Ionic Mechanism of Long-Lasting Discharges of Action Potentials Triggered by Membrane Hyperpolarization in the Medial Lateral Habenula

The electrical stimulation of the stria medullaris produces hyperpolarizing postsynaptic potentials that induce long-lasting, intense discharges of action potentials

Along the way of projections, the axon fibers arising from various areas of the forebrain join together and form fiber bundles that travel in the stria medullaris to reach the habenula. The connection between the stria medullaris and the habenula was well preserved in sagittal habenular slices (Fig. 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Photomicrographs of a sagittal (A) and a frontal habenular slice (B). The habenula (Hb) receives major input from the striatum and limbic forebrain through the stria medullaris (stm). The efferents from the habenula travel via the fascicles retroflexus (rf) to various nuclei of the midbrain. Scale bars, 1 mm. HP, Hippocampus; TH, thalamus; SC, superior colliculus.

In sagittal habenular slices, electrical stimulation of the stria medullaris produced complex patterns of responses in the cells located in the medial part of the lateral habenular nucleus. The representative examples are shown in Figure 2. Our extracellular recordings from the lateral habenular cells that are illustrated in Figure 2, Aa and Ab, showed that they generated action potentials spontaneously at frequencies of ∼9 and 11 Hz, respectively. In these cells, electrical stimulation of the stria medullaris resulted in a brief inhibition of spontaneous activity for ∼0.7-1 sec. In Figure 2Aa, the inhibitory period was followed by increases in firing up to ∼17 Hz for the initial 3-4 sec. Then, the firing rate declined gradually over the next 4-5 sec and returned to the spontaneous level. In another cell shown in Figure 2Ab, the inhibitory period was followed first by three or four cycles of brief bursts and then by an increase in tonic firing up to 40-50 Hz. The increase in tonic firing lasted for 14-18 sec before the discharge rate returned to the spontaneous level.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Electrical stimulation of the stria medullaris produces a brief inhibition of habenular activity that is followed by persistent increases in firing. Aa, Ab, Extracellular recordings from two different habenular cells. A train of 10 electrical stimuli at 100 Hz was delivered to the stria medullaris. The stimulation resulted in a brief inhibition of spontaneous activity for ∼700 msec to 1 sec. The inhibition was followed by marked increases in firing. Ba, Whole-cell patch intracellular recording from a lateral habenular cell. A single electrical stimulation to the stria medullaris (100 μA) generated a hyperpolarizing postsynaptic potential of ∼10 mV in amplitude and 1 sec in duration. This synaptic hyperpolarization transiently inhibited spontaneous firing. Bb, The increase in the amplitude of electrical stimulation to 150 μA increased the size of the hyperpolarizing synaptic potential to 25-30 mV. This synaptic hyperpolarization was followed by burst and tonic increases in firing. The increases in tonic firing lasted for ∼40 sec. Action potentials are truncated in Ba and Bb.

The cells that exhibited the transient inhibition of spontaneous activity and the ensuing increase in tonic firing in response to electrical stimulation of the stria medullaris were most concentrated in the medial part of the lateral habenular nucleus. In this area, all tonically active cells recorded with extracellular electrodes showed a similar change in firing as was shown above in response to electrical stimulation of the stria medullaris (n = 6 cells). In those six cells, the increase in tonic firing ensuing the transient inhibition persisted 7-26 sec (average 15.8 ± 6.3 sec). Therefore, all our intracellular recordings shown below in this study were conducted in the medial part of the lateral habenula.

Our whole-cell intracellular recordings revealed that the brief inhibition of spontaneous activity in response to electrical stimulation of the stria medullaris results from the activation of hyperpolarizing postsynaptic potentials (n = 9 cells) (Fig. 2B). The size of these hyperpolarizing synaptic potentials grew as the current amplitude of electrical stimulation was increased. The decay of large size hyperpolarizing postsynaptic potentials led to the rebound generation of a brief burst of action potentials that was followed by persistent, tonic increases in firing (Fig. 2Bb). This change in firing was consistent with that demonstrated above with extracellular electrodes in Figure 2A.

