Voltage- and Ca²⁺-Gated Currents in Zebrafish Olfactory Receptor Neurons

Recently, our understanding of the physiology, biochemistry and molecular biology of olfaction has progressed rapidly. Genes coding for putative olfactory receptor proteins have been cloned (Buck and Axel, 1991; Ngai et al. 1993), several transduction pathways have been described (for reviews, see Breer and Boekhoff, 1992; Firestein, 1992; Shepherd, 1994; Zufall et al. 1994) and the electrical events that these transduction cascades initiate are becoming better understood (for a review, see Firestein, 1992; see also Miyamoto et al. 1992; Dubin and Dionne, 1993; Firestein et al. 1993; Ivanova and Caprio, 1993; Kleene, 1993; Kurahashi and Yau, 1993; Lowe and Gold, 1993a,b; Morales et al. 1994; Okada et al. 1994). Additional progress in our understanding of olfaction is likely, but may be particularly swift if an animal model is chosen that is particularly useful for genetic analysis. The zebrafish Danio rerio is an increasingly popular animal for the study of vertebrate development and its genetic control (Rossant and Hopkins, 1992; Kimmel, 1993; Strähle and Blader, 1994). The characterization of mutants is facilitated because zebrafish are easy to rear and maintain in the laboratory and their embryos develop quickly and externally. Additionally, transgenic zebrafish have been successfully produced (Stuart et al. 1990). Recently, the development of the zebrafish olfactory epithelium (Hansen and Zeiske, 1993) and the organization of the olfactory bulb (Baier and Korsching, 1994) have been described and partial gene sequences coding for putative odorant receptors and the locations of receptor cells expressing mRNAs for these receptors have been determined (Byrd et al. 1994; Korsching and Baier, 1992; Baier et al. 1994).

While these reports constitute significant progress towards a better understanding of olfactory development and differentiation, and may have important implications regarding the encoding of olfactory signals, little is known regarding the physiology of the zebrafish olfactory system. A study employing the underwater electro-olfactogram technique (EOG; Silver et al. 1976) has shown that the zebrafish olfactory epithelium responds to a wide range of odorants and is generally similar to that of other fishes in its specificity and sensitivity (Michel and Lubomudrov, 1995). However, it remains to be shown that the physiology of single zebrafish olfactory receptor neurons (ORNs) is similar to that of ORNs from other species. Voltage-and Ca²⁺-activated currents have been described for ORNs from a number of animals and, while there are many similarities between species in their complement of currents, some differences have been noted. For example, ORNs from some species lack transient outward currents while ORNs from others lack inward rectifying currents (see Discussion). To evaluate the usefulness of the zebrafish for the study of olfactory physiology, we have undertaken a study of gated currents in the olfactory receptor cells of the zebrafish. Here we report that zebrafish ORNs possess a complement of voltage-and Ca²⁺-activated currents similar but not identical to the gated currents reported for ORNs from other species.

Zebrafish [Danio rerio, formerly Brachydanio rerio (Hamilton-Buchanan) (Meyer et al. 1993)] were purchased from a commercial supplier (Steve Lambourne Co, Inc., Sylmar, CA, USA) and housed in recirculating 40–80 l aquaria at approximately 26 °C. Fish were fed Tetramin flake food daily and used within 6 months of their arrival in the laboratory.

To isolate olfactory rosettes, fish were first immersed in artificial fresh water (see below) containing the general anesthetic MS-222 (20 mg l^-1) until gentle pinches to the tail failed to elicit escape behavior. Fish were then rapidly decapitated. Olfactory rosettes were dissected free of the head in cold fish Ringer and kept at 4 °C until needed. Following decapitation, the sex of the fish was determined by opening the abdominal cavity and examining its contents under a dissecting microscope. The presence of oocytes 0.3–0.8 mm in diameter indicated a mature female fish (Selman et al. 1993, 1994).

