Ion permeation and conduction in a human recombinant 5-HT₃ receptor subunit (h5-HT_3A)

Abstract

1. A human recombinant homo-oligomeric 5-HT₃ receptor (h5-HT_3A) expressed in a human embryonic kidney cell line (HEK 293) was characterized using the whole-cell recording configuration of the patch clamp technique.

2. 5-HT evoked transient inward currents (EC₅₀ = 3.4 μm; Hill coefficient = 1.8) that were blocked by the 5-HT₃ receptor antagonist ondansetron (IC₅₀ = 103 pm) and by the non-selective agents metoclopramide (IC₅₀ = 69 nm), cocaine (IC₅₀ = 459 nm) and (+)-tubocurarine (IC₅₀ = 2.8 μm).

3. 5-HT-induced currents rectified inwardly and reversed in sign (E_5-HT) at a potential of −2.2 mV. N-Methyl-d-glucamine was finitely permeant. Permeability ratios P_Na/P_Cs and P_NMDG/P_Cs were 0.90 and 0.083, respectively.

4. Permeability towards divalent cations was assessed from measurements of E_5-HT in media where Ca²⁺ and Mg²⁺ replaced Na⁺. P_Ca/P_Cs and P_Mg/P_Cs were calculated to be 1.00 and 0.61, respectively.

5. Single channel chord conductance (γ) estimated from fluctuation analysis of macroscopic currents increased with membrane hyperpolarization from 243 fS at −40 mV to 742 fS at −100 mV.

6. Reducing [Ca²⁺]_o from 2 to 0.1 mm caused an increase in the whole-cell current evoked by 5-HT. A concomitant reduction in [Mg²⁺]_o produced further potentiation. Fluctuation analysis indicates that a voltage-independent augmentation of γ contributes to this phenomenon.

7. The data indicate that homo-oligomeric receptors composed of h5-HT_3A subunits form inwardly rectifying cation-selective ion channels of low conductance that are permeable to Ca²⁺ and Mg²⁺.

The 5-hydroxytryptamine type 3 (5-HT₃) receptor is a transmitter-gated ion channel belonging to the Cys-loop superfamily of pentameric proteins that includes the nicotinic acetylcholine, γ-aminobutyric acid type A and glycine receptors (Maricq, Peterson, Brake, Myers & Julius, 1991). Neuronally located 5-HT₃ receptors mediate a rapidly desensitizing depolarizing response contingent upon the gating of a cation channel that conducts monovalent metals in a non-selective manner (Derkach, Surprenant & North, 1989; Lambert, Peters, Hales & Dempster, 1989; Yang, 1990). Unusually, the elementary conductance of the channel may vary by approximately 60-fold across preparations (i.e. 0.3-19 pS; Lambert et al. 1989; Yang, 1990; Peters, Malone & Lambert, 1993; Hussy, Lukas & Jones, 1994) and indirect evidence suggests that receptors with distinct conductances can exist within the same cell type (Derkach et al. 1989; Yang, Mathie & Hille, 1992; Hussy et al. 1994).

Diversity in channel properties appears to extend to the conduction of Ca²⁺, which from measurements of relative permeability has been claimed to be impermeant (Yakel, Shao & Jackson, 1990; Gilon & Yakel, 1995; Glitsch, Wischmeyer & Karschin, 1996), or, to an approximation, equipermeant with Na⁺ (Yang, 1990; Yang et al. 1992). Furthermore, although extracellular Ca²⁺ and Mg²⁺ have consistently been found to block macroscopic currents mediated by 5-HT₃ receptor activation, the effect can be either voltage independent (Peters, Hales & Lambert 1988; Yang, 1990; Yang et al. 1992; Peters et al. 1993; Gill, Peters & Lambert, 1995; Glitsch et al. 1996), or markedly enhanced by membrane hyperpolarization (Maricq et al. 1991; Eiselé, Bertrand, Galzi, Devillers-Thiéry, Changeux & Bertrand, 1993; Kawa, 1994).

The basis of the heterogeneous properties of the 5-HT₃ receptor channel is currently obscure. A solitary 5-HT₃ receptor subunit (5-HT_3A) has been isolated from mouse (m5-HT_3A; Maricq et al. 1991; Hope, Downie, Sutherland, Lambert, Peters & Burchell, 1993), rat (r5-HT_3A; Isenberg et al. 1993; Miyake, Mochizuki, Takemoto & Akuzawa, 1995) and human (h5-HT_3A; Belelli, Balcarek, Hope, Peters, Lambert & Blackburn, 1995; Miyake et al. 1995) tissues. Although two splice variants of m5-HT_3A have been identified (Maricq et al. 1991; Hope et al. 1993), their pharmacological and biophysical properties are essentially identical (Hussy et al. 1994; Gill et al. 1995; Peters, Hope, Sutherland & Lambert, 1997). In common with the related nicotinic α7 and α8 subunits, the 5-HT_3A subunit functions efficiently as a homo-oligomeric complex. The homo-pentameric nature of the 5-HT₃ receptor may facilitate attempts to identify elements of the channel domain influencing ionic conduction and selectivity, as in studies performed on the nicotinic α7 subunit (see review by Galzi & Changeux, 1995).

The present study characterizes the 5-HT_3A subunit cloned from the human amygdala and stably expressed in human embryonic kidney (HEK 293) cells (Belelli et al. 1995; Hope, Peters, Brown, Lambert & Blackburn, 1996). We focus upon ion selectivity and single channel conductance and demonstrate that the latter is modulated by extracellular divalent cations. The results represent the first description of such processes for the h5-HT_3A receptor and additionally provide information that might form a useful comparison with the data obtained from structurally homologous homo-oligomeric cation-selective ion channels. A preliminary account of a part of this work has appeared in abstract form (Brown, Hope, Peters & Lambert, 1996)

METHODS

Cell maintenance

The isolation of a cDNA encoding h5-HT_3A and the subsequent stable expression of the cDNA in a mammalian cell line (HEK 293 h5-HT_3A cells) have been reported in detail elsewhere (Belelli et al. 1995; Hope et al. 1996). Cells were maintained in minimal essential medium (MEM), supplemented with 10% (v/v) heat-inactivated fetal bovine serum, penicillin (1 × 10³ i.u. ml⁻¹), streptomycin (100 μg ml⁻¹) and geneticin (1 mg ml⁻¹) in 25 cm² tissue culture flasks held at 37°C in an atmosphere of 95% air- 5% CO₂ and 100% relative humidity. Cells were passaged by aspiration of the growth medium followed by two rinses with Ca²⁺- and Mg²⁺-free Hanks’ balanced salt solution. Detachment of the cells from the substratum was achieved by incubation with 1 ml of Hanks’ solution containing trypsin (500 μg ml⁻¹) and EDTA (200 μg ml⁻¹) in combination with mechanical agitation and trituration. Enzyme action was terminated by the addition of 10 ml of growth medium. For electrophysiological experiments, approximately 10 000 cells were seeded into 35 mm diameter ‘Nunclon’ Petri dishes and maintained in 2 ml of growth medium which was refreshed every 3 days. Cells were used in electrophysiological experiments 1–7 days after replating.

