Asynchronous Ca²⁺ current conducted by voltage-gated Ca²⁺ (Ca_V)-2.1 and Ca_V2.2 channels and its implications for asynchronous neurotransmitter release

Abstract

We have identified an asynchronously activated Ca²⁺ current through voltage-gated Ca²⁺ (Ca_V)-2.1 and Ca_V2.2 channels, which conduct P/Q- and N-type Ca²⁺ currents that initiate neurotransmitter release. In nonneuronal cells expressing Ca_V2.1 or Ca_V2.2 channels and in hippocampal neurons, prolonged Ca²⁺ entry activates a Ca²⁺ current, I_Async, which is observed on repolarization and decays slowly with a half-time of 150–300 ms. I_Async is not observed after L-type Ca²⁺ currents of similar size conducted by Ca_V1.2 channels. I_Async is Ca²⁺-selective, and it is unaffected by changes in Na⁺, K⁺, Cl⁻, or H⁺ or by inhibitors of a broad range of ion channels. During trains of repetitive depolarizations, I_Async increases in a pulse-wise manner, providing Ca²⁺ entry that persists between depolarizations. In single-cultured hippocampal neurons, trains of depolarizations evoke excitatory postsynaptic currents that show facilitation followed by depression accompanied by asynchronous postsynaptic currents that increase steadily during the train in parallel with I_Async. I_Async is much larger for slowly inactivating Ca_V2.1 channels containing β_2a-subunits than for rapidly inactivating channels containing β_1b-subunits. I_Async requires global rises in intracellular Ca²⁺, because it is blocked when Ca²⁺ is chelated by 10 mM EGTA in the patch pipette. Neither mutations that prevent Ca²⁺ binding to calmodulin nor mutations that prevent calmodulin regulation of Ca_V2.1 block I_Async. The rise of I_Async during trains of stimuli, its decay after repolarization, its dependence on global increases of Ca²⁺, and its enhancement by β_2a-subunits all resemble asynchronous release, suggesting that I_Async is a Ca²⁺ source for asynchronous neurotransmission.

Voltage-gated Ca²⁺ (Ca_V) channels are activated by depolarization and conduct inward Ca²⁺ currents that initiate many cellular responses to electrical signaling (1). Ca_V2 channels conduct P/Q-, N-, and R-type currents in presynaptic nerve terminals, and Ca²⁺ entry through these channels initiates neurotransmitter release at conventional synapses (1–3). P/Q- and N-type Ca²⁺ currents conducted by Ca_V2.1 and Ca_V2.2 channels, respectively, are more efficiently coupled to neurotransmitter release than R-type Ca²⁺ currents conducted by Ca_V2.3 channels (4, 5). Neurotransmitter release occurs in two phases: a fast synchronous (phasic) component and a slow asynchronous (tonic) component (6). The slower asynchronous component of release is proposed to result from residual Ca²⁺ remaining after an action potential acting on a different Ca²⁺ sensor than the sensor for synchronous neurotransmitter release (6–8). Remarkably, when synchronous release is prevented by deletion of its Ca²⁺ sensor synaptotagmin, the asynchronous release process can cause exocytosis of the entire readily releasable pool, suggesting a functional competition between synchronous and asynchronous release processes for the same pool of docked and primed synaptic vesicles (7). In contrast to synchronous release, both the source of Ca²⁺ entry and the intracellular Ca²⁺ sensor for asynchronous neurotransmitter release remain unknown, although Doc2 has emerged as a strong candidate for the Ca²⁺ sensor (9).

Ca_V2.1 channels are regulated by binding of calmodulin (CaM) and related calcium sensor proteins to a bipartite binding site in the C-terminal domain, leading to Ca²⁺-dependent facilitation (CDF) and Ca²⁺-dependent inactivation (CDI) during trains of depolarizing stimuli (10–13). Recent work has shown that these forms of Ca²⁺-dependent regulation of presynaptic Ca²⁺ channels modulate synaptic transmission to produce short-term facilitation and short-term depression (14–17). To explore the possible role of presynaptic Ca²⁺ channels as a source of Ca²⁺ entry that triggers asynchronous neurotransmitter release, we searched for a form of Ca²⁺-dependent regulation of Ca_V2 channels that could contribute to asynchronous neurotransmitter release. Here, we show that prolonged Ca²⁺ entry through Ca_V2.1 or Ca_V2.2 channels caused by long depolarizations or trains of depolarizations at 100 Hz elicits an inward Ca²⁺ current on membrane repolarization. We have termed this current I_Async, because it is activated asynchronously with respect to membrane depolarization. The asynchronous appearance of I_Async, which occurs on repolarization rather than being synchronous with depolarization, the rise of I_Async during repetitive pulse trains, its rate of decay after repolarization, and its sensitivity to EGTA closely resemble asynchronous transmitter release. This previously unrecognized Ca²⁺ signal may contribute to the residual Ca²⁺ that triggers asynchronous release of neurotransmitters.

Results

Prolonged Ca²⁺ Entry Activates an Asynchronous Ca²⁺-Dependent Inward Current.

To identify previously unrecognized forms of Ca²⁺-dependent regulation of Ca_V2.1 channels expressed in human embryonic kidney tsA-201 cells, the membrane potential was depolarized from −80 to +30 mV for 1 s to generate global rises in intracellular Ca²⁺ and induce CDI (11); then, the membrane potential was repolarized to −80 mV for 1 s. This voltage protocol activated a previously unrecognized inward current that was present immediately on membrane repolarization and decayed with a time constant of ∼200 ms (Fig. 1A, black). Even our highest resolution recordings did not reveal a clear rising phase for this current, consistent with a rise time of <1 ms (Fig. S1A). We have termed this current I_Async, because its activation is asynchronous with membrane depolarization. A 1-s depolarization did not induce I_Async with Ba²⁺ substituted for Ca²⁺ as the charge carrier, indicating that I_Async is a Ca²⁺-dependent current (Fig. 1A, red).

