An estimate of the resting membrane resistance of frog olfactory receptor neurones

Abstract

The ability of a frog olfactory receptor neurone (ORN) to respond to odorous molecules depends on its resting membrane properties, including membrane resistance and potential. Quantification of these properties is difficult because of a shunt conductance at the membrane–pipette seal that is in parallel with the true membrane conductance. In physiological salines, the sum of these two conductances averaged 235 pS. We used ionic substitution and channel blockers to reduce the membrane conductance as much as possible. This yielded a lower limit for the membrane conductance of 158 pS. The upper limit of resting membrane resistance, then, is 6 GΩ. The membrane is permeable to K⁺ and, to a lesser extent, other cations. No resting Cl⁻ conductance was detectable. Correcting measured zero-current potentials for distortion by the shunt suggests that the resting membrane potential is no more negative than −75 mV. The present results help to explain why frog ORNs are excitable at rest.

The ability of a cell to respond to a stimulus is dependent on its state and its sensitivity. To be responsive to a weak stimulus, a receptor neurone must maintain a high resting membrane resistance. This allows a larger depolarization for a given receptor current. However, it has proven difficult to measure the membrane resistance accurately. In olfactory receptor neurones (ORNs), reported values for the zero-current potential range from −90 to −30 mV (reviewed by Schild & Restrepo, 1998). During whole-cell electrical recording from an isolated neurone, the measured current reaches the pipette through two parallel conductance pathways (Fenwick et al. 1982; Fischmeister et al. 1986; Lynch & Barry, 1991). The first of these is the membrane conductance itself, which consists of any open membrane channels. The second (shunt) pathway allows ions to flow into the recording pipette through the tiny gap between the pipette and the membrane. What can be measured directly is the sum of these two conductances. It is quite possible, though, that only a tiny fraction of this input conductance is attributable to the membrane.

The two conductances can be distinguished if one assumes that they have different ionic selectivities. Typically it is assumed that the shunt pathway permits free diffusion of all ions in physiological recording solutions (Lynch & Barry, 1991). In rat olfactory receptor neurones (ORNs), the two pathways were distinguished by assuming that the resting membrane is only permeable to K⁺ (Lynch & Barry, 1991). In frog ORNs, though, evidence suggests an additional small resting conductance through cyclic-nucleotide-gated (CNG) channels (Pun & Kleene, 2003). These channels conduct Na⁺, K⁺ and Ca²⁺. A model to distinguish membrane and shunt conductances should account for this non-selective cationic conductance as well as any possible resting Cl⁻ conductance.

We have used an alternative approach to estimate the resting membrane conductance of frog ORNs. Ionic substitution and channel blockers were used to decrease both the shunt and membrane conductances. Effects of these reagents on the shunt could be determined directly. The remaining decreases in conductance were attributed to reduction of current through membrane channels. We estimate that the resting membrane conductance is at least 158 pS. The resting membrane potential, corrected for effects of the shunt, is no more negative than −75 mV. K⁺ is the most permeant ion, but there is also some permeability to other cations. No resting conductance to Cl⁻ was detectable.

Methods

Electrophysiology

Northern grass frogs (Rana pipiens) were decapitated and pithed as approved by the University of Cincinnati Institutional Animal Care and Use Committee. Physiological recordings were made from freshly dissociated cells prepared as described elsewhere (Kleene & Gesteland, 1991a). Cells were suspended in a Na-Hepes-buffered medium (standard bath solution), transferred into a chamber seated on the stage of an inverted microscope and used for experiments. Olfactory receptor neurones (ORNs) were identified by their motile cilia. The standard bath solution contained the following (mm): NaCl, 120; KCl, 3; CaCl₂, 0.1; Hepes, 5. The pH of the solution was adjusted to 7.2 with NaOH and the osmolarity to about 250 mosmol l⁻¹ with sucrose when necessary.