Electrical recordings

Agonist-evoked currents were recorded from whole cells under voltage clamp using a List Electronics EPC7 amplifier and converter headstage utilizing standard techniques. Cells were selected on the basis of small size (< 30 μm) and lack of visible attachment to other cells in order to facilitate spatial control of voltage. Unless stated otherwise, currents were recorded at a holding potential of −60 mV. The series resistance, which was generally less than 10 MΩ, was compensated by 50% or greater in all experiments. Currents were digitized (sampling rate, 50–1000 Hz as appropriate to the application) using a Digidata 1200 interface (Axon Instruments) and recorded on-line using pCLAMP 6.02 (Axon Instruments) or the Strathclyde Electrophysiology Software package (Dempster, 1993), or low-pass filtered (500 Hz, Bessel characteristic) and stored on magnetic tape using a Racal Store 4DS FM tape recorder for subsequent analysis. Current recordings illustrated were filtered at 100 Hz: voltage ramps were filtered at 50 Hz. Hard copy was recorded using a Gould TA240 chart recorder.

Cells were continually superfused (4 ml min⁻¹) with an extracellular solution (E1) containing (mm): NaCl, 140; KCl, 2.8; MgCl₂, 2.0; CaCl₂, 1.0; Hepes, 10; adjusted to pH 7.2 with 1 m NaOH. Patch electrodes were pulled on a Narishigi PB7 two stage electrode puller using Corning type 7052 glass (Garner Glass Co., Claremont, CA, USA). When filled with a standard intracellular solution (I1), comprising (mm): CsCl, 140; MgCl₂, 2.0; CaCl₂, 0.1; EGTA, 1.1; Hepes, 10; adjusted to pH 7.2 with 1 m CsOH, pipette resistances were in the range 3–5 MΩ. The intracellular free calcium concentration ([Ca²⁺]_i) was estimated to be to be 10 nm (Fenwick, Marty & Neher, 1982). For ion selectivity experiments, MgCl₂ was omitted from the internal solution, and the extracellular solution (E2) was simplified by replacing KCl with NaCl, and reducing the concentrations of CaCl₂ and MgCl₂ to 0.1 mm each, in order to minimize their potential influence upon measurements of the reversal potential of 5-HT-evoked currents (E_5-HT; see below). The permeation of other monovalent cations, was assessed by either partially or wholly substituting the test ion X, as XCl, for NaCl in solution E2. Permeability towards Ca²⁺ was evaluated with extracellular solutions with an elevated Ca²⁺ concentration ([Ca²⁺]_o), which were prepared as 100 mm CaCl₂ buffered with 5 mm l-histidine (pH 7.2), or 20 mm CaCl₂ and 100 mm N-methyl-d-glucamine, also buffered with l-histidine. Permeability towards Mg²⁺ was examined with extracellular Mg²⁺ elevated to 100 mm buffered with 5 mm l-histidine (pH 7.2). The potential contribution of chloride ions to the 5-HT-induced current was examined using a modified E2 solution, in which NaCl was totally replaced by sodium isethionate. To minimize changes in reference electrode potential during the superfusion of media with altered ionic composition, a bridge containing 3 m KCl in agar (4 % w/v) was employed. Liquid junction potentials arising at the tip of the patch pipette were measured as described by Fenwick et al. (1982) and potential measurements corrected post hoc.

E_5-HT was determined by two methods. In the first, the membrane potential was clamped (V_hold) at desired values and the current responses to three consecutive pressure applications of 10 μm 5-HT were digitized and computer averaged. The steady-state current-voltage (I-V) relationship and interpolated value of E_5-HT were derived from the plots of average peak current amplitude against V_hold. In the second method, a voltage ramp, sweeping from −150 to +50 mV at a rate of 0.25 mV ms⁻¹ generated by the Clampex program of the pCLAMP suite, was imposed at the peak of the current response induced by bath-applied 5-HT (1 μm). Over the time necessary to complete the ramp, the agonist-induced current at a constant V_hold did not change appreciably. Subtraction of the leakage current recorded in the absence of 5-HT from the current recorded in the presence of the agonist yielded the current specifically due to the 5-HT-evoked conductance increase. In control experiments, currents resulting from either depolarizing or hyperpolarizing rectangular step changes in V_hold were not associated with discernible relaxations following the settling time of the clamp (approximately 2 ms). It is thus likely that the relatively slowly changing ramp generates an I-V relationship that is essentially steady state. As anticipated from this conclusion, I-V relationships determined for an individual cell by either method, were, upon scaling, adequately superimposed (see Results).

Agonists were applied to cells in one of three ways. For routine pharmacological characterization of the response, repetitive applications of 5-HT (10 μm) were made by pressure ejection (1.4 × 10⁵ Pa, 10–50 ms, 0.033-0.020 Hz) from a second patch pipette placed within 100 μm of the cell. In order to construct the concentration-response relationship for 5-HT, concentrations of the agonist in the range 0.6-30 μm were applied to single cells using a computer-controlled solution exchanger (DAD, Adams and List Associates Ltd, Westbury, NY, USA). Solutions were delivered from a 100 μm diameter outlet tube which was placed about 200 μm from the cell. The solution exchange time was approximately 100 ms. For experiments involving fluctuation analysis (see below) of 5-HT-induced whole-cell currents, the agonist was applied via the inlet introducing perfusate to the bath. The position of the inlet, relative to the cell, was adjusted to yield a slowly rising and decaying agonist-induced response suitable for stationary analysis of membrane current fluctuations. Antagonist compounds were applied via the superfusate. All experiments were conducted at ambient temperature (17–23°C).