Olfactory receptor neurons (ORNs) were isolated by first placing rosettes in divalent cation-free Ringer containing L-cysteine-activated (1.25 mg ml^-1) papain (0.25 mg ml^-1) and allowing digestion to occur at room temperature for 15 min. Rosettes were then placed in divalent cation-free Ringer and cells were liberated by trituration onto coverslips coated with concanavalin A.

Whole-cell voltage-clamp recordings were made using borosilicate patch pipettes (resistance approximately 7 MΩ) that had been placed onto ORN somata under visual control. Light suction was applied to achieve a gigaohm seal and greater, maintained suction was applied to achieve the whole-cell configuration. Gated currents were amplified with Axopatch 200 or 200A amplifiers (Axon Instruments, Foster City, CA, USA), low-pass-filtered at 5 kHz (except when recording capacitance transients, see below), digitized on-line, and stored on an IBM-compatible 486 computer using PClamp software. Leak subtraction was performed on-line unless otherwise specified. Data were analyzed using primarily Webfoot software (Salt Lake City, UT, USA) with some analyses performed using Quattro (Borland International, Inc., Scotts Valley, CA, USA).

Cell capacitance (C_m), series resistance (R_s) and input resistance were determined by averaging records obtained from 50 steps from -84 mV to -74 mV (low-pass-filtered at 50 kHz). The difference between the holding currents recorded at these two potentials was noted to calculate input resistance. Capacitance was determined by integrating the area under the averaged capacitance transient to determine net charge. Series resistance was calculated by fitting the decay of the averaged capacitance transient to a single exponential function with time constant τ and applying the formula R_s= τ/C_m. The average R_s value obtained was 36.0±20.0 MΩ (mean ± S.D.). In some cases, on-line R_s compensation was applied at 40–62 %. In cases where large currents were recorded, series resistance error often exceeded 5 mV, so in these cases the magnitude of the error was determined and data were plotted in current–voltage relationships at potentials that had been corrected for this error. Specifically, since every different peak current value that was used to compute a mean current value was actually recorded at a slightly different potential (owing to differences in R_s error), each mean current value was plotted at an average potential that differed by no more than 5 mV from the R_s-corrected potentials at which each of the individual current values were recorded.

Junction potentials between the pipette and bath solutions were measured using a patch pipette filled with 3 mol l^-1 KCl as the reference electrode (Neher, 1992). These potentials measured +4 mV in magnitude, bath relative to pipette, with both the normal and Cs⁺ pipette solutions (see below). All references to membrane potentials have been corrected for the error introduced by junction potentials (Barry and Lynch, 1991).

Equilibrium potentials for ions were calculated with the Nernst equation using concentrations when the concentration of an ion was less than 100 mmol l^-1 and activities when the concentration was greater than 100 mmol l^-1. To determine activities, published values for activity coefficients (Weast, 1968) were fitted to fourth-order polynomial equations and those equations were used to predict the activity coefficients that corresponded to particular ion concentrations through interpolation. Since the activity coefficient of an ion depends on the other ionic species present (Vaughan-Jones and Aikin, 1995), and since we did not attempt to correct for this effect, the resulting calculated equilibrium potentials are only estimates.

