Glycine Receptor β Subunits Play a Critical Role in Potentiation of Glycine Responses by ICS-205,930

Materials and Methods

Heterologous Expression of GlyRs in X. LaevisOocytes.

Human GlyR α1, α2, α1-G254A, and rat β subunit cDNAs (generous gifts from H. Betz's laboratory; Grenningloh et al., 1990a,b; Bormann et al., 1993) were transcribed in vitro using the appropriate (T7, T3, or SP6) mMessage-mMachine kits (Ambion, Austin, TX). cRNA (50 nl) were injected into X. laevis oocytes using a nanoliter injector (World Precision Instruments, Sarasota, FL) to reach a final amount of cRNA close to 50 ng in the case of homomeric GlyRs. For the expression of heteromeric GlyRs, the cRNA for the α and β subunits were mixed in a ratio of 1:4 and the final amount of injected cRNA coding for the α subunit was between 25 and 35 ng.

Defolliculated oocytes were isolated from X. laevis ovaries after 1 h of shaking incubation in a Ca²⁺-free medium (84.5 mM NaCl, 5 mM HEPES, and 1 mM MgCl₂; pH adjusted to 7.6 with KOH) containing 1 mg/ml of collagenase type II (GIBCO, Grand Island, NY). Oocytes were kept at 19°C in individual wells containing 200 μl of Barth's medium (88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO₃, 0.82 mM MgSO₄, 0.33 mM Ca(NO₃)₂, 0.41 mM CaCl₂, 10 mM HEPES, pH 7.6 adjusted with NaOH) with 50 μg/ml of gentamycin (GIBCO). Experiments were performed at room temperature (19–24°C), 3 to 5 days after cRNA injection.

Recording and Data Analysis.

Using a two-electrode voltage-clamp and a OC-725C amplifier (Warner Instrument, Hamden, CT), currents were recorded in oocytes held at −70 mV. Oocytes were rapidly superfused using a closed chamber as described in (Supplisson and Bergman, 1997). All tubing was made of Teflon and changes in solution were achieved using a computer-controlled motorized valve (Omnifit, Cambridge, UK). The extracellular recording solution contained 100 mM NaCl, 1.8 mM CaCl₂, 1 mM MgCl₂, 5 mM HEPES, and its pH was adjusted to 7.4 with KOH. Repetitive applications of Gly in the absence of any modulator controlled the stability of the recording. The interval between successive Gly applications was chosen to allow complete recovery from desensitization, and was usually 40 s for Gly concentrations ≤50 μM and 90 or 120 s for higher Gly concentrations. When several Gly concentrations were tested on the same oocyte, the Gly concentration was increased progressively, to avoid possible irreversible desensitization of the receptors by the highest Gly concentrations. Experiments showing cumulative desensitization during repetitive applications were eliminated. The currents were filtered at 20 Hz and digitized at 100 Hz using a Digidata 1200A or 1320 and pClamp 6-8 software (Axon Instrument, Foster, CA). The leakage currents were subtracted on display. Glycine concentration response curves (Fig. 1) were fitted with Kaleidagraph (Abelbeck/Synergy Software, Reading, PA), using the equation:Equation 1R is the amplitude of the response, and the parameters R_max, EC₅₀, andn_H are freely determined by the program. To clearly illustrate the bottom of these curves, the data and their fit are presented using logarithmic scales.

Drugs.

A fresh stock solution of ICS (3-tropanyl-indole-3-carboxylate hydrochloride; RBI, Sigma) was prepared in distilled water at 1 mM and was kept on ice during the afternoon for the experiments using concentrations lower than 20 μM. For the experiments using higher concentrations of ICS, a 1 mM stock solution was prepared directly in the extracellular saline containing Gly. Picrotoxin (Sigma, St. Louis, MO) was diluted at 50 mM in ethanol. For testing the effect of picrotoxin, the same final concentration of ethanol (6/10,000) was present in all the solutions and did not affect the response. A given stock solution of Gly (Sigma), prepared at 100 mM in distilled water and kept at −20°C, was used for several months.