Analysis of results

Agonist concentration-response curves were iteratively fitted (Microsoft Excel) with the logistic equation:

(1)

where I is the peak inward current evoked by agonist at concentration A; I_max is the maximal inward current evoked by a saturating concentration of agonist; EC₅₀ is the concentration of agonist inducing a half-maximal current response and n is the Hill coefficient. An equation of the same form was used to analyse the concentration dependency of antagonist-induced blockade of the 5-HT-evoked response. The parameters derived were the concentration of antagonist producing a 50% block of the control response to 5-HT (IC₅₀) and the associated interaction coefficient (n).

Permeability ratios (relative to Cs⁺) were calculated from measurements of E_5-HT in media of varied ionic composition utilizing calculated ion activities. To analyse the data where only a single monovalent cation species was present extracellularly ([X]_o), the Goldman-Hodgkin-Katz (voltage) equation (Hille, 1992) was employed:

(2)

where R, T and F have their usual meaning, [X]_o and [Cs⁺]_i are the calculated activities of extracellular monovalent and internal Cs⁺ ions, and P_X/P_Cs is the permeability of X relative to Cs⁺. This equation ignores the small error anticipated due to the presence of low concentrations of permeant divalent ions (e.g. Ca²⁺) within the extra- or intracellular media. For binary mixtures of extracellular monovalent cations, terms for P_Y/P_Cs and [Y]_o were included in the above equation. Permeability ratios for Ca²⁺ were calculated from modifications to the Goldman-Hodgkin-Katz (voltage) equation introduced by Lewis (1979), namely:

(3)

where [Ca²⁺]_o and [Ca²⁺]_i are the external and internal activities of Ca²⁺ and P′_Ca/P_Cs is a modified term relating the permeability of Ca²⁺ to Cs⁺. The ion activity coefficient of Ca²⁺ was taken as the square of the calculated mean ion activity of the calcium salt. Under the present recording conditions, eqn (3) simplifies to:

(4)

which can be written as:

(5)

In experiments where Ca²⁺ was the sole extracellular cation the term for P_X/P_Cs was omitted in the calculations. The relative permeability of Mg²⁺ was calculated in an identical manner.

Fluctuation analysis

Fluctuations in slowly activating whole-cell currents evoked by 1 μm 5-HT were used to estimate the single channel conductance of h5-HT_3A. Signals were recorded on-line using the programme SPAN 3.0 (Dempster, 1993). The signal was divided into a DC-coupled signal low-pass filtered at a cut-off frequency of 1 kHz and an amplified (× 10) AC-coupled signal with a 1–1000 Hz bandpass frequency (Butterworth characteristic). Records of mean DC andAC fluctuations both before (for baseline noise calculations) and in the presence of the agonist were digitized at a rate of 2 kHz using a Data translation DT2801A laboratory interface and stored on a computer. The digitized record was composed of blocks with a duration of 1/f_res s, where f_res (Hz) was the desired resolution of the recording (usually 1 Hz). The variance method was used to estimate the current, i, flowing through a single channel using the equation (Dempster, 1993):

(6)

where σ² is the variance, I_m is the mean current; I_j current sample j and N is the number of A/D samples in the block. The relationship between σ² and I_m is parabolic rising from zero to a maximum at P = 0.5 and falling to zero again as P approaches 1:

(7)

where n is the number of channels within the cell. When P is small, (< 0.1) eqn (7) simplifies to the linear function:

(8)

Plots of σ²vs.I_m were obtained from blocks recorded during the entire exposure to microperfused 5-HT (1 μm). Single channel current was estimated mostly from a linear function (eqn (8)) but occasionally a parabolic function (eqn (7)) resulted in a better fit. Background variance, attributable to sources other than the 5-HT-induced conductance increase, was removed by subtracting the mean value of sixteen to thirty-two blocks recorded prior to the application of the agonist. Single channel chord conductances (γ) were calculated as:

(9)

All quantitative data are reported as the arithmetic mean ± the standard error of the mean.

Drugs and reagents

Drugs were obtained from the following sources: cocaine hydrochloride, 5-hydroxytryptamine creatine sulphate complex, (+)-tubocurarine chloride, 5-methoxytryptamine hydrochloride, metoclopramide hydrochloride (Sigma); methysergide hydrogen maleate (Sandoz (Novartis), Basel, Switzerland); ketanserin tartrate (Janssen, Beerse, Belgium); ondansetron hydrochloride, m-chlorophenylbiguanide dihydrochloride (Glaxo); 2-methyl-5-hydroxytryptamine maleate, 1-phenylbiguanide (Research Biochemicals, Natick, MA, USA). Stock solutions (10 mm) of all the drugs were prepared on the day of the experiment in twice distilled de-ionized water and diluted with the appropriate recording solution. All cell culture media were purchased from Gibco. Analytical grade reagents were obtained from Sigma or BDH (Poole, UK).

RESULTS

Pharmacological characterization of h5-HT_3A

Although a pharmacological characterization of the h5-HT_3A receptor was not a prime objective of the present study, several key ligands which have revealed heterogeneity in the properties of 5-HT₃ receptors (Peters et al. 1997) were examined to confirm that the operational properties of the receptor expressed in HEK 293 cells and Xenopus laevis oocytes (Belelli et al. 1995; Miyake et al. 1995) are similar.

At a holding potential of −60 mV, the majority (> 95%) of cells studied (n = 270) responded to pressure-applied 5-HT (10 μm) with a transient inward current response. It has previously been demonstrated that untransfected HEK 293 cells do not respond to 5-HT (Hope et al. 1996). Rapidly superfused 5-HT (0.3-30 μm) induced concentration-dependent currents that displayed desensitization in the continued presence of the agonist (Fig. 1A). The EC₅₀ and Hill coefficient calculated from the mean data depicted in Fig. 1C and fitted with eqn (1) were 3.4 ± 0.2 μm and 1.8 ± 0.3 μm, respectively (n = 9). The apparent affinity and co-operativity of action of 5-HT, which may be inferred from the macroscopic concentration-effect relationship, agree with results obtained with h5-HT_3A expressed in Xenopus laevis oocytes (Belelli et al. 1995; Miyake et al. 1995). Responses to 5-HT were mimicked by the selective 5-HT₃ receptor agonists 2-methyl-5-HT (10 μm; n = 5), 1-phenylbiguanide (80 μm; n = 5) and m-chlorophenylbiguanide (1 μm; n = 5; Fig. 1B). 5-Methoxytryptamine (100 μm), which is an agonist at many G-protein-coupled 5-HT₃ receptor subtypes but lacks activity at 5-HT₃ receptors (Hoyer et al. 1994), neither activated a response in cells responsive to 5-HT (n = 3), nor blocked the 5-HT-evoked current (n = 2; data not shown).