Fig. 1.

Activation of a Ca²⁺-dependent asynchronous inward current, I_Async, on membrane repolarization. (A) Voltage protocol and averaged currents for Ca_V2.1 channels with β_2a-subunits normalized to the peak of the current during depolarization. (B) Mean values measured from currents in A. Black, Ca²⁺; red, Ba²⁺. (Upper Left) I_Async measured at −80 mV after a 1-s depolarization. (Upper Right) Total charge accumulation during the depolarization measured by integrating the current before normalizing. (Lower) I_Async normalized to either the peak of the inward current during the depolarization (Lower Left) or the integrated current (Lower Right). *P < 0.05; **P < 0.01. (C) Example currents recorded during a train of depolarizations from −80 to +20 mV at 100 Hz. (Upper) The examples are taken from the indicated times in the train. I_Async is the inward current present at −80 mV. The red line indicates 0 pA current. (Scale bars: 400 pA, 5 ms.) (Lower) Black lines indicate Ca²⁺ currents. The timescale does not permit resolution of individual currents. However, peak current amplitudes show facilitation followed by inactivation. The red area represents I_Async. (D) Averaged I_Async measured at −80 mV after each depolarization and normalized to the peak of current during the first depolarization to +20 mV and plotted against time during the train. (E) Mean values from currents in D. Black, Ca²⁺; red, Ba²⁺. (Upper Left) I_Async after the last tail current of the train. (Upper Right) Total charge accumulation during the train measured by integrating all currents before normalizing. (Lower) I_Async at the end of the train normalized to either the peak current during the first depolarization to 20 mV during the train (Lower Left) or the total charge accumulation during the train (Lower Right). *P < 0.05. (F) Reversal potential of I_Cav2.1 and I_Async. (Upper) Currents elicited by a 1-s depolarization to +30 mV followed by a ramp from +30 to +110 mV to measure I_Cav2.1 and activate I_Async. I_Async followed by I_Cav2.1 was then measured during steps to potentials from +10 to +70 mV separated by 5-ms repolarizations. The 5-ms repolarizations deactivate I_Cav2.1 but not I_Async. I_Cav2.1 and I_Async were then blocked by 2 mM Co and 0.2 mM Cd in the absence of CaCl₂, and I_Gating was measured in response to the same protocol. (Lower) Protocol schematic. Magnification of box from Upper with I_Async (red arrowhead) and I_Cav2.1 (blue arrowhead) after subtraction of I_Gating. I_Async was measured as the transient current before the rise of the large inward Ca_V2.1 current. (G, Upper) I to V relationship for mean data of I_Async (red) and I_Cav2.1 (blue) after subtraction of I_Gating during steps to the indicated potentials. (G, Lower) Mean reversal potentials of I_Cav2.1 (black) measured using the ramp from +30 to +110 mV after the 1-s depolarization, and I_Cav2.1 (blue) and I_Async (red) were measured using the steps from +10 to +70 mV separated by 5-ms repolarizations.

Ca²⁺ entry accumulating during repetitive depolarizations to 20 mV for 5 ms at 100 Hz elicited CDF followed by CDI (Fig. 1C, black and Fig. S1B), which was previously reported (11). Surprisingly, we also observed an inward current, I_Async, that was present at the holding potential and increased in amplitude during the train (Fig. 1 C, red, D, and E). Moreover, I_Async was not present when Ba²⁺ was the charge carrier, indicating that it is Ca²⁺-dependent (Fig. 1 D and E), similar to CDF and CDI (Fig. S1B) (11).

Because I_Async is Ca²⁺-dependent, it is necessary to normalize it to account for cell to cell variability in Ca²⁺ channel expression. To determine the best measure for normalization, we plotted I_Async vs. the peak of the Ca²⁺ current, the integral of the Ca²⁺ current, or the extent of inactivation at the end of the 1-s depolarization (Fig. S2), which is dependent on global intracellular Ca²⁺ (11). We found that I_Async correlates better with the peak current and integrated current than with percent inactivation (Fig. S2). Normalizing to peak current or integrated current produced similar results (Fig. 1 B and E); therefore, we have only shown data normalized to peak current in the subsequent figures.

Ion Selectivity of I_Async.

The reversal potential of I_Async was measured after subtracting the gating current caused by outward movement of the voltage sensors of the Ca_V2.1 channels, because the capacitative current generated by gating charge movement was comparable in magnitude with I_Async near the reversal potential for Ca²⁺ current. The reversal potential for I_Async was 35.0 ± 2.4 mV, identical to I_Cav2.1 (37.7 ± 2.5 mV) (Fig. 1 F and G), showing that I_Async is a Ca²⁺-selective current. There are no Na⁺ or K⁺ ions in our solutions, but protons are present, and there is a high concentration of Cl⁻. Neither replacing Cl⁻ with SO₄²⁻ (Fig. 2 A–D) nor lowering the pH of the extracellular or intracellular solution (Fig. 2 E–H) altered I_Async, eliminating contributions from Ca²⁺-activated Cl⁻ channels (18) and voltage-gated proton channels (19) to I_Async. Replacing Tris⁺ in the extracellular medium with Na⁺ and K⁺ had no effect (Fig. 2I), indicating that Na⁺ or K⁺ channels do not contribute significantly to I_Async. In addition, specific blockade of Ca²⁺-activated K⁺ channels (20), transient receptor potential channels (21), hyperpolarization-activated cyclic nucleotide-gated channels (22), or ryanodine receptors (23) did not affect I_Async (Fig. 3 A–E and G–K). I_Async was also unaffected by depletion of intracellular stores with thapsigargin (Fig. 3 F and L), indicating that Ca²⁺ release-activated Ca²⁺ currents do not contribute to I_Async. However, application of Zn²⁺ blocked I_Async and the Ca_v2.1 tail current with similar affinity (Fig. 4 A and B). Together, these results provide strong evidence that I_Async is a Ca²⁺-dependent and Ca²⁺-selective current that is conducted by the transfected Ca_V2.1 channels themselves and is blocked by Zn²⁺.