The cells in the medial part of the lateral habenula generate long-lasting, intense discharges of action potentials in response to the injection of a brief hyperpolarizing current pulse

Based on the pattern of spontaneous activity at rest, we classified cells in the medial part of the lateral habenular nucleus into three groups. The first group of cells exhibited spontaneous, tonic firing of action potentials in the frequency range of 1-20 Hz (Fig. 3Aa) (n = 69 of total 132 cells). The average discharge rate was 5.7 ± 2.6 Hz. This pattern of spontaneous firing in the slice preparation is consistent with that of the habenular activity recorded in vivo from the cat with extracellular electrodes in a previous study (Meier and Herrling, 1993). To measure cell input resistance, we hyperpolarized the cells just below the threshold for action potentials by passing a steady negative current and superimposed a brief hyperpolarizing current pulse (100 msec, 10 pA). The average input resistance was 1070 ± 382 MΩ (n = 14 cells). In contrast, cells of the second group were silent at rest (Fig. 3Ab) (n = 31 out of 132 cells). The resting membrane potentials of these cells (- 59 ± 4 mV) were negative to the threshold for action potentials (-52 ± 1 mV) that was estimated as the onset of the regenerative phase of action potentials. The last group of cells generated spontaneous oscillations of burst discharges (Fig. 3Ac) (n = 32 of 132 cells). It is noteworthy that a subpopulation of cells in the first group altered the mode of spontaneous firing from tonic to burst oscillations when the membrane potential was hyperpolarized with the injection of steady negative current (Fig. 3B) (n = 7 cells). Input resistances of cells in the three groups were not significantly different.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Three distinct patterns of spontaneous activity recorded from neurons of the medial part of the lateral habenula. Aa, One group of cells generated tonic firing of action potentials at rest. Dotted line indicates membrane potential of -60 mV. Ab, Another group of cells were silent at rest. Ac, The last group generated burst oscillations in resting conditions. B, A subpopulation of cells in the first group changed its spontaneous firing from tonic to burst oscillations in response to membrane hyperpolarization by a steady negative current. Action potentials are truncated.

A common characteristic of cells of the first and second groups in the medial part of the lateral habenula is that they generate prolonged, intense discharges of burst and tonic mixtures of action potentials in response to a brief hyperpolarizing current pulse. A representative example is shown in Figure 4. In this cell, the injection of a 100-msec-long hyperpolarizing current pulse resulted in a dramatic rebound generation of action potentials (Fig. 4A). At the break of current pulse, it generated a burst of four or five high-frequency action potentials at 80-90 Hz. This burst was followed by a tonic series of action potentials that continued, albeit gradually declining in discharge rate, over the next 15-20 sec. In a total of 40 cells analyzed, the duration of tonic action potential generation in response to a 100 msec hyperpolarizing current pulse ranged from 3 to 39 sec (average 14.8 ± 8.5 sec). In these cells, current pulses that produced ∼30-40 mV membrane hyperpolarization were used to activate prolonged discharges of action potentials.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

A brief membrane hyperpolarization induced prolonged discharges of action potentials. A, At membrane potentials of more than -65 to -70 mV, a hyperpolarizing current pulse triggered long-lasting discharges of action potentials, whereas a depolarizing current pulse activated similar discharges at less than -70 to -75 mV. The initial bust discharge is expanded in the inlet. The action potentials are truncated. B, The block of action potentials by bath infusion of 1 μm TTX revealed a persistent membrane depolarization that underlay the long-lasting discharges of action potentials. C, A reduction of external Na⁺ from 151 to 26 mm by the substitution of external NaCl with NMDG-Cl blocked most of the persistent depolarization and uncovered a transient rebound depolarization in response to a hyperpolarizing current pulse. Resubstitution of NMDG-Cl with NaCl restored the persistent membrane depolarization in response to current pulse injection. D, Expansion of the rebound depolarization recorded 3 min after the infusion of NMDG-Cl shown in C. The initial peak of the rebound depolarization is followed by the afterdepolarization of ∼1 sec in duration. E, The injection of hyperpolarizing current pulses at regular intervals showed a progressive increase in cell input resistance, but had little effects on the duration of the persistent depolarization of the membrane.