Figure 1. 5-HT and selective 5-HT₃ receptor agonists evoke inward current responses in HEK 293 cells stably expressing h5-HT_3A receptors.

A, traces from one cell depicting the concentration dependence of the peak current response evoked by rapidly superfused 5-HT (0.6-30 μm). Note that the rate of activation and desensitization of the current response increases with agonist concentration. B, inward current responses from one cell comparing the agonist activity of rapidly superfused 5-HT (3 μm) and the 5-HT₃ receptor-selective compounds 2-methyl-5-HT (10 μm), 1-phenylbiguanide (PBG; 80 μm) and m-chlorophenylbiguanide (mCPBG; 1 μm). Currents in A and B were recorded at a holding potential of −60 mV. Horizontal bars above the current traces indicate the period of agonist application. C, graph depicting the concentration dependency of the inward current response evoked by 5-HT (0.3-30 μm). To combine the data from nine different cells, the peak current amplitude evoked by a saturating concentration of 5-HT was assigned a value of 100% and the currents elicited by all other concentrations of agonist are expressed relative to that value. The curve fitted according to eqn (1) yielded an EC₅₀ value of 3.4 ± 0.2 μm and a Hill coefficient of 1.8 ± 0.3 for 5-HT. Vertical bars, where they exceed the size of the symbol, indicate the standard error of the mean.

Responses to pressure-applied 5-HT were suppressed in a concentration-dependent manner by the selective 5-HT₃ receptor antagonist ondansetron (10 pm-1 nm; Fig. 2A). Due to the slow reversal of blockade by this drug, it was not feasible to restore the control response between the application of several concentrations of the antagonist. For this reason, the concentration inhibition curve illustrated in Fig. 2E was obtained with cumulative additions of ondansetron and yielded an IC₅₀ of 103 ± 11 pm (n = 6). Less potent antagonism was observed for metoclopramide (IC₅₀ = 69 ± 5 nm; n = 5), cocaine (IC₅₀ = 459 ± 9 nm: n = 5) and (+)-tubocurarine (IC₅₀ = 2.8 ± 0.1 μm: n = 6; Fig. 2B-E) which have previously been documented to block 5-HT₃ receptor-mediated responses (Peters et al. 1997). Antagonism by those agents was readily reversible, and the full recovery was demonstrated between successive applications of several concentrations of each antagonist. Ketanserin (n = 3) and methysergide (n = 3), each at a concentration (1 μm) far higher than their equilibrium dissociation constants for occupancy and blockade of a variety of G-protein-coupled 5-HT receptors (Hoyer et al. 1994), had no effect upon the current response evoked by 5-HT (not illustrated). The rank order of antagonist potencies (ondansetron > metoclopramide > cocaine > (+)-tubocurarine) agrees with the data obtained in the oocyte expression system (Belelli et al. 1995), and also equates to that found in radioligand binding studies utilizing [³H]granisetron, a selective 5-HT₃ receptor antagonist, to label either the h5-HT_3A receptor expressed in membrane homogenates prepared from stably transfected HEK 293 cells (Hope et al. 1996), or the receptor endogenous to human brain membranes (Bufton, Steward, Barber & Barnes, 1993).

Figure 2. Inhibition of h5-HT_3A receptor-mediated currents by 5-HT₃ receptor antagonists.

Inward current responses evoked by pressure-applied 5-HT (10 μm) were reduced by the bath application of: ondansetron (100 pm) (A), metoclopramide (Met; 100 nm) (B), cocaine (300 nm) (C) and (+)-tubocurarine ((+)-TC; 3 μm) (D). All records are the computer-generated average of 3 consecutive responses to 5-HT recorded at a holding potential of −60 mV. Each antagonist was tested on a different cell. E, graphical depiction of the concentration-dependent inhibition of the 5-HT-evoked current by 5-HT₃ receptor antagonists. Peak current amplitude is expressed as a percentage of that observed in the absence of antagonist. The fitted curves (eqn (1)) give IC₅₀ values of: ondansetron (▪; 103 pm), metoclopramide (•; 69 nm), cocaine (▾; 459 nm) and (+)-tubocurarine (♦; 2.2 μm). Each data point is the mean of 5–6 experiments conducted upon different cells. Vertical bars indicate the standard error of the mean.

Ionic dependency and rectification of 5-HT-induced currents

Employing standard extracellular (E1) and intracellular (I1) recording media, the peak amplitude of currents evoked by pressure-applied 5-HT, recorded at holding potentials in the range −80 to +40 mV, demonstrated inward rectification and reversed in sign (E_5-HT) at a potential of −0.2 ± 0.5 mV (n = 7). The whole-cell chord conductance at a holding potential of +40 mV amounted, on average, to 38% of the value obtained at −80 mV. The combination of total replacement of extracellular K⁺ by Na⁺, a reduction in the concentration of extracellular Ca²⁺ and Mg²⁺ to 0.1 mm each (solution E2), and omission of Mg²⁺ from the pipette solution, had no effect upon E_5-HT (−1.6 ± 0.5 mV; n = 8). Inward rectification persisted, and the average chord conductance at +40 mV as a percentage of that observed at −80 mV (36%) was essentially unchanged. Similar results were obtained when the Mg²⁺-deficient pipette solution was employed in combination with solution E1 (data not shown). Blockade of h5-HT_3A channels by internal Mg²⁺ is thus unlikely to contribute to the observed inward rectification of the 5-HT-evoked whole-cell current.

In subsequent experiments, the I-V relationship of the macroscopic current response to 5-HT was more conveniently generated by a voltage ramp (−150 to + 50 mV) imposed at the peak of the current response to bath-applied 5-HT (1 μm). The I-V relationship obtained with this protocol in the presence of solution E2 and Mg²⁺-deficient pipette solution yielded a value for E_5-HT (−2.2 ± 0.4 mV, n = 11) and a pattern of rectification indistinguishable from those previously obtained from measurement of peak current responses at different holding potentials. This is exemplified in Fig. 3A, where the appropriately scaled I-V relationships obtained by the two methods in the same cell are shown to be superimposable.

Figure 3. Ionic dependency of h5-HT_3A receptor-mediated currents.