Fig. 2.

Ion selectivity of I_Async. (Left) Averaged currents elicited by a 1-s depolarization to +30 mV normalized to the peak of the current during depolarization as in Fig. 1A (A, E, G, and I) or averaged I_Async elicited by a train of depolarizations from −80 to +20 mV at 100 Hz measured after the tail current at −80 mV normalized to the peak of the first current at +20 mV and plotted against time during the train (C). Right. Mean values (± SEM) measured from currents in left panels (B, D, F, and H). I_Async normalized to the peak of the inward current during the depolarization (Upper). Total charge accumulation during the depolarization measured by integrating the current before normalizing (Lower). (A–D) Black, CaCl₂; red, CaSO₄. (E and F) Black, extracellular (EC) pH 7.3; red, pH 7. (G and H) Black, intracellular (IC), pH 7.3; red, IC pH 7.0. (I) Currents were evoked by a 2-s depolarization from −80 to +20 mV in control extracellular solution containing 10 mM Ca²⁺ (black). Cells were allowed to recover for 2 min, and 145 mM Na⁺ and 5 mM K⁺ were added to the extracellular solution (red). The Ca²⁺ concentration remained at 10 mM. (I, Inset) Percent change of I_Async in extacellular solution containing Na⁺ and K⁺ compared with control.

Fig. 3.

I_Async in the presence of channel and pump inhibitors. (A–F) Averaged currents elicited by a 1-s depolarization to +30 mV normalized to the peak of the current during depolarization in control (black) and the presence of a specific ion channel or ion pump inhibitor (red). (G–L) Mean (±SEM) values measured from currents in A–F. Black, control; red, with inhibitors. (A and G) 10 mM tetraethylammonium (TEA). (B and H) 100 nM apamin. (C and I) 100 μM 2-APB. (D and J) 100 μM ZD-7288. (E and K) 10 μM ryanodine. (F and L) 2 μM thapsigargin.

Fig. 4.

Blockade of I_Async. (A) Averaged currents elicited by a 1-s depolarization to +30 mV normalized to the peak of the current during depolarization in control (black) and Zn²⁺ (red). (Left) Peak of I_Cav2.1 tail current (black arrow) is significantly reduced in the presence of 100 μM Zn²⁺ (red arrow). (Right) Magnification of I_Async showing similar level of block as for the I_Cav2.1 tail current in Left measured in the presence of 100 μM Zn²⁺. I_Cav2.1 during the depolarization is unaffected by 100 μM Zn²⁺, and therefore, Ca²⁺ entry is spared. (B) Bar graphs showing the percent change of the peak I_Cav2.1 (yellow, I_peak), tail I_Cav2.1 (blue, I_tail), and I_Async (green) compared with control (black) in the indicated concentrations of Zn²⁺.

Ca²⁺ Entry Through the Ca_V1.2 Channel Does Not Induce I_Async.

To develop additional support for the hypothesis that the transfected Ca_V2.1 channels themselves both induce and conduct I_Async without a contribution from other channels endogenous to tsA-201 cells, we tested whether similar amounts of Ca²⁺ entry into tsA-201 cells through a Ca_V1.2 channel can induce I_Async. Ca_V1.2a_1800stop channels containing a stop codon at amino acid 1800, where these channels are proteolytically cleaved in vivo (24), produce larger Ca²⁺ currents than the full-length Ca_V1.2 channels (24, 25), which should favor activation of I_Async. However, like WT Ca_V1.2 channels, Ca_V1.2a_1800stop inactivates more rapidly than Ca_V2.1, resulting in decreased Ca²⁺ entry (Fig. 5A, compare black with blue). To prolong Ca²⁺ entry through Ca_V1.2a_1800stop channels, we blocked CDI by coexpressing a mutant CaM (CaM₃₄) containing alanine substitutions that impair Ca²⁺ binding to its two C-terminal EF hands (Fig. 5A, red) (26). Coexpression of CaM₃₄ does not alter inactivation of Ca_V2.1 channels (Figs. 1A and 2A, P = 0.32), because EF hands 3 and 4 of CaM support facilitation but not inactivation of Ca_V2.1 channels (12, 13).

Fig. 5.

I_Async conducted by Ca_V2.1 or Ca_V1.2 channels. Black, Ca_V2.1 coexpressed with CaM₃₄; blue, Ca_V1.2a_1800stop; red, Ca_V1.2a_1800stop coexpressed with CaM₃₄. *P < 0.05; **P < 0.01. (A) Mean normalized currents and I_Async elicited by a 1-s depolarization to 30 mV. (B, Upper) I_Async normalized to the peak of the inward current during the depolarization from A. (B, Lower) Total charge accumulation during the depolarization measured by integrating the current in A before normalizing. (C) Mean I_Async elicited by a train of depolarizations from −80 to 20 mV at 100 Hz measured at −80 mV, normalized to the peak of the first current at 20 mV, and plotted against time during the train. (D, Upper) I_Async measured at the end of the train normalized to the peak of the first current at 20 mV during the train as in C. (D, Lower) Total charge accumulation during the train measured by integrating all currents from data in C before normalizing.

