INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Classical conditioning in the mollusk, Hermissenda crassicornis, involves the presentation of light as the conditioned stimulus and turbulence as the unconditioned stimulus. The result is a change not only in this animal's phototactic behavior (Crow and Alkon 1978; Farley and Alkon 1982), but also in its inherent cellular characteristics. One of these cellular correlates identified after associative learning is found in the ocular photoreceptors. Following conditioning, the excitability of type B photoreceptors and their response to light are enhanced (Alkon et al. 1982, 1985; Blackwell 2002; Crow 1985), while the type A cells exhibit a decrease in light-evoked response (Farley and Han 1997; Richards et al. 1984). The learning-induced photoresponse in these cells is intrinsic (Crow and Alkon 1980; Farley and Alkon 1982; McPhie et al. 1993), with a corresponding increase in input resistance of type B cells (Crow and Alkon 1980). Furthermore, voltage-clamp studies have linked the enhanced excitability of type B photoreceptors in conditioned animals to a reduction in the amplitude of two K⁺ currents, the transient (I_A) and Ca²⁺-activated K⁺ (I_K-Ca) currents (Alkon et al. 1985).

Extensive studies on neuronal plasticity have also been done in other model systems, using both vertebrates and other invertebrates. In Aplysia californica, studies of its defensive siphon and tail withdrawal reflex have demonstrated the role of presynaptic facilitation and a Ca²⁺-dependent postsynaptic enhancement (Bao et al. 1998; Hawkins et al. 1983; Murphy and Glanzman 1996), as well as the involvement of activity-dependent plasticity after associative conditioning (Antonov et al. 2001). Moreover, the associative changes in membrane properties and in the synaptic strength of neurons such as in long-term potentiation (LTP) of Aplysia's sensorimotor neurons, are similar to those in hippocampal cells (Lynch et al. 1990; Staubli and Rogers 1994; Williams et al. 1989) and cerebellar cells (Linden and Ahn 1999).

Studies have demonstrated that learning-induced changes involve the influx of Ca²⁺, activation of Ca²⁺-dependent protein kinases, and the resultant phosphorylation of transmembrane ionic channels. However, such cellular changes as observed in Hermissenda have mostly been studied in the cell body of photoreceptors (Crow and Alkon 1980; Yamoah and Crow 1994, 1995; but see Tamse and Yamoah 2002). Among the diverse ionic currents that have been identified in B photoreceptors is depolarization-induced inward rectifier current (I_h) that is activated by membrane hyperpolarization (Yamoah et al. 1998). Thus inhibitory inputs from the Hermissenda hair cells can be manifested as excitatory output in the B photoreceptors. Several ionic currents in hair cells were described previously (Yamoah 1997). However, it still remains unclear what modifications ensue in the hair cells during conditioning. We hypothesize that activity-dependent modification of the biophysical properties of hair cells can increase the strength of the hair cell-photoreceptor synapse.

Here, we show that Hermissenda hair cells express a P-type Ca²⁺ current that exhibits voltage-dependent facilitation. The single-channel conductance of the P-type channel in hair cells is approximately 20 pS. The current is enhanced by cAMP analogs, and by an activator of adenylyl cyclase, forskolin. We demonstrate that the voltage-dependent facilitation of the P-type current is dependent on the activation of protein kinase A (PKA) using specific inhibitors of the enzyme. These results suggest that repetitive stimulation can accentuate Ca²⁺ influx into hair cells through the P-type channels. Because of the nonlinear relation between Ca²⁺ influx and neurotransmitter release (Katz and Miledi 1967), we expect that voltage-dependent facilitation of the P-type current in hair cells would produce a robust change in synaptic efficacy.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Hair cell isolation