Artificial fresh water contained (in mmol l^-1): 3.0 NaCl, 0.2 KCl, 0.2 CaCl₂, 1.0 Hepes (pH 7.2). Normal fish Ringer (bath solution) contained (in mmol l^-1): 137.0 NaCl, 2.0 KCl, 1.8 CaCl₂, 1.6 MgCl₂, 5.0 Hepes, 10.0 glucose, approximately 4.0 NaOH (pH 7.4). The composition of the divalent-cation-free fish Ringer used in cell dissociation was the same as that of the normal Ringer except that it lacked CaCl₂ and MgCl₂. In experiments where Ca²⁺ currents were blocked, CoCl₂ replaced CaCl₂ and MgCl₂ in the Ringer’s solution on an equimolar basis (Co²⁺ Ringer, 3.4 mmol l^-1 CoCl₂). This allowed us to use the highest possible concentration of CoCl₂ while maintaining a normal concentration of divalent cations. In cases where tetrodotoxin (TTX), 4-acetamido-4’-isothiocyanatostilbene-2,2’-disulfonic acid (SITS) or 4-aminopyridine was used, these compounds were added to the Ringer at final concentrations of 1 μmol l^-1, 5 mmol l^-1 and 5 mmol l^-1, respectively, without compensation for changes in osmolarity. The composition of the normal pipette solution was (in mmol l^-1): 140.0 KCl, 0.1 CaCl₂, 2.0 MgCl₂, 1.1 EGTA, 10.0 Hepes, approximately 10.1 KOH (pH 7.5). In experiments where K⁺ currents were blocked, the pipette solution contained 140.0 mmol l^-1 CsCl instead of KCl and approximately 7.7 mmol l^-1 NaOH instead of KOH (Cs⁺ pipette solution). The free Ca²⁺ concentration in the normal pipette solution was calculated to be 10 nmol l^-1. The compositions of the solutions used in this study are summarized in Table 1. All chemicals were obtained from Sigma Chemical Co. (St Louis, MO, USA) or Baxter Diagnostics Inc. (McGaw Park, IL, USA).

Mean values are cited throughout with either the standard deviation (S.D.) or standard error (S.E.M.) where the sample sizes were large and range when the sample sizes were small.

We report results from 56 zebrafish ORNs. Olfactory receptor neurons appeared as round somata with diameters of approximately 4 μm and possessing a single thick process of approximately 2 μm in diameter and of 6–15 μm in length (the dendrite). Apical cilia were evident on approximately 10 % of these dendrites. Twenty-three per cent of the cells (13 out of 56) also possessed a much thinner process of variable length (but generally less than 10 μm) which was presumably the axon. Upon achieving the whole-cell recording configuration, all cells assumed a rounded morphology. The average cell capacitance was 0.66±0.18 pF (S.D.) and the average apparent input resistance was 8.0±7.5 GΩ (S.D.). With normal Ringer and pipette solution, depolarizing steps from a holding potential of -94 mV elicited a rapid, transient inward current followed by outward currents with transient and sustained components (Fig. 1). Presumably as a result of the small cell size and rounded morphology during recording, we saw no evidence of poor voltage control due to space-clamp problems.

The transient inward current was isolated using the Cs⁺ pipette solution and Co²⁺ Ringer in the bath. Depolarizing steps elicited transient inward currents that first activated between –74 mV and –64 mV (15 cells; Fig. 2A). Current amplitude increased with further depolarization to about –20 mV and then decreased at more positive test potentials. The peak transient inward current ranged from –15 to –307 pA and occurred at test potentials ranging from –24 to –4 mV. For each of these cells, current amplitudes were converted to conductance measurements using a routine that corrected the membrane potentials for the error introduced by the series resistance and allowed the reversal potential to float to achieve the best fit to the Boltzman function. Using this method, the average potential where conductance was half-maximal was determined to be –28±11 mV (S.D.) and the average predicted reversal potential was determined to be +57±22 mV (S.D.; Fig. 2B). We conclude that this current is a Na⁺ current (I_Na) in part because this value for the average reversal potential is close to the estimated value of E_Na (which is +67 mV) but also because the current was never observed when the bath contained TTX and because the current was insensitive to the Co²⁺ in the Co²⁺ Ringer.

With I_Na isolated using the agents described above, the voltage-dependence of its steady-state inactivation was determined in four cells by first holding cells at –114 mV and then stepping to various potentials to achieve steady-state inactivation of I_Na followed by a step to –24 mV to elicit the residual I_Na that had not been inactivated. The resulting data were fitted with a Boltzman distribution and the subsequent fits were used to determine the potential at which inactivation was half-maximal (Fig. 2B). The average potential for half-maximal inactivation was –78 mV (range –69 to –82 mV).

