The bag cell neurons of Aplysia are a cluster of cells that control egg laying behavior. After brief synaptic stimulation, they depolarize and fire spontaneously for up to 30 min. During the first few seconds of this afterdischarge, the action potentials of the bag cell neurons undergo pronounced broadening. Single bag cell neurons in culture also show spike broadening in response to repeated depolarizations. This broadening is frequency-dependent and associated with the induction of a depolarizing afterpotential lasting minutes. In some neurons the depolarizing afterpotential is sufficient to trigger spontaneous firing. To test the possibility that spike broadening during stimulation is required to trigger the depolarizing afterpotential, we eliminated frequency-dependent broadening by heterologous expression of the Kv3.1 potassium channel. This channel has rapid activation and deactivation kinetics and no use-dependent inactivation. Expression of Kv3.1 prevented spike broadening and also eliminated the depolarizing afterpotential. Measurements of the integral of calcium current during voltage commands, which simulated the action potentials of the control neurons and those expressing Kv3.1, indicate that spike broadening produces up to a fivefold increase in calcium entry. Manipulations that limit calcium entry during action potentials or chelation of intracellular calcium using BAPTA AM prevented the induction of the depolarizing afterpotential. We conclude that spike broadening is essential for the induction of the depolarizing afterpotential probably by regulating calcium influx and suggest that one of the physiological roles of spike broadening may be to regulate long-term changes in neuronal excitability.
- depolarizing afterpotential
- spike broadening
- long-term excitability
- potassium channel
- expression vector
The duration of individual spikes often increases during a train of action potentials (Aldrich et al., 1979; Coates and Bulloch, 1985; Jackson et al., 1991). This progressive spike broadening can enhance transmitter release, probably by regulating the calcium signal that triggers secretion (Jackson et al., 1991). Nevertheless the increase in calcium influx that is a consequence of action potential broadening (Augustine, 1990; Ma and Koester, 1995; Sabatini and Regehr, 1997) may play additional roles. For example many long-term changes in neuronal excitability are known to be calcium-dependent (Malenka et al., 1992; Linden, 1994).
The predominant mechanism underlying spike broadening seems to be the inactivation of potassium currents (Aldrich et al., 1979; Jackson et al., 1991), although complex interactions between several currents can determine the extent and rate of broadening (Ma and Koester, 1995,1996). In one cell type, the bag cell neurons of Aplysia, the gene for the channel whose inactivation underlies broadening has been cloned (Quattrocki et al., 1994). These neurons are normally silent but depolarize after a brief synaptic input and fire spontaneously for up to 30 min [the “afterdischarge” (Kupfermann and Kandel, 1970)]. During this time they secrete a variety of neuropeptides that control egg laying behavior. During the first few seconds of the afterdischarge, the action potentials of the bag cell neurons undergo a progressive increase in duration attributable, at least in part, to the use-dependent inactivation of AplysiaKv2.1, a delayed-rectifier potassium current (Quattrocki et al., 1994).
Although the role of the initial phase of spike broadening in the bag cell neurons is not fully understood, one possibility is that it triggers a slow depolarizing afterpotential that is probably responsible for initiating the full afterdischarge (Kaczmarek and Kauer, 1983; Brown and Mayeri, 1989). Other neurons that undergo sustained fluctuations in activity, such as the peptidergic neurons of the hypothalamus, also undergo spike broadening and can exhibit depolarizing afterpotentials (Andrew and Dudek, 1984; Bourque and Renaud, 1985).
A determination of the role of spike broadening requires techniques for its experimental manipulation. One successful approach in other neurons has been to use trains of broadened and nonbroadened spikes as voltage-clamp commands (Ma and Koester, 1995). However this technique may not be feasible in neurons (such as the bag cell neurons) that have a large neuritic tree and in which space clamp is likely to be imperfect. To overcome this problem, we have used a novel approach that involves the heterologous expression of a potassium channel, Kv3.1, in single neurons using the pNEX expression plasmid (Kaang et al., 1992;Zhao et al., 1994). The Kv3.1 channel was chosen because its rapid activation and deactivation kinetics were expected to provide a repolarizing drive that would limit the activity-dependent increase in spike duration. This strategy effectively eliminated spike broadening and allowed us to test its functional consequences. We find that spike broadening is essential for the induction of the depolarizing afterpotential (and hence the afterdischarge), probably by regulating calcium influx.