The mollusk, Hermissenda crassicornis, was obtained from Sea Life Supply (Sand City, CA). Animals were held in modified 50-ml tubes and maintained in a tank with recirculating artificial seawater (ASW) kept at 12-14°C. Hermissenda were fed scallops and a 12/12-h light/dark cycle was instituted. Dissection protocols of the CNS were followed as described previously (Yamoah 1997; Yamoah and Crow 1994; Yamoah et al. 1994). Briefly, the CNSs were dissected in ASW and allowed to stabilize at 4°C for 10 min. These were then treated with an enzyme cocktail consisting of protease XXIV (Sigma, St. Louis, MO) and dispase II (Boehringer Mannheim, Mannheim, Germany) at 1 mg/ml ASW and 3 mg/ml ASW, respectively. Isolated CNSs were digested for 15-20 min at 4°C and for another 10 min at room temperature. ASW wash was done several times at 4°C, for a total of 10-15 min. The statocyst was then excised from the CNS, and the transparent capsule around the statocyst torn gently apart to expose the underlying hair cells.

Solutions

All chemicals were obtained from Sigma unless indicated otherwise. ASW used for dissection and CNS wash was comprised of the following (in mM): 400 NaCl, 10 KCl, 10 CaCl₂, 50 MgCl₂, and 15 HEPES. The solution was sterile-filtered and the pH adjusted to 7.8 with 1N NaOH. The extracellular or bath solution during recording of whole cell Ca²⁺ currents consisted of (in mM) 300 choline chloride, 50 MgCl₂, 20 CaCl₂, 10 glucose, 5 4-aminopyridine (4-AP), 100 tetraethylammonium chloride (TEA-Cl), and 15 HEPES, sterile-filtered and adjusted to pH 7.7 with 1 M TEA-OH. The pipette solution was made up of the following (in mM): 400 CsCl, 20 NaCl, 2 MgCl₂, 5 EGTA, 20 TEA-Cl, 10 reduced glutathione, and 40 HEPES. This was also sterile-filtered and adjusted to pH 7.4 with 1 M TEA-OH. Stock solutions of Ca²⁺ channel blockers Cd²⁺ (100 mM), Co²⁺ (100 mM), nimodipine (100 mM in DMSO), and -agatoxin-IVA (1 mM, CalBiochem, La Jolla, CA) were made, and final concentrations of 1 mM, 5 mM, 5-20 µM, and 0.001-100 µM, respectively, were used. Stock solutions of adenosine 3',5'-cyclic monophosphate, N⁶,O^2'-dibutyryl-, sodium salt (dibutyryl-cAMP, 5 mM, CalBiochem), forskolin (5 mM in DMSO, Sigma), H-89 (1 mM in DMSO, CalBiochem), and the synthetic peptide inhibitor of PKA [PKAI(6-22)] (500 µM in distilled water, GIBCO BRL, Rockville, MD) were made and stored at 20°C before use. For experiments in which DMSO was used as the solvent for pharmacological agents, the final concentration of the solvent was approximately 0.001%. The corresponding control experiments were performed using similar concentrations of DMSO (0.001%). For single-channel recordings, the bath solution contained (in mM) 300 KCl, 100 TEA-Cl, 50 MgCl₂, 10 D-glucose, 10 CaCl₂, 5 4-AP, and 10 HEPES, and was adjusted to pH 7.4 with TEA-OH, to shift the resting potential to approximately 0 mV. Patch electrodes were filled with (in mM) 250 Ba²⁺, 100 TEA-Cl, 5 4-AP, and 10 HEPES at pH 7.4 (adjusted with TEA-OH). For all recording solutions, TEA-Cl was used to maintain an osmolarity of approximately 1,000 mosmol.

