trains suggests that recovery from inactivation between pulses will be inhibited much earlier than during a 14 Hz train, and this is consistent with the faster time course of Ca²⁺ current inactivation seen during a 20 Hz train (Fig. 1B). It is still difficult to reconcile such a large change in Ca²⁺-dependent Ca²⁺ current inactivation with such small differences in [Ca²⁺]_i, and possible explanations are considered in Discussion.

Ca²⁺-dependent enzymes

The Ca²⁺-dependent protein phosphatase calcineurin (protein phosphatase 2B) has been proposed to mediate Ca²⁺-dependent inactivation of Ca²⁺ channels by dephosphorylation of the channel or of a closely associated protein (Chad and Eckert, 1986; Armstrong, 1989). To evaluate the role of calcineurin in the inactivation of neurohypophysial Ca²⁺ channels, we examined Ca²⁺ current during 20 Hz trains in the presence of two enzyme inhibitors. The choice of 20 Hz trains was based on the fact that this stimulus protocol revealed Ca²⁺-dependent Ca²⁺ channel inactivation especially clearly (Fig.1B vs 8A). After 15 sec, peak Ca²⁺ current was inactivated by 40 ± 4% (n = 11) in control recordings (Fig.1B), and intracellular BAPTA reduced this inactivation to 20.0 ± 1.5% (n = 5) (Fig.8A). We first tested RS-20, a 20-amino-acid peptide that binds calmodulin with nanomolar affinity and blocks most of the known Ca²⁺-calmodulin-dependent enzymes (Lukas et al., 1986), including calcium/calmodulin-dependent protein kinase II and calcineurin. Because the concentration of calmodulin in cells can be quite high, we added 50 μm RS-20 to the patch pipette filling solution and waited 5 min to allow this compound to diffuse into the cell. Ca²⁺ channels were seen to inactivate by 35 ± 6% (n = 6), which was indistinguishable from controls (p = 0.24), and different from that seen with intracellular BAPTA (p = 0.027).

Calcineurin can be activated weakly by direct binding of Ca²⁺ to its regulatory domain (Stemmer and Klee, 1994; Perrino et al., 1995). Therefore, cyclosporin A, a direct inhibitor of calcineurin (Liu, 1993), also was tested. With 50–100 nm cyclosporin A in the patch pipette, Ca²⁺ current still inactivated 35% ± 9% after 15 sec (n = 5; p = 0.28 vs control andp = 0.07 vs intracellular BAPTA). Experiments with cyclosporin A must be interpreted in light of the fact that the receptor for this agent is a small soluble protein (Liu, 1993), which could wash out in whole-terminal patch recordings such as these. However this drug was effective in experiments of similar design in hippocampal neurons (Tong et al., 1995), suggesting that the negative result here is meaningful. These results therefore suggest a mechanism of Ca²⁺-dependent inactivation of Ca²⁺ current that is independent of calmodulin and calcineurin.

Calcium- and voltage-dependent inactivation

The experiments presented here show that Ca²⁺channel inactivation in neurohypophysial nerve terminals is sensitive to both voltage and intracellular Ca²⁺. The effect of intracellular Ca²⁺ was very prominent at negative potentials at which an inhibition of the recovery of Ca²⁺ channels was seen (Figs. 6, 7). Inhibition of recovery from inactivation by Ca²⁺ has been reported for other Ca²⁺ channels as well (Brehm et al., 1980;Yatani et al., 1983; Gutnick et al., 1989). Intracellular Ca²⁺ also influenced Ca²⁺ channel inactivation at positive potentials, but the effect at positive potentials was overshadowed in large part by voltage-dependent inactivation (Fig. 4). At positive potentials we were surprised to see that removing Ca²⁺ made inactivation faster. In many other preparations Ba²⁺ substitution and intracellular chelators reduced the rate of current decay during sustained depolarization (Chad and Eckert, 1984; Chad, 1989). Therefore, the increase in the rate of Ca²⁺ channel inactivation shown here for both Ba²⁺ substitution and intracellular BAPTA (Figs. 2A,B) represents a property unique to the Ca²⁺ channels of the neurohypophysis. Other unique properties of neurohypophysial N-type Ca²⁺ channels have been described, including lower single channel conductance and more rapid inactivation, as compared with N-type channels in cell bodies (Lemos and Nowycky, 1989; Wang et al., 1993).