With I_Na isolated as above, recovery of I_Na from inactivation was investigated in eight zebrafish ORNs by holding cells at –104 mV, stepping to –4 mV to inactivate I_Na completely, stepping back to -104 mV for various times to allow some degree of recovery, and finally stepping to –24 mV to elicit I_Na. The resulting current magnitudes were plotted as a function of the recovery interval and were fitted well by double exponential functions (Fig. 2C). At –104 mV, the average time constants of recovery were 23±21 ms and 532±474 ms (S.D.). Recovery was much faster at -124 mV (two cells; Fig. 2C), but it was not possible to hold cells for the full duration of an experiment at this potential. This made it impossible to obtain accurate measurements of the second, slower time constant of recovery but the faster time constants were 7 ms and 5 ms in these two cells.

A sustained inward current was investigated in four cells by including TTX in the Ringer and using the Cs⁺-based pipette solution. Depolarizing steps from a holding potential of –94 mV elicited a very small inward current that in three cells appeared to be sustained except at +16 mV (Fig. 3), while in a fourth cell the inward current appeared to be transient at all potentials. Because this current was highly labile, it was not possible to demonstrate its Co²⁺-sensitivity by first observing the current, then blocking it with Co²⁺ Ringer, and then showing recovery in normal Ringer. Therefore, in two of these cells, the current was first observed in normal Ringer (with TTX) and then blocked by changing the bath to Co²⁺ Ringer (with TTX), while in the other two cells this order of bath presentation was reversed. Results were similar with either order of bath presentation. A very small Co²⁺-sensitive current, which was presumably I_Ca, was activated between –44 and –34 mV (Fig. 4). Its amplitude increased at more positive potentials to reach a peak at approximately -10 mV and then declined at still more positive potentials. In the one cell where this current appeared transient at all potentials, contaminating currents were observed in the presence of Co²⁺ Ringer. When contaminating currents were removed by subtracting them from the currents originally observed in normal Ringer (with 1 μmol l^-1 TTX), the isolated Co²⁺-sensitive I_Ca was found to be largely sustained (not shown). Therefore, the transient nature of the currents originally observed in this cell was due to the presence of contaminating currents. In contrast, the transient nature of I_Ca at +16 mV was observed in all cells examined (Fig. 4) and was not due to contamination as blocking I_Ca with Co²⁺ Ringer did not reveal contaminating currents. In three other cells where contaminating currents were observed, they were unaffected by changing the bath to Ringer containing the Cl^- channel blocker SITS (5 mmol l^-1), so it is unlikely that these currents were carried by Cl^-.

Ca²⁺-independent outward currents were elicited from eight cells by applying depolarizing steps using the normal pipette solution and with Co²⁺ Ringer containing TTX in the bath. In five of these eight cells, depolarizing steps were also applied with normal Ringer (with TTX) in the bath to elicit total outward currents and the Ca²⁺-dependent component was isolated by subtracting traces obtained with Co²⁺ Ringer from those obtained with normal Ringer. Ca²⁺-independent outward currents appeared first between –34 and –24 mV and increased at more positive potentials (Fig. 5B,D). Ca²⁺-dependent outward currents were first measurable between –44 and –34 mV, reached a peak at approximately 0 mV, and declined at more positive potentials (Fig. 5C,D). Both Ca²⁺-dependent and Ca²⁺-independent outward currents displayed inactivation.

The ion-carrying outward current was determined by analyzing tail currents. With Ca²⁺-independent outward currents isolated as described above, seven cells were first held at –74 mV, then stepped to +16 mV to activate outward currents, and finally stepped to test potentials ranging from –134 to –94 mV to elicit tail currents (Fig. 6). The average reversal potential of the Ca²⁺-independent outward currents was –116 mV (range –127 to –107 mV), which is near the estimated value for E_K of –107 mV. In two cells, the reversal potential of the Ca²⁺-dependent outward current was determined by first eliciting total tail currents using normal Ringer with TTX and then again with Ca²⁺-dependent currents blocked (using Co²⁺ Ringer with TTX). In a third cell, the order of bath presentations was reversed. Subtraction yielded Ca²⁺-dependent tail currents. The average reversal potential of the Ca²⁺-dependent tail currents was -102 mV (range -117 to -85 mV; Fig. 6) which is close to the estimated value of E_K of -107 mV. Since the estimated value for E_Cl is -1 mV, it is clear that the Ca²⁺-dependent outward currents were carried largely by K⁺ and not by Cl^-.