Electrophysiology

WHOLE CELL CA²⁺ CURRENT RECORDINGS. Whole cell recordings were performed using standard patch-clamp recordings with the Axopatch 200B amplifier (Axon Instruments, Foster City, CA) (Hamill et al. 1981). A horizontal electrode puller (Sutter Instrument Model P-97) was used to make patch pipettes from borosilicate glass capillaries (1.5 mm OD and 1 mm ID; World Precision Instruments, Sarasota, FL). Pipette tips were then fire-polished using a micro-forge (MF-830, Narishige, Tokyo, Japan) to obtain tip diameters approximately 1 µm. Using pipette solutions with ionic strength equivalent to seawater (internal solution), the pipette resistances were approximately 1 M, and only cells in experiments with seal resistances >1.2 G were accepted for analysis. A 3% agar bridge with 3 M KCl was used for reference electrode. Currents were digitized through an A/D converter (Digidata 1200, Axon Instruments). Data collection was controlled through pClamp software (version 8.0, Axon Instruments), and experiments were carried out at room temperature. Data analysis of recorded currents was carried out using Clampfit 8.1 (Axon Instruments) and Origin 6.0 (Microcal Software, Northampton, MA).

SINGLE-CHANNEL RECORDING. The cell-attached configuration was used. Patch pipettes were made from borosilicate glass with 2.0 mm OD and 1 mm ID. The tips of electrodes were fire-polished, and to reduce the capacitance of electrodes, regions close to the tips (approximately 10 µm) were coated with Sylgard (Dow Corning, Midland, MI). Patch pipettes filled with single-channel recording solution had resistances of 1.1 ± 0.7 M (n = 51). Single-channel currents were filtered at 1-2 kHz using a low-pass Bessel filter, sampled at 10-40 kHz, and stored in a personal computer. The channels were activated at a frequency of 0.2 Hz. Analysis was carried out using custom-written software, which was linked to Origin software (MicroCal., Northampton, MA). Leak and capacitative currents were corrected off-line by fitting smooth templates to null traces and subtracting it from active traces. Open-close transitions were detected using half-height threshold analysis criteria. Idealized records were used to generate amplitude histograms and then fitted to a single Gaussian distribution using a Levenberg-Marquardt algorithm to obtain the mean single-channel amplitude and SD. We used a minimum of five voltage steps and their corresponding single-channel currents to determine the unitary conductance. Single-channel current-voltage relation was fitted by a linear least-square regression line, and single-channel conductance obtained from the slope of the regression line. Experiments were carried out at room temperature (approximately 21°C). Where appropriate, pooled data were presented as means ± SD. Significant differences between groups were tested using Student's t-test and the statistical significance was set at P < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Whole cell Ca²⁺ currents in Hermissenda hair cells

The outward K⁺ currents responsible for the repolarizing phase of membrane action potentials of hair cells in the statocysts of Hermissenda have been described in detail (Yamoah 1997). In contrast, phenotypes of the inward currents that shape the depolarization phase of the membrane action potential are unknown. Here, we first studied the inward Ca²⁺ current in hair cells of Hermissenda and determined the properties of the current that may contribute toward presynaptic mechanisms of associative conditioning in this organism. Outward K⁺ currents were suppressed using external TEA-Cl, 4-AP, and internal Cs⁺, and inward Na⁺ currents were eliminated by replacing the external sodium with choline ions (Yamoah et al. 1994). Figure 1A shows a family of inward current traces elicited from a holding potential of 80 mV and stepped to test voltages from 50 to 60 mV. For clarity, a few of the current traces were removed from the illustration. The corresponding current-voltage (I-V) relations from pooled data of 15 cells showed that the current activated at approximately 30 mV and peaked at approximately 20 mV (Fig. 1B). Figure 1C shows a family of instantaneous tail currents elicited using two-pulse voltage-clamp protocol. The instantaneous tail I-V curve could be fitted with a regression line, with zero current at approximately 100 mV (Fig. 1D). By method of exclusion and the reversal potential of the current, we inferred that the current was carried predominantly by Ca²⁺ ions. Furthermore, we established that the inward current was a Ca²⁺ current by examining its sensitivity toward Cd²⁺ and Co²⁺. Whereas 1 mM Cd²⁺ was sufficient to completely block approximately 97% of the current (at a test potential of 10 mV, control current, 0.46 ± 0.11 nA; Cd²⁺, 0.013 ± 0.008 nA; P < 0.05, n = 7), a higher dosage of Co²⁺ (5 mM) was required to block approximately 95% of the current (at a test potential of 10 mV, control current, 0.43 ± 0.16 nA; Co²⁺, 0.02 ± 0.01 nA; P < 0.05, n = 5). The effects of Cd²⁺ and Co²⁺ were reversible (data not shown).