MATERIALS AND METHODS
Cell culture. Cells were isolated and maintained using previously described techniques (Schacher and Proshansky, 1983;Whim et al., 1997). Briefly, abdominal ganglia from Aplysia californica (∼150 gm from Marine Specimens Unlimited, Marinus Inc. or the National Institutes of Health/University of MiamiAplysia Resource Facility) were incubated at 34°C for 3 hr in 1% dispase (Boehringer Mannheim, Indianapolis, IN) in sterile normal artificial seawater (nASW) (in mm): 460 NaCl, 10.4 KCl, 55 MgCl2, 11 CaCl2, and 15 HEPES, pH 7.8. Bag cell neurons were removed using finely pulled glass probes and were maintained in sterile culture medium [30%Aplysia hemolymph/70% ASW containing penicillin (50 U/ml), streptomycin (50 μg/ml), vitamins (0.5 × MEM), and nonessential (0.2 × MEM) and essential (0.2 × MEM) amino acids withoutl-glutamine] at room temperature.
Neurons were generally plated into poly-l-lysine–coated culture dishes (Falcon #1006) and allowed to extend neurites. Individual cells were positioned so that their neurites did not subsequently overlap. In experiments in which the whole-cell calcium current was measured (see below), neurons were individually maintained in droplets of culture medium on dishes that had been made nonadherent by coating them with 5% BSA. After 24 hr the primary neurites had been reabsorbed, and when required the neurons were plated into poly-l-lysine–coated dishes and used immediately. Cells cultured in these conditions also show depolarizing afterpotentials (data not shown). This procedure minimized neurite outgrowth and improved the conditions for voltage clamp. All neurons were generally used within 5 d after isolation.
Measurement of action potentials and potassium currents.Recordings were made with an Axoclamp 2A amplifier (Axon Instruments, Foster City, CA). Bag cell neurons were impaled with a single microelectrode (∼10 MΩ) filled with 2 m potassium acetate, 0.5 m KCl, and 10 mm HEPES, pH 7.2. Before impalement the tips were dipped into Sigmacoat (Sigma, St. Louis, MO) to reduce electrode capacitance. Cells either were held under current clamp in discontinuous current-clamp mode or were voltage clamped with the switching voltage-clamp technique. Voltage-clamp gain was generally 0.3–0.8 nA/mV, and switching frequencies of 5 kHz were routinely obtained. Recordings were stored on videotape or on the computer hard disk. Voltage protocols were generated using Indec BASIC-FASTLAB software (Sunnyvale, CA). Many current-clamp experiments used a stimulation paradigm consisting of 20 action potentials evoked at frequencies ranging from 0.1 to 5 Hz. Unless otherwise stated, action potentials were evoked from a holding potential of −50 to −55 mV with 50 msec depolarizing current steps. This relatively long duration was found necessary to evoke action potentials reproducibly (probably because in vitro, action potentials are initiated in the cell body rather than in the periphery). In some figures, voltage records were filtered at 20 Hz that slightly reduced the action potential amplitude. Quantitative measurements of action potential kinetics were therefore made from records filtered at 2 kHz. Most current-clamp experiments were performed in nASW. Some experiments used a low Ca2+(0.5 mm) and high Mg2+ (110 mm) ASW. To chelate intracellular calcium, we pretreated a subset of neurons with 20 μm BAPTA AM for 30 min and then washed the neurons with ASW [this treatment is expected to buffer the action potential-induced rise in intracellular calcium with only minimal effects on the resting levels (Girod et al., 1995)]. Control neurons were treated with vehicle (0.1% DMSO). To isolate potassium currents, we bathed neurons in an extracellular medium containing (in mm): 460 N-methyl-d-glucamine, 10.4 KCl, 55 MgCl2, 11 CoCl2, and 15 HEPES, pH 7.8. BAPTA AM was purchased from Calbiochem (La Jolla, CA); most other chemicals were from Sigma.
Whole-cell voltage clamp and measurement of calcium currents. Whole-cell recordings were made with pipettes that were fire polished to 1 MΩ. The pipette solution contained (in mm): 100 tetraethylammonium (TEA)-Cl, 470 CsCl, 30 HEPES, 5 MgCl2, 5 Na2ATP, 20 EGTA, 4.14 CaCl2, and 5.5 glucose, pH 7.3. The nominal free calcium concentration was ∼30 nm. The extracellular medium was nASW (use of an extracellular solution in which NaCl was replaced by TEA and KCl was replaced by CsCl did not improve the isolation of the calcium current). Recordings were made with an EPC-7 (List-Electronics, Darmstadt-Eberstadt, Germany) using pClamp 6 software (Axon Instruments) from neurons that had been cultured in a nonadhesive environment (see above). Currents were digitized at 5 kHz and filtered at 3 kHz. Cell capacitance was canceled after breakthrough, and series resistance was compensated (70%).