The activation of the long-lasting discharge of action potentials was dependent on the membrane potential of habenular cells. Whereas hyperpolarizing current pulse triggered long-lasting discharges of action potentials at membrane potentials between -55 and -65 mV, it failed to do so when the membrane potential was hyperpolarized at less than -70 mV. At potentials less than -70 mV, brief depolarizing current pulse (100 msec) was effective in activating the prolonged discharges (Fig. 4A).

These results suggest that the long-lasting increase in tonic firing in response to synaptic stimulation of the stria medullaris (Fig. 2) is mediated by the intrinsic properties of habenular cells that are activated in response to transient membrane hyperpolarization.

A tonic entry of Na⁺ underlies long-lasting discharges of action potentials

To characterize the ionic mechanism that underlies the long-lasting discharges of action potentials, we blocked action potential generation in an attempt to uncover any persistent change in the membrane potential that is associated with the prolonged generation of action potentials. In 12 cells analyzed, after bath infusion of 0.5-1 μm TTX, a hyperpolarizing current pulse (100 msec) at membrane potentials of more than -65 mV resulted in a persistent depolarization of the membrane in duration ranging from 5.2 to 55.7 sec. The average duration from the 12 cells was 19.6 ± 16.2 sec. One example of such cases is shown in Figure 4, A and B. In this cell, the injection of a hyperpolarizing current pulse in the presence of TTX activated a rebound depolarization, the peak of which reached the membrane potential of -20 to -30 mV immediately after the break of current pulse (Fig. 4B). The membrane potential then remained depolarized and steady for the initial 18-20 sec before gradually declining to baseline over the next 20 sec. At potentials less than -70 mV, a brief depolarizing current pulse activated a similar persistent depolarization of the membrane. These results indicate that the long-lasting discharge of action potentials in response to synaptic stimulation of the stria medullaris that was shown in Figure 2 is mediated by the generation of a persistent membrane depolarization in response to a brief synaptic hyperpolarization of habenular cells.

To characterize the ionic mechanism underlying the TTX-insensitive, persistent membrane depolarization, we tested whether a tonic influx of Na⁺ is involved. For this test, we lowered external Na⁺ and examined its effect on the generation of the persistent membrane depolarization. A reduction in external Na⁺ from 151 to 26 mm by replacing NaCl with equimolar N-methyl-d-glucamine chloride (NMDG-Cl) eliminated most of the persistent membrane depolarization that had been produced in response to hyperpolarizing current pulse (Fig. 4C). For example, in five cells tested, a hyperpolarizing current pulse in the presence of TTX and 151 mm external Na⁺ generated membrane depolarization, respectively, for average duration of 7.2 ± 2.4, 11.3 ± 2.7, 17.7 ± 3.2, 26.3 ± 4.0, and 43 ± 5.3 sec. After the reduction of external Na⁺ to 26 mm, the same hyperpolarizing current pulse resulted in only a transient rebound depolarization in all the five cells. This transient rebound depolarization was composed of an initial rebound spike and an ensuing afterdepolarization (ADP) of ∼1-2 sec in duration (Fig. 4C,D).

During the persistent membrane depolarization, we injected a small hyperpolarizing current pulse at regular intervals (n = 7 cells) (Fig. 4D). The current pulse injection revealed a progressive increase in cell input resistance over time during the persistent membrane depolarization, indicating that the decay of the persistent depolarization is associated with a decrease in membrane conductance. The current injection, however, did not halt the persistent membrane depolarization, suggesting that the TTX-insensitive, persistent depolarization is not generated by such regenerative potentials as plateau potentials that can be repolarized abruptly by opposing hyperpolarizing influences. We refer to the underlying depolarizing potential of the TTX-insensitive, persistent depolarization of habenular cells as a “long-lasting depolarizing afterpotential (LDAP).”