Ca²⁺ entry through Ca_V2.1 channels modulated by CaM₃₄ generates I_Async after a 1-s depolarization (Fig. 5 A and B, Upper) that is similar to I_Async in cells expressing only Ca_V2.1 channels (Fig. 1, P = 0.1). In contrast, although 1-s depolarizations produce a similar amount of Ca²⁺ entry, Ca_V1.2a_1800stop channels plus CaM₃₄ produce little or no I_Async (Fig. 5A). The small amount of apparent I_Async seen in the mean data (Fig. 5B, Upper) arises from residual tail current of the slowly deactivating Ca_V1.2 channel rather from than from a distinct I_Async, which is shown in experiments with repetitive depolarizing pulses (Fig. 5 C and D). During trains of stimuli, Ca_V2.1 channels plus CaM₃₄ conduct a similar level of I_Async as in the absence of CaM₃₄ (Fig. 5C, P = 0.5). In contrast, there is no accumulation of I_Async during the train for Ca_V1.2a_1800stop plus CaM₃₄ (Fig. 5C), and the small residual tail current of the Ca_V1.2 channels does not increase during the train (Fig. 5C). Although there is a small but significant (P < 0.05) reduction in Ca²⁺ entry during trains in cells expressing Ca_V1.2a_1800stop and CaM₃₄ (Fig. 5D), this decrease is insufficient to explain the lack of I_Async. The lack of I_Async after a 1-s depolarization in cells expressing Ca_V1.2a_1800stop channels (Fig. 5A) and the absence of a time-dependent change in I_Async during trains despite similar Ca²⁺ influx (Fig. 5 C and D) indicate that Ca²⁺ entry through these channels does not induce I_Async in tsA-201 cells. These results support the conclusion that Ca_V2.1 channels themselves conduct I_Async rather than other channels endogenous to the tsA-201 cells.

Both Ca_V2.1 and Ca_V2.2 Channels Conduct I_Async.

If I_Async is important for neurotransmitter release, it may be conducted by other Ca_V2 channels that initiate neurotransmitter release. To examine this point, we measured I_Async following N-type Ca²⁺ currents conducted by Ca_V2.2 channels. Ca²⁺ entry through Ca_V2.2 channels was not significantly different from Ca²⁺ entry through Ca_V2.1 channels during 1-s depolarizations (Fig. 6 A and B, P = 0.26). I_Async was observed on repolarization after N-type Ca²⁺ currents conducted by Ca_V2.2 channels, similar to our experiments with Ca_V2.1 channels (Fig. 6 A and B, P = 0.15). During 100-Hz trains, the levels of Ca²⁺ entry and I_Async conducted by Ca_V2.2 channels were both somewhat smaller but not significantly different than I_Async conducted by Ca_V2.1 channels (Fig. 6 C and D, Upper [P = 0.22] and Lower [P = 0.07]).

Fig. 6.

I_Async conducted by Ca_V2.1 or Ca_V2.2 channels. Black, Ca_V2.1; red, Ca_V2.2. (A) Mean normalized currents and I_Async elicited by a 1-s depolarization to 30 mV. (B, Upper) I_Async normalized to the peak of the inward current during the depolarizations from A. (B, Lower) Total charge accumulation during the depolarizations measured by integrating the current during the depolarizations in A before normalizing. (C) Mean I_Async elicited by a train of depolarizations from −80 to 20 mV at 100 Hz measured at −80 mV, normalized to the peak of the first current at 20 mV, and plotted vs. time during the train. (D, Upper) I_Async measured at the end of each train normalized to the peak of the first current at 20 mV during the train. (D, Lower) Total charge accumulation during each train measured by integrating all of the currents from data in C before they were normalized.

I_Async and Synaptic Transmission in Cultured Hippocampal Neurons.

Single cultured hippocampal neurons form synapses on themselves called autapses that can be activated by generation of action potentials by depolarization of the neuronal cell body (27). We measured somatic P/Q- and N-type Ca²⁺ and Ba²⁺ currents from these neurons under whole-cell voltage clamp in the presence of tetrodotoxin to block Na⁺ currents and nimodipine to block L-type Ca²⁺ currents (Fig. 7A). I_Async was observed on repolarization after activation of P/Q- and N-type Ca²⁺ currents, and it was blocked by substitution of Ca²⁺ with Ba²⁺ as for transfected Ca_V2.1 and Ca_V2.2 channels (Fig. 7 A and B). Repetitive Ca²⁺ entry through P/Q- and N-type Ca²⁺ currents increased I_Async during a 40-Hz train of depolarizations, and the increase was completely blocked by substitution of Ca²⁺ with Ba²⁺ (Fig. 7 C and D). These results show that I_Async is a conserved property of these two Ca_V2 channel subfamily members in neurons as well as in transfected cells.

Fig. 7.

I_Async conducted by Ca_V2.1 and Ca_V2.2 channels in hippocampal neurons. (A) I_Async recorded from endogenous Ca_v2.1 and Ca_v2.2 channels normalized to the peak of the inward current during the 1-s depolarization to 30 mV. (B, Upper) I_Async normalized to the peak of the inward current during the depolarizations from A. (B, Lower) Total charge accumulation measured by integrating the current during the depolarizations in A before normalizing. *P < 0.05. (C) I_Async measured on membrane repolarization after each pulse of a 40-Hz train normalized to the peak of the first pulse in Ca²⁺ (black) and Ba²⁺ (red). (D) Charge_Async (Q_Async) of each pulse during a 40-Hz train normalized to the peak of the first pulse in Ca²⁺ (black) and Ba²⁺ (red). (E) Example of an action current and EPSC. (F) Example of EPSCs recorded during a train of depolarizations from −80 to +30 mV at 20 Hz. Peak current amplitudes show initial facilitation followed by depression. The red area represents asynchronous release. (G) Phasic neurotransmitter release normalized to the peak of the first EPSC at 30 mV during a 20-Hz train. (H) Asynchronous neurotransmitter release as a percentage of the first EPSC at 30 mV during a 20-Hz train.

