Corresponding author J. J. Lambert: Neurosciences Institute, Department of Pharmacology and Neuroscience, Ninewells Hospital and Medical School, Dundee University, Dundee DD1 9SY, UK. Email: firstname.lastname@example.org
1The influence of a transmembrane (TM2) amino acid located at a homologous position in human β1 (S290) and β3 (N289) GABAA receptor subunits and the RDL GABA receptor of Drosophila (M314) upon allosteric regulation by general anaesthetics has been investigated.
2GABA-evoked currents mediated by human wild-type (WT) α6β3γ2L or WT RDL GABA receptors expressed in Xenopus laevis oocytes were augmented by propofol or pentobarbitone. High concentrations of either anaesthetic directly activated α6β3γ2L, but not RDL, receptors.
3GABA-evoked currents mediated by human mutant GABAA receptors expressing the RDL methionine residue (i.e. α6β3N289Mγ2L) were potentiated by propofol or pentobarbitone with ≈2-fold reduced potency and, in the case of propofol, reduced maximal effect. Conspicuously, the mutant receptor was refractory to activation by either propofol or pentobarbitone.
4Incorporation of the homologous GABAAβ1-subunit residue in the RDL receptor (i.e. RDLM314S) increased the potency, but not the maximal effect, of GABA potentiation by either propofol or pentobarbitone. Strikingly, either anaesthetic now activated the receptor, an effect confirmed for propofol utilizing expression of WT or mutant RDL subunits in Schnieder S2 cells. At RDL receptors expressing the homologous β3-subunit residue (i.e. RDLM314N) the actions of propofol were similarly affected, whereas those of pentobarbitone were unaltered.
5The results indicate that the identity of a homologous amino acid affects, in a complementary manner, the direct activation of human (α6β3γ2L) and RDL GABA receptors by structurally distinct general anaesthetics. Whether the crucial residue acts as a regulator of signal transduction or as a component of an anaesthetic binding site per se is discussed.
The molecular mechanisms that underlie the rapid depression of central nervous system function by general anaesthetics remain to be identified. The strong correlation between simple indices of anaesthetic potency and the solubility of anaesthetics in fatty solvents established by Meyer and Overton has traditionally been held to indicate a hydrophobic site of action, commonly equated with the lipid bilayer of the nerve cell membrane. A relatively non-specific physicochemical interaction involving an alteration in the volume, thickness or phase behaviour of the lipid bilayer might, in principle, explain how agents that range from inert gases to complex steroidal structures can exert a similar end-effect (Little, 1996). However, not all anaesthetics act similarly. The pharmacological spectrum of an anaesthetic can include amnesia, immobility in response to a noxious stimulus, unconsciousness, analgesia and suppression of the stress response, but the balance of such effects are properties of the individual agent, an observation that is not readily explained by a unitary theory of anaesthesia (Eger et al. 1997). At a more fundamental level, general anaesthetics produce only modest changes in membrane structure that can be mimicked by small changes in temperature (Franks & Lieb, 1994). When coupled with the cut-off phenomenon and recently established violations of the Meyer-Overton rule, lipid theories of general anaesthesia appear less tenable (Franks & Lieb, 1994; Eger et al. 1997).
Neural proteins, particularly the major transmitter-gated channels mediating central inhibition (GABAA and glycine) and excitation (glutamate), have recently received considerable attention as alternative sites of anaesthetic action (Franks & Lieb, 1994; Harris et al. 1995). The spectrum of actions that contribute to a state of anaesthesia could potentially be explained by differential effects upon multiple transmitter systems, or subtypes of receptor responsive to a particular transmitter. A particularly persuasive case can be made for the involvement of the GABAA receptor in certain aspects of anaesthesia, such as the suppression of movement in response to noxious stimulation (Eger et al. 1997; Quinlan et al. 1998). Indeed although structurally diverse, the majority of clinically useful and experimental general anaesthetics at appropriate concentrations share the common feature of potentiating the actions of GABA at the GABAA receptor (Tanelian et al. 1993; Franks & Lieb, 1994; Lambert et al. 1998). Additionally, the enantioselectivity of a number of the general anaesthetics, such as isoflurane, etomidate, pentobarbitone and neuroactive steroids, is mirrored in their anaesthetic potency (Lambert et al. 1998). A direct link between anaesthetic mechanisms and GABAA receptor function has been established by the use of mice genetically engineered to lack the β3 GABAA receptor subunit. In comparison with wild-type mice, the null-allele animals demonstrated a change in anaesthetic requirement that was dependent upon the identity of the anaesthetic agent and the response quantified (i.e. loss of righting reflex or nocifensive reflex) (Quinlan et al. 1998).
In mammals, the GABAA receptor is composed of five polypeptide subunits drawn from the products of a multigene family (α1-6, β1-3, γ1-3, δ and ε). These are differentially expressed within the central nervous system (Smith & Olsen, 1995; Whiting et al. 1995) and recombinant expression studies have revealed that subunit composition influences both the physiological and pharmacological properties of the receptor (Sieghart, 1995). Determining the actions of anaesthetics at different GABAA receptor isoforms may help explain brain region-selective effects of general anaesthetics and may aid the identification of protein domains that form the anaesthetic binding pocket, or that are required for the alteration of GABA-gated chloride channel function by the anaesthetic (Smith & Olsen, 1995; Belelli et al. 1997; Mihic et al. 1997; Peters & Lambert, 1997).
In addition to enhancing the actions of GABA, a number of general anaesthetics possess GABA-mimetic activity (Tanelian et al. 1993; Lambert et al. 1998). This effect generally occurs at concentrations greater than those required for GABA enhancement. The differing concentration dependencies of the GABA-modulatory and GABA-mimetic activities of general anaesthetics are compatible with their respective mediation by high and low affinity anaesthetic binding sites on the GABAA receptor protein. The concept of distinct binding sites appears to be supported by our previous studies concerning the actions of the general anaesthetics propofol and pentobarbitone on an invertebrate recombinant GABA receptor (RDL) isolated from Drosophila melanogaster (ffrench-Constant et al. 1991; Chen et al. 1994) where these anaesthetics enhanced the actions of GABA but, in contrast to mammalian GABAA receptors, did not directly activate the receptor (Belelli et al. 1996). Therefore, either the RDL receptor truly lacks an anaesthetic activation site, or alternatively, the site is present but occupation by the anaesthetic is not transduced into channel opening. Independent of the underlying cause, the RDL subunit may be instructive in identifying those amino acids that are required for the GABA-mimetic actions of certain anaesthetics. Here we demonstrate that the nature of a single transmembrane amino acid governs, in a complementary manner, the GABA-mimetic actions of the general anaesthetics pentobarbitone and propofol at both mammalian and invertebrate GABA receptors. The results are discussed in relation to whether this amino acid contributes to the anaesthetic binding site, or is essential for transduction.