The LDAP was abolished by bath infusion of 50 μm flufenamic acid (FFA). This concentration of FFA has been routinely used to block nonspecific cationic currents (Shaw et al., 1995; Partridge and Valenzuela 2000; Ghamari-Langroudi and Bourque, 2002). In six cells tested, before FFA infusion, a hyperpolarizing current pulse in the presence of TTX triggered the LDAP in duration ranging from 5 to 36 sec (Fig. 5A,B). The bath infusion of 50 μm FFA blocked the generation of the LDAP and uncovered the transient rebound depolarization that consisted of an initial spike and an ADP of ∼1-2 sec in duration (Fig. 5C). These spike and ADPs were similar to those shown above with the substitution of external NaCl with NMDG-Cl (Fig. 4C,D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Flufenamic acid blocks the generation of the LDAP. A, A hyperpolarizing current pulse induced persistent discharges of action potentials. B, Bath infusion of 1 μm TTX revealed the LDAP underlying the persistent discharges of action potentials. C, The subsequent bath infusion of 50 μm FFA blocked the LDAP and uncovered a transient spike and the ensuing ADP in response to current pulse injections. D, The LDAP was regenerated after FFA washout.

Taken altogether, those results above from TTX application, NMDG substitution, and FFA infusion suggest that the activation of a tonic Na⁺ influx possibly through nonselective cationic channels is critical for the generation of the LDAP in response to a brief membrane hyperpolarization.

Ca²⁺ entry triggers the generation of the LDAP

The generation of the transient rebound spike in response to a hyperpolarizing current pulse that was uncovered by NMDG substitution or bath infusion of FFA resembles that of the Ca²⁺ spike through the activation of low-threshold T-type Ca²⁺ channels, as was shown in numerous areas of the brain (Llinas and Yarom, 1981; Jahnsen and Llinas, 1984a,b; Llinas and Muhlethaler, 1988; Huguenard, 1996). In fact, a previous study demonstrated the presence of T-type Ca²⁺ current in rat lateral habenular cells (Wilcox et al., 1988; Huguenard et al., 1993).

The generation of the transient spike and the ADP that appear to precede and underlie, respectively, the LDAP raises the question of whether Ca²⁺ entry during the transient rebound depolarization plays a role in the activation of the persistent membrane depolarization. To address this question, we removed Ca²⁺ from the external solution and examined its effect on the LDAP. A representative case is shown in Figure 6. This cell was spontaneously active and continued to generate action potentials at ∼5-6 Hz (Fig. 6A). The injection of a hyperpolarizing current pulse at membrane potentials more than -65 mV resulted in long-lasting discharges of action potentials for ∼15-20 sec. After bath infusion of 1 μm TTX, the hyperpolarizing current pulse generated the LDAP of 30-35 sec in duration. In this cell, the removal of external Ca²⁺ eliminated not only the transient spike and ADP but also the LDAP in response to either a hyperpolarizing or depolarizing current pulse (Fig. 6B) (n = 8 cells). This result indicates, first, that the transient rebound spike and ADP are generated by Ca²⁺ entry and, second, that they are important for the subsequent activation of the LDAP.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Ca²⁺ entry during the transient spike is crucial for the induction of the LDAP. A, The injection of a hyperpolarizing current pulse at rest increased spontaneous firing of action potentials. At -62 mV, the hyperpolarizing current pulse triggered prolonged discharges of action potentials. The block of action potentials by 1 μm TTX revealed the LDAP. B, At 20-30 sec after bath infusion of 0 mm external Ca²⁺, the LDAP was eliminated. Only a prominent transient rebound spike was generated. C, At 3-5 min after bath infusion of 0 mm external Ca²⁺, the transient spike was also blocked. D, In 0 mm external Ca²⁺, the simulation under voltage-clamp conditions (VC) of the rebound depolarization failed to induce the LDAP when the recording was switched to current-clamp mode (CC) when the simulated rebound spike declined to -15 mV past the peak (top). The bottom is the current recorded during the simulation of the rebound depolarization under voltage-clamp conditions (VC) and after switching the amplifier to current-clamp mode (CC).