Because P/Q- and N-type Ca²⁺ currents initiate neurotransmitter release in hippocampal autapses, we measured monosynaptic excitatory post synaptic currents (EPSCs) and asynchronous neurotransmission elicited by action potentials generated by depolarization of the neuronal cell body through the patch pipette. Typical recordings showed an initial action current from the action potential followed by a monosynaptic EPSC (Fig. 7E). Trains of action potentials at 20 Hz elicited trains of EPSCs that showed facilitation followed by depression (Fig. 7 F and G). Asynchronous synaptic transmission increased steadily during the train (Fig. 7 F and H), similar to the increase in I_Asynch during 40-Hz trains of stimuli (Fig. 7 C and D). Phasic neurotransmission decreased by 60% by the end of the train, whereas asynchronous neurotransmission increased to 20% of the phasic EPSC amplitude (Fig. 7 G and H). These results reveal I_Async in hippocampal neurons and indicate that the increase in I_Async during trains of stimuli parallels the increase in asynchronous neurotransmission. A quantitative comparison of these results is not possible because the size and time course of Ca²⁺ transients would be different in the cell body and synapse, and the power law for dependence of asynchronous release on Ca²⁺ is unknown.

Regulation of I_Async by Ca_Vβ-Subunits.

Ca²⁺ channel β-subunits have a major influence on the cell surface expression, kinetics, and voltage dependence of gating of Ca²⁺ channels (28). To determine how Ca²⁺ channel β-subunit composition alters I_Async, we compared I_Async conducted by slowly inactivating Ca_V2.1 channels containing the β_2a-subunit with rapidly inactivating channels containing the β_1b-subunit expressed in tsA-201 cells. The enhanced voltage-dependent inactivation (VDI) of Ca_V2.1 channels containing β_1b-subunits (Fig. 8A) resulted in a substantial decrease in Ca²⁺ entry compared with the more slowly inactivating channels containing β_2a-subunits (Fig. 8B). I_Async measured after a 1-s depolarization was significantly smaller in channels containing β_1b-subunits than in channels containing β_2a-subunits (Fig. 8 A and B). Ca_V2.1/β_1b channels also inactivate more rapidly during 100-Hz trains than Ca_V2.1/β_2a channels, which prevents significant facilitation of Ca_V2.1/β_1b channels (29). As expected, Ca_V2.1/β_1b channels produced significantly less I_Async during trains than Ca_V2.1/β_2a channels (Fig. 8 C and D). These data show that Ca_Vβ-subunits, which alter the kinetics of voltage-dependent inactivation and Ca²⁺ entry (28), regulate activation of I_Async substantially. These results suggest that rapid voltage-dependent inactivation of I_Ca reduces the intracellular Ca²⁺ transient and opposes activation of I_Async.

Fig. 8.

Effect of β-subunit composition on I_Async. (A) Mean currents elicited by a 1-s depolarization to 30 mV normalized to the peak of the current during depolarization. (B) Data measured from currents in A. Black, β_2a; red, β_1b. (B, Upper) I_Async normalized to the peak of the inward current during the depolarization. (B, Lower) Total charge accumulation during the depolarization measured by integrating the current before normalizing. **P < 0.01. (C) Mean I_Async elicited by a train of depolarizations from −80 to 20 mV at 100 Hz measured at −80 mV and plotted against time during the train. (D) Data measured from currents in C. Black, β_2a; red, β_1b. (D, Upper) I_Async measured at the end of the train normalized to the peak of the first inward current during the train. (D, Lower) Total charge accumulation during the train measured by integrating all currents before normalizing. *P < 0.05; **P < 0.01.

Global Ca²⁺ Transients Induce I_Async.

Local rises in intracellular Ca²⁺ cause CDF of Ca_V2.1 channels, whereas global rises in intracellular Ca²⁺ induce CDI (11–13, 30). To determine whether I_Async depends on global rises in intracellular Ca²⁺, we increased the concentration of the slow Ca²⁺ chelator EGTA in the recording pipette from 0.5 to 10 mM, a concentration known to block induction of CDI by global rises in intracellular Ca²⁺ but not block induction of CDF by local rises in intracellular Ca²⁺ (11). We found that EGTA (10 mM) blocked I_Async without altering Ca²⁺ entry (Fig. 9 A and B). These data indicate that global rises in intracellular Ca²⁺ are required to induce I_Async.

Fig. 9.

Effects of Ca²⁺ and CaM on I_Async. (A and B) Effect of EGTA. Black, Ca_V2.1 with 0.5 mM EGTA in the recording pipette; red, Ca_V2.1 with 10 mM EGTA in the recording pipette. **P < 0.01. (A) Mean I_Async elicited by a train of depolarizations from −80 to +10 mV at 100 Hz measured at −80 mV and plotted against time during the train. (B, Upper) I_Async measured at the end of the train normalized to the peak of the first current at 20 mV during the train from A. (B, Lower) Total charge accumulation during the train measured by integrating all currents from data in A before normalizing. (C) Effect of pulse duration. (C, Upper) Normalized smoothed Ca²⁺ currents from an example cell showing activation of I_Async with increasing duration of the depolarization to 20 mV. Open circles mark the peak of I_Async. Triangles point to the end of the pulse at the point of maximal inactivation. Cells were allowed to recover for 2 min after each depolarization. Each colored trace illustrates the tail current and I_Async for repolarization at the time indicated on the abscissa. (C, Lower) Mean I_Async (n = 5) normalized to the peak of the inward current during the depolarization and plotted against length of the depolarization. The smooth curve is a fit to the kinetic model (Eq. 1)