Electrodes were fabricated from thin-walled borosilicate glass capillary tubes (outer diameter 1.5 mm) on a two-stage electrode puller (BB-CH-PC, Mecanex). The pipettes were filled with various solutions; see Tables 1 and 2 for their compositions. The pH of the intracellular medium was adjusted to 7.2 with the appropriate hydroxide, and the osmolarity was adjusted to 250 mosmol l⁻¹ when necessary. When filled with recording solutions, pipettes had resistances of 6–10 MΩ.

Table 1.

Composition of cytoplasmic solutions (mm)

	KCl	K2SO4	CsCl	Cs2SO4	MeSO3H	CsOH	NaCl	TEACl	EGTA	Hepes	MgCl2	Glucose
Perforated-patch recording
Soln A (KCl)	64	28	—	—	—	—	12	—	0.5	20	1	5
Soln B (CsCl)	—	—	64	28	—	—	12	—	0.5	20	1	5
Whole-cell recording
Soln C (CsCl + TEA)	—	—	90	—	—	—	—	25	1.1	10	1	5
Soln D (CsMeSO3 + TEA)	—	—	—	—	120	∼120	10	5	1.1	5	1	—

CsOH was added in amounts sufficient to reach pH 7.2. MeSO₃H, methanesulphonic acid; TEACl, tetraethylammonium chloride; CsMeSO₃, caesium methanesulphonate.

Table 2.

Composition of external solutions (mm)

	NaCl	KCl	MeSO3H	Choline	TEACl	Hepes	CaCl2	MgCl2	Amiloride	TTX	CoCl2
Soln A (NaCl)	120	3	—	—	—	5	2	1	—	—	—
Soln B (NaCl, TEA, amiloride, TTX, Co2+)	120	3	—	—	5	5	2	1	1	0.001	4
Soln C (cholineMeSO3, amiloride, TTX, Co2+)	—	—	120	∼120	—	10	—	—	1	0.001	4

Choline was added in amounts sufficient to reach pH 7.2 (shown as ∼120 mm). Abbreviations used: MeSO₃H, methanesulphonic acid; TEACl, tetraethylammonium chloride; TTX, tetrodotoxin.

A 5 mV voltage step of duration 20 ms was repetitively passed through the pipette and the current amplitude monitored. When the pipette touched the surface of a cell, gentle suction was applied to the end of the pipette. As the seal resistance increased, as evidenced by a drop in the current-step amplitude, a pipette potential of −50 mV was imposed. This appeared to facilitate the formation of a high-resistance seal between the pipette and the membrane. The seal was considered sufficient when the holding current at −60 mV was less than 5 pA. For whole-cell recording, additional suction was applied to rupture the membrane. Intracellular recording was evidenced by the sudden appearance of capacitive current transients. The transients were then nulled by adjustment of the series resistance and access resistance controls of the amplifier (Axopatch 1A, Union City, CA, USA). Pipette potential was then set to −80 mV and voltage steps from −50 to +20 mV of duration 100 ms were applied. If depolarization resulted in an inward current followed by an outward current, the neurone was used for further studies. Similar procedures were used for perforated-patch whole-cell recording with 100–250 μg ml⁻¹ (final concentration) nystatin in the pipette solution. Following the formation of a high-resistance seal, the negative pressure was released and no further suction was applied. Recordings only proceeded when the access resistance dropped below 40 MΩ and inward currents followed by outward currents could be elicited by depolarization.

Voltage and duration of the ramps were controlled with the program pCLAMP (version 5.5.1, Axon Instruments) via an A/D D/A board and an IBM-compatible computer. Voltage ramps of duration 1 s from −100 to 0 or +20 mV were applied at 0.5 Hz. Current records were sampled at 1 kHz. No data were taken until it was confirmed that the neurone had voltage-activated currents. Thereafter, runs were taken every minute. For each run, four ramps were averaged. A minimum of three runs were taken in a control solution. Then a pipette of tip diameter 5−10 μm containing an external blocking solution was brought within 20 μm of the cell. The compositions of the various external solutions are shown in Table 2. Gentle pressure was applied to the pipette to cause an outflow of the blocking solution. This method of application is sufficient to alter the ionic environment surrounding the small ORN (see MacDermott & Westbrook, 1986). Another three runs were taken during the application of the blocking solution. Then the pipette containing the blocking solution was removed and the time course of recovery followed. The access resistance was monitored throughout the studies to ensure that the reduction in slope conductance did not result from a resealing of the membrane under the recording pipette. Between runs, depolarizing voltage steps were also applied occasionally during the recovery phase to ensure that excitability was still intact. Following recovery (as judged by an increase in slope conductance), the blocking solution was applied once more and the effects monitored.