Voltage-clamp commands to produce mock action potentials were constructed from a series of voltage ramps modeled from recordings made from bag cell neurons in culture. Measurements were made from two cells that we selected as being most typical of their class (we did not attempt to control for minor differences in rise time, etc., within a class). The “Kv3.1-like” action potential was composed of three ramps from a holding potential of −55 mV. The first ramp traveled +53 mV in 26 msec, the second increased 25 mV in 3 msec, and the third descended 78 mV in 4 msec. The “Kv3.1-like + amplitude-adjusted” action potential was similar except the second ramp increased 45 mV and the third descended 98 mV. The“fully broadened control” action potential was composed of four ramps from a holding potential of −55 mV. The first ramp traveled +40 mV in 50.6 msec, the second increased 58 mV in 3.8 msec, the third descended 42 mV in 24.6 msec, and the fourth descended 56 mV in 15 msec (see Results). Comparison of theKv3.1-like and fully broadened control waveforms mimicked the amount of broadening seen at 1 Hz in recorded action potentials (∼2.6 times).
The resulting currents were leak-subtracted using inverted and scaled (P/4) mock action potentials. Net calcium current was then measured as the difference current after subtraction of records made in the presence of Co-ASW (11 mm CoCl2 substituted for CaCl2).
Construction of plasmids. To make pNEX-Kv3.1, we removed aSalI-Spe1 fragment containing the Kv3.1 coding sequence from pGEM-A-Kv3.1 (Luneau et al., 1991) and ligated this fragment into SalI-XbaI–digested pNEX (Kaang et al., 1992) (kindly provided by B-K. Kaang and E. R. Kandel). To make pNEX-green fluorescent protein (GFP), we ligated aHindIII (blunted)-XbaI fragment containing the wild-type green fluorescent protein coding sequence (Chalfie et al., 1994) (kindly provided by M. Chalfie) into SphI (blunted)-XbaI–digested pNEX. Plasmid DNA was prepared using a commercial kit (Qiagen midi columns, Hilden, Germany) and stored at −30°C in sterile H2O.
Injection of plasmid DNA. Before injection, cells were washed three times with 10 ml of ASW in each wash. Injections were made with pressure pulses (typically 1–10 pulses; each pulse ∼300 msec in duration) using a Picospritzer (General Valve, Fairfield, NJ) and an injection pressure of 14 psi. Microelectrodes had tip diameters of <1 μm and had resistances of ∼2 MΩ when filled with 3 mKCl. Injection solutions contained 11 mm Tris-Cl, pH 7.2, 530 mm potassium acetate, 4.3 mg/ml dextran rhodamine (10,000 molecular weight), and plasmid DNA (0.1–0.9 mg/ml). Before injection, the plasmid DNA was centrifuged at 15,500 ×g for 15 min. After injection, cells were stored in the dark and examined 12 hr later for GFP expression. Neurons were then used within 2 d. No differences in physiology were observed between neurons that were (1) uninjected, (2) injected with pNEX-GFP but showed no expression, or (3) injected with pNEX-GFP and expressed GFP. Thus the data from these control neurons were pooled. For each figure, however, the phenotype of the neuron is given. These data also strongly suggest that the observed effects after Kv3.1 expression were specific and not simply caused by the overexpression of a foreign protein. Overexpression of a membrane protein (a glutamate receptor) also did not replicate the effects seen after expression of Kv3.1 (M. D. Whim and L. K. Kaczmarek, unpublished observations). However these data do not eliminate the possibility that the Kv3.1 subunits may have additional, subtle effects on cell physiology. For the purposes of statistical analysis, comparisons between treatments were made using a two-tailed Student’s t test. Ap value of 0.05 was considered significant. A one-way ANOVA with a post hoc Bonferroni multiple comparisons test was used to determine the effect of mock action potentials on net calcium current.