The gene encoding the Drosophila GABA receptor (Rdl) gives rise to transcripts that are alternatively spliced at two locations producing variants of exon 3, termed ‘a’ and ‘b’, and exon 6, denoted ‘c’ and ‘d’ (ffrench-Constant & Rocheleau, 1993). The receptor used in the present study incorporates variants ‘a’ and ‘c’ and is hence termed RDLac, but here we will refer to it simply as RDL. In this report, the effects of general anaesthetics on the RDLac receptor are occasionally compared with previous studies conducted upon an RDL receptor containing the ‘b’ and ‘d’ splice variants. The latter additionally differs from RDLac in respect of five residues encoded by exons 1, 2, 4 and 8 (Chen et al. 1994; Belelli et al. 1996; Hosie & Sattelle, 1996) and hence we employ the original appellation DRC 17-1-2. The numbering of amino acid residues in the human GABAAβ1- and β3-subunits is that of Haddingham et al. (1993) and includes the entire coding region (putative signal peptide and mature protein). Thus, β1 (S290) and β3 (N289) are equivalent to β1 (S265) and β3 (N264), respectively, in schemes that number residues from the putative signal peptide cleavage site.
Cell culture and transfection of Drosophila S2 cells
The Drosophila Expression System (Invitrogen B.V., Groningen, The Netherlands) was used to express the recombinant invertebrate GABA receptors. The Drosophila melanogaster cell line Schnieder 2 (S2) was cultured as a semi-adherent monolayer in complete Drosophila Expression System (DES) medium (Invitrogen), supplemented with 10 % (v/v) fetal bovine serum, penicillin (50 i.u. ml−1) and streptomycin (50 μg ml−1) at room temperature without CO2.
The cDNAs encoding the invertebrate RDL GABA receptor subunits (i.e. RDL, RDLM314N and RDLM314S) in the vector pNB40 (McGurk et al. 1998) were amplified by the polymerase chain reaction (PCR) using two specific oligonucleotides (forward primer, 5′-CCGCTCGAGCGGAAGCTTGCTTGTTC-3′ with XhoIrestriction site shown in bold; reverse primer, 5′-TGCTCTAGAGCAAGCATTCGTTTTTT-3′ with XbaI restriction site shown in bold). The PCR product was cloned into the DES-inducible expression vector, pDS47/V5-HisB, under the control of the inducible metallothionein (MT) promoter and sequenced (fmol DNA Sequencing System, Promega, Southampton, UK). The DNA was prepared for transfection using Hybaid Recovery Quick Flow Maxi Kits (Hybaid, Ashford, UK).
Prior to transfection, S2 cells were cultured on polylysine-coated plates (to improve adherence) at an approximate density of 2 × 106 cells per 35 mm2 dish for 24 h. Transient transfection of the cells was performed according to the manufacturer's recommendations using calcium phosphate. Briefly, 19 μg of cDNA was mixed with 36 μl of 2 M CaCl2 and the volume adjusted to 300 μl with sterile water. This solution was added dropwise, with vigorous vortexing, to 300 μl 2 × HBS buffer (Gibco, Paisley, UK) and incubated at room temperature for 40 min, to allow the precipitate to form. The calcium phosphate-DNA precipitate was subsequently added dropwise to the cells and incubated for a further 18 h. Post-transfection, the cells were washed twice with complete DES medium. Expression of the RDL GABA receptor subunits was induced by the addition of 500 μM copper sulphate to the culture medium for a minimum of 24 h prior to electrophysiological analysis (Bunch et al. 1988).
Electrophysiological recordings from S2 cells
The whole-cell recording mode of the patch clamp technique was employed to record agonist-evoked currents from S2 cells 1-5 days after the induction of expression of either wild-type or mutant RDLM314N and RDLM314S GABA receptors. In all experiments, the cells were superfused at a rate of 2-4 ml min−1 with an extracellular solution containing (mM): 140 NaCl, 2.8 KCl, 1 CaCl2, 2 MgCl2, 10 glucose, 20 sucrose and 10 Hepes (pH 7.2). GABA (10 μM), or propofol (10-100 μM) was applied locally by pressure ejection (1.4 × 105 Pa; 3-60 ms) from a modified patch pipette. All other agents were applied via their inclusion within the superfusate. Patch electrodes were filled with a solution comprising (mM): 140 CsCl, 1 CaCl2, 2 MgCl2, 11 EGTA and 10 Hepes (pH 7.2); they had resistances in the range of 2-4 MΩ. All experiments were performed at room temperature (18-21°C). Series resistance compensation up to 70 % was applied when appropriate. The specific resting membrane conductance of S2 cells expressing wild-type or mutant RDL GABA receptors, was determined by measuring the amplitude of the leakage current evoked by a 20 mV hyperpolarizing voltage step. The steady state current-voltage relationship for GABA or anaesthetic agents acting upon S2 cells was obtained by averaging four to seven agonist-evoked responses at holding potentials within the range -80 to +40 mV using the Strathclyde Electrophysiology Software WinWCP program (Dempster, 1997).