View larger version (26K): [in this window] [in a new window]   Fig. 1. Whole cell Ca2+ currents recorded from hair cells of Hermissenda. A: representative traces for a family of Ca2+ currents obtained from a holding potential of 80 mV and stepped 60 mV. Outward potassium currents were inhibited with bath TEA, 4-AP, and pipette Cs+, and inward sodium current was removed by substituting choline for sodium ions. B: current-voltage curve obtained from 15 hair cells (mean ± SD). C: representative tail current traces and the corresponding protocol used to elicit the current. For clarity, some of the current traces were not plotted. D: reversal potential of the Ca2+ current was measured by clamping back to voltages ranging from 70 to +130 mV after stepping at 10 mV for approximately 25 ms. In n = 9 cells, the mean reversal potential was estimated from the regression line to be 100 ± 3 mV, which is close to that of Ca2+, indicating Ca2+ as the key charge carrier.

To identify the Ca²⁺ current subtype that is expressed in hair cells, we used known organic Ca²⁺ channel blockers. Nimodipine, at micromolar concentrations (5-20 µM), produced no effect on whole cell Ca²⁺ currents (at a test potential of 10 mV, control current, 0.44 ± 0.05 nA; 10 µM nimodipine, 0.43 ± 0.07 nA; P = 0.2, n = 4). In contrast, -agatoxin-IVA blocked the Ca²⁺ current (Fig. 2A). Although the washout of -agatoxin-IVA appeared incomplete (Fig. 2A), this is likely due to the unavoidable rundown of the whole cell current over time. The half-blocking concentration of -agatoxin-IVA estimated from the dose response curve was approximately 0.5 µM (Fig. 2B). Thus the pharmacology of the current suggests that it may belong to the P/Q-subtype.

View larger version (16K): [in this window] [in a new window]   Fig. 2. Pharmacology of the Ca2+ current in hair cells. A: Ca2+ currents were elicited at 2-min intervals from a holding potential of 80 mV using a step potential of 20 mV. The current magnitude was normalized against the peak current and the time-course of run-down is shown (). The current remained stable for approximately 30 min after whole cell configuration was attained. -Agatoxin-IVA (10 µM) blocked the Ca2+ current (see representative traces in inset). The time course of the reversible effects of the toxin is shown (). B: -agatoxin-IVA block of the Ca2+ current is dose-dependent. The percentage inhibition of the current is plotted against -agatoxin-IVA concentration (µM). The continuous line represents a fit of the data to the logistic function: (A1  A2)/(1 + (X/Xo)p) + A2, where A1 and A2 are the initial and final current amplitudes, respectively, X is the concentration of -agatoxin-IVA, Xo is the half-blocking concentration, and p is the Hill coefficient. Numbers in parentheses represent the number of cells. The half-blocking concentration of -agatoxin-IVA was estimated to be 0.5 µM.

The steady-state activation and inactivation properties of the current were examined using standard voltage protocols (Yamoah and Crow 1994; Yamoah et al. 1994). Normalized, steady-state activation and inactivation curves for the Ca²⁺ current are presented in Fig. 3. Using the Boltzmann distribution to fit the two curves, the estimated half-activation voltage (V_m1/2) and the maximum slope, k_m, were 1.2 ± 0.6 and 7.6 ± 0.6 mV (n = 8), respectively. The inactivation curve generated from currents at a test potential of 10 mV, and at several conditioning potentials was sigmoidal in shape, and the calculated half-inactivation voltage (V_h1/2) and the slope factor, k_h, were 14.0 ± 3.0 and 13.2 ± 1.6 mV (n = 7), respectively. The presence of a window current from approximately 20 to 10 mV suggest that the Ca²⁺ current may contribute substantially toward the action potential of hair cells.