where Eq. 1 represents the effect of Ca²⁺ concentration on I_Async and Eq. 2 represents the estimate of intracellular Ca²⁺ that depends on the Ca²⁺ current and a Ca²⁺-dependent decline in intracellular Ca²⁺ caused by Ca²⁺ binding and clearance (∝ = 0.224 ± 0.040). (D and E) Effect of mutant CaM. Black, Ca_V2.1; red, Ca_V2.1 coexpressed with CaM₁₂₃₄. (D) Mean I_Async elicited by a train of depolarizations from −80 to 20 mV at 100 Hz measured after the tail current at −80 mV, normalized to the peak of the first current at 20 mV, and plotted vs. time during the train. (E, Upper) I_Async measured at the end of the train normalized to the peak of the first current at 20 mV during the train from D. (E, Lower) Total charge accumulation during the train measured by integrating all currents from data in D before normalizing. (F and G) Effect of mutations in Ca_V2.1 CaM binding site. Black, Ca_V2.1; red, mutant IM-AA/ΔCBD Ca_V2.1. *P < 0.05. (F) Mean I_Async elicited by a train of depolarizations from −80 to 20 mV at 100 Hz measured at −80 mV, normalized to the peak of the first current at 20 mV, and plotted vs. time during the train. (G, Upper) I_Async measured at the end of the train normalized to the peak of the first current at 20 mV during the train from F. (G, Lower) Total charge accumulation during the train measured by integrating all currents from data in F before normalizing.

To determine the time course of induction of I_Async by changes in global Ca²⁺, we measured I_Async activation during depolarizations of varying durations to 20 mV in transfected tsA-201 cells (Fig. 9C). Although the peak Ca²⁺ current was the same in each depolarization, the amount of Ca²⁺ entry increased with the length of depolarization, which is illustrated by the traces of different color (Fig. 9C, Upper). I_Async exhibited a U-shaped dependence on the length of depolarization, with an initial increase leading to a peak after a 3-s depolarization followed by a significant decrease (P < 0.01) to the end of the 5-s depolarizing pulse (Fig. 9C, Upper, open circles). This time course is consistent with a model based on the assumption that I_Async depends on the rise of intracellular Ca²⁺ concentration caused by Ca²⁺ entry through the Ca_V2.1 channels followed by Ca²⁺-dependent Ca²⁺ clearance (Fig. 9). A least-squares fit of the experimental data to this model yields a numerical solution that describes the data well (Fig. 9C, Lower), and this model provided a much better fit than one that did not include Ca²⁺-dependent Ca²⁺ clearance. These results suggest that the increase in intracellular Ca²⁺ induces I_Async steadily up to 3 s; then, I_Async begins to decrease as Ca_V2.1 channels inactivate, and the intracellular Ca²⁺ concentration falls during depolarizations longer than 3 s.

Although the amount of I_Async increases and decreases with intracellular Ca²⁺ during long depolarizations, there was no significant difference in the mean decay time constant for I_Async, which varied from 177 ± 46 ms after a 500-ms depolarization to 231 ± 24 ms after a 5-s depolarization. These results indicate that the decay of I_Async is not dependent on Ca²⁺ in the range of concentrations generated by channel opening, consistent with dissociation of Ca²⁺ from its regulatory site(s) as the rate-limiting step in decay of I_Async.

CaM Is Not the Ca²⁺ Sensor for I_Async.

CaM is a resident Ca²⁺ sensor on Ca_V2.1 channels, mediating Ca²⁺-dependent regulation of Ca_V2.1 channels by binding to two closely spaced sites in the intracellular C terminus of the channel, the IQ-like motif and the CaM-binding domain (CBD) (10, 12, 13). We tested the hypothesis that Ca²⁺ binding to CaM induces I_Async by cotransfecting Ca_V2.1 channels with mutant CaM (CaM₁₂₃₄ containing alanine substitutions that prevent Ca²⁺ binding to all four EF hands). These mutations block both CDF and CDI (12, 13). CaM₁₂₃₄ expression slightly but not significantly increased Ca²⁺ entry during a train of stimuli, but it did not alter I_Async (Fig. 9 D and E). These results show that Ca²⁺ binding to CaM is not required to generate I_Async. To further exclude the possibility that CaM binding to Ca_V2.1 channels mediates I_Async, we induced I_Async in cells transfected with the Ca_V2.1 channel mutant IM-AA/ΔCBD, which lacks a functional IQ-like motif and CBD and therefore, lacks both CDF and CDI (13). Although these mutations decreased Ca²⁺ entry into tsA-201 cells, they did not affect I_Async (Fig. 9 F and G). From these results, we conclude that Ca²⁺/CaM binding to the known sites in the C terminus of Ca_V2.1 channels is not required to generate I_Async.

Discussion

Our data reveal an unexpected behavior of Ca_V2.1 and Ca_V2.2 channels in transfected cells and cultured hippocampal neurons: after Ca²⁺ entry, these channels continue to conduct Ca²⁺ when the membrane potential is repolarized. This conductive state is clearly distinguishable from the tail current, which has much faster decay kinetics and is Ca²⁺-independent. I_Async is Ca²⁺-selective, not detectably permanent to Na⁺, K⁺, H⁺, and Cl⁻, and not affected by inhibitors of a broad range of ion channels. I_Async is not observed in tsA-201 cells expressing Ca_V1.2 channels with Ca²⁺ currents similar in size to those cells produced by Ca_V2 channels, indicating that I_Async is both induced and conducted by Ca_V2.1 and Ca_V2.2 channels without involvement of other channels present endogenously in tsA-201 cells.