Results

Expression of Homomeric and Heteromeric GlyRs in X. laevis Oocytes: Glycine Sensitivity and Picrotoxin Sensitivity.

To study modulations of responses to low concentrations of Gly, we expressed a high density of GlyRs (maximum response >10 μA). Such an expression would be expected to approach the situation encountered in native neurons, where the GlyRs are packed at high density in clusters (Béchade et al., 1996). Concentration-response curves obtained for homomeric GlyRs, constituted by either α1 or α2 human subunits, are illustrated in Fig. 1, A and B, respectively. A complete curve was recorded from a single oocyte (see under Materials and Methods) and fitted with a Hill equation (see figure legend for the values derived from each individual fit). The current traces illustrated in Fig. 1E (obtained from the oocyte expressing α1 GlyRs) show that Gly concentrations higher than 100 μM induced some rapid desensitization during applications of only 4 s duration.

Similar experiments were performed in oocytes expressing heteromeric GlyRs, after injection of a mixture of cRNAs coding for the rat β subunit and the human α1 or α2 subunit (see under Materials and Methods). The efficacy of the method was checked by measuring the picrotoxin sensitivity of the response to 100 μM Gly. Table1 gives the mean percentages of reduction induced by 10 μM and 30 μM picrotoxin, for oocytes injected only with α subunits (α1 or α2) cRNA, and for oocytes coinjected with β cRNA. The picrotoxin sensitivity seen in oocytes coinjected with β cRNA was significantly lower from that of the corresponding homomeric α GlyRs (P ≤ .001), indicating that heteromeric GlyRs were successfully expressed. Glycine responses recorded in the absence or presence of 10 μM picrotoxin from oocytes expressing homomeric or heteromeric GlyRs are illustrated in Fig. 1F. Glycine concentration-response curves obtained for heteromeric α1/β GlyRs and α2/β GlyRs are illustrated in Fig. 1, C and D, respectively. Table 2 gives the mean values of the Gly EC₅₀ and the Hill coefficient, calculated for each type of GlyR by averaging the results obtained from the fit of each individual concentration-response curve.

Modulation of Homomeric α1 GlyRs by ICS.

To maximize potentiation of Gly responses by ICS, most experiments were performed using low concentrations of Gly (≤20 μM), as was the case in the previous evaluation of this modulation in spinal neurons (see also Fig.3).

In the experiments performed here using human homomeric α1 GlyRs, no potentiation was ever detected using concentrations of ICS lower than 0.4 μM, in contrast to previous findings from rat neuronal receptors. ICS concentrations ranging from 1 to 40 μM did, however, potentiate the response of homomeric α1 GlyRs evoked by 10 to 20 μM Gly. In the experiment of Fig. 2A, as in most other experiments, ICS was applied only in the presence of Gly (15 μM). The potentiating effect of ICS was rapidly established, rapidly reversible, and increased with increasing concentrations of ICS up to 20 μM (Fig. 2, A, B, and D). When the ICS concentration was further increased (from 20 μM to 200 μM), the potentiation was progressively replaced by an inhibition, which was also rapidly established and rapidly reversible (Fig. 2C). Figure 2D gives the mean percentage of modulation of responses to 15 μM Gly as a function of the ICS concentration. The potentiation induced by a given concentration of ICS (10 μM) decreased with increasing concentrations of Gly and was no longer detectable when using 100 μM Gly (Fig.3).

On a given oocyte, we compared the potentiations of Gly responses induced by brief applications of ICS (without preincubation, as in Fig.2) with those induced by continuous applications of ICS (between and during successive Gly applications). The results (not shown) were not significantly different: the effective range of ICS concentrations was unchanged (ICS concentrations tested with both protocols on a same oocyte: 0.4, 1, and 10 μM) and ICS did not affect the current recorded in the absence of Gly.

Modulation of Homomeric α2 GlyRs.