To determine whether the Ca²⁺-independent K⁺ current consisted of a single inactivating current or an inactivating current (such as an A-current; I_A) and a delayed rectifier current (I_DR), Ca²⁺-independent outward currents were isolated from eight cells using the agents described above and then cells were subjected first to a protocol that elicited total Ca²⁺-independent outward currents and then to a protocol that inactivated I_A and selectively activated I_DR. To elicit total Ca²⁺-independent outward currents, cells were held at –74 mV and then stepped to –124 mV to remove any steady-state inactivation, and finally stepped to various test potentials (Fig. 7A). To elicit the delayed rectifier current selectively, cells were subjected to the same protocol but, instead of being stepped to –124 mV, they were stepped to –44 mV to inactivate I_A (Fig. 7B). Subtraction yielded the current inactivated at –44 mV (Fig. 7C). Results are consistent with the presence of two Ca²⁺-independent K⁺ currents, I_A and I_DR. Both currents activated between –34 and –24 mV and increased in magnitude with further depolarization (Fig. 7D). The observation that isolated I_DR displayed some inactivation (Fig. 7B) suggests that the two currents were not completely separated by this method; I_A was probably not completely inactivated at –44 mV. Therefore, the rate of I_A inactivation was determined by fitting total Ca²⁺-independent K⁺ current traces (e.g. Fig. 7A) rather than fitting traces of isolated I_A. At approximately 40 mV (the value varied slightly owing to differences between cells in series resistance error), the rate of inactivation was best fitted by a single exponential function with an average time constant of 76±57 ms (eight cells, S.D.).

This is similar to values reported for I_A in other systems (Rudy, 1988). Inactivation was not analyzed at other potentials, but inspection of current traces (e.g. Fig. 7A) suggests that this rate of inactivation was independent of membrane potential.

In some neurons, I_A can be discriminated from I_DR pharmacologically using 4-aminopyridine (4-AP; Hille, 1992). To determine whether I_A in zebrafish ORNs is 4-AP-sensitive, total Ca²⁺-independent currents were recorded from four ORNs before and during exposure to 5 mmol l^-1 4-AP. Subtraction of the relatively sustained 4-AP-insensitive current (I_DR, Fig. 8B) from the total Ca²⁺-independent outward current (Fig. 8A) left a rapidly inactivating 4-AP-sensitive outward current (I_A, Fig. 8C).

To determine whether zebrafish ORNs possessed currents activated by hyperpolarization, six cells were held at –54 mV and stepped to potentials ranging from –69 to –129 mV for 875 ms. One of these cells displayed slow, inward rectification (not shown) but the others did not. Hyperpolarization-activated currents were also not observed in the eight cells described above that had been stepped from –74 to –124 mV for 518 ms in experiments designed to separate I_A from I_DR. Inward rectifying currents were not investigated further.

Michel and Lubomudrov (1995) have reported sex differences in olfactory sensitivity of zebrafish using the EOG technique. With the limited number of cells in this study, we found no obvious sex differences in the magnitudes or thresholds of gated currents. This is consistent with the belief that voltage-and Ca²⁺-gated currents do not contribute to the EOG, which reflects instead the summed receptor potentials of many ORNs (Getchell, 1986).