A, comparison of I-V relationships determined by a voltage ramp (−150 to +50 mV; 0.25 mV ms⁻¹), applied at the peak of the current response to 5-HT (1 μm; curve), and peak current responses evoked by pressure-applied 5-HT (10 μm; ▪) recorded at steady holding potentials in the range −100 to +40 mV from a single cell. The relationships were superimposed by applying a constant scaling factor to all currents evoked by pressure-applied 5-HT. Currents were recorded in the presence of the Na⁺-based extracellular solution E2 and a nominally Mg²⁺-free Cs⁺-containing pipette solution. E_5-HT was estimated to be approximately 0 mV. Leakage currents determined in the absence of 5-HT have been subtracted. B, I-V relationship of the 5-HT (1 μm)-induced current determined by voltage ramp from a single cell exposed to extracellular media containing 143 mm (1), 75 mm (2) or 20 mm Na⁺ (3). Equimolar substitution of Na⁺ was performed with NMDG. The inset displays portions of the I-V relationships on expanded voltage and current axes to clarify the potential at which there is zero net current (i.e. E_5-HT). E_5-HT in the exemplar cell was −2 mV in 143 mm Na⁺, −14 mV in 75 mm Na⁺ and −46 mV in 20 mm Na⁺. Leakage currents determined in the appropriate solutions were subtracted. C and D, I-V relationships for 5-HT-evoked currents obtained from cells bathed in extracellular media in which either Ca²⁺ (100 mm; C) or Mg²⁺ (100 mm; D) was totally replaced Na⁺. E_5-HT was estimated to be −10 and −20 mV, respectively.

The total replacement of Cl⁻ within solution E2 by isethionate anions had essentially no effect upon E_5-HT (+1.5 ± 0.8 mV; n = 3), confirming the results of Belelli et al. (1995). Inward rectification was unaltered in the Cl⁻-deficient solution (data not shown). The above results suggest the 5-HT-evoked current to be mediated exclusively by cations. Employing E_5-HT measured in solution E2 in conjunction with eqn (2) suggests the permeability of Na⁺ relative to Cs⁺ (i.e. P_Na/P_Cs) to be 0.90. Modified E2 solution, in which Na⁺ was totally replaced by the relatively large organic cation N-methyl-d-glucamine (NMDG), supported small current responses to 5-HT that reversed in sign at a potential of −63.5 ± 2.0 mV (n = 8). Neglecting small errors that might be contributed by the very low concentrations of permeant divalent cations (see below) present extra- and intracellularly, the permeability of NMDG relative to Cs⁺ can be calculated from eqn (2) to be 0.083.

Intermediate levels of substitution, where [Na⁺]_o was reduced to 75 mm and 20 mm by partial replacement with NMDG, displaced E_5-HT to −17.5 ± 0.7 mV (n = 5) and −39.8 ± 1.5 mV (n = 4), respectively (Fig. 3B). These values are within one millivolt of those predicted by eqn (2) employing a P_Na/P_Cs ratio of 0.90 and modified to include a P_NMDG/P_Cs ratio of 0.083. Collectively, the results of the above experiments are in agreement with previous investigations showing that both native and mouse recombinant 5-HT₃ receptors discriminate weakly between monovalent metals (Lambert et al. 1989; Yakel et al. 1990; Yang et al. 1990; Yang et al. 1992; Malone, Peters & Lambert, 1993; Gill et al. 1995, but see Glitsch et al. 1996). The finite permeability of NMDG (see also Yakel et al. 1990; Yang, 1990; Malone et al. 1993) suggests a relatively large minimal pore size for the receptor formed from h5-HT_3A subunits, which would be compatible with estimates in the range between 0.76 and 0.81 nm made for 5-HT₃ receptors endogenous to N18 neuroblastoma cells (Yang, 1990) and rabbit nodose ganglion neurones (Malone et al. 1993).

Calcium and magnesium are permeant

The permeability of h5-HT_3A towards Ca²⁺ and Mg²⁺ was examined using extracellular media containing 100 mm Ca²⁺ or 100 mm Mg²⁺ as charge carriers. In either solution, inward current responses to 5-HT were clearly discernible at negative holding potentials, although they were of considerably smaller amplitude than those observed with the Na⁺-containing solutions E1 and E2, an observation in qualitative agreement with results obtained with mouse N18 neuroblastoma cells (Yang, 1990) or rat superior cervical ganglion neurones (Yang et al. 1992). h5-HT_3A is thus quite permeant to both Ca²⁺ and Mg²⁺, in addition to monovalent cations. Employing voltage ramps to estimate E_5-HT resulted in values of −11.9 ± 0.7 mV (n = 5) and −21.1 ± 0.8 mV (n = 6) for cells bathed in 100 mm CaCl₂ or 100 mm MgCl₂, respectively (Fig. 3C and D). A P_Ca/P_Cs ratio of 1.00 and a P_Mg/P_Cs ratio of 0.61 were calculated from these values substituted into eqn (5). As a further test of the accuracy of the voltage ramp method, E_5-HT in the presence of 100 mm Ca²⁺ was additionally determined by measurement of peak current responses evoked by pressure-applied 5-HT in cells clamped at steady holding potentials. The interpolated value of E_5-HT (−11.7 ± 1.3; n = 6) agreed with that determined by voltage ramp. A further estimate of P_Ca/P_Cs was made from measurements of E_5-HT in an extracellular solution containing 100 mm NMDG and 20 mm CaCl₂. Under such conditions, E_5-HT was −31.3 ± 1.5 mV (n = 5) and P_Ca/P_Cs was calculated to be 1.39. Reasons for the differing estimates of P_Ca/P_Cs are considered in the Discussion.

Extracellular divalent cations modulate the amplitude of currents mediated by h5-HT_3A

The effect of extracellular Mg²⁺ and Ca²⁺ was investigated using voltage ramps over the potential range −150 mV to +50 mV. Figure 4 illustrates the results obtained with one cell, which was successively exposed to 5-HT (1 μm) applied within an Na⁺-containing medium with Ca²⁺ and Mg²⁺ each present at a concentration of 2.0 mm, or media in which [Ca²⁺]_o alone was reduced to 0.1 mm, or both [Ca²⁺]_o and [Mg²⁺]_o were reduced to 0.1 mm (i.e. E2).

Figure 4. Extracellular divalent cations suppress h5-HT_3A receptor-mediated currents.