To determine whether it is Ca²⁺ entry or the change in the membrane voltage during the transient rebound depolarization that induces the LDAP, we generated a voltage command that simulated the temporal change in the membrane voltage that was observed during the rebound generation of the transient spike in response to a hyperpolarizing current pulse (Fig. 6D). We applied the voltage command to seven different habenular cells in 0 mm external Ca²⁺ to examine whether the simulated membrane-rebound depolarization that mimics the generation of the transient spike would elicit the LDAP in the absence of external Ca²⁺. In the presence of 2 mm external Ca²⁺, those seven cells tested generated the LDAP in duration ranging from 7 to 35 sec in response to a hyperpolarizing current pulse. Then, after external Ca²⁺ was lowered to 0 mm, the cells were held at -50 mV under voltage-clamp conditions, and the voltage command was delivered. When the simulated rebound depolarization proceeded past the peak at 0 mV and declined to -15 mV, the intracellular amplifier was swiftly switched to the current-clamp recording mode to examine whether the simulated rebound depolarization triggered the LDAP. In 0 mm external Ca²⁺, after switching the amplifier to the current-clamp mode, the membrane potential immediately declined to baseline, indicating that Ca²⁺ entry into the cell, not the change in membrane voltage, during the rebound transient spike is critical for the activation of the LDAP. In addition, in the absence of external Ca²⁺, when the habenular cells were hyperpolarized at less than -70 mV by steady negative current, the injection of a depolarizing current pulse (0.1-1 sec) that depolarized the cells to -20-0 mV also failed to elicit the LDAP (Fig. 6C).

The chelation of intracellular Ca²⁺ by including 10 mm bis(2-aminophenoxy)ethane-N,N,N′,N′-tetra-acetic acid (BAPTA) in the pipette solution also resulted in a progressive decrease in the amplitude and duration of the LDAP with time in response to hyperpolarizing current pulse (Fig. 7) (n = 6 cells). After 4 min, the injection of a hyperpolarizing current pulse elicited a similar transient rebound depolarization that was composed of an initial spike and ensuing ADP as was shown with NMDG substitution and bath infusion of FFA (Figs. 4C,D, 5C). This result also supports that Ca²⁺ entry is critical for the activation of the LDAP.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Intracellular infusion of 10 mm BAPTA gradually blocks the activation of the LDAP. After 4 min of BAPTA infusion, a hyperpolarizing current pulse generated a rebound depolarization of 2-3 sec. The external solution contained 0.5 μm TTX.

Taken altogether, the results shown so far suggest that Ca²⁺ entry during the transient rebound depolarization triggers the activation of a tonic influx of Na⁺. This tonic influx of Na⁺ drives the long-lasting depolarization of habenular cells that promotes persistent discharges of action potentials.

Ca²⁺ entry through L-type Ca²⁺ channels triggers the LDAP

Ca²⁺ entry through T-type Ca²⁺ channels produces the transient spike in the habenular cells. During the generation of the transient spike, Ca²⁺ is also likely to enter the cell through high-threshold Ca²⁺channels when the membrane potential depolarizes at more than -30 mV. In particular, the ADP ensuing the initial spike during rebound depolarization could be mediated by Ca²⁺ entry through high-threshold Ca²⁺ channels.

To determine which Ca²⁺ entry leads to the activation of the LDAP, we blocked L-type Ca²⁺ channels by bath infusion of nifedipine in a total of 10 cells. In one half of the 10 cells, we applied 5 μm nifedipine in the bathing medium. In the other half, we increased nifedipine to 10-15 μm. The bath infusion of 5 μm nifedipine decreased the duration of the LDAP by 65.1 ± 11.8% (p < 0.05) (Fig. 8A). In those five cells, 5 μm nifedipine shortened the duration of the LDAP on average from 9.7 to 3.9, 10.1 to 5.3, 14.3 to 3.2, 16.8 to 4.7, and 17.2 to 5.4 sec, respectively, when generated in response to hyperpolarizing current pulse. In the other five cells, bath infusion of 10-15 μm nifedipine blocked most of the LDAP (Fig. 8B). In these cells, before nifedipine infusion, a hyperpolarizing current pulse generated the LDAP in average durations of 7.2, 15.0, 18.2, 18.7, and 22.3 sec, respectively. After bath infusion of nifedipine, the hyperpolarizing current pulse generated a transient spike without the ADP. These results indicate, first, that Ca²⁺ entry through L-type Ca²⁺ channels plays a dominant role in the activation of the LDAP and, second, that the ADP is mediated mostly by activation of L-type Ca²⁺ channels.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

The block of Ca²⁺ entry though L-type Ca²⁺ channels decreases the duration of the LDAP. A, Bath infusion of 5 μm nifedipine reduced the duration of the LDAP by 65 ± 12%. B, A higher concentration of nifedipine (10-15 μm) blocked the LDAP and uncovered a transient rebound spike in response to current pulse injection. Nifedipine effect did not wash out after re-infusion of ACSF for 20 min.