Induction of a depolarizing afterpotential in a single bag cell neuron in culture
Stimulation of a single cultured bag cell neuron with 20 action potentials at a frequency of 1 Hz often produced a depolarizing afterpotential that could last for many minutes (range, 1 to >6 min;n = 33; Fig.1 A). Similar depolarizing afterpotentials can be recorded from neurons in the intact bag cell cluster after high-frequency stimulation of the afferent input (Kupfermann and Kandel, 1970; Kaczmarek and Kauer, 1983). To determine the change in conductance during the slow depolarization, we held the neuron shown in Figure 1 B under voltage clamp and stepped the membrane potential at regular intervals from −50 to −60 mV before and after a series of 20 voltage steps at 1 Hz (to induce a train of unclamped action potentials). After stimulation, a slow inward current was observed that had a time course similar to that of the depolarizing afterpotential. The increase in the size of the current steps after stimulation indicated an apparent increase in membrane conductance (Fig. 1 B).
The induction of the depolarizing afterpotential was found to depend on stimulation frequency. When trains of 20 action potentials were applied at frequencies from 0.1 to 5 Hz [which is the physiological range of firing frequencies (Kaczmarek et al., 1982)], depolarizing afterpotentials were typically not observed until stimulation frequencies of 0.5–1 Hz (Fig. 2). In many cases the evoked potentials were very long-lasting (over 30 min), so for practical purposes the resting membrane potential was reset to prestimulus levels before the next stimulus paradigm. Although increasing the stimulation frequency led to an increase in the size of the depolarizing afterpotential, the amplitude varied between individual neurons, ranging from <5 to >15 mV (n = 33).
Frequency-dependent spike broadening accompanies the induction of the depolarizing afterpotential
Dramatic changes in spike duration were observed using stimulation paradigms that evoked a depolarizing afterpotential. Conversely, much less spike broadening was observed with paradigms that did not evoke a depolarizing afterpotential. The dissimilarity between the amount of spike broadening at 0.1 and at 1 Hz can be seen by comparing the 1st and 19th action potentials in the train at the two frequencies. For the neuron shown in Figure 3, 20 action potentials at a frequency of 1 Hz evoked a depolarizing afterpotential of 4 mV, and the duration of the spike more than doubled. Much less spike broadening was observed with the same number of action potentials at 0.1 Hz, a paradigm that did not evoke a depolarizing afterpotential. Because of the long-lasting nature of the depolarizing afterpotential and the variability between neurons, it was not possible to assess the correlation between intermediate amounts of spike broadening and the size of the evoked potential. In contrast to the changes in spike duration, no activity-dependent changes in spike amplitude were seen in these experiments.
A Kv3.1-like current is expressed after pNEX-Kv3.1 injection
The correlation of marked spike broadening with the induction of the depolarizing afterpotential suggested that these two factors might be causally linked. Because spike broadening in bag cell neurons involves the activity-dependent inactivation of a potassium current (Quattrocki et al., 1994), we reasoned that spike broadening might be reduced by the heterologous expression of a delayed-rectifier potassium channel that did not inactivate with repetitive depolarizations. For this purpose the Kv3.1 potassium channel (Luneau et al., 1991) has all the required characteristics, because it generates a noninactivating current that rapidly activates and deactivates (and thus should not have any long-term consequences on membrane excitability). In addition, because Kv3.1 activates only at relatively depolarized potentials, its expression would not be expected to alter the resting membrane potential significantly. We therefore subcloned the Kv3.1 coding sequence into the pNEX expression vector (Kaang et al., 1992) and then injected single bag cell neurons with pNEX-Kv3.1 plus pNEX-GFP (which encodes the green fluorescent protein and serves as a visual marker for plasmid expression). Control neurons were injected with pNEX-GFP alone or were uninjected. One day later a variable percentage of injected neurons was GFP positive (this ranged from 10 to 55%;n = 243 cells in 10 separate experiments). A typical GFP-positive neuron is shown in Figure4 A.
To determine whether the GFP-positive neurons that had been coinjected with pNEX-Kv3.1 also expressed a Kv3.1-like current, we measured total outward potassium current under voltage clamp. Control neurons had a net outward current that showed some inactivation during a 1 sec voltage step (particularly at depolarized potentials). In contrast, the net outward current in neurons that had been coinjected with pNEX-GFP and pNEX-Kv3.1 activated and deactivated more rapidly, showed little inactivation (Fig. 4 B), and was over 10 times the amplitude of control currents measured at −20 mV (Fig. 4 C). These observations are consistent with the expression of a Kv3.1-like current. GFP expression was found to be a reliable marker of channel synthesis because every GFP-positive cell was also found to have a characteristic “Kv3.1-like physiology” (see below;n = 21).