Preparation of transcripts and oocyte injection
cDNAs encoding the human α6, β3, β3N289M and γ2L GABAA receptor subunits in the pCDM8 vector were provided by Dr Paul Whiting (Merck, Sharp and Dohme, Harlow, UK). Receptors composed of α6β3γ2L subunits were specifically chosen for study because we have previously demonstrated this combination to be particularly sensitive to the allosteric effects of the anaesthetic etomidate, whereas the Drosophila RDL GABA receptor is etomidate insensitive (Belelli et al. 1997; Hill-Venning et al. 1997; McGurk et al. 1998). The cDNAs for α6 and γ2L GABAA receptor subunits were linearized at unique XhoIsites, those for the β3, RDL, RDLM314N and RDLM314S at a NotI site and that for β3N289M at an XbaIsite. cRNA transcripts were prepared according to standard protocols (Hope et al. 1993). Denaturing gel electrophoresis was used to check the integrity of transcripts prior to injection. cRNA transcripts (30-50 nl; 1 mg ml−1) were injected into Xenopus laevis oocytes (stage V-VI) which had previously been defollicated by treatment with 2 mg ml−1 collagenase ‘A’ (Boehringer-Mannheim) for 3 h at room temperature (20-23°C) in nominally calcium-free Barth's saline. Injected oocytes were individually maintained at 19-20°C for up to 12 days in 96-well plates containing 200 μl of standard Barth's saline (composition, mM: 88 NaCl, 1 KCl, 2.4 NaHCO3, 15 Hepes, 0.5 Ca(NO3)2, 0.5 CaCl2 and 1.0 MgS04; adjusted to pH 7.6 with NaOH). The solution was supplemented with 0.1 mg ml−1 gentamicin.
Electrophysiological recordings from Xenopus laevis oocytes
Recordings were made from oocytes 2-12 days after cRNA injection. Unless specified otherwise, agonist-evoked currents were recorded at a holding potential of -60 mV using either an Axoclamp 2A or a Gene Clamp 500 amplifier (Axon Instruments) in the two-electrode voltage-clamp mode. Oocytes were held in a recording chamber (0.5 ml) and continually superfused (7-10 ml min−1) with frog Ringer solution (composition, mM: 120 NaCl, 2 KCl, 1.8 CaCl2, 5 Hepes; adjusted to pH 7.4 with NaOH). Current-passing and voltage-sensing intracellular electrodes had resistances of 0.5-1.3 MΩ (when filled with 3 M KCl and measured in frog Ringer solution). Current and voltage signals were both acquired (at the digitization rate of 20 Hz) and analysed using the WinWCP program (Dempster, 1997) using an Axon Instruments Digidata 1200 interface and a Dell Dimension XPS PC. Timed pulses of drugs dissolved in Ringer solution were applied to oocytes via a BPS-4 bath perfusion system (Adams and List Associates, New York) with a four-way manifold. Drug application was automatically controlled by the computer program during recording sweeps via the laboratory interface TTL digital outputs and solenoid-controlled pinch valves attached to the perfusion lines. For each oocyte a maximally effective concentration of GABA (1 mM and 3 mM for the invertebrate and mammalian receptor, respectively) was applied once every 20 min until the peak inward current response produced was stable (Belelli et al. 1996). Typically such agonist-evoked currents had a rise time of 0.5-1 s. The magnitude of the GABA enhancing actions of positive allosteric modulators is dependent upon the concentration of GABA used (Harris et al. 1995; Belelli et al. 1996). Hence, the concentration of GABA which evoked a response amounting to 10 % of the current produced by a saturating concentration of GABA (i.e. EC10) was determined for each oocyte and used to evaluate the modulation of GABA-evoked currents by the modulatory agent. The latter was quantified as the increase in the peak amplitude of the GABA-evoked current and data were normalized by expressing the observed response as a percentage of the maximal GABA response, denoted Emax. Modulators were pre-applied for 20-60 s before their co-application with the appropriate concentration of GABA.
Current-voltage curves for responses induced by GABA or modulators (at mutated RDL receptors) were obtained using concentrations of agonist that produced currents displaying little overt desensitization. During the maximal activation of such currents, a voltage ramp generated by the WinWCP program sweeping from -90 to +10 mV (70 mV s−1) was applied (see Fig. 5). Digital subtraction of the leakage current recorded in the absence of agonist, or agonist applied in the presence of a saturating concentration of picrotoxin, yielded the current component specifically due to GABA receptor activation. The ramp protocol was used to minimize the relief of block by divalent cations of a non-selective cation conductance endogenous to oocytes that occurs relatively slowly at positive holding potentials. All measurements were performed at ambient temperature (18-22°C).
Concentration-effect relationships for the GABA-enhancing actions of positive allosteric modulators were fitted iteratively by use of FigP version 6c (Biosoft, Cambridge, UK), with the Hill equation:
where I is the amplitude of the GABA-evoked current in the presence of the modulator at concentration [A], Imax is the amplitude of the response in the presence of a maximally effective concentration of the modulator, EC50 is the concentration of the modulator producing half-maximal enhancement and nH is the Hill coefficient. Concentration-effect relationships for the GABA-mimetic effects of propofol, pentobarbitone and δ-HCH were similarly fitted where I now represents the amplitude of the current evoked by the modulator concentration [A], Imax is the amplitude of the response in the presence of a maximally effective concentration of modulator and EC50 is the concentration of modulator producing a half-maximal response. In instances where the concentration-response relationship for agonist modulation and GABA-mimetic effects were clearly bell shaped, curve fitting (to determine EC50 and nH) was restricted to the ascending limb and apparent maximum. The illustrated curves were fitted by eye and have no theoretical significance. Quantitative data are presented as means ±s.e.m.
Drugs and reagents
γ-Aminobutyric acid (GABA), 5α-pregnan-3α-ol-20-one (5α3α), sodium pentobarbitone, δ-hexachlorocyclohexane (δ-HCH) and picrotoxin were purchased from Sigma. Propofol (2,6-diisopropylphenol) was from Aldrich. Stock solutions of GABA (10 mM) pentobarbitone (10 mM) and picrotoxin (1 mM) were prepared directly in extracellular recording medium. Over the range of pentobarbitone concentrations investigated (1 μM to 6 mM), only the highest influenced the normal pH (7.4) of the frog Ringer solution (pH 8.1). In control experiments, Ringer solution adjusted to this pH had no direct effects on wild-type or mutant RDL receptors and influenced neither the direct nor the GABA modulatory effects of this anaesthetic (see Fig. 7). The concentration of propofol was adjusted to 100 or 200 mM in ethanol. For experiments using propofol within the range 100 nM to 100 μM, the concentration of ethanol was 17 mM (i.e. 0.1 % v/v). At higher concentrations of the anaesthetic (300-600 μM) the concentration of ethanol was 51 mM (i.e. 0.3 % v/v). For all invertebrate and mammalian receptors tested, such concentrations (17-51 mM) of ethanol influenced neither GABA receptor function nor its modification by positive allosteric modulators. δ-HCH and 5α3α were dissolved in DMSO at the concentrations of 30 mM and 10 mM, respectively. Over the range of modulator concentrations studied (δ-HCH, 100 nM to 300 μM; 5α3α, 10 nM to 10 μM), the concentration of DMSO was 0.1 % v/v, which by itself had no influence upon the direct or the modulatory actions of the compounds studied. Cell culture reagents were obtained from either Gibco or Invitrogen. Analytical grade reagents were purchased from Sigma or BDH.