The activation of K⁺ currents plays a role in terminating the LDAP

The block of action potentials by TTX disclosed the LDAP that was often longer than the duration of the promoted discharge of action potentials in response to current pulse injection (Figs. 4, 5, 6). One possible explanation for this phenomenon is that the action potential generation results in the activation of various K⁺ currents, and these outward K⁺ currents exert an influence toward the termination of the LDAP.

To verify the role of K⁺ currents in the termination of the LDAP, we selected one particular type of K⁺ current, the Ca²⁺-activated K⁺ current. We examined how the block of the Ca²⁺-activated K⁺ current influences the LDAP. The bath infusion of 100 nm apamin, the specific blocker of small conductance Ca²⁺-activated K⁺ current, resulted in the prolongation of the LDAP by 285 ± 110% (p < 0.01) (Fig. 9). In four cells tested, apamin increased the average durations of the LDAP from 4.5 to 20.2, 13.7 to 32.3, 22.8 to 47.6, and 40.3 to 109.4 sec, respectively.

• Download figure
• Open in new tab
• Download powerpoint

Figure 9.

A, Bath infusion of 100 nm apamin increased the duration of action potential generation in response to a hyperpolarizing current pulse. The effect of apamin did not wash out after re-infusion of ACSF for 20 min.

Dual role of intracellular Ca²⁺ mobilization by the ER in the generation of the LDAP

The sensitivity of the LDAP to Ca²⁺ entry raises the possibility that intracellular Ca²⁺ regulation may also play an important role in controlling the duration of the persistent discharge of action potentials induced by synaptic hyperpolarization. In neurons, the endoplasmic reticulum (ER) is considered to be the major organelle that regulates intracellular Ca²⁺ concentration in physiological conditions (Svoboda and Mainen, 1999; Berridge et al., 2000; Rose and Konnerth, 2001). On the one hand, in response to either signaling stimuli or Ca²⁺ entry from the external space, the ER releases Ca²⁺ into the cytoplasm and increases intracellular Ca²⁺ levels. On the other hand, the ER takes up Ca²⁺ from the cytoplasm for cell homeostasis and maintains a low intracellular Ca²⁺ concentration.

To examine the role of intracellular Ca²⁺ regulation by the ER in the generation of the LDAP, we first examined the possibility that Ca²⁺ entry from the external space induces Ca²⁺ release from the ER and thereby enhances the activation of the LDAP. The bath infusion of 70-100 μm dantrolene, the blocker of Ca²⁺ release from the ER (Fruen et al., 1997; Zhao et al., 2001), significantly decreased the duration of the LDAP by 68.3 ± 12.5% (p < 0.01) (n = 5 of five cells tested) (Fig. 10A). In those four cells tested, 10 min after dantrolene infusion, the average duration of the LDAP was decreased from 13.6 to 4.9, 18.0 to 9.0, 19.2 to 5.9, 24.7 to 6.2, and 29.4 to 4.9 sec, respectively.

• Download figure
• Open in new tab
• Download powerpoint

Figure 10.

The release of Ca²⁺ from the ER enhances the generation of the LDAP, whereas the block of Ca²⁺ release from the ER prolongs it. A, The block of Ca²⁺ uptake into the ER by bath infusion of 1 μm thapsigargin prolongs the duration of the LDAP. Note the change in time scale in Ab. B, The block of Ca²⁺ release from the ER by bath infusion of 70-100 μm dantrolene significantly decreases the amplitude and duration of the LDAP.

In contrast, the block of Ca²⁺ uptake into the ER by bath infusion of 1 μm thapsigargin resulted in a prolongation of the LDAP by 315 ± 263% (p < 0.01) in five of five cells tested. It has been shown that this concentration of thapsigargin blocks Ca²⁺ uptake into the ER (Thastrup et al., 1990; Treiman et al., 1998). In those cells tested, 1 μm thapsigargin increased the average duration of the LDAP from 3.4 to 7.2, 7.5 to 15.4, 20.2 to 49.1, 39.3 to 82.2, and 55.2 to 95.6 sec, respectively (Fig. 10B).