Digitized recordings were imported into the graphics program Origin (version 6.0, Microcal, Northampton, MA, USA). The slope conductance for each run was obtained by fitting the points between −85 and −55 mV by linear regression. Each data point presented is the mean of two to four runs. A minimum of five different cells were measured for each condition. Results are presented as the mean ± s.e.m. Comparisons within individual cells (control versus blocking solution versus recovery) were performed with Student's t test, and comparisons between populations of cells were done with analysis of variance. A P value of < 0.05 was taken as statistically significant.

Corrections for the shunt conductance

The input conductance measured is the sum of two parallel conductances: the neuronal membrane conductance and a shunt conductance through the membrane-pipette seal (Fenwick et al. 1982; Fischmeister et al. 1986; Lynch & Barry, 1991). The shunt conductance is a function of the free solution conductances of the bath (external) and pipette (cytoplasmic) solutions. Neuronal input conductances were measured in a series of bath and pipette solutions (Table 3). At each step, just one of the two solutions was changed. In some cases, this was expected to lead to a substantial change in the shunt conductance. It was assumed that the solution present in the shunt was an equal mixture of the bath and pipette solutions. For each such mixture, the free solution conductance was measured with a conductance cell and an impedance bridge. Ratios of the estimated shunt conductances through successive solution changes are shown in Table 3.

Table 3.

Successive blocking of resting ORN conductance

	Cytopl. soln	External soln	G (pS)	▵G (pS)	n	▵Shunt
Perforated-patch recording
1 Internal K+	A	A	235 ± 12		12
2 Internal Cs+	B	A	182 ± 13	53 ± 17*	21	0.98
Whole-cell recording
3 Block outward K+	C	A	111 ± 10	71 ± 16*	17	0.98
4 Block cationic influx	C	B	77 ± 6	34 ± 12*	17	0.99
5 Block Cl− efflux	D	B	59 ± 7	18 ± 9	14	0.92
6 Block all	D	C	54 ± 4	5 ± 8	18	0.80

For each pair of internal and external solutions shown, input conductance G was measured. Solution compositions are defined in Tables 1 and 2. ▵G shows the decrease in input conductance compared to the value in the line immediately preceding.

^*Conductance is significantly different from that in the line immediately preceding (P < 0.05 by two-way analysis of variance). ▵Shunt shows the free solution conductance of an equal mixture of the given internal and external solutions relative to that measured in the solutions shown in the line immediately preceding. Each row represents a different population of neurones, except that rows 3 and 4 are from a single population.

The fractional change in shunt conductance was estimated by a second, independent method: measuring the conductances of an open pipette before and after a given solution change (data not shown). In physiological salines, the range of pipette conductances was 50–150 nS (corresponding to resistances between 7 and 20 MΩ). The open-pipette current–voltage relation was nearly linear, even in asymmetric solutions. The two estimation methods yielded similar results. The overall decrease in shunt conductance between the first and last pair of solutions tested (rows 1 and 6 of Table 3) was estimated to be 0.70 by using the conductance cell and 0.59 by measuring open-pipette conductances.

At the start of each experiment, the amplifier current was zeroed when the open pipette was immersed in the first bath solution. This compensated for the changes in half-cell potential that occurred at either Ag–AgCl electrode when most of the Cl⁻ was replaced by methanesulphonate⁻. A liquid junction potential of up to 8 mV at the tip of the open pipette was compensated initially but was absent after forming a seal against the cell (Barry & Lynch, 1991). This voltage error was not compensated. Since the current–voltage relations were nearly linear over the range of voltages used to measure the conductance, a small voltage error did not affect the slope conductance.