General properties of mammalian and invertebrate wild-type and mutant GABA receptors
The mutation of methionine 314, located within the second transmembrane domain (TM2) of RDL, to serine or asparagine has previously been shown to produce an approximately fourfold reduction of the EC50 for GABA (from 20 to 5 μM) with no effect on the associated Hill coefficient (McGurk et al. 1998; see Table 1). In a reciprocal manner, the mutation of the homologous residue (asparagine 289) within GABAA receptor β3-subunit to methionine caused a modest increase in the EC50 for GABA, from 14 μM for the wild-type α6β3γ2L receptor to 31 μM for the mutant α6β3N289 Mγ2L receptor, in the absence of any change in the Hill coefficient (McGurk et al. 1998; see Table 1). In the present study, we have further characterized the impact of these mutations on the physiological and pharmacological properties of the receptor.
Table 1. The GABA-modulatory and GABA-mimetic activities of intravenous anaesthetics acting at wild-type and mutant mammalian and invertebrate (RDL) GABA receptors expressed in Xenopus laevis oocytes
Data are listed as the means ± S.E.M. of 4–7 observations. Drug concentrations in parentheses refer to the maximum tested. Emax is the maximal effect of GABA as a percentage. n.d., not determined due to limited action or absence of clear maximal effect. Data are from †Belelli et al. 1997 and *McGurk et al. 1998.
14 ± 10.8*
1.0 ± 0.05*
8.7 ± 0.5
180 ± 26
41 ± 4(300 μM)
38 ± 2
143 ± 20
175 ± 9(nH= 2 ± 0.2)
135 ± 52
0.37 ± 0.01
90 ± 10
5 ± 1(3 μM)
0.7 ± 0.06
135 ± 7
23 ± 2
96 ± 24
31 ± 1*
0.9 ± 0.03*
17 ± 2
39 ± 2
95 ± 7
143 ± 4
10 ± 3(1 mM)
0.29 ± 0.04
125 ± 21
9 ± 7(3 μM)
11.8 ± 0.6(100 μM)
3.5 ± 0.4(600 μM)
20 ± 0.4*
1.7 ± 0.1*
25 ± 2
85 ± 6
837 ± 64
72 ± 7
25 ± 4(10 μM)
16 ± 1(600 μM)
3.1 ± 0.1
98 ± 9
4.8 ± 0.2*
1.7 ± 0.1*
5.9 ± 0.5
71 ± 3
63 ± 4(nH= 1.8 ± 0.3)
25 ± 2
394 ± 47
53 ± 2
16 ± 1(10 μM)
64 ± 3
68 ± 2
3.9 ± 0.1
56 ± 2
6 ± 2(30 μM)
4.8 ± 0.2*
2.1 ± 0.1*
1.5 ± 0.1
83 ± 5
14.6 ± 0.3(nH= 2.1 ± 0.1)
104 ± 11
179 ± 12
81 ± 6
1540 ± 46(nH= 2.8 ± 0.1)
79 ± 4
18 ± 1(10 μM)
39 ± 5(600 μM)
1.1 ± 0.05
57 ± 9
10.5 ± 0.3(nH= 3.3 ± 0.1)
79 ± 9
The local application of GABA (10 μM, 1.4 × 105 Pa, 3-20 ms, 0.033 Hz) evoked an inward current response (holding potential, Vh= -60 mV) on 44 % (n= 50), 24 % (n= 85) and 48 % (n= 70) of S2 cells previously transfected with RDL, RDLM314S or RDLM314N receptor subunit cDNAs, respectively (Fig. 1). For all receptor constructs examined, the reversal potential (Vrev) of the GABA-evoked current was close to zero millivolts (RDL, 0.8 ± 0.6 mV, n= 6; RDLM314S, 1 ± 0.5 mV, n= 7; RDLM314N, 0.9 ± 0.8 mV, n= 6) as expected for an anion selective channel under the present recording conditions ([Cl−]o= 148.8 mM; [Cl−]i= 146 mM). Similarly, the Vrev values determined for GABA acting at RDL (-34 ± 2 mV, n= 6), RDLM314S (-33 ± 3 mV, n= 6) and RDLM314N (-32 ± 2 mV, n= 5) receptors expressed in oocytes (not shown) are in accord with the equilibrium potential for chloride ions in these cells (Chen et al. 1994). The similarity in Vrev values across wild-type and mutant receptors suggests that the permeabilty of Cl− relative to other ions present is not significantly affected by the mutations.
Changing the nature of TM2 located amino acids may result in a receptor which exhibits an increased probability of spontaneous channel openings in the absence of GABA. Such spontaneously open channels may be subject to positive allosteric regulation by certain modulators, as documented for homomeric receptors formed by murine β3 GABAA receptor subunits (Wooltorton et al. 1997). Under such circumstances, the distinction between the GABA-modulatory and GABA-mimetic activities of anaesthetics is blurred. However, in the present study, the average resting conductance of stage V or VI oocytes expressing RDL (0.81 ± 0.08 μS, n= 4), RDLM314S (0.83 ± 0.09 μS, n= 6), or RDLM314N (1.15 ± 0.19 μS, n= 6) was not significantly different (two-way ANOVA, P= 0.21) from that of sham injected oocytes (1.08 ± 0.2 μS, n= 5). Similarly, the average resting membrane conductances of S2 cells expressing RDL (1.88 ± 0.56 nS, n= 6), RDLM314S (2.16 ± 0.77 nS, n= 5), or RDLM314N (2.0 ± 0.18 nS, n= 6) receptors, did not differ appreciably. Moreover, the application of picrotoxin (1 μM) to oocytes or to S2 cells expressing wild-type or mutant RDL receptors did not induce an outward current (Vh= -60 mV), or reduce the resting membrane conductance of the cell. Collectively, these observations suggest that upon expression, both wild-type and mutant invertebrate receptors have a low probability of channel opening in the absence of GABA and therefore that GABA-mimetic activity can be unambiguously discerned (see below).