I-V relationships, obtained from one cell by voltage ramps (−150 to +50 mV; 0.25 mV ms⁻¹), applied at the peak of the current response to 5-HT (1 μm) in Na⁺-based extracellular media containing 2 mm Ca²⁺ and 2 mm Mg²⁺ (1), 2 mm Mg²⁺ and 0.1 mm Ca²⁺ (2), or 0.1 mm Ca²⁺ and 0.1 mm Mg²⁺ (3). Note that inward rectification of the 5-HT-induced current is observed under all conditions and that suppression of the current by extracellular divalent cations appears to be voltage independent. The inset shows that E_5-HT is insensitive to the small changes in either [Ca²⁺]_o or [Mg²⁺]_o.

The reduction in [Ca²⁺]_o alone resulted in a potentiation of the amplitude of the 5-HT-induced current without any discernible change in E_5-HT (−2.3 ± 1.8 mV, n = 5). Potentiation was apparent over the entire range of voltages examined, and the increase in 5-HT-evoked chord conductance at −150 mV (i.e. 157.1 ± 17.6% of control, n = 5) and +50 mV (i.e. 178.6 ± 13.4% of control, n = 5) indicate the effect to be essentially voltage independent. The simultaneous reduction in [Ca²⁺]_o and [Mg²⁺]_o to 0.1 mM again had no impact upon E_5-HT (−2.8 ± 1.4 mV, n = 5), but an additional, voltage-independent, potentiation of the 5-HT-induced current was observed, with chord conductances at −150 and +50 mV increasing to 248.6 ± 23.1% (n = 5) and 222.4 ± 23.3% (n = 5) of control, respectively (Fig. 4).

Fluctuation analysis of the 5-HT-induced current response

The amplitude of the single channel current activated by 5-HT was estimated by fluctuation analysis utilizing extracellular solutions with differing concentrations of divalent cations. In each instance, slowly rising and desensitizing inward current responses, suited to stationary fluctuation analysis, were elicited by a low concentration of bath-applied 5-HT (1 μm). Exemplar data, derived from one cell bathed in medium in which both Ca²⁺ and Mg²⁺ were present at a concentration of 2 mm, are illustrated in Fig. 5. In all cases, the macroscopic current response to 5-HT was associated with an increase in ‘noise’, which was most clearly discerned in high gain AC-coupled records (Fig. 5). In the majority of instances, the relationship between the noise variance (σ²) and mean amplitude (I_m) of the inward current response to 5-HT was linear, allowing an estimate of unitary current (i) be made from the slope of current-variance plots (eqn (8)). Estimates of i varied non-linearly with holding potential. For example, the data shown in Fig. 5 yielded estimates of i of 5.9, 12.0, 31.4 and 60.1 fA at holding potentials of −40, −60, −80 and −100 mV, respectively. This trend, which is indicative of inward rectification at the single channel level, was observed in seven additional cells. Mean values of i as a function of holding potential are presented in Table 1, along with single channel chord conductances.

Figure 5. Fluctuation analysis whole-cell currents mediated by h5-HT_3A receptors.

Examples from a single cell of low gain DC-coupled records (top panel) and high gain AC-coupled (1–1000 Hz bandwidth) records (middle panel) of inward current responses evoked by bath-applied 5-HT (1 μm; horizontal bar) at holding potentials of −40 (A), −60 (B), −80 (C) and −100 mV (D) in the presence of standard extracellular (E1) and intracellular media (I1). Note that the development and desensitization of the inward current response are paralleled by respective increases and decreases in membrane current fluctuations. The latter are most pronounced at hyperpolarized potentials. The lower panels illustrate the relationship between the variance of the AC-coupled current and DC amplitude throughout the 5-HT-induced response. The slope of the line fitted to the data points, by linear regression analysis, provides estimates of single channel current amplitudes of 5.9, 12.0, 31.4 and 61.1 fA at holding potentials of −40, −60, −80 and −100 mV, respectively. Background current variance recorded in the absence of 5-HT at each holding potential was subtracted.

Table 1.

Summary of the voltage dependence of single channel current amplitude and chord conductance of h5-HT_3A

	2 mm Ca2+-2 mm Mg2+		2 mm Mg2+- 0.1 mm Ca2+		0.1 mm Ca2+- 0.1 mm Mg2+
Vh (mV)	i (fA)	γ (fS)	i (fA)	γ (fS)	i (fA)	γ (fS)
−40	10.1 ± 1.0	243 ± 23	11.7 ± 1.0	278 ± 25	18.4 ± 2.4	430 ± 57
−60	19.1 ± 1.7	310 ± 28	29.4 ± 0.6	473 ± 9	42.1 ± 3.1	671 ± 50
−80	40.0 ± 1.3	453 ± 16	50.5 ± 1.1	614 ± 13	83.4 ± 2.0	1008 ± 24
−100	75.4 ± 4.1	742 ± 40	99.1 ± 4.4	969 ± 48	154.6 ± 7.2	1505 ± 71

Single channel current amplitude was estimated by fluctuation analysis of the macroscopic I-V relationship over the voltage range −40 to −100 mV in the presence of the indicated extracellular divalent cation concentrations. The chord conductance (γ) was calculated using eqn (9) with the appropriate E_5-HT.

Reducing the extracellular concentration of either Ca²⁺ or Mg²⁺ augmented the amplitude of the single channel current activated by 5-HT over the potential range (−40 to −100 mV) examined. Mean data derived from seven cells in which [Ca²⁺]_o was reduced to 0.1 mm and four cells in which both [Ca²⁺]_o and [Mg²⁺]_o were decreased to 0.1 mm are presented in Table 1, along with chord conductances calculated utilizing E_5-HT determined under the same ionic conditions. The percentage increase in single channel conductance produced by the reducing the extracellular concentration of divalent cations was little affected by holding potential (−40 to −100 mV; Fig. 6A).

Figure 6. Extracellular divalent cations decrease the single channel conductance of h5-HT_3A receptors.