Results

Using patch-clamp recording, we measured the input conductance of resting isolated frog olfactory receptor neurones. Initially the input conductance of a sample population of neurones was measured by nystatin perforated-patch recording in physiological solutions (Horn & Marty, 1988; Pun & Kleene, 2003). This method largely preserves the normal cytoplasm. Small cations, and to a much lesser extent Cl⁻, still equilibrate with the pipette solution. In physiological solutions, the input conductance averaged 235 pS (Table 3, row 1; Fig. 1A). This input conductance is the sum of two parallel conductances, the neuronal membrane conductance and a shunt conductance through the membrane–pipette seal (Fenwick et al. 1982; Fischmeister et al. 1986; Lynch & Barry, 1991). The goal of the studies that follow was to estimate how much of this input conductance is attributable to the membrane. To this end, a series of ionic replacements and channel blockers were applied to successively block as much of the membrane conductance as possible.

Figure 1. Effects of ionic substitutions on current-voltage (I–V) relations in olfactory receptor neurones.

A, representative I–V relations from two different ORNs obtained under perforated-patch recording. The recording shown in black was from a cell recorded with K⁺ in the pipette solution (Table 1, cytoplasmic solution A). The grey trace was recorded from another cell with Cs⁺ in the pipette solution (Table 1, cytoplasmic solution B). A large outward current was blocked when K⁺ was replaced with Cs⁺ in the recording pipette solution, and the I–V relation became nearly linear. In other cells, a small inward current could be recorded at positive potentials (as in B). Slope conductances at negative potentials were 234 pS with K⁺ in the pipette and 182 pS with Cs⁺ in the pipette. A physiological external solution was used (Table 2, external solution A). B, ramp currents obtained from a neurone under whole-cell recording with Cs⁺ in the pipette solution (Table 1, cytoplasmic solution C). The recording shown in black was from a cell placed in a physiological external solution (Table 2, external solution A). The grey trace shows the current when an external solution containing TEA, amiloride, TTX and Co²⁺ (Table 2, external solution B) was applied to the same neurone. The inward current at depolarized potentials was completely blocked. Slope conductances measured between −80 and −60 mV were 107 pS in external solution A and 80 pS in external solution B. Extrapolations of these conductances are shown as straight lines. Note that the current axis is expanded in B compared to A.

As a first step, input conductance was measured in a second population of neurones with Cs⁺ in the cytoplasmic (pipette) solution in place of K⁺. Replacement of cytoplasmic K⁺ with Cs⁺ is expected to reduce any resting K⁺ efflux through the neuronal membrane. The perforated-patch recording method was still used, and the standard external (bath) solution was retained. The mean conductance at negative potentials was 53 pS less than in the first population of neurones (Table 3, row 2). In physiological solutions, a large outward current was present at depolarized potentials. This was virtually absent in the neurones studied with Cs⁺ in the pipette solution (Fig. 1A). Replacement of cytoplasmic K⁺ with Cs⁺ does not affect the free solution conductance (▵shunt = 0.98, Table 3, row 2). Thus the 53 pS decrease in input conductance can be attributed to the neuronal membrane rather than the shunt.

To further block membrane currents, it was necessary to introduce large, impermeant ions into the cytoplasm. For this purpose, the patch was ruptured to give the whole-cell recording configuration in a third population of neurones. Tetraethylammonium⁺ (TEA) was added to the cytoplasmic solution, which already contained Cs⁺ instead of K⁺. Adding TEA was expected to eliminate any remaining K⁺ efflux but to have little effect on the shunt conductance (▵shunt = 0.98, Table 3, row 3). The input conductance was reduced by 71 pS. With the large outward K⁺ current blocked, it was possible to see a smaller inward current at depolarized potentials (Fig. 1B). With Cs⁺ in the pipette solution, this inward current was always apparent in the whole-cell configuration but only sometimes in the perforated-patch configuration.