View larger version (18K): [in this window] [in a new window]   Fig. 3. Steady-state activation and inactivation of Ca2+ currents in hair cells of Hermissenda. The activation curve was obtained by plotting the normalized peak Ca2+ currents (I/Imax) at each test potential and fitted by the Boltzmann equation {I/Imax = [1 + exp(V1/2  V)/km]}1, where V1/2 is the half-activation voltage and km is the maximum slope. The V1/2 and km as calculated from the activation curve were 1.2 ± 0.6 and 7.6 ± 0.6 mV (n = 7), respectively. The inactivation curve was plotted against the test potentials after normalizing to the peak Ca2+ currents (I/Imax), and fitted by the Boltzmann equation {I/Imax = [1 + exp(V  V1/2,)/kh]}1, where V1/2, is the half-inactivation voltage and kh is the slope factor. The estimated V1/2 and kh from the inactivation curve were 14.0 ± 3.0 and 13.2 ± 1.2 mV (n = 7), respectively. Inset: representative current traces used to generate the inactivation curve.

Single-channel currents in hair cells

To further identify Ca²⁺ currents in Hermissenda hair cells, we measured the unitary current from cell-attached patches. Figure 4A shows a family of single-channel currents traces carried by 250 mM Ba²⁺. The amplitude histograms of the unitary current obtained at a test potential of 15 mV is shown (Fig. 4B). Illustrated in Fig. 4C is the unitary current amplitude plotted against the test potentials. The slope conductance of the channel was approximately 20 pS.

View larger version (27K): [in this window] [in a new window]   Fig. 4. Single-channel Ba2+ currents in hair cells. A: a family of single-channel Ba2+ currents recorded in the cell-attached configuration from a holding potential of 80 mV to the step potentials indicated beside each trace. Ba2+ (250 mM) was the charge carrier. B: an example of amplitude histograms (step potential = 15 mV) used to determine the mean magnitude of the unitary current at each step potential. C: unitary current-voltage relation. The slope conductance of the regression line is approximately 20 pS. Group data from 6 patches had a mean of 19.6 ± 1.8 pS.

Ca²⁺ currents show voltage-dependent facilitation

Figure 5A shows Ca²⁺ current traces generated using a test potential of 0 mV from a holding potential of 80 mV. In Fig. 5B, however, the initial test pulse was followed by a conditioning depolarization pulse to 70 and 90 mV for approximately 700 ms with a gap of approximately 5 ms, followed by a second test potential to 0 mV. The Ca²⁺ current was visibly enhanced following the conditioning depolarization pulses, increasing by approximately 1.2- and 1.5-fold using 70 and 90 mV 700-ms conditioning depolarization pulses, respectively (Fig. 5C). For example, control current at preconditioning test potential of 0 mV = 0.35 ± 0.07 nA; postconditioning (90 mV) current was at 0.52 ± 0.10 nA (P < 0.05, n = 7). Voltage-dependent facilitation of the Ca²⁺ current was influenced by the duration of the conditioning pulse. As shown in Fig. 5D, the magnitude of the test currents increased with the duration of the conditioning pulses.

View larger version (24K): [in this window] [in a new window]   Fig. 5. Voltage-dependent facilitation of Ca2+ currents in hair cells. A: exemplary Ca2+ current trace elicited from a holding potential of 80 mV to a step potential of 0 mV for approximately 5 ms. Note that the time scale of a and b are different; the gaps shown in the protocol and trace indicate the end and beginning of the clock. B: cell membrane was held at 80 mV and stepped to 0 mV, and then stepped back to 80 mV for 5 ms, followed by a conditioning prepulse to 70 and 90 mV for 0.7 s and then a gap of 5 ms at the holding potential before the test pulse at 0 mV for 5 ms (note that the pulse protocol shown is not drawn to scale). There was an enhancement of the current, after the conditioning pulse. C: normalized group data expressed as the percentage facilitation from 11 cells are plotted. D: increasing the duration of the conditioning pulse further enhanced the current, which can be described with time constants of 1.2 ± 0.2, 1.1 ± 0.5, and 0.42 ± 0.06 s at 50-, 70-, and 90-mV conditioning pulses, respectively.