TEA partially rescues the control phenotype
Because Kv3.1 is blocked by low concentrations of external TEA (Kanemasa et al., 1995), we tested whether the control phenotype could be “rescued” from pNEX-Kv3.1–injected neurons. For these experiments we used neurons that had particularly high levels of Kv3.1-like channel expression, resulting in a decrease in the amplitude of individual action potentials (Fig. 5). For simplicity we monitored the effects of TEA on spike amplitude. Although 1–10 mm external TEA had relatively little effect on the amplitude of single spikes (or their duration) in the control cells (Fig. 5 A,C), it markedly increased the spike amplitude (and, to a lesser extent, the duration) of the pNEX-Kv3.1–injected neurons (Fig.5 B,C). However even in 10 mm TEA, the amplitude of the spike was still significantly lower in the Kv3.1-injected than in the untreated controls (88 ± 5 vs 105 ± 5 mV, mean ± SD; p < 0.01;n = 4 for both cell types). Thus multiple lines of evidence indicate that the expression of a Kv3.1-like channel in pNEX-Kv3.1–injected neurons produces the large increase in the outward potassium current.
Kv3.1 expression reduces repetitive firing and increases action potential threshold
We next examined the effect of the increase in Kv3.1 outward current on the physiology of the bag cell neurons. Kv3.1 expression was found to reduce neuronal excitability (as monitored by the number of action potentials evoked by a series of depolarizing current steps). A step depolarization of 0.5 nA evoked seven action potentials in a control neuron, whereas in a pNEX-Kv3.1–injected cell, the same current pulse evoked only a single action potential (Fig.6 A). This depression of excitability was observed over a range of current injections (Fig.6 B). On average, over three times as much current was required to evoke a single action potential from a cell injected with pNEX-Kv3.1 compared with a control neuron [0.4 ± 0.13 nA (n = 5) vs 0.14 ± 0.01 nA (n = 5), mean ± SD; **p < 0.01; Fig.6 C]. The resting membrane potential of pNEX-Kv3.1–injected neurons was more hyperpolarized than was that of control cells [−39 ± 4 mV (n = 5) vs −32 ± 3 mV (n = 5), mean ± SD; *p < 0.05; Fig. 6 D]. This may be because the high levels of Kv3.1 expression result in partial activation of the Kv3.1 current even at the resting potential. Nevertheless, the input resistance of pNEX-Kv3.1–injected and control cells (measured at −55 mV) was not significantly different [372 ± 61 MΩ (n = 5) vs 364 ± 72 MΩ (n = 5), mean ± SD;p = 0.85; Fig. 6 E], indicating that Kv3.1 expression per se does not appear to have a deleterious effect on these neurons.
Kv3.1 expression blocks spike broadening and the induction of the depolarizing afterpotential
Even at frequencies as high as 5 Hz, there was essentially no spike broadening in the Kv3.1-expressing neurons. The response of a typical pNEX-Kv3.1–injected neuron to stimulation with a train of 20 action potentials at 1 Hz is shown in Figure7 B. Control cells showed a normal degree of spike broadening (Fig. 7 A). Pooled data from several neurons indicated that the Kv3.1-expressing neurons showed no significant increase in spike duration during the train [duration of first spike, 9 ± 3 msec, vs duration of 19th spike, 10 ± 3.0 msec (n = 4), mean ± SD; p = 0.62; Fig. 7 C].
The expression of Kv3.1 significantly depressed the amplitude of individual action potentials when compared with that of control neurons [67 ± 11 mV (n = 4) vs 91 ± 7 mV (n = 5), mean ± SD (p < 0.01), for the first action potential in a 1 Hz train], but as in the control neurons, there was no activity-dependent change in spike amplitude (Fig. 7 D).
Expression of Kv3.1 also resulted in the complete suppression of the depolarizing afterpotential in all neurons tested. Even a stimulation paradigm of 20 action potentials at 5 Hz (Fig.8) was ineffective in triggering any long-term change in membrane potential. Both uninjected neurons and those that had been injected with only pNEX-GFP showed normal depolarizing afterpotentials (compare with Fig. 2). Thus the presence of spike broadening seems to be critical for the induction of the depolarizing afterpotential.