With TEA and Cs⁺ still in the pipette, TEA, amiloride, tetrodotoxin (TTX) and Co²⁺ were added to the external bath in a fourth population of neurones. TEA was predicted to block any remaining K⁺ current. Co²⁺ was expected to block any influx of Ca²⁺, while amiloride and TTX would eliminate Na⁺ influx. Amiloride also blocks the olfactory cyclic nucleotide-gated (CNG) channels (Frings & Lindemann, 1988; Frings et al. 1992; Kleene, 1994). The mixture, then, should block most cationic influx. In this bath, the input conductance decreased by 34 pS (Table 3, row 4). Addition of these blockers does not affect the free solution conductance (▵shunt = 0.99, Table 3, row 4), so the shunt conductance should not change. Thus the 34 pS decrease in input conductance can be attributed to the neuronal membrane. Addition of external TEA, amiloride, TTX and Co²⁺ eliminated the inward current seen at depolarized potentials (Fig. 1B).

The solutions just described were expected to block all neuronal cationic fluxes. However, it was still possible that a membrane Cl⁻ conductance remained. To test this, the solutions just described were modified further. First, most cytoplasmic Cl⁻ was replaced with methanesulphonate⁻ (MeSO₃⁻). In this population of neurones, the mean input conductance was smaller by just 18 pS (Table 3, row 5). This decrease was not statistically significant. As a final attempt to block any remaining membrane currents, the external bath was also modified. Cl⁻ was replaced with MeSO₃⁻ and Na⁺ was replaced with choline⁺. The 5 pS decrease in input conductance was not statistically significant (Table 3, row 6).

Other evidence confirmed that the resting neurone has little conductance to Cl⁻. Two populations of neurones were compared, each bathed in the physiological external solution (Table 2, external solution A). When the pipette solution contained CsCl and TEA (Table 1, cytoplasmic solution C), the resting conductance averaged 111 ± 10 pS (n = 17). In another population of neurones, CsMeSO₃ and TEA were used to make the pipette solution (Table 1, cytoplasmic solution D). In this case, the resting conductance averaged 112 ± 8 pS (n = 34). Resting conductances measured with the Cl⁻-containing and MeSO₃⁻-containing pipette solutions were not significantly different, suggesting that there is little resting Cl⁻ conductance.

Estimates of the neuronal and shunt conductances

In physiological solutions, the input conductance averaged 235 ± 12 pS. It is possible to estimate the separate contributions of the membrane and shunt conductances by assuming that the ionic substitutions and channel blockers ultimately eliminated all membrane conductance. The remaining conductance of 54 pS (Table 3, row 6) is then attributed solely to the shunt. However, the shunt conductance would be larger in physiological solutions, which have higher free solution conductances than the solutions made with ionic substitutions. It was found that the free solution conductance is reduced by a factor of 0.70 in going from the physiological solutions (cytoplasmic solution A and external solution A) to the substituted solutions (cytoplasmic solution D and external solution C). In physiological solutions, then, the shunt would be 77 ± 6 pS (i.e. 54 pS/0.70). The remainder of the input conductance, 158 ± 13 pS, is an estimate of the true membrane conductance in physiological solutions.

Discussion

The purpose of this study was to estimate the resting membrane conductance of an isolated frog olfactory receptor neurone (ORN), i.e. the conductance in the absence of odorous stimulation. An input conductance can be measured by recording from an ORN in the perforated-patch or whole-cell recording configuration. In this situation, current measured at the pipette arises from two parallel conductance pathways (Fenwick et al. 1982; Fischmeister et al. 1986; Lynch & Barry, 1991). The first pathway consists of current passing through the neuronal membrane. The second pathway is a shunt passing through the membrane-pipette seal. The two conductances can be discriminated by assuming that they have different ionic permeabilities. We assumed that the shunt allows free diffusion and has no ionic selectivity. In this case its conductance depends only on the free solution conductances of the salines used. We also assumed that the membrane conductance arises from ion-selective channels. Current through these channels can be reduced by substituting less permeant ions or by adding channel blockers.