A, graph depicting the influence of [Ca²⁺]_o and [Mg²⁺]_o upon the single channel chord conductance of h5-HT_3A estimated from fluctuation analysis of macroscopic current responses recorded in the presence of 2 mm Ca²⁺-2 mm Mg²⁺ (▪); 0.1 mm Ca²⁺-2 mm Mg²⁺ (▴) or 0.1 mm Ca²⁺- 0.1 mm Mg²⁺ (♦). Inward rectification is apparent under all conditions and is little affected by the extracellular concentrations of divalent cations. Data points are the means of 4–7 determinations of single channel chord conductance performed on different cells. Vertical bars indicate the standard error of the mean. B, C and D, comparison of the voltage dependence of macroscopic and single channel currents evoked by 5-HT in 3 different cells bathed in media containing 2 mm Ca²⁺ and 2 mm Mg²⁺ (B); 2 mm Mg²⁺ and 0.1 mm Ca²⁺ (C) or 0.1 mm Ca²⁺ and 0.1 mm Mg²⁺ (D). The I-V relationship of the macroscopic current response to 5-HT (1 μm) was determined by a voltage ramp (−150 to +50 mV; 0.25 mV ms⁻¹), applied at the peak of the current response to 5-HT (1 μm), whereas single channel current amplitudes were estimated by stationary fluctuation analysis at holding potentials of −40, −60, −80 and −100 mV. Superimposition of the macroscopic and single channel current amplitudes was achieved by scaling the latter by a constant factor. The data suggest that, at least within the voltage range −40 to −100 mV, inward rectification under each ionic condition examined can be attributed to the conductance properties of the h5-HT_3A receptor channel.

In three different cells, the macroscopic I-V relationship for the 5-HT-induced whole-cell current, recorded in the presence of 2 mm Ca²⁺- 2 mm Mg²⁺, 0.1 mm Ca²⁺-2 mm Mg²⁺, and 0.1 mm Ca²⁺- 0.1 mm Mg²⁺ was directly compared with the voltage dependence of the single channel current amplitude determined in the same cells under the identical ionic conditions. Figure 6B-D shows the superimposition of the appropriately scaled single channel currents, determined at holding potentials in the range −40 to −100 mV, upon the macroscopic I-V relationship. The similarity in the voltage-dependence of the macroscopic and elementary I-V relationships suggests that the observed inward rectification is an inherent property of the elementary conductance activated by 5-HT.

DISCUSSION

The present results indicate that h5-HT_3A subunits expressed in HEK293 cells form homo-oligomeric, inwardly rectifying, cation-selective ion channels of low conductance. Single channel and whole-cell current responses exhibited a similar degree of inward rectification, and both the elementary and macroscopic currents were influenced in an essentially identical manner by divalent cations at physiologically relevant concentrations. These features are discussed below.

Divalent cations permeate the h5-HT_3A receptor

Ca²⁺ and Mg²⁺ permeate 5-HT₃ receptor channels of N18 cells (Yang, 1990) and rat peripheral neurones (Yang et al. 1992). In agreement, the h5-HT_3A channel exhibited apparent P_Ca/P_Cs ratios of 1.0 and 1.39 when [Ca²⁺]_o equalled 100 or 20 mm, respectively. The discrepancy between these values might be attributed to a more negative surface potential in the presence of the 20 mm Ca²⁺-120 mm NMDG mixture, which the present analysis neglects (Lewis, 1979; Yang, 1990). In apparent contradiction, several studies of native and m5-HT_3A receptors suggest Ca²⁺ to be impermeant (Yakel et al. 1990; Gilon & Yakel, 1995; Glitsch et al. 1996). However, Ca²⁺ permeability was often assessed in the presence of monovalent species and over a limited range of divalent cation concentrations, potentially obscuring an underlying permeability to Ca²⁺. Consistent with this suggestion, 5-HT₃ receptor-mediated increases in [Ca²⁺]_i imaged in single N1E-115 cells, or mammalian cells expressing m5-HT_3A receptors, are modest in the presence of extracellular Na⁺ and K⁺, but their replacement by NMDG greatly augments the Ca²⁺ signal (Hargreaves, Lummis & Taylor, 1994). Such an effect is independent the inhibition of the plasma membrane Na⁺-Ca²⁺ exchanger caused by the replacement of Na⁺ by NMDG (Hargreaves et al. 1994).

The P_Ca/P_Cs ratios calculated here approximate to those obtained for neuronal, α-bungarotoxin-insensitive, nicotinic acetylcholine receptors (reviewed by McGehee & Role, 1995). For such receptors, the fraction of the inward current carried by Ca²⁺ is sufficient to raise [Ca²⁺]_i to levels at which Ca²⁺-dependent processes are activated (e.g. Vernino, Rogers, Radcliffe & Dani, 1994). However, permeability ratios do not predict ion flux ratios. Whether the inward flux of Ca²⁺ through h5-HT_3A receptor channels under physiological conditions is important requires evaluation. 5-HT₃ receptor activation elevates cytosolic Ca²⁺ in clonal cell lines (Hargreaves et al. 1994) and striatal synaptosomes (Nichols & Mollard, 1996) and facilitates Ca²⁺-dependent processes such as transmitter release (e.g. Blandina, Goldfarb, Craddock-Royal & Green, 1989). However, in most cases, the relative contribution of Ca²⁺ fluxes through 5-HT₃- and voltage-activated calcium channels has not been dissected and may, indeed, be hampered by the fact that l-type calcium channel antagonists also modulate 5-HT₃ receptor activity (Hargreaves et al. 1994). The simultaneous measurement of 5-HT₃ receptor-evoked current and [Ca²⁺]_i in individual cells (e.g. Vernino et al. 1994) may provide definitive information.

Single channel conductance of the h5-HT_3A receptor

As found for the two splice variants of the m5-HT_3A receptor (i.e. m5-HT_3A(a) and m5-HT_3A(b); Hope et al. 1993) expressed in mammalian cells, and the receptor endogenous to neuroblastoma cell lines (Lambert et al. 1989; Yang, 1990; Hussy et al. 1994; Gill et al. 1995), the h5-HT_3A receptor gates an inwardly rectifying single channel of sub-picosiemen conductance. Indeed, for both m5-HT_3A (Hussy et al. 1994) and h5-HT_3A, the non-linearity of the single channel current appears to account for the rectification of whole-cell currents. Supporting this conclusion, Barann, Göthert, Bönisch, Dybek & Urban (1997) noted 5-HT₃ receptor-mediated currents evoked by 5-HT applied rapidly to outside-out patches of N1E-115 cells to display quantitatively similar rectification (i.e. doubling of current amplitude between −80 and −100 mV) in the absence of any change in the kinetics of the current.