Propofol (0.3-60 μM) produced a concentration-dependent enhancement of GABA-evoked currents recorded from oocytes expressing α6β3γ2L receptors with a calculated EC50 of 8.7 ± 0.5 μM and an Emax of 180 ± 26 % (n= 7; Fig. 2 and Table 1). Higher concentrations (100-300 μM) of propofol were associated with a reduced enhancement of the GABA-evoked response resulting in a bell-shaped concentration- response relationship (Fig. 2). That the peak current in the presence of a maximally effective concentration of propofol exceeded that evoked by a saturating concentration of GABA alone may imply that the latter is a partial agonist. However, it should be recognized that the kinetics of agonist application in the oocyte system are relatively slow (see Methods). Receptor desensitization during the rising phase of the inward current response to GABA would be likely to result in the truncation of the response to GABA at high concentrations, decreasing the apparent efficacy of the agonist. Whether the bell-shaped concentration-response curve to propofol (and indeed other anaesthetics, see below) can be attributed to enhanced desensitization or channel blockade cannot be resolved from the present data.
In the absence of GABA, propofol (10-300 μM) induced a concentration-dependent inward current response which we have previously demonstrated to result from the anaesthetic directly gating the GABAA receptor-channel complex (Belelli et al. 1996). At the highest concentration tested (300 μM), propofol induced a current amounting to 41 ± 4 % (n= 7) of the current induced by a saturating concentration of GABA (Fig. 2).
Propofol (EC50= 8 μM) is also a positive allosteric modulator of a splice variant form of the RDL GABA receptor (DRC 17-1-2) expressed in oocytes (Chen et al. 1994; Belelli et al. 1996). Acting at the RDL receptor, this anaesthetic is approximately threefold less potent (EC50= 25 ± 2 μM; n= 4) than at DRC 17-1-2 and human α6β3γ2L receptors and exhibits a reduced maximum effect (Emax= 85 ± 6 %) in comparison with the human receptor (Fig. 3; Table 1). In contrast to its effects on mammalian GABAA receptors, propofol (≤ 300 μM) does not directly gate the invertebrate RDL GABA receptor (Fig. 3). This result is in accord with that previously reported for the DRC 17-1-2 receptor (Belelli et al. 1996). The interaction of the general anaesthetic etomidate with both mammalian GABAA receptors and the Drosophila GABA receptor is greatly influenced by the nature of a single amino acid located within the TM2 domain, the putative channel forming region of mammalian β-subunits (β1, S290; β2, N289; β3, N289) and the RDL receptor (RDL M314) (Belelli et al. 1997; McGurk et al. 1998). In the present study, the co-expression in oocytes of α6- and γ2L-subunits with a β3-subunit in which the transmembrane asparagine residue had been mutated to methionine (the homologous residue in the RDL receptor) resulted in a receptor at which propofol as a GABA modulator was modestly less potent (EC50= 17 ± 2 μM, n= 4) and substantially less efficacious (Emax= 39 ± 2 %, n= 4) than with the wild-type (α6β3γ2L) receptor (Fig. 2 and Table 1). Furthermore, high concentrations of propofol (≤ 300 μM) did not directly gate the α6β3N289Mγ2L receptor (Fig. 2). Hence, in the latter respect, the mutant mammalian receptor now resembled the wild-type RDL receptor.
The reciprocal mutation of the methionine residue of the invertebrate receptor to an asparagine (RDLM314N) or a serine residue (RDLM314S; the homologous residue of the mammalian β1-subunit) increased the potency of propofol as a GABA receptor modulator (assessed in oocytes) by approximately 4- and 17-fold, respectively (RDLM314N, EC50= 5.9 ± 0.5 μM; RDLM314S, EC50= 1.5 ± 0.1 μM; versus RDL EC50= 25 ± 2 μM; all n= 4), but had little or no effect on the magnitude of the maximum potentiation produced (RDLM314N, Emax= 71 ± 3 %; RDLM314S, Emax= 83 ± 5 %; versus RDL Emax= 85 ± 6 %; all n= 4; see Fig. 3 and Table 1). However, these mutations produced a more fundamental change in the effects of propofol at the RDL GABA receptor. Hence, when applied to oocytes expressing mutant RDL receptors, propofol (3-300 μM) evoked concentration-dependent inward currents in the absence of GABA (Fig. 3). In the case of the RDLM314S receptor, the maximal response to propofol was comparable to that evoked by a saturating concentration (1 mM) of GABA (EC50= 14.6 ± 0.3 μM, Emax= 104 ± 11 %, nH= 2.1 ± 0.1, n= 4). The effects of propofol at the RDLM314N receptor were less pronounced, both in terms of potency and maximal effect (EC50= 63 ± 4 μM, Emax= 25 ± 2 %, nH= 1.8 ± 0.3, n= 4; see Table 1 and Fig. 3). The Hill slopes associated with anaesthetic activation of these mutant receptors suggest the co-operative involvement of at least two molecules of propofol in this effect. The ability of propofol to activate the mutant RDL receptors was confirmed in additional experiments conducted upon S2 cells. In this system, the brief local application of propofol (10 μM for RDLM314S, 100 μM for RDLM314N, 1.4 × 105 Pa, 0.017 Hz, Vh= -60 mV) evoked transient inward currents from cells directed to express either the RDLM314S or the RDLM314N receptor, whereas the wild-type receptor was refractory to activation by the anaesthetic (Fig. 4). In both oocytes and S2 cells, the anaesthetic-induced currents mediated by the mutant RDL receptors were blocked by the co-application of the GABA antagonist picrotoxin (1 μM; Fig. 5). We have previously demonstrated that the GABA-evoked current associated with the activation of the RDL receptor in oocytes is mediated by the transmembrane flux of chloride ions (Chen et al. 1994). For oocytes expressing the RDLM314S (Fig. 5) or the RDLM314N receptor (not shown), the reversal potential of the currents induced by propofol (10-100 μM) was -34 ± 0.05 mV (n= 4) and -27 ± 3 mV (n= 4), respectively. These values are very similar to those determined for the appropriate control obtained with 2 μM GABA (i.e. RDLM314S= -33 ± 3 mV, n= 6; RDLM314N= -32 ± 2 mV, n= 5; Fig. 5). Additionally, the reversal potential of propofol-evoked currents recorded from S2 cells expressing either RDLM314S or RDLM314N receptors (-0.25 ± 1 mV, n= 4 and 0.6 ± 0.6 mV, n= 4, respectively) corresponded to those of GABA-evoked currents (1 ± 0.5 mV, n= 7 and 0.9 ± 0.8 mV, n= 6, respectively) and are consistent with chloride ions mediating these anaesthetic-evoked currents (see above). Collectively, these observations suggest that propofol can directly gate the mutant receptors.