Activation and Decay of I_Async.

The relationship between I_Async and pulse duration is well-fit by a model based on the assumption that I_Async is induced as the intracellular Ca²⁺ concentration increases dependent on the Ca²⁺ current; it declines as the Ca²⁺ current inactivates; and the intracellular Ca²⁺ concentration is decreased by a Ca²⁺-dependent extrusion process. Block of I_Async by dialysis of 10 mM EGTA, a concentration that blocks Ca²⁺ channel regulation by global rises in intracellular Ca²⁺ (11, 30), also supports this model. These results indicate that I_Async increases in response to the level of intracellular Ca²⁺ and thus, amplifies short-term increases in global Ca²⁺ by mediating additional Ca²⁺ entry with a long half-life.

Our results do not reveal the mechanistic basis for I_Async. However, because I_Async has a similar reversal potential to I_Cav2.1 and is blocked by Zn²⁺ to a similar extent as Ca_v2 tail current, it is likely that I_Async reflects the activity of Ca_V2 channels that are induced by an increase in global Ca²⁺ to either remain open or reopen on repolarization. These two possible mechanisms are distinct at the level of channel gating mechanisms. Because we observe no rising phase for I_Async (Fig. S1), our results are most consistent with a model in which the slowly decaying I_Async is induced progressively and conducts Ca²⁺ influx continuously during depolarizing pulses but is not clearly observed at that time because of the much larger conventional Ca²⁺ current. On repolarization, a tail current representing I_Async is observed as a slowly decaying inward current with a half-life of 200 ms at −80 mV. Most Ca²⁺ influx conducted by I_Async would occur during repolarization because of the larger driving force and long half-life. In the alternative model, Ca_V2 channels could open rapidly to conduct I_Async immediately on repolarization. However, this opening process must occur within 1 ms, because we have been unable to resolve a rising phase of I_Async (Fig. S1A), which would be observable if channels open after a longer delay. Ca_V2.1 channels deactivate within this time frame (Fig. 1), and therefore, a conformational change that activates I_Async could also occur within 1 ms. The lack of a rising phase of I_Async (Fig. S1A) is compatible with either a rapid, hyperpolarization-activated conformational change that mediates opening or a rapid, voltage-dependent dissociation of an intracellular blocking molecule that instantaneously allows current flow. Additional biophysical studies will be required to distinguish between these two distinct mechanisms for generation of I_Async.

I_Async seems to be similar, in some respects, to the resurgent current reported for voltage-gated Na⁺ channels -1.1 and -1.6 (31–33), but it is clearly distinguished by its strong regulation by Ca²⁺ and its nearly instantaneous appearance on repolarization. Additional structure-function work will be required to determine whether this mode of conductance reflects a conserved function of some types of voltage-gated Na⁺ and Ca²⁺ channels.

Site of Ca²⁺-Dependent Activation of I_Async.

Increases in global Ca²⁺ often activate CaM, which regulates Ca_V2.1 and Ca_V2.2 channels (10–13, 30). However, our experiments show that mutations that block CaM regulation of Ca_V2.1 channels (10, 12, 13) do not reduce I_Async. Thus, it is likely that an unrecognized Ca²⁺ sensor for regulation of Ca_V2.1 channels detects the global rise in intracellular Ca²⁺ that induces I_Async. All Ca_V1 and Ca_V2 channels contain an EF hand motif at the beginning of their C-terminal domains. We have found that this EF hand in Ca_V1.2 channels binds Mg²⁺ and mediates Mg²⁺-dependent inhibition of L-type Ca²⁺ currents (34, 35), but it is conceivable that it may function differently in Ca_V2 channels and induce I_Async. Other Ca²⁺-dependent signaling molecules, such as CaMKII (36), calcineurin (37), or other unidentified Ca²⁺ binding sites on Ca_V2.1 channels, could also be the Ca²⁺ sensor for I_Async.

Inactivation and I_Async.

Our data indicate that differences in VDI can profoundly alter I_Async activation. A major determinant of VDI is the Ca²⁺ channel β-subunit (28). Slowly inactivating Ca_V2.1 channels containing β_2a-subunits allow robust accumulation of I_Async compared with rapidly inactivating channels containing β_1b-subunits, which have markedly reduced I_Async induced by trains or 1-s depolarizations (Fig. 8). Reduced Ca²⁺ entry caused by enhanced inactivation with β_1b complicates interpretation of these results. Whether the reduction in I_Async accumulation reflects the difference in Ca²⁺ entry or a difference in the fraction of channels that are inactivated remains an open question.

I_Async and Asynchronous Neurotransmitter Release.

A central question raised by these results is what is the role of I_Async at the synapse? Ca²⁺ entry through Ca_V2 channels triggers the fusion of synaptic vesicles, initiating synaptic transmission that occurs in two phases: a fast synchronous component and a slower asynchronous component that builds during trains of action potentials. The fraction of synaptic vesicle exocytosis that is mediated by the slower asynchronous release process varies from synapse to synapse (38–44). In some synapses, asynchronous neurotransmitter release can exceed the size of synchronous release at the end of long trains of stimuli. It is thought that asynchronous release depends on residual Ca²⁺ that accumulates during trains of action potentials, because the Ca²⁺ chelator EGTA blocks asynchronous release (38–40, 42, 45).