Induction of the depolarizing afterpotential is calcium-dependent
To test the hypothesis that the changes in action potential shape produced by expression of Kv3.1 blocked the induction of the depolarizing afterpotential by altering calcium influx, we performed experiments to test the calcium dependence of the depolarizing afterpotential. We first tested the effects of lowering extracellular calcium. In the experiment illustrated in Figure9 A, stimulation of a control cell with 20 action potentials at 1 Hz evoked a depolarizing afterpotential. When the bathing medium was exchanged for a low calcium and high magnesium ASW, this stimulation paradigm no longer induced a depolarizing afterpotential but revealed an underlying afterhyperpolarization. Stimulating for a third time, but now in nASW, evoked a depolarizing afterpotential, although one somewhat reduced in amplitude.
Although lowering the extracellular calcium concentration (and raising magnesium) blocked the induction of the depolarizing afterpotential, it also suppressed spike amplitude and slowed the kinetics of spike repolarization (data not shown; n = 2). In a second series of experiments, therefore, bag cell neurons were treated with the calcium chelator BAPTA AM (a membrane permeable form of BAPTA). This treatment prevented the induction of the depolarizing afterpotential that was observed in control neurons after a train of 20 action potentials at 1 Hz [−2 ± 2 mV (n = 3) vs 4 ± 1.4 mV (n = 4), mean ± SD; **p < 0.01; BAPTA AM vs control; Fig.9 B]. In contrast to the effect of lowering extracellular calcium, the spike amplitude and degree of spike broadening was not significantly different between the control and BAPTA AM-treated neurons (Fig. 9 C). These experiments suggest that a rise in intracellular calcium levels is required for the induction of the depolarizing afterpotential.
Changes in action potential duration alter calcium influx
Because expression of Kv3.1 in bag cell neurons decreases the amplitude of action potentials as well as inhibiting spike broadening, the elimination of the depolarizing afterpotential could result from reduced calcium influx as a result of reduced spike height rather than from reduced spike broadening. To test the effect of the changes in action potential shape on calcium influx, we measured the whole-cell calcium current during a series of voltage ramp commands that mimicked three types of action potentials: (1) action potentials measured after Kv3.1 expression (Kv3.1-like); (2) action potentials after Kv3.1 expression but with no reduction in spike amplitude (Kv3.1-like + amplitude-adjusted); and (3) fully broadened action potentials from control cells stimulated at 1 Hz (fully broadened control).
As shown in Figure10 A,C, increasing the amplitude of the action potential (by 20 mV) did produce an increase in the integral of the calcium current, but this was not statistically significant [−12.6 ± 4.8 nA·msec (n = 3) vs −15.1 ± 5.4 nA·msec (n = 3), mean ± SD; p = 0.061). In contrast, the calcium influx during a fully broadened control action potential was over fivefold larger than that during the Kv3.1-like + amplitude-adjusted andKv3.1-like action potentials (−78.4 ± 25.2 nA·msec; n = 3; **p < 0.01 for both forms of action potential compared with the fully broadened control action potential; Fig.10 B,C). The most economical explanation of these results is that the suppression of spike amplitude by Kv3.1 has a relatively minor effect on calcium influx whereas the suppression of spike duration substantially reduces calcium influx, although these results do not rule out more complicated interactions between low levels of calcium influx and membrane voltage. Hence Kv3.1 is likely to inhibit the induction of the calcium-dependent depolarizing afterpotential by reducing spike broadening.
Spike broadening and DAP induction are reduced in refractory neurons
Although stimulation of individual bag cell neurons in culture typically evoked a subthreshold depolarizing afterpotential, in some cases the membrane depolarization was sufficient to evoke spontaneous action potentials. In one cell a full afterdischarge-like event was observed (Fig. 11 A). This indicates that the ability to evoke an afterdischarge is not dependent on neuron–neuron interactions [although autaptic self-innervation may play a role, as suggested for bag cell neuronsin situ (Brown and Mayeri, 1989)]. After the termination of spontaneous activity, the membrane potential was reset to the preafterdischarge level. Stimulating again with the same paradigm (20 action potentials at 1 Hz) no longer evoked an afterdischarge but evoked a small afterhyperpolarization. Increasing the stimulation frequency to 2 Hz also failed to re-evoke an afterdischarge (data not shown). Evidently the neuron had become refractory, a condition that lasts for several hours (Kaczmarek and Kauer, 1983).