Acting at wild-type α6β3γ2L GABAA receptors expressed in oocytes, pentobarbitone (3 μM to 100 μM) produced a concentration-dependent enhancement of GABA (EC10)-evoked currents with a calculated EC50 of 38 ± 2 μM and an Emax of 143 ± 20 % (n= 7; see Fig. 6 and Table 1). Higher concentrations of pentobarbitone (300-600 μM) were associated with a reduced enhancement of the GABA-evoked response, resulting in a bell-shaped concentration- effect relationship (Fig. 6). In the absence of GABA, pentobarbitone (30 μM to 1 mM) induced a concentration-dependent inward current (Fig. 6) which we have previously demonstrated to result from the barbiturate directly gating the receptor-channel complex (Belelli et al. 1996). The calculated EC50 for the agonist effect of the barbiturate was 175 ± 9 μM (n= 7) with an Emax of 135 ± 52 % and an nH of 2 ± 0.2 indicative of co-operative binding (see Table 1).
It is already documented that pentobarbitone is also a positive allosteric modulator of the invertebrate RDL GABA receptor expressed in Xenopus laevis oocytes (Chen et al. 1994; Belelli et al. 1996). However, pentobarbitone does not directly gate the RDL receptor, a feature which differs from the well established GABA-mimetic effect of this anaesthetic at the mammalian GABAA receptor (Chen et al. 1994; Belelli et al. 1996). In the present study, the potency with which pentobarbitone enhanced GABA-evoked currents mediated by α6β3N289Mγ2L GABAA receptors expressed in oocytes was slightly reduced (EC50= 95 ± 7 μM, n= 4) compared with the wild-type receptor (Fig. 6 and Table 1). Maximal enhancement of GABA by the anaesthetic at wild-type and mutant (Emax= 143 ± 4 %) receptors was identical. However, the β-subunit mutation exerted a profound effect upon the GABA-mimetic actions of pentobarbitone. At α6β3N289Mγ2L receptors, pentobarbitone (1 mM) produced a current amounting to only 10 ± 3 % (n= 4) of that evoked by a saturating concentration of GABA. By comparison, current activation to 135 ± 52 % (n= 7) of the GABA maximum was produced by 600 μM pentobarbitone acting at α6β3γ2L receptors. Hence, in this respect this mammalian mutant GABAA receptor now more closely resembles the RDL GABA receptor.
For the RDL receptor, the exchange of the transmembrane methionine to asparagine (RDLM314N) influences the actions of propofol and etomidate, but has no effect on the actions of pentobarbitone (McGurk et al. 1998). However, the interaction of pentobarbitone with RDLM314S receptors was quite distinct from that observed for either RDL or RDLM314N receptors. This mutation enhanced the GABA-modulatory potency of pentobarbitone approximately fivefold, but had little effect on the maximum potentiation produced (RDLM314S, EC50= 179 ± 12 μM, Emax= 81 ± 6 %, n= 4; versus RDL, EC50= 837 ± 64 μM, Emax= 72 ± 7 %, n= 4; see Fig. 7 and Table 1).
Additionally, pentobarbitone acquired agonist activity, evoking concentration-dependent (30 μM to 6 mM) inward current responses which, at the highest concentrations of anaesthetic tested, were comparable in amplitude to those elicited by a saturating concentration (1 mM) of GABA (pentobarbitone EC50= 1.5 ± 0.05 mM, Emax= 79 ± 4 %, n= 4; see Fig. 7 and Table 1). As for propofol this effect was associated with a Hill coefficient greater than 1 (nH= 2.8 ± 0.01). In oocytes, such pentobarbitone-evoked currents were blocked by the co-application of picrotoxin (1 μM) and reversed in sign at a potential (-28.5 ± 2.2 mV, n= 4) similar to that of GABA-evoked responses (-33 ± 3 mV; not shown). Hence pentobarbitone can now gate this mutated receptor.
5α-Pregnan-3α-ol-20-one (5α3α) and δ-hexachlorocyclohexane (δ-HCH)
Propofol and pentobarbitone are more potent allosteric modulators of mammalian GABAA than invertebrate RDL receptors, and etomidate is inactive at the latter (McGurk et al. 1998). To determine whether the crucial residue may additionally influence the actions of other positive allosteric modulators, we quantified the actions of 5α3α, an anaesthetic neurosteroid which exhibits a clear preference for mammalian versus invertebrate receptors, and an isomer of the insecticide γ-hexachlorocyclohexane (lindane), namely δ-HCH, which we have previously demonstrated to have a greater apparent efficacy at invertebrate than at mammalian GABA receptors (Belelli et al. 1996).
Acting at wild-type α6β3γ2L receptors, 5α3α (10 nM to 3 μM) produced a concentration-dependent enhancement of currents evoked by GABA at EC10 with a calculated EC50 of 370 ± 10 nM and an Emax of 90 ± 10 % (n= 7, see Table 1). By contrast, we have previously reported that this steroid is much less effective at the RDL receptor, with the relatively high concentration of 10 μM increasing the amplitude of the current induced by an EC10 concentration of GABA to only 25 % of the GABA maximum (McGurk et al. 1998; Table 1). However, mutation of the β3 asparagine to the homologous methionine residue of the RDL receptor had little effect on the GABA-modulatory effects of the neurosteroid at α6β3N289 Mγ2L receptors (EC50= 290 ± 40 nM, Emax 125 ± 21 %, n= 5). Similarly, the mutation of the transmembrane methionine of the RDL receptor to asparagine or serine did not influence the modest GABA-enhancing actions of 10 μM 5α3α (McGurk et al. 1998 and Table 1).