In physiological solutions, the measured input conductance of an ORN averaged 235 pS. To reduce current through membrane channels as much as possible, we subsequently modified both the external and internal solutions. Na⁺ and K⁺ were replaced with choline⁺ and Cs⁺, respectively; Cl⁻ was replaced with methanesulphonate⁻; and channel blockers TEA, amiloride, TTX and Co²⁺ were added. An operational assumption is that these solutions block all membrane current, and so the remaining conductance (54 pS) is a measure of the shunt conductance. In physiological solutions, which contain ions of higher mobility, the shunt conductance would be 77 pS. The portion of the input conductance in physiological solutions not attributable to the shunt (158 pS) is taken to be the resting membrane conductance. In fact, it is possible that the substituted solutions did not block all of the membrane current. If so, then a greater proportion of the input conductance would be attributable to the membrane, and the true membrane conductance would be greater than 158 pS. Thus 158 pS is a lower limit of the membrane conductance. The upper limit is the measured input conductance, 235 pS. The corresponding range of membrane resistances is 4–6 GΩ. This range is similar to that reported for other small cells (Fenwick et al. 1982).

Our results allow some inferences about the ionic selectivity of the resting neuronal membrane. Of the total conductance (158 pS), little or none was attributable to Cl⁻ channels. Eliminating Cl⁻ from either side of the membrane caused no significant change in input conductance. Replacing cytoplasmic Cl⁻ required whole-cell recording, and the resulting exchange of cytoplasm and pipette solution could have affected the resting conductance. A Ca²⁺-activated Cl⁻ current plays a prominent role in olfactory transduction (Kurahashi & Yau, 1994). Since this current is activated only when [Ca²⁺]_i exceeds ∼1 μm (Kleene & Gesteland, 1991b), it is probably not active at rest. To facilitate Cl⁻ homeostasis, the resting neurone may have one or more Cl⁻ transport processes. However, these were not detectable as a resting Cl⁻ conductance in the present study.

Distinguishing between K⁺ and Na⁺ conductances is more difficult. It is likely that most of the neuronal resting conductance is due to K⁺-selective channels. The largest decreases in conductance came when internal K⁺ was replaced with Cs⁺ and/or TEA. This should block K⁺-selective channels but is not expected to be completely selective for them. Our estimate of the neuronal resting potential (see below) is within 20 mV of the equilibrium potential for K⁺, which also suggests that K⁺ has the greatest conductance at rest. However, perforated-patch recordings suggest that frog ORNs also have a small resting conductance through the cyclic nucleotide-gated (CNG) channels, which conduct Na⁺, K⁺ and Ca²⁺ (Pun & Kleene, 2003). Isolated frog olfactory cilia in physiological solutions have conductances similar to those reported here: no detectable permeability to Cl⁻ and a cationic permeability that favours K⁺ over Na⁺ by a ratio of 2: 1 (Kleene, 1992).

Lynch & Barry (1991) used a different method to estimate the membrane resistance in ORNs while correcting for shunt conductance. They found that the mean resistance of rat ORNs averaged 26 GΩ (38 pS expressed as a conductance). This resistance exceeds our estimate, and there are three likely explanations. First, as judged by cellular capacitance, ORNs of the rat have about half the surface area of those of the frog (see Table 1 of Schild & Restrepo, 1998). A twofold reduction in surface should lead to a doubling of the resistance. Second, in the rat model it was assumed that the resting membrane potential equals the equilibrium potential for K⁺. Assuming that the resting potential is more positive would lead to a lower estimate of membrane resistance (by equations 2 and 3 of Lynch & Barry, 1991). Finally, the extracellular solutions bathing the rat ORNs contained choline⁺ instead of Na⁺. This was done on the assumption that the membrane of the resting rat ORN is permeable only to K⁺. Whether this is true in the rat has not been tested. However, in ORNs of the frog, a resting conductance through the CNG channels (Pun & Kleene, 2003) gives the membrane some permeability to Na⁺ and Ca²⁺.