Based on comparison of the properties of I_Async with the properties of asynchronous neurotransmitter release, we propose that I_Async may be a potential Ca²⁺ source for asynchronous neurotransmission. Asynchronous neurotransmitter release builds during trains of action potentials similar to the growth of I_Async during trains of depolarizations in transfected cells or neurons (38–40, 42, 45). Asynchronous neurotransmitter release decays over 100 ms or more after a train of stimuli (6) similar to I_Async. Asynchronous release is blocked by EGTA (38–41, 45) similar to I_Async. Overexpression of β_2a-subunits increases asynchronous neurotransmitter release in autaptic synapses formed by cultured hippocampal neurons (46) similar to the increased I_Async that we have observed for Ca_V2.1 channels containing β_2a-subunits (Fig. 8). Thus, our results suggest that I_Async, an unexpected mode of Ca²⁺-dependent modulation of Ca_V2 channels, may contribute to asynchronous synaptic transmission by providing a source of prolonged asynchronous Ca²⁺ entry after the presynaptic action potential that substantially increases residual Ca²⁺, engages a distinct Ca²⁺ sensor, and initiates asynchronous neurotransmitter release.

Materials and Methods

Cell Culture and Transfection of tsA-201 Cells.

Before transfection, tsA-201 cells were grown to ∼70% confluence in DMEM/Ham's F-12 with 10% FBS (Life Technologies) and 100 units/mL penicillin and streptomycin at 37 °C in 10% CO₂. Cells in 35-mm dishes were transfected with cDNAs encoding Ca²⁺ channel subunits-α₁ (2 μg), -β_2a (1.5 μg) or -β_1b (1.5 μg), -α₂δ (1 μg) using the Ca²⁺ phosphate method or Fugene 6 (Roche). Where indicated, 1 μg cDNA encoding mutant CaM was cotransfected. CD8 cDNA (0.3 μg) was included to identify transfected cells. The α₁-subunits used were the rbA isoform of α₁2.1, the e37a splice variant of α₁2.2 (47) (gift from Diane Lipscombe, Brown University, Providence, RI), and the rabbit cardiac muscle α₁1.2 truncated at amino acid 1800 (24).

Cell Culture of Single Hippocampal Neurons.

Neurons isolated from the hippocampi of postnatal day 0-1 WT (C57BL/6;129SvJ) mice were cultured on small microislands as previously described (48). Neurons were plated onto a feeder layer of astrocytes laid down 1–7 d earlier and used for recordings at 13–16 d in culture.

Electrophysiological Recording and Data Analysis.

Whole-cell voltage-clamp recordings were obtained from tsA-201 cells at room temperature 2–3 d after transfection. Unless stated otherwise, the extracellular solution contained 10 mM CaCl₂ or 10 mM BaCl₂, 150 mM Tris, 1 mM MgCl₂, and anti-CD8 beads (Dynal) to allow visualization of transfected cells. The intracellular solution consisted of 120 mM N-methyl-d-glucamine, 60 mM Hepes, 1 mM MgCl₂, 2 mM Mg-ATP, and 0.5 mM EGTA unless otherwise noted. The pH of intracellular and extracellular solutions was adjusted to 7.3 with methanesulfonic acid. TEACl (Sigma), ZnCl₂ (Sigma), 2-APB (Sigma), ZD7288 (Tocris), apamin (Tocris), ryanodine (Tocris), and thapsigargin (Tocris) were added to the extracellular solution. Recordings were made using an EPC10 patch-clamp amplifier with PULSE software (HEKA Elektronik) and filtered at 2.9 kHz. Leak and capacitive transients were subtracted using a P/−4 protocol. Because extracellular Ba²⁺ shifts the voltage dependence of activation by −10 mV, voltage protocols were adjusted to compensate for this difference. Cells were allowed to recover for 2 min between voltage protocols as previously discussed (29). Data analysis was performed using Igor Pro (Wavemetrics). Data from 1-s depolarizations were smoothed with a binomial algorithm with a 20-point window before averaging. Normalized I_Async from 100-Hz trains was smoothed within four points before averaging.

Whole-cell voltage-clamp recordings were obtained from isolated autaptic hippocampal neurons after 13–16 d in microisland culture using a Multiclamp 700A amplifier (Molecular Devices). The extracellular solution for EPSC recordings contained 119 mM NaCl, 5 mM KCl, 1 mM CaCl₂, 1.5 mM MgCl₂, 30 mM glucose, 20 mM Hepes, and 1 μM glycine, whereas the intracellular pipette solution contained 148.5 mM K-gluconate, 9 mM NaCl, 1 mM MgCl₂, 10 mM Hepes, and 0.2 mM EGTA. To elicit EPSCs, the membrane potential was held at −60 mV, and synaptic responses were evoked by triggering unclamped action currents with a 20-Hz train of 40 stimuli (1-ms steps from −60 to +30 mV) every 60 s; each EPSC in the train was normalized to the peak amplitude of the first EPSC. Calcium currents were recorded from single neurons using a P/4 protocol and an EPC10 patch-clamp amplifier with PULSE software (HEKA Elektronik) and filtered at 2.9 kHz. The internal solution contained 30 mM CsCl, 20 mM tetraethylammonium Cl, 50 mM cesium methanesulfonate, 0.2 mM EGTA, 30 mM Hepes, 1 mM MgCl₂, 5 mM TrisATP, 10 mM Tris phosphocreatine, 0.3 mM NaGTP, pH 7.2 (295 mosm). Similar external solution to EPSC recordings was also used for the calcium current recording supplemented with 1 μM tetrodotoxin to block sodium channels and 3 μM nimodipine to block L-type calcium channels. Series resistance was monitored, and only cells with stable series resistance were included in the data analysis. Series resistance was compensated 75–85%. All averaged data represent the mean ± SEM. Statistical significance was determined using Student t and one-way ANOVA tests.