These results suggest that intracellular Ca²⁺ mobilization by the ER plays two different roles depending on the stage of the development of the LDAP. At early stages, the induction of Ca²⁺ release from the ER enhances the generation of the LDAP, whereas, at later stages, the uptake of Ca²⁺ into the ER decreases contributes to its termination.

Ca²⁺ entry from the external space triggers the generation of the LDAP

Ca²⁺ entry through T- and L-type Ca²⁺ channels plays distinct roles in the activation of the LDAP. A brief hyperpolarization of habenular cells triggers the rebound generation of a transient spike that is mediated by Ca²⁺ entry through T-type Ca²⁺ channels. The depolarization of the membrane by the transient spike in turn results in the activation of high threshold Ca²⁺ channels. Ca²⁺ entry through L-type high-threshold Ca²⁺ channels then leads to the activation of a tonic influx of Na⁺ that generates an LDAP. The persistent depolarization driven by the LDAP stimulates prolonged, intense discharges of tonic action potentials.

We propose that the LDAP is most likely mediated through the activation of CAN currents for the following reasons. First, the substitution of external Na⁺ with NMDG⁺ abolished the LDAP, indicating that the LDAP is generated by a tonic entry of Na⁺. Second, the LDAP is not blocked by TTX, indicating that the tonic Na⁺ entry is not mediated through voltage-gated Na⁺ channels. Third, the LDAP is triggered by Ca²⁺ entry into the cell. Fourth, the LDAP is blocked by low concentrations of FFA that have been routinely used to block nonselective cationic currents. These characteristics of the LDAP are consistent with those of the depolarizing potentials that were reported to be generated by Ca²⁺-activated nonselective cationic (CAN) currents (Partridge and Swandulla, 1993; Wilson et al., 1996; Morisset and Nagy, 1999; Partridge and Valenzuela, 1999, 2000). Although we still cannot rule out the possibility that the LDAP of habenular cells is generated by some other currents yet to be characterized, it is most likely that it is generated by CAN currents.

Whereas Ca²⁺ entry through T-type channels is indispensable for the activation of the LDAP as it generates a rebound spike in response to synaptic hyperpolarization and thereby activates L-type Ca²⁺ channels, Ca²⁺ entry through T-type channels alone appears to be insufficient to activate the LDAP. In our results, the block of L-type Ca²⁺ channels by bath infusion of 10-15 μm nifedipine significantly blocked the LDAP while uncovering the rebound spike that is presumably mediated by T-type Ca²⁺ currents. This result indicates that Ca²⁺ entry through L-type channels is critical for the activation of the LDAP. It is not clear why Ca²⁺ entry through T-type channels alone is insufficient for the activation of the LDAP. Possibly, the subcellular distribution of T-type Ca²⁺ channels is different from that of those channels mediating a tonic entry of Na⁺ for generation of the LDAP, or Ca²⁺ conductance through T-type channels is not large enough to increase local Ca²⁺ levels above the threshold concentration for the activation of the channels mediating the tonic entry of Na⁺.

Another intriguing difference between L- and T- type Ca²⁺ channels is that Ca²⁺ entry through L-type channels induces Ca²⁺ release from the ER. This Ca²⁺-induced Ca²⁺ release from the ER amplifies the generation of the LDAP. Indeed, the block of Ca²⁺ release from the ER by dantrolene markedly reduced the duration of the LDAP by 68 ± 13% (Fig. 10B). Based on these results, it seems reasonable to propose that Ca²⁺ entry through L-type channels results in the activation of the LDAP to a larger extent via inducing Ca²⁺ release from the ER than through the direct activation of the Ca²⁺-activated tonic Na⁺ entry.

The maintenance and termination of the LDAP

The duration of the LDAP appears to be determined by the sum of opposing depolarizing and hyperpolarizing influences that change in their relative strength during the time course of the depolarizing potential generation. The major current that activates and sustains the LDAP is a tonic influx of Na⁺ ions probably through CAN channels.