We have characterized five gated currents in zebrafish ORNs: a transient TTX-sensitive I_Na, a sustained I_Ca, I_DR, a Ca²⁺-dependent K⁺ current (I_K(Ca)) and I_A. The I_Na found in zebrafish ORNs appears to be generally similar to that found in other species. The current is activated first between –74 and –64 mV (Fig. 2A), and similar thresholds have been reported for ORNs from Xenopus laevis (Schild, 1989), rat (Rajendra et al. 1992; Murrow et al. 1995) and catfish (Miyamoto et al. 1992). Others have reported more positive thresholds (Trotier, 1986; Firestein and Werblin, 1987; McClintock and Ache, 1989; Suzuki, 1989; Trombley and Westbrook, 1991; Zufall et al. 1991; Lucero et al. 1992; Nevitt and Moody, 1992; Delgado and Labarca, 1993). The fact that, while in the gigaseal cell-attached configuration, we only rarely observed action potentials indicates that the resting membrane potential of zebrafish ORNs may be more negative than spike threshold, i.e. more negative than -74 to -64 mV. Another possibility is that the resting membrane potential is sufficiently positive that Na⁺ channels are inactivated at rest. It was not possible to measure the membrane potential directly because of problems inherent in the study of such small cells; the seal conductance contributes significantly to the resting conductance of the cell, leading to erroneous determinations of both membrane potential and input resistance (Barry and Lynch, 1991).

Recovery of I_Na from inactivation was very slow in zebrafish ORNs (Fig. 2C). Similar results have been reported for rat ORNs (Rajendra et al. 1992), and in ORNs from both species the rate of recovery from inactivation is voltage-dependent, being much faster at more negative potentials. Slow recovery has also been reported in salmon ORNs while, in contrast, recovery from inactivation is relatively rapid in ORNs from catfish (Miyamoto et al. 1992) and lobster (McClintock and Ache, 1989). Slow recovery from inactivation could profoundly affect firing properties by causing progressive spike amplitude decrement and spike frequency adaptation. Spike amplitude decrement during firing has been reported for ORNs from a large number of species using whole-cell, conventional intracellular, and extracellular recording techniques (Suzuki, 1977, 1989; Trotier and MacLeod, 1986; Firestein and Werblin, 1987; Firestein et al. 1990; Frings and Lindemann, 1991).

The extremely small I_Ca that we observed in zebrafish ORNs is activated at a threshold between –44 and –34 mV (Fig. 4) and appears to be largely sustained except at +16 mV (Fig. 3). By recording four steps to each test potential, averaging each of the four steps and smoothing the resulting data, we were able to detect I_Ca in zebrafish ORNs that usually measured less than –10 pA in peak amplitude. It is difficult to compare our results with those of others because of the common practice of increasing the extracellular Ca²⁺ concentration, or adding Ba²⁺ to the bath, to increase the amplitude of the currents observed. Since this practice usually results in an increase in the extracellular concentration of divalent cations, it would be expected to shift the voltage-dependence towards more positive potentials (Hille, 1992). In some ORNs that have been studied using normal divalent cation concentrations, thresholds for I_Ca activation were found to be around –20 mV (Firestein and Werblin, 1987; Suzuki, 1989; Delgado and Labarca, 1993), which is more positive than we found in zebrafish ORNs. However, in lobster and salmon ORNs (McClintock and Ache, 1989; Nevitt and Moody, 1992), I_Ca threshold was more negative, ranging from -40 to –30 mV. Because the threshold of I_Ca activation in zebrafish ORNs is intermediate between that described for low-threshold (T-type, thresholds between –65 and –50 mV) and high-threshold (L-and N-type, thresholds positive from -10 mV) Ca²⁺ currents (see Tsien et al. 1988, for a review), we cannot group the I_Ca that we describe into any of these categories without further study, such as testing for dihydropyridine-sensitivity. It may be that I_Ca in zebrafish ORNs is passed by several different channel types or, alternatively, that I_Ca does not fit into any of these categories; this is certainly true for some other Ca²⁺ currents that have been described (Tsien et al. 1988).

We describe three outward K⁺ currents: I_DR, I_K(Ca) and I_A (Figs 5–7). In general, these currents are found in ORNs from other species as well. Exceptions include ORNs from lobster (McClintock and Ache, 1989), neonatal rat (Trombley and Westbrook, 1991) and the toad Caudiverbera caudiverbera (Delgado and Labarca, 1993), which lack transient outward currents. We characterize the transient outward current that we describe as an A-current largely on the basis of its sensitivity to 4-aminopyridine and its inactivation rate; the average time constant at +40 mV is 76 ms, which is well within the range of A-currents described in other species (Rudy, 1988). It is likely that I_A in zebrafish ORNs plays roles similar to those described in other cells, such as delaying the onset to the first spike, lengthening the interspike interval (Hille, 1992) or assisting in spike repolarization (Sacchi and Belluzzi, 1993).