Aside from voltage-dependent facilitation of the Ca²⁺ current, the current was enhanced on application of dibutyryl-cAMP (500 µM; Fig. 6, A and C). Analysis of the group data of the peak current, as elicited by a voltage step to 20 mV from a holding potential of 80 mV, revealed that the cAMP analog resulted in a significant enhancement of the Ca²⁺ current (control mean = 0.60 ± 0.03 nA; cAMP = 0.84 ± 0.05 nA; P < 0.05, n = 7). The effects of forskolin, an activator of adenylyl cyclase, suggest that the increase in Ca²⁺ current was mediated by PKA (Fig. 6, B and C). In the presence of forskolin, the Ca²⁺ current elicited by a voltage step to 20 mV from 80 mV was increased (control mean = 0.50 ± 0.09 nA; cAMP = 0.80 ± 0.08 nA; P < 0.05, n = 6). Cells that were dialyzed with the PKA inhibitor, PKI, produced a substantial reduction of the Ca²⁺ current (data not shown, but see Fig. 7C). Likewise, administration of H-89 produced a significant reduction of the current, suggesting PKA was constitutively active in the hair cells (control mean = 0.51 ± 0.04 nA; H-89 = 0.29 ± 0.10 nA; P < 0.05, n = 7).

View larger version (23K): [in this window] [in a new window]   Fig. 6. Modulation of Ca2+ currents in Hermissenda hair cells using dibutyryl-cAMP. A: Ca2+ current traces were generated by holding the membrane at 80 mV and stepping to voltages from 80 to 40 mV. Some of the traces were removed for clarity. The experiments were conducted before and after application of 500 µM dibutyryl-cAMP. Dibutyryl-cAMP produced a noticeable increase in the magnitude of the current. B: similarly, forskolin enhanced the Ca2+ current as shown. C: current-voltage relationships before () and after addition of cAMP (), showing increased current amplitude in the presence of this 2nd messenger (n = 7). The group data showing the effects of forskolin before () and after () application indicate a significant enhancement of the current (n = 6).

View larger version (23K): [in this window] [in a new window]   Fig. 7. Voltage-dependent facilitation is mediated by cAMP. A: Ca2+ current trace elicited from a holding potential of 80 mV to a step potential of 0 mV for approximately 4 ms. The current protocols were similar to that described in Fig. 5B. Note the different time scale for the test pulse to 0 mV and the conditioning depolarizing pulse to 80 mV. B: cAMP enhanced the current, postconditioning. On application of 500 µM dibutyryl-cAMP, the voltage-dependent facilitation of the current was further enhanced. C: in the presence of 5 µM PKI in the recording pipette, voltage-dependent facilitation was completely abolished. D: normalized group data, expressed as percentage facilitation from 6 cells in each set of experiments, are plotted.