At lease two cellular systems work together toward the termination of the LDAP. One is the activation of outward K⁺ currents. In particular, we showed that Ca²⁺ entry into the cytoplasm increases the activation of K⁺ currents that oppose the persistent membrane depolarization. In our experiments, the block of small conductance Ca²⁺-activated K⁺ currents by apamin resulted in the prolongation of the LDAP. This result indicates a dual, contrasting action of Ca²⁺ entry in the generation of the LDAP. At early stages of its generation, the dominant action of Ca²⁺ is to trigger and sustain the LDAP probably through the activation of CAN currents. At later stages, the relative influence of Ca²⁺ on the activation of K⁺ currents increases and thereby contributes to the termination of the LDAP. Although not examined in this study, the activation of other types of K⁺ currents is also likely involved in the termination of the LDAP.

The other cellular component that contributes to the termination of the LDAP is Ca²⁺ uptake by the ER. This action of the ER is contrasted with its contribution to the generation and enhancement of the LDAP at early stages when the ER releases Ca²⁺ into the cytoplasm in response to Ca²⁺ entry from the external space. Based on our observation that the block of Ca²⁺ uptake into the ER by thapsigargin markedly prolonged the LDAP (Fig. 10A), we propose that at later stages, the uptake of Ca²⁺ into the ER lowers intracellular Ca²⁺ levels and thus expedites the termination of the Ca²⁺-activated tonic Na⁺ entry.

Functional significance of the long-lasting depolarization of habenular cells

As described in the Introduction, based on anatomical circuitry, the habenula is generally divided into three parts: medial nucleus, medial part of the lateral nucleus, and lateral part of the lateral nucleus (Herkenham and Nauta, 1977, 1979; Sutherland, 1982; Ellison, 1994; but see Andres at al, 1999; Geisler et al., 2003).

The present study and other previous studies showed that the three areas of the habenula exhibit different electrical properties. In our recordings, most neurons in the medial nucleus of the habenula generate tonic series of spontaneous action potentials, whereas cells in the lateral part of the lateral habenula mostly generate burst discharges of action potentials (data not shown). These patterns of electrical activity are consistent with those reported in previous studies (McCormick and Prince, 1987; Wilcox et al., 1988; Huguenard et al., 1993). The medial part of the lateral habenula contains both types of neurons that generate action potentials in tonic and burst modes. More distinctively, the major population of neurons in this area generates prolonged discharges of action potentials in response to a brief membrane hyperpolarization. Those neurons with such distinct pattern of action potential generation are densely populated in particular near the border to the medial habenular nucleus.

The previous anatomical studies showed that the medial part of the lateral habenula, while receiving input from the limbic regions of the forebrain, gives rise to projections to various areas of the hypothalamus, substantia nigra pars compacta, ventral tegmental area, dorsal raphe nucleus, etc. In agreement with the anatomical circuitry, the electrical activity of the habenula has been shown to exert a powerful control over the activity of serotonin cells in the nucleus of dorsal raphe and dopamine cells in the ventral tegmental area and substantia nigra (Matsuda and Fujimura, 1992; Ferraro et al., 1996; Morris et al., 1999; Conley et al., 2002).

What would be the possible advantage of the striatum and limbic forebrain to exert influences on the midbrain structures via the medial part of the lateral habenula? Although we have not yet characterized the neurotransmitter type of the synaptic afferents from the striatum and limbic forebrain, the major effect of synaptic activation of these afferents appears to be a transient hyperpolarization of cells in the medial part of the lateral habenula. The intrinsic property of the habenular cells that produces long-lasting discharges in response to transient synaptic hyperpolarization may render the medial part of the lateral habenula to function as an interface that converts inhibitory influences of the striatum and limbic forebrain into excitatory ones and conveys these excitatory influences to the midbrain. There is evidence that lateral habenular cells that project to the midbrain use excitatory amino acids as neurotransmitter (Kalen et al., 1985, 1986). The habenula may also convert a transient increase in the activity of the striatum and limbic forebrain into a persistent drive that prolongs action potential generation in downstream midbrain target neurons. Such long-lasting discharge of action potentials induced by a brief synaptic hyperpolarization may increase the versatility of inhibitory synapses in the regulation of neural activity.