The decrease in steady-state inactivation with prepulse potentials above 10 mV (Fig. 3) was correlated with the decrease in Ca²⁺ current (Figs. 2, 4). A U-shaped steady-state inactivation curve is a hallmark of Ca²⁺-dependent inactivation (Chad and Eckert, 1984; Imredy and Yue, 1994), but in other preparations this shape has been attributed to voltage-dependent facilitation (Slesinger and Lansman, 1991). Using voltage pulses too brief to inactivate channels, we saw no evidence for facilitation. Thus, the fact that intracellular BAPTA reduced the upward slope in the steady-state inactivation plot (Fig. 5B) strengthens the case for a Ca²⁺-dependent component of inactivation. Neither Ba²⁺ substitution nor intracellular BAPTA removed Ca²⁺-dependent inactivation completely, and this is consistent with studies in other preparations showing that Ba²⁺ could substitute weakly for Ca²⁺ in Ca²⁺-dependent inactivation (Brehm and Eckert, 1978; Tillotson, 1979; Gutnick et al., 1989; Mazzanti et al., 1991; Imredy and Yue, 1994) and that intracellular EGTA reduced but did not abolish the effect (Brehm and Eckert, 1978; Bechem and Pott, 1985).

Mechanism of Ca²⁺-dependent inactivation

Early work in Helix neurons suggested dephosphorylation of Ca²⁺ channels by calcineurin as the mechanism of Ca²⁺-dependent Ca²⁺ channel inactivation (Chad and Eckert, 1986; Armstrong, 1989). However, more recent studies favor a mechanism involving a direct action of Ca²⁺ ions. Studies with enzyme inhibitors failed to reveal a role for protein phosphatases in Ca²⁺channel inactivation (Fryer and Zucker, 1993; Imredy and Yue, 1994). Further, Ca²⁺ channels reconstituted into artificial bilayers could be cycled through Ca²⁺-dependent inactivation and reactivation without ATP (Haack and Rosenberg, 1994). Finally, deletion of a segment of the Ca²⁺ channel that contains a putative Ca²⁺ binding site removes Ca²⁺-dependent inactivation (de Leon et al., 1995). Our finding that RS-20 and cyclosporin A had no influence on Ca²⁺ channel inactivation in neurohypophysial nerve terminals during trains of pulses is consistent with these diverse results on other Ca²⁺ channels and argues against an enzymatic mechanism for inactivation.

An enzymatic mechanism seems unlikely, but a simple binding process seems to be at odds with our finding that 14 and 20 Hz trains raised [Ca²⁺]_i to similar levels (Fig.8B), because these two train frequencies produced very different amounts of Ca²⁺ channel inactivation (Fig. 1B). Stuenkel (1994) saw increases in [Ca²⁺]_i with train frequency, but his plots showed relatively small changes in the 14–20 Hz range as well. Although the [Ca²⁺]_i levels were similar for these two frequencies, the dynamics of the rises and falls in [Ca²⁺]_i were quite different. The more rapid rise in [Ca²⁺]_i may be responsible for the rapid component of Ca²⁺ channel inactivation seen with 20 Hz trains. Furthermore, these dynamic aspects lead one to consider the possibility that [Ca²⁺]_i around channels rises and falls very rapidly during a train and that the time course of these changes varies with frequency. Such changes would not be evident in our measurements of the spatial average of [Ca²⁺]. It is well known that [Ca²⁺] is very different within domains around single channels and clusters of channels (Chad and Eckert, 1984; Augustine and Neher, 1992; Llinás et al., 1995;Tucker and Fettiplace, 1995). The domains around individual Ca²⁺ channels are not likely to be relevant in the present study because they collapse within <1 msec after channel closure and therefore would disappear at negative potentials at which a large effect of Ca²⁺ on recovery from inactivation is seen. However, domains around clusters of Ca²⁺channels are thought to be relevant to Ca²⁺-dependent Ca²⁺ channel inactivation (Imredy and Yue, 1992), and these larger domains should decay more slowly. Fixed Ca²⁺ buffers (Zhou and Neher, 1993) and Ca²⁺-dependent cytoskeletal interactions (Johnson and Byerly, 1993) also could help Ca²⁺ domains to persist after channel closure. Estimates of [Ca²⁺]_i levels that inactivate Ca²⁺ channels vary widely among preparations and are often highly model-dependent (Gutnick et al., 1989; Romanin et al., 1992; Fryer and Zucker, 1993; Hirano and Hiroaki, 1994), but many of these studies suggest that [Ca²⁺]_i is >1 μm and well above the spatial averages we observed during trains (∼0.5 μm). If the effect of Ca²⁺ on Ca²⁺ channels has a kinetic delay and if the time course of [Ca²⁺] within domains depends on train frequency, then this could help to explain why 20 Hz trains produce greater Ca²⁺ channel inactivation than 14 Hz trains while raising average [Ca²⁺]_i to similar levels.