We failed to demonstrate the existence of a Ca²⁺-dependent Cl^- current (I_Cl(Ca)) in zebrafish ORNs. A Ca²⁺-dependent Cl^--current would have been evident in two sets of experiments. First, when I_Ca was isolated, I_Cl(Ca) would have led to the development of small outward currents at +16 mV. Outward currents were occasionally observed under these conditions but these currents could not be blocked by replacing the bath with Co²⁺ Ringer, so they were not Ca²⁺-dependent, and they were not blocked by SITS, so they were probably not carried by Cl^-. Second, I_Cl(Ca) should have caused the reversal potential of the Ca²⁺-dependent outward currents to be well positive of E_K since the estimated value of E_Cl is -1 mV. In one cell, the Ca²⁺-dependent tail currents reversed at -85 mV, which suggests that the tail currents were not carried exclusively by K⁺. Obviously Ca²⁺ is a potential current carrier in this case, in addition to Cl^-. In the other two cells examined, the Ca²⁺-dependent tail currents reversed quite close to E_K. Ca²⁺-dependent Cl^- currents have recently been described in ORNs from other species, where they participate in generating the receptor potential (Kleene, 1993; Kurahashi and Yau, 1993; Lowe and Gold, 1993a). It is likely that a Cl^- current plays a similar role in zebrafish ORNs because EOG recordings have shown that odor responses can be markedly reduced by application of the Cl^- channel blocker SITS (W. C. Michel, unpublished observations). Our failure to demonstrate the presence of a Ca²⁺-dependent Cl^- conductance could have resulted from the loss of cilia during the dissociation procedure if this conductance is found exclusively on the cilia. Approximately 10 % of the cells used in the present study had obvious cilia. Another possibility is that the cilia lack a voltage-gated Ca²⁺ current and so depolarizing voltage steps would not be expected to activate an exclusively ciliary I_Cl(Ca). The voltage-and Ca²⁺-activated currents that we describe appear well suited for encoding a graded receptor potential into a train of action potentials. The Ca²⁺ current is probably too small to contribute significantly to generating the action potential but, by gating I_K(Ca), probably contributes to spike repolarization and perhaps spike frequency adaptation. The A-current may assist in spike repolarization and probably lengthens the interspike interval to allow encoding of a broader range of receptor potential amplitudes than would be possible otherwise. The general absence of inward rectifying currents suggests that the resting membrane potential is set by the equilibrium potential of leak currents rather than resulting from a balance of leak and inward rectifying currents as is found in some other cells (Hille, 1992).

With the emergence of the zebrafish as an important model for investigations of the molecular basis of olfactory development, it becomes increasingly important to characterize the olfactory capabilities of this species functionally. Behavioral studies have shown that zebrafish respond to chemosensory stimuli; both amino acids and putative pheromones elicit behavioral responses (Bloom and Perlmutter, 1977; Van Den Hurk and Lambert, 1983; Van Den Hurk et al. 1987; Steele et al. 1990, 1991). Michel and Lubomudrov (1995) have recently shown that the zebrafish olfactory organ has a response specificity similar to that of most other fishes. The current investigation demonstrates that isolated zebrafish ORNs are amenable to electrophysiological characterization. High-resistance seals formed easily and, in 25 % of the cells studied, a seal resistance greater than 20 GΩ (measured as input resistance) was maintained after achieving the whole-cell configuration. The small size of zebrafish ORNs allows for excellent voltage control and for rapid perfusion of the cell with the pipette solution in the whole-cell configuration. Rapid perfusion should allow efficient targeting of biochemical probes to transduction sites in the cilia of ORNs. All of these factors suggest that zebrafish ORNs have particular advantages for the study of olfactory physiology.

We thank Dr Mary T. Lucero for a critical reading of an earlier version of this manuscript. This work was supported by NIDCD DC-01418.