Previous studies have shown that voltage-dependent facilitation of the L-type Ca²⁺ currents resulted from PKA phosphorylation of the channels (Dolphin 1998; Kamp et al. 2000; Sculptoreanu et al. 1993). We examined the effects of dibutyryl-cAMP on Ca²⁺ current pre- and postdepolarizing conditioning pulses. Consistent with the data shown in Fig. 6, the cAMP analog increased the Ca²⁺ current (comparing currents preceding the conditioning pulse; Figs. 7, A-C). In addition, the postdepolarization conditioning current was further enhanced in the presence of the cAMP analog. Analysis of the group data revealed a statistically significant increase in Ca²⁺ currents after the application of dibutyryl-cAMP. In contrast, for cells that were dialyzed with PKI, voltage-dependent facilitation of the current was completely abolished (Fig. 7C).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The neuronal network that confers associative conditioning-induced plasticity in Hermissenda involves polysynaptic connections (Crow and Tian 2002). However, until recently, only the monosynaptic connection between the photoreceptors and hair cells have been examined (Alkon and Fuortes 1972; Crow and Tian 2000; Frysztak and Crow 1994, 1997). However, the roles of the presynaptic hair cells remain most uncertain. The study was motivated, in part, by the established role of presynaptic neurons in associative conditioning-induced plasticity (Schuman and Clark 1994). This is the first detailed description of the voltage-dependent Ca²⁺ currents in the presynaptic hair cells, and we found that the currents exhibit a voltage-dependent facilitation. This characteristic could potentially facilitate Ca²⁺ influx, increasing neurotransmitter release as well as activating a variety of Ca²⁺-dependent mechanisms that may be required for the cellular mechanisms associated with classical conditioning in Hermissenda. Similar to the modulation of ionic currents and presynaptic mechanisms of neuronal plasticity in Aplysia sensory neurons, facilitation of Ca²⁺ currents in Hermissenda hair cells is mediated by PKA activation. However, it appears that in hair cells, the Ca²⁺ current is modulated directly by PKA, which is different from Aplysia sensory neurons wherein K⁺ currents are the main targets for PKA modulation (Baxter and Byrne 1990; Byrne and Kandel 1996; Goldsmith and Abrams 1992; Sugita et al. 1997).

Identity of the Ca²⁺ current in Hermissenda hair cells

Studies of Ca²⁺ channels in neurons and other cells in vertebrates have led to the classification of at least six subtypes of Ca²⁺ channels, namely, T-, N-, L-, P-, Q-, and R-type channels (Randall and Tsien 1995; Tottene et al. 1996; Tsien et al. 1988). The heterogeneity of the channels results from the expression of distinct ₁-subunits (Snutch and Reiner 1992). Molecular cloning techniques have been used to identify at least six types of ₁-subunits, which are classified as A, B, C/D, E, and H and correspond to the P/Q-, N-, cardiac and neuronal L-, R-, and T-type channels, respectively (Catterall 2000). There are striking similarities between the Ca²⁺ currents described in presynaptic hair cells of Hermissenda to that of the P-type currents, the latter of which are expressed in presynaptic terminals (Mulligan et al. 2001). The whole cell Ca²⁺ current in Hermissenda hair cells was insensitive to nimodipine and -conotoxin GVIA (data not shown), ruling out the presence of L- and N-type currents, respectively. However, the Ca²⁺ current was blocked markedly by -agatoxin-IVA, which suggest strongly that hair cells express the P-type current. Typically, the activation-voltage and the peak P-current in mammalian neurons occur at approximately 50 and 0 mV, respectively (Regehr and Atluri 1995; Tsujimoto et al. 2002) compared with that observed in Hermissenda hair cells, i.e., approximately 30 and 20 mV, respectively. However, we expect a shift in the voltage-dependent activation, arising from surface charge effects imposed by the high (approximately 60 mM) divalent cations in the seawater recording conditions (Rodriguez-Contreras et al. 2002). Taken together, we can infer that the P-type channel carries the predominant Ca²⁺ current in Hermissenda hair cells.