Calcium currents in nerve terminals and the relevance to secretion

The Ca²⁺ currents described in the present study in slices of the neurohypophysis were similar in many respects to those described in dissociated neurohypophysial nerve terminals (Lemos and Nowycky, 1989; X. Wang et al., 1992, 1993). In this preparation the slow time constant of inactivation ranged from 0.07–1.25 sec and theV_1/2 of steady-state inactivation ranged from −75 to −60 mV. Our slow time constant of inactivation fell in the range of values from neurosecretosomes, but ourV_1/2 value of −30 mV was significantly different. This difference may be attributable in part to experimental conditions. In neurosecretosomes longer inactivating pulses were used, and exogenous Ca²⁺ buffer was added to the patch pipette solutions. The Ca²⁺ dependence of Ca²⁺ channel inactivation has yet to be investigated in neurosecretosomes, but Ca²⁺ channels in two other nerve terminal preparations, retinal bipolar nerve terminals (von Gersdorff and Matthews, 1996) and rat brain synaptosomes (Tareilus et al., 1994), have been shown to inactivate by a Ca²⁺-dependent mechanism.

Neurosecretosomes contain channels closely related to N-type and L-type Ca²⁺ channels (Lemos and Nowycky, 1989; Wang et al., 1993). Evidence also has been presented for P-type Ca²⁺ channels in the neurohypophysis (Salzberg et al., 1990; Wang and Lemos, 1994). On the basis of these other studies we can assign the rapidly inactivating component of Ca²⁺ current to N-type Ca²⁺channels and the slowly inactivating component to L-type and P-type Ca²⁺ channels. Most studies showing Ca²⁺-dependent inactivation were conducted on L-type channels (Chad, 1989), but Ca²⁺-dependent inactivation has been reported in N-type (Cox and Dunlap, 1995) and P-type (Tareilus et al., 1994) Ca²⁺ channels as well. N-type Ca²⁺ channels are thought to play the greater role in secretion from the neurohypophysis (Dayanithi et al., 1988; Obaid et al., 1989; von Spreckelsen et al., 1990), as well as from most other nerve terminals (Dunlap et al., 1995). However, L-type Ca²⁺ channels also contribute to secretion, not only in the neurohypophysis (Nowycky, 1991; Bicknell et al., 1993; Stuenkel and Nordmann, 1993; Raji and Nordmann, 1994) but in a number of other preparations (Perney et al., 1986; Artalejo et al., 1994; Loechner et al., 1996). L-type Ca²⁺ channels will be especially important during trains because of the rapid inactivation of N-type Ca²⁺ channels. The variable contributions made by different Ca²⁺ channel types to secretion partly may reflect dependence on stimulation protocol. Ca²⁺ channel inactivation behavior such as that described here may account for some of this.

Ca²⁺ channel inactivation was evident during physiological trains of action potential-like pulses. A higher train frequency produced more inactivation (Fig. 1B), and intracellular BAPTA abolished this frequency dependence (Fig.8A). Vasopressin and oxytocin have optimal frequencies for stimulation of secretion (Dutton and Dyball, 1979;Poulain and Wakerley, 1982; Gainer et al., 1986) and exhibit use-dependent depression of secretion as the frequency is increased (Gainer et al., 1986; Hobbach et al., 1988). The results presented here suggest that Ca²⁺ current inactivation would vary significantly, depending on stimulation frequency and duration, and these variations in Ca²⁺ channel activity could influence neuropeptide secretion. Other membrane processes also have been proposed to contribute to the fatigue of secretion, including the activation of a Ca²⁺-dependent K⁺channel (Bielefeldt et al., 1992; G. Wang et al., 1992; Bielefeldt and Jackson, 1993), the activation of a slowly activated K⁺ channel termed the D channel (Bielefeldt et al., 1992), and accumulation of extracellular K⁺ (Leng and Shibuki, 1987). It is likely that a variety of mechanisms comes into play, acting together to determine how neurosecretion will vary with different forms of electrical activity. Diverse mechanisms such as these can serve to integrate many features of a complex electrical stimulus, including burst frequency, duration, and interburst interval, to determine the amount of hormone released.