We extended the model of Lynch & Barry to include K⁺ and Na⁺ conductances as well as the non-membrane shunt. Although this extended method is sound in principle, the results of the numerical analyses of our data were not significant at even the 90% confidence level. We tried other variations of this approach. Complete current–voltage (I–V) relations were measured with high external Na⁺ and then with high external K⁺. The difference between these two relations should have very little dependence on the non-membrane shunt conductance. We tried to fit this difference I–V to a sum of I–V relations predicted by the Goldman-Hodgkin-Katz current equation. Again, though, the fit could not resolve the individual K⁺ and Na⁺ conductances with confidence (data not shown). In fact, the measured I–V relation showed stronger outward rectification than could be accounted for by a pure K⁺ conductance modelled with the Goldman-Hodgkin-Katz current equation.

We have reported that the zero-current potential while recording from unstimulated frog ORNs averages −51.7 ± 2.3 mV (Pun & Gesteland, 1991). However, the true resting membrane potential must be more negative. When the amplifier measures zero current,

where V_p is the pipette potential; E_s is the junction potential at the shunt; E_m is the membrane potential; G_s is the conductance of the shunt; G_m is the membrane conductance (Barry & Lynch, 1991; Qu et al. 2000). The membrane potential E_m can be estimated after substituting the measured values for V_p (−51.7 ± 2.3 mV), E_s (−3 mV in physiological solutions), G_s (77 ± 6 pS), and G_m (158 ± 13 pS). E_m is estimated to be −75 ± 6 mV. Due to the influence of the resting non-selective cationic conductance, the resting potential is more positive than the equilibrium potential for K⁺ in our physiological solutions (−95 mV). As discussed above, our method may underestimate the membrane conductance. On assuming that a greater portion of the input conductance is attributable to the membrane, the estimated membrane potential becomes less negative, approaching the measured zero-current potential of −51.7 mV. It must be acknowledged that the estimated resting potential depends on the choice of physiological solutions. Should the K⁺ gradient between the cytoplasmic and interstitial fluids be unusual, our estimate would be in error. In addition, we have not tested whether a pump might contribute to the resting membrane potential. In vomeronasal chemoreceptor neurones of the frog, for example, the Na⁺, K⁺-ATPase allows the membrane potential to be more negative than the equilibrium potential for K⁺ (Trotier & Døving, 1996).

An interesting open question is the fraction of voltage-gated Na⁺ channels that are inactivated in a resting ORN. In the frog, the mean voltage for half-inactivation was initially reported to be −82 mV (Pun & Gesteland, 1991), which is near our estimate of the resting membrane potential. Having half of the Na⁺ channels inactivated at rest would reduce the neurone's ability to generate action potentials at a high frequency. However, the half-inactivation potential was shifted to −62 mV when GTP was included in the cytoplasmic solution or when perforated-patch recording was done (Pun et al. 1994). In this situation, 84% of the Na⁺ channels would be available for activation at −75 mV. In rat, the half-inactivation potential is −110 to −100 mV regardless of whether GTP is present (Rajendra et al. 1992; Qu et al. 2000). In principle, the currents measured when half-inactivation potentials are determined are also in error due to the shunt conductance. However, near half-inactivation, the membrane conductance is much greater than the shunt conductance, so this error is probably negligible.

Based on an uncorrected estimate of the resting membrane potential, we previously suggested that most of the voltage-gated Na⁺ channels might be inactivated at rest (Pun et al. 1994). However, our present results suggest that this is not the case. Our study does confirm the considerable effect of the shunt resistance on the measured zero-current potential. The more positive zero-current potentials reported (e.g. −30 mV) are difficult to account for with the Goldman-Hodgkin-Katz equations, given that they were measured under physiological conditions. More likely, they reflect the effect of the shunt on a cell with high input resistance. This effect should be especially pronounced for a small cell such as the frog ORN, which has a cell capacitance of about 6 pF.