When the durations of action potentials were compared between the first and second stimulus trains, it was clear that spike broadening was essentially absent in the refractory condition (Fig.11 B). This result indicates that spike broadening is closely associated with the ability of the bag cell neurons to undergo an afterdischarge. Furthermore, it suggests that the degree of spike broadening can be physiologically modulated.
Spike broadening has been observed in a wide variety of neurons and is generally assumed to be a mechanism that potentiates transmitter release. Our results are consistent with an additional role: that of controlling the subsequent induction of prolonged, activity-dependent changes in membrane potential. Stimulation of individual bag cell neurons at frequencies close to 1 Hz evoked a slow depolarizing afterpotential similar to that observed in situ (Kaczmarek and Kauer, 1983; Brown and Mayeri, 1989). The current(s) that underlies the depolarizing afterpotential is likely to contribute to the induction of a long-lasting period of spontaneous firing (the afterdischarge).
To investigate whether there was a causal link between spike broadening and induction of the depolarizing afterpotential, we expressed the Kv3.1 potassium channel in individual bag cell neurons in culture. This channel is found in high levels in fast-firing neurons that show rapid action potential repolarization (Perney et al., 1992; Weiser et al., 1995; Massengill et al., 1997). The expression of Kv3.1 in the bag cell neurons effectively prevented spike broadening and eliminated the induction of the depolarizing afterpotential. Kv3.1 was well suited to this role because it activates and deactivates rapidly and undergoes little inactivation.
The large outward current that was evident after injection of pNEX-Kv3.1 was determined to be Kv3.1-like on the basis of several criteria, namely, a sensitivity to low concentrations of extracellular TEA and a characteristic voltage dependence and kinetics. Although most of the changes observed matched those seen when Kv3.1 is heterologously expressed in mammalian cells (Kanemasa et al., 1995), two differences were noted. First, although 10 mm TEA inhibits over 90% of the Kv3.1 current when this channel is expressed in fibroblasts (Kanemasa et al., 1995), it did not completely rescue the control phenotype from Kv3.1-expressing bag cell neurons. This may result from changes in the affinity of TEA binding in the high salt conditions of ASW. Alternatively, the Kv3.1b subunit may associate with other endogenous subunits in bag cell neurons to produce a channel that has slightly different characteristics than those of Kv3.1 expressed in fibroblasts.
Given this result it is not surprising that depolarizing afterpotentials could not be evoked in Kv3.1-expressing neurons after TEA treatment. It is likely that this is caused by the inability of low concentrations of TEA to restore the duration of the action potential to that of control cells (data not shown). Higher concentrations of TEA could not be used because endogenous currents in bag cell neurons are also TEA-sensitive (Strong and Kaczmarek, 1986). Alternatively, because the experiments with BAPTA AM indicate that calcium influx is critical for the induction of the depolarizing afterpotential, it is possible that the spatial pattern of calcium entry was altered in neurons expressing Kv3.1.
Also unanticipated was the observation that Kv3.1 expression reduced the amplitude of individual action potentials. Although not predicted from computer models of rapidly firing vertebrate cells (Kanemasa et al., 1995), this is most readily explained when one considers the slow kinetics of the bag cell neuron action potential, which can take several milliseconds to peak. This is sufficient time for substantial activation of Kv3.1, which would further slow the action potential trajectory and provide sufficient hyperpolarizing drive to reduce its final amplitude.
In spite of these caveats, the expression of Kv3.1 was sufficient to prevent spike broadening in bag cell neurons and to suppress the induction of the depolarizing afterpotential. Even after stimulation with 20 action potentials at 5 Hz, there was no detectable long-term change in membrane potential. Because Kv3.1 expression also reduced spike amplitude, it is not possible to conclude formally that spike broadening is the only aspect of the action potential that regulates the induction of the depolarizing afterpotential. However as the experiments with mock action potentials indicate, the suppression of spike amplitude had a relatively minor effect on the calcium current compared with the suppression of spike duration. As found in other preparations (Llinas et al., 1982; McCobb and Beam, 1991; Yazejian et al., 1997), most calcium influx occurred during the downstroke of the broadened action potential. Restoring spike amplitude increased the time integral of the calcium current by <25%, whereas restoring the spike duration increased the time integral by >400%. We conclude that spike broadening (and hence calcium influx) is a major regulator of the induction of the depolarizing afterpotential.