Mechanisms of voltage-dependent facilitation in P-type current in hair cells

Whereas voltage-dependent facilitation of Ca²⁺ current has been observed in several preparations (Dolphin 1996, 1998), and in recombinant Ca²⁺ channels expressed in HEK-293 cells (Kamp et al. 2000), others have failed to detect voltage-dependent facilitation (Meza and Adams 1998; Zong et al. 1995). The diverse outcomes of these studies suggest that voltage-dependent facilitation of Ca²⁺ currents may be regulated by a variety of mechanisms (Kamp et al. 2000). Still, there are some uncertainties regarding a given mechanism. For example, there are contrasting reports on whether PKA-mediated phosphorylation of Ca²⁺ channels is necessary for voltage-dependent facilitation (Eisfeld et al. 1996; Kamp et al. 2000; Sculptoreanu et al. 1993). However, a common feature for voltage-dependent facilitation is that it is prevalent in the L-type Ca²⁺ currents. In contrast, the Ca²⁺ current observed in hair cells was of the P-type, and showed voltage-dependent facilitation that is mediated by PKA activation (Fig. 7). Another distinct feature of the P-type current facilitation in hair cells is that its kinetics can be described by a monophasic time constant (approximately 0.4-1.2 s), which is relatively slow compared with the more rapid voltage-dependent facilitation reported for the L-type current (approximately 10 ms: Kamp et al. 2000). Thus the slow kinetics of the voltage-dependent facilitation of Ca²⁺ currents in Hermissenda hair cells would be consistent with the requirement for PKA activation. Facilitation of the P/Q-type Ca²⁺ current has been demonstrated in presynaptic neurons of the calyx of Held synapse in vertebrates, with a relatively fast time constant (approximately 15 ms). Moreover, in that system, neuronal calcium sensor 1 mediates activity-dependent facilitation of the current (Tsujimoto et al. 2002). In addition, the P/Q-type current in vertebrate neurons exhibits G-protein-mediated inhibition that is relieved by strong depolarizations, producing an apparent voltage-dependent facilitation (Jones and Elmslie 1997).

Functional implication for synaptic plasticity in Hermissenda

Since neurotransmitter release is mediated by Ca²⁺ influx via voltage-gated Ca²⁺ channels and because of the nonlinear relation between Ca²⁺ influx and neurotransmitter release (Katz and Miledi 1967), we predict that voltage-dependent facilitation of the P-type current in hair cells would produce robust change in synaptic efficacy. Thus prolonged stimulation of hair cells can enhance the efficacy of the hair cell-photoreceptor synapse. Although synaptic plasticity is one of the hallmarks for mechanisms of learning and memory, it remains unclear how the present observation could be incorporated in the mechanisms of classical conditioning in Hermissenda. However, it is also conceivable that inhibitory inputs from hair cells can be manifested as excitatory output in the B photoreceptors. Moreover, this can be mediated through activation of a depolarization-induced inward rectifier current (I_h) that is activated by membrane hyperpolarization in type B cells (Yamoah et al. 1998). The prediction would rely on the temporal synchrony between inhibitory inputs from hair cells, [unconditioned stimulus (US)] and activation of I_h in the B photoreceptors, which receive the conditioned stimulus, CS. Nonetheless, voltage-dependent potentiation of Ca²⁺ currents may serve as one of the mechanisms of synaptic plasticity in Hermissenda neurons.

Modulation of synaptic connections between neurons that are involved in both associative and nonassociative conditioning has also been well documented (Glanzman 1995). The mechanisms underlying associative and nonassociative learning in the gill and siphon withdrawal reflexes involve changes in synaptic efficacy, mediated in part, by 5-HT-mediated facilitation of the sensory neurons (Byrne and Kandel 1996; Clark et al. 1994) and a subsequent enhancement of the release of glutamate at the sensory-motor neuron synapse (Chin et al. 2001; Chitwood et al. 2001; Levenson et al. 2000; Marinesco and Carew 2002). These studies have shown that synaptic facilitation results in activation of adenylyl cyclase, resulting in increased levels of cAMP, activation of PKA, which in turn, leads to the phosphorylation of proteins (Klein 1993; Wu et al. 1995). The enhanced cAMP levels and the ensuing PKA activation and phosphorylation of ion channels result in the reduction of outward K⁺ currents, leading to the broadening of action potential duration, and enhancement of Ca²⁺ influx (Bao et al. 1998; Kandel 2001; Sugita et al. 1997). The aforementioned studies reported that K⁺ currents are the primary ionic currents subject to PKA modification. However, our present findings demonstrate that modulation of the P-type Ca²⁺ current in Hermissenda hair cells may contribute toward neuronal plasticity associated with learning and memory. Thus neuronal plasticity in these two invertebrate systems may share common features, which involves the requirement for enhanced Ca²⁺ influx.