δ-HCH enhances GABA at both mammalian and invertebrate GABA receptors, but is not GABA mimetic (Belelli et al. 1996). In the present study, δ-HCH enhanced GABA (EC10)-evoked currents recorded from oocytes expressing the RDL receptor with an EC50 of 3.1 ± 0.1 μM and an Emax of 98 ± 9 % (n= 3, Fig. 8). These values are similar to those reported for the DRC 17-1-2 receptor and hence this compound does not discriminate between the splice variants of the invertebrate receptor (Belelli et al. 1996). Relatively high concentrations of δ-HCH did not induce an inward current.
Mutation of the transmembrane methionine to a serine residue produced a dramatic change in the effects of δ-HCH acting at the Drosophila receptor. Concentrations of δ-HCH ≤ 1 μM acting upon oocytes expressing the RDLM314S receptor evoked a concentration-dependent (1-300 μM) inward current response (EC50= 10.5 ± 0.3 μM, nH= 3.3 ± 1.1; Emax= 79 ± 9 %, n= 3) that was completely blocked by picrotoxin (10 μM) suggesting that δ-HCH has a GABA-mimetic action (Fig. 8). Similar to the effects of propofol and pentobarbitone (Figs 3 and 7), the direct activation of this receptor by δ-HCH was associated with a steep Hill coefficient, indicative of channel activation requiring more than one drug molecule. The acquisition of the GABA-mimetic activity by pentobarbitone and propofol acting at the mutant RDL receptors was accompanied by an increased apparent affinity of these anaesthetics for the ‘GABA-modulatory site’. However, by contrast this mutation had little effect on the GABA-modulatory actions of δ-HCH (RDL wild-type, EC50= 3.1 ± 0.1 μM, Emax= 98 ± 9 %, n= 3; RDLM314S, EC50= 1.1 ± 0.05 μM, Emax= 57 ± 9, n= 3), the modest change in EC50 and the reduction of the apparent Emax being a consequence of the juxtaposition of the GABA-modulatory and GABA-mimetic actions (Fig. 8).
The GABAA modulatory EC50 of δ-HCH acting at the RDLM314N receptor was little changed (EC50= 3.9 ± 0.1 μM, n= 3) versus the wild-type RDL receptor, although the maximum potentiation produced was reduced (Emax= 56 ± 2 %, n= 3; Fig. 8 and Table 1). Acting at oocytes expressing RDLM314N receptors, δ-HCH, in the absence of GABA, induced a small inward current (for 30 μM δ-HCH, Emax= 6 ± 2 %, n= 3) which was completely blocked by the co-application of 1 μM picrotoxin (Fig. 8 and Table 1). Hence, δ-HCH can directly activate this form of the receptor but with a much reduced apparent efficacy versus the RDLM314S receptor.
A number of general anaesthetics exhibit both GABA-modulatory and GABA-mimetic actions at the GABAA receptor. The distinct concentration dependencies of these two effects may be explained by the presence of high (GABA-modulatory) and low (GABA-mimetic) affinity anaesthetic binding sites on the GABAA receptor protein (Harris et al. 1995). Consistent with this concept, the GABA-modulatory, but not the GABA-mimetic, effects of certain anaesthetics are reduced by the introduction of a δ- or ε-subunit into binary complexes of αβ-GABAA subunits (Zhu et al. 1996; Davies et al. 1997; cf. Whiting et al. 1997). Similarly, α4-subunit-containing receptors support GABA potentiation by propofol and pentobarbitone, but not direct receptor activation by these anaesthetics (Wafford et al. 1996). Further support for the existence of distinct modulatory and activation sites derives from observations made on the Drosophila GABA receptor, where the anaesthetics pentobarbitone and propofol clearly enhance the actions of GABA but are not GABA mimetic (Chen et al. 1994; Belelli et al. 1996).
The intriguing question remains as to how a variety of general anaesthetics, which exhibit little or no commonality in chemical structure to GABA, or between themselves, can nevertheless activate the GABAA receptor-channel complex. The results of interaction studies with binary combinations of anaesthetics, such as steroids and barbiturates (Belelli et al. 1996), indicate that despite sharing a common channel-gating effect, such agents may bind to distinct non-overlapping sites. Certainly, the direct gating effect occurs through a site different from that occupied by GABA, since numerous anaesthetics potentiate, rather than occlude, the binding of agonist ligands to the GABAA receptor (reviewed by Tanelian et al. 1993; Sieghart, 1995). In addition, mutations of the β-subunit that greatly diminish the potency of GABA have no effect upon the direct effects of pentobarbitone (Amin & Weiss, 1993). Moreover, the expression of β1- or β3-subunits in isolation results in the formation of homo-oligomeric receptor-channel complexes that are largely unresponsive to GABA, yet are activated by general anaesthetics that include etomidate, propofol and pentobarbitone (Sanna et al. 1995; Krishek et al. 1996; Wooltorton et al. 1997; reviewed in Lambert et al. 1997). However, the interpretation of those experiments is complicated by the fact that, in many cases, β-homomeric receptors appear spontaneously active in the absence of GABA (Krishek et al. 1996; Wooltorton et al. 1997).