Influx of calcium ions during the afterdischarge regulates several phenomena. First, as we have shown, calcium influx through broadened action potentials is required to trigger the depolarizing afterpotential (and probably the afterdischarge). The increased calcium current is also likely to support the shoulder of the action potential (Aldrich et al., 1979; Acosta-Urquidi and Dudek, 1981; Ma and Koester, 1995). Second, the secretion of neuropeptides from the bag cell neurons is calcium-dependent (Arch, 1972). Third, the refractory period, which follows the afterdischarge, requires previous calcium influx (Kaczmarek et al., 1982; Kaczmarek and Kauer, 1983). Experimental evidence indicates a substantial rise in intracellular calcium during this sequence of events. Using calcium-sensitive electrodes, Fisher et al. (1994) reported a rise in calcium of several hundred nanomolar during the initial phase of the afterdischarge.
Experiments to measure the degree of spike broadening in excitable and refractory neurons in the intact nervous system have been inconclusive. This is because action potentials fail to propagate toward the somata during the refractory period [type I refractoriness (Kaczmarek et al., 1978)]. Moreover electrical coupling between bag cell neurons in situ prevents a rigorous control of membrane potential that is important when measuring spike duration (Acosta-Urquidi and Dudek, 1981).
Nevertheless our present findings from cells in culture suggest that spike broadening is reduced in refractory neurons. It is known that action potentials evoked from the caudodorsal cells ofLymnaea (which are homologous to the bag cell neurons ofAplysia) become less sensitive to calcium channel blockers during the refractory period (Kits and Bos, 1982). At this time, spike trains do not evoke a detectable rise in intracellular calcium (Kits et al., 1997). These observations could be partly explained if refractory neurons showed less spike broadening and reduced calcium influx. Because however we observed that spike broadening did not always lead to the induction of a depolarizing afterpotential, it seems to be necessary but not sufficient for induction of the depolarizing afterpotential. It is likely that other factors downstream of calcium entry, such as the activation of calcium-sensitive kinases (Roeper et al., 1997) or the autaptic detection of peptide secretion (Brown and Mayeri, 1989; Brussaard et al., 1990), are also required.
The endogenous factors that determine the long-term modulation of spike broadening are not known, but two potential mechanisms can be proposed. First, an upregulation of delayed-rectifier–like currents could suppress spike broadening in a manner analogous to the experimental expression of Kv3.1. The physiological mechanism however is likely to involve covalent modification of channels or associated proteins rather than channel expression, given the time course of the switch from afterdischarge to refractoriness (∼30 min). Second, the kinetics ofAplysia Kv2.1, the current whose inactivation underlies spike broadening, could be modulated (Quattrocki et al., 1994).
As regards the physiological role of the depolarizing afterpotential, it seems likely that it regulates the ability of the bag cell neurons to undergo an afterdischarge. In excitable neurons, repetitive stimulation evokes a depolarizing afterpotential that will trigger an afterdischarge. In refractory neurons, repetitive stimulation evokes an afterhyperpolarization that inhibits afterdischarge. Because the refractory period is thought to prevent onset of egg-laying behavior within 18 hr of a previous egg-laying episode, the afterhyperpolarization impedes the inappropriate induction of an afterdischarge during this time. This “yin–yang” model emphasizes that a reciprocal interaction between these two potentials contributes to the ability of the bag cell neurons to undergo an afterdischarge. The caudodorsal cells of Lymnaea also show a similar regulation of the depolarizing afterpotential (Brussaard et al., 1988).
In conclusion, we have used a novel method to investigate the physiological role of spike broadening. We find that in addition to its well known ability to modulate transmitter secretion, spike broadening can also play a critical role in regulating long-term changes in neuronal excitability. Similar mechanisms may underlie sustained changes in activity that have been observed in other neurons, such as the magnocellular neurons of the hypothalamus (Wakerley and Lincoln, 1973).
This work was supported by National Institutes of Health Grant NS-18492 to L.K.K. We thank Drs. Si-Qiong Liu and Neil Magoski for critically reading this manuscript and for helpful technical advice.
Correspondence should be addressed to Dr. Leonard K. Kaczmarek, Department of Pharmacology, Yale University School of Medicine, 333 Cedar Street, New Haven, CT 06520-8066.
Dr. Whim’s present address: Department of Pharmacology, University College London, Gower Street, London, UK.