Consistent with the low single channel conductance inferred by fluctuation analysis, single channel events cannot be detected during inward current responses to 5-HT₃ receptor recorded from outside-out membrane patches excised from clonal cell lines (Lambert et al. 1989; Hussy et al. 1994; Barann et al. 1997). In contrast, resolvable channel events, yielding substantial channel conductances, are mediated by 5-HT₃ receptors of guinea-pig submucous plexus (9–15 pS, Derkach et al. 1989) and rabbit nodose ganglion neurones (19 pS, Peters et al. 1993). Direct observation of inwardly rectifying 5-HT₃ receptor channels in rat and mouse superior cervical ganglion cells reveals conductances within the ranges 10–15 pS and 8–13 pS, respectively. Yet fluctuation analysis applied to the same cells yields far lower values, a result which may suggest the co-expression of resolvable ‘high-conductance’ and very low conductance 5-HT₃ receptor isoforms of the type characterized here (Yang et al. 1992; Hussy et al. 1994).

Why recombinant 5-HT_3A receptors exhibit such a low single channel conductance is an enigma. Amino acid residues implicated in ion conduction and permeation that bracket and line the channel-forming M2 domain of nicotinic acetylcholine receptors are largely conserved in species homologues of 5-HT_3A (Galzi & Changeux, 1995; Peters et al. 1997). Thus, two major influences, an acidic amino acid forming an intermediate ring of negative charge and a hydroxylated residue forming a polar ring at positions denoted −1′ and 2′, respectively, are conserved in all 5-HT_3A subunit sequences (Imoto et al. 1988; Imoto, Konno, Nakai, Wang, Mishina & Numa, 1991; Galzi & Changeux, 1995; Peters et al. 1997). The commonality also extends to an extracellular ring of negative charge at position 20′ and a polar ring located at position 6′, which in nicotinic receptors are documented to exert a less profound effect upon conduction. Similarly, the equatorial leucine ring at position 9′ and a valine residue located at position 13′, either of which when modified by mutagenesis greatly affect the properties of nicotinic receptors (Galzi & Changeux, 1995), are present within the 5-HT_3A subunit. However, a third hydroxyl ring formed by a serine residue at position 10′ in nicotinic α-subunits is replaced by glycine in the 5-HT_3A subunit. Mutation of the 10′ serine to asparagine, valine or tyrosine in the nicotinic receptor of Torpedo produces a modest reduction in channel conductance (Imoto et al. 1991). Since glycine does not possess a side chain whereby steric hinderance to ion flow could occur, it seems most unlikely that this substitution can account for the very low conductance of 5-HT_3A subunits in comparision to nicotinic receptors.

A unique feature of the M2 sequence of 5-HT_3A receptor subunits is the presence of a basic lysine residue at position 4′ which, a priori, could depress channel conductance by repulsion of permeant cations. However, alignment with a nicotinic acetylcholine receptor α-subunit suggests that the lysine faces away from the channel lumen (Akabas, Kaufmann, Archdeacon & Karlin, 1994). Moreover, mutation of the lysine to glycine, serine, glutamine or arginine residues does not influence the conductance of m5-HT_3A receptors (M. Gunthorpe, J. Peters, C. Gill, S. Lummis & J. Lambert, unpublished observations).

An inner ring of negative charges, located at the cytoplasmic vestibule of the channel, provided by aspartate residues in m5-HT_3A, is absent from h5-HT_3A, where the homologous residue is a polar asparagine (Belelli et al. 1995; Miyake et al. 1995). Fixed negative charges at the channel vestibules of nicotinic acetylcholine receptors increase channel conductance (Imoto et al. 1988), in part by electrostatic interactions which raise the local concentration of permeant cations at the entrance to a narrower region of the pore (Dani, 1986). Superficially, the present data do not support a similar role for the inner ring in 5-HT_3A receptor function, since the conductance of human and mouse homologues are identical (cf. Hussy et al. 1994). However, the conditions of neither study were optimal for detecting an electrostatic effect, because the intracellular solutions were of moderate ionic strength and contained 2 mm Mg²⁺ (both reducing the influence of fixed charges) and outward single channel currents were not estimated.

Blockade of h5-HT_3A by divalent cations

The suppression of macroscopic currents mediated by recombinant and endogenous 5-HT₃ receptors by extracellular Ca²⁺ and Mg²⁺ is reported to be either voltage dependent (Maricq et al. 1991; Eiseléet al. 1993; Kawa, 1994) or voltage independent (Peters et al. 1988, 1993; Yang, 1990; Gill et al. 1995; Glitsch et al. 1996). Here, the suppression of whole-cell currents by these ions was essentially voltage independent. A region of negative slope conductance at hyperpolarized potentials was never observed, even when all extracellular Na⁺ was completely replaced by either Ca²⁺ or Mg²⁺. Indeed, inward rectification persisted at potentials negative to the reversal potential under such conditions. Extracellular Ca²⁺ and Mg²⁺ reduced the conductance of the h5-HT_3A receptor, and for Ca²⁺, the effect was of the same order of magnitude as that found for nicotinic acetylcholine receptors (Decker & Dani, 1990). The decrease in single channel conductance in the presence of divalent cations represents a major contribution to the decrease in the macroscopic current response observed under identical ionic conditions. Potential additional factors, such as alterations in agonist affinity, single channel kinetics or desensitization, are likely be of secondary importance under the present experimental conditions.

The depression of channel conductance produced by divalent cations could result from the masking of fixed charges in the outer vestibule of the channel and competition between Ca²⁺, Mg²⁺ and monovalent cations for conduction within the pore. Qualitatively, the data are compatible with a scheme wherein Ca²⁺ (and Mg²⁺) exhibit an affinity for a binding site within the channel which exceeds that of monovalent cations. By their relatively slow dissociation from the site, the divalents would retard the flux of monovalent ions through the channel. Simplistically, this mechanism predicts that the effect of extracellular of Ca²⁺ and Mg²⁺ should be most pronounced upon inward currents, and that the single channel conductance measured at potentials negative to E_5-HT with Ca²⁺ (or Mg²⁺) present extracellularly as the sole charge carrier should be considerably reduced. We could not test these predictions because increased background ‘noise’ at depolarized potentials and a reduced current in the presence of high concentrations of divalent cations hindered assessment of channel conductance by fluctuation analysis. Our results over a limited range of negative holding potentials give little indication of voltage-dependent block, but this is not unexpected, since the driving force upon both Ca²⁺ and Mg²⁺ was consistently inward.

In conclusion, receptors constructed from h5-HT_3A subunits demonstrate pharmacological (see Peters et al. 1997) and biophysical properties which are essentially identical to those of some 5-HT₃ receptors native to neuronal tissues, providing a strong indication that homo-oligomeric assemblies of 5-HT_3A subunits occur in nature. However, the disparate single channel conductances reported in the literature may indicate that structural diversity within this receptor class has yet to be revealed.