The cardinal finding of the present study is that the direct gating of both a mammalian GABAA receptor (α6β3γ2L) and an invertebrate GABA receptor (RDL) by the structurally unrelated anaesthetics pentobarbitone and propofol is influenced in a complementary manner by a single transmembrane amino acid. Hence, these anaesthetics enhance the actions of GABA at both receptors, but only directly activate the chloride channel of the mammalian receptor (Belelli et al. 1996). However, the mutation of a single transmembrane amino acid residue (asparagine 289) of the β3-subunit to methionine (the homologous residue of the RDL receptor) abolished direct activation by propofol and substantially reduced the GABA-mimetic activity of pentobarbitone. In addition, the mutation reduced the GABA-modulatory effects of both pentobarbitone and propofol. The effects of this mutation could, of course, be non-specific, causing changes in the secondary, tertiary, or quaternary structure of the protein. Compelling evidence against such an interpretation is provided by the properties of the RDL receptor, where the transmembrane methionine residue was changed to the equivalent asparagine of the β2 and β3, or the serine of the β1-GABAA receptor subunit. Acting at RDLM314S receptors, either propofol or pentobarbitone, in the absence of GABA, directly activated the receptor channel complex in a picrotoxin-sensitive manner. This fundamental change was not restricted to general anaesthetics, as δ-HCH also directly activated the RDLM314S receptor, but was inert in this respect at the wild-type receptor. Indeed, the maximal current produced by these agents was similar to that evoked by a saturating concentration of GABA. Furthermore, analysis of their concentration-response relationships revealed Hill slopes substantially greater than 1, suggesting that co-operative drug binding was required to induce channel opening. Propofol and δ-HCH also activated the RDLM314N receptor-channel complex, although with a reduced maximal effect compared with that observed at the RDLM314S receptor. By contrast, pentobarbitone does not activate the RDLM314N receptor (McGurk et al. 1998). Interestingly, the appearance of the GABA-mimetic effects of propofol and pentobarbitone was also accompanied by an increase in the potency of these anaesthetics in enhancing the actions of GABA, suggesting these two effects are coupled, whereas the serine or asparagine mutation (RDLM314S, RDLM314N) revealed a GABA-mimetic action of δ-HCH, but with little effect on GABA-modulation. Should mutation of the TM2-located amino acid have resulted in an RDL receptor with an increased propensity for spontaneous channel openings, then the direct effects of these agents might merely reflect the promotion of background channel opening. Such a mechanism may contribute to the apparent direct effects of anaesthetics reported for homo-oligomeric receptors composed of β1- or β3-subunits (Krishek et al. 1996; Wooltorton et al. 1997). However, in the present study, the resting membrane conductance of both oocytes and S2 cells was not influenced by the expression of RDLM314S and RDLM314N receptors in comparison with the wild-type receptor. Furthermore, picrotoxin was without effect on such cells (cf. Krishek et al. 1996; Wooltorton et al. 1997) and therefore the direct gating of the channel by the anaesthetics does not result from this mechanism.
Therefore, the nature of this TM2-located amino acid exerts a considerable influence on the allosteric effects of at least three general anaesthetics. For etomidate, positive allosteric actions at both mammalian and invertebrate receptors are favoured by asparagine, reduced for serine and absent for methionine (Belelli et al. 1997; McGurk et al. 1998; see Table 1). By contrast, for propofol and pentobarbitone acting at RDL receptors, activity is favoured by serine, reduced for asparagine, but again is least active for the methionine residue. Similarly, the actions of these anaesthetics at mammalian receptors are limited by the methionine residue. However, propofol and pentobarbitone do not discriminate between β1 (serine)- and β2- or β3 (asparagine)-containing receptors (Hill-Venning et al. 1997) and hence the preference for the serine residue exhibited with the invertebrate receptor does not extend to the mammalian receptor.
For GABAA and glycine receptor α-subunits, residues occupying an equivalent position in TM2 exert a profound influence upon allosteric regulation by both volatile and intravenous anaesthetic agents. For the glycine receptor α1-subunit, modulation of receptor function by ethanol appears to be partly determined by the molecular volume of the amino acid which occupies this crucial position, potentiation being favoured by small amino acids and inhibition by relatively bulky residues (Ye et al. 1998). Whether the size of the amino acid residue is also an important determinant of anaesthetic action at the GABAA receptor has not been systematically investigated to date. The allosteric effects of general anaesthetics at both GABAA and glycine receptors are additionally influenced by the nature of an amino acid residue within TM3, which again is thought to be located near to the extracellular aspect of the membrane (Mihic et al. 1997; Krasowski et al. 1998a,b). Of particular interest is the observation that expression of the GABAAα2-subunit with a β1-subunit in which the methionine residue in TM3 of the wild-type protein has been exchanged for a tyrosine residue results in a receptor which is insensitive to the GABA potentiating effects of propofol, but the direct-gating of the receptor channel complex by the anaesthetic is unchanged (Krasowski et al. 1998a). Therefore, the GABA-modulatory and GABA-mimetic actions of propofol are dramatically influenced by amino acid residues in both TM2 and TM3.
In aggregate, the actions of a number of chemically disparate compounds acting at GABAA, RDL and glycine receptors are influenced by the nature of two transmembrane (TM2 and TM3) amino acids. This permissiveness of both receptors and drugs would appear incompatible with the identified amino acids contributing to the anaesthetic binding site and favours these residues playing a key role in transduction. In this respect it is of interest that mutation of a conserved leucine to a threonine residue in the TM2 domain of the nicotinic α7-subunit produced a dramatic change in the pharmacology of this transmitter-gated ion channel, since a number of structurally distinct antagonists acquire agonist character (Bertrand et al. 1992; Palma et al. 1998). However, it is important to note that the actions of steroidal anaesthetics at GABAA and RDL receptors are not affected and therefore not all classes of anaesthetics are similarly influenced (Belelli et al. 1997; McGurk et al. 1998; Rick et al. 1998).
In conclusion, both the GABA-modulatory and GABA-mimetic effects of two structurally distinct general anaesthetics at both mammalian and invertebrate GABA receptors are greatly influenced by the nature of an amino acid located within TM2. However, whether or not this amino acid participates directly in the binding of the anaesthetic to the receptor protein remains to be clarified. In this respect, two recent reports highlighting the importance of the N-terminal extracellular domain of the GABAA receptor α-subunit for the GABA-modulatory effects of propofol (Uchida et al. 1997) and the GABA-mimetic actions of pentobarbitone (Fisher et al. 1997) may prove to be pertinent. In addition to GABA and glycine receptors, key amino acids which dictate the interaction of certain anaesthetics with glutamate (Minami et al. 1998) and nicotinic receptors (Forman et al. 1995) have also been identified. Collectively, these observations demonstrate that far from being indiscriminate, the interaction of anaesthetics with transmitter-gated ion channels is highly specific. In the future, the generation of mice carrying such anaesthetic resistant mutations should permit a better evaluation of the molecular targets that mediate the behavioural effects of general anaesthetics.
We are grateful to Drs P. Whiting and R. Roush for providing the human and Drosophila GABA receptor cDNAs, respectively. This work was supported by grants from the MRC (M. P. and D. B) and Organon Teknika (K. M.).