In the case of human homomeric α2 GlyRs, no potentiation of Gly responses was observed, whatever the ICS or Gly concentrations tested (between 0.04 and 20 μM for ICS, between 7.5 and 50 μM for Gly). For example, we tested the effects of 0.04 to 1 μM ICS on the responses to low Gly concentrations (between 7.5 and 20 μM) in 16 different oocytes and observed either no modulation or, at 1 μM, a very small inhibitory effect (Fig. 4, A and D). Above 1 μM, ICS clearly reduced Gly responses in a concentration-dependent manner (Fig. 4, A, B, and D). This inhibitory effect was rapidly reversible (Fig. 4A) and was independent of the method used to apply ICS, without or with preincubation [compare Fig. 4B, middle traces, with Fig. 4C, left traces, from different oocytes; result confirmed by comparison of the effects induced without and with preincubation on a given oocyte (not shown)]. As illustrated by Fig. 4, C and E, the inhibitory effect of ICS increased slightly with increasing Gly concentrations between 10 and 50 μM.

Modulation of Heteromeric GlyRs.

The modulation by ICS of the responses of heteromeric GlyRs to 20 μM Gly was studied for ICS concentrations ranging from 20 nM to 20 μM. Figure5, A, B, and E, describes the results obtained from oocytes expressing α1 and β subunits, whereas Fig. 5, C, D, and F, describes the results obtained from oocytes expressing α2 and β subunits. Low concentrations of ICS, below 1 μM, potentiated the responses of both types of heteromeric GlyRs: for example, 20 nM ICS induced a potentiation of 20.3% ± 4.5% (n = 8) (mean ± S.E.M.; maximum 33.4%) in oocytes expressing α1 and β subunits and a potentiation of 12.2% ± 4.2% (n = 3) (mean ± S.E.M.; maximum 20.5%) in oocytes expressing α2 and β subunits. For both types of heteromeric GlyRs, the potentiation, which became more pronounced when the ICS concentration was increased, was maximal between 0.2 and 1 μM ICS (see Fig. 5, E and F, for the mean values of the percentage of potentiation; the maximum values of the percentage of potentiation induced by 0.2 and 1 μM ICS were 137% and 117.8%, respectively, for α1/β and 67.1% and 126.0%, respectively, for α2/β; see figure legend for the variability of the results). Increasing the ICS concentration above 1 μM revealed an opposite effect observed more easily in oocytes expressing α2 and β subunits than in oocytes expressing α1 and β subunits. In the experiment illustrated by Fig.5, A and B (α1/β), potentiation was no longer detected when the concentration of ICS was increased to 20 μM, whereas a potentiation was transiently detected (Fig. 5A, arrow) during the washout of this high concentration (which can be understood because of the potentiating effect of submicromolar ICS concentrations on α1/β GlyRs). In the experiment illustrated by Fig. 5, C and D (α2/β), the percentage of potentiation was close to maximum already for 0.1 μM ICS, and decreased between 1 and 4 μM; a net inhibition of the Gly response was induced by 10 and 20 μM ICS, whereas a potentiation was observed during the washout of these high ICS concentrations (see arrows in Fig.5C). Comparison of the mean results obtained from all the oocytes expressing heteromeric α1/β GlyRs (Fig. 5E) or α2/β GlyRs (Fig.5F) with the results obtained from oocytes expressing the respective homomeric GlyRs clearly demonstrates the influence of β subunits on the modulation of Gly responses by ICS. For α1-containing GlyRs, β subunits markedly shift the ICS concentration range inducing potentiation toward lower values. For α2-containing GlyRs, β subunits allow potentiation by submicromolar concentrations of ICS, modulation never occurring in the absence of β subunits.

As already shown for homomeric α1 GlyRs (Fig. 3), the potentiation of Gly responses by a given concentration of ICS decreased with increasing Gly concentrations both in oocytes expressing heteromeric α1/β GlyRs (Fig. 5G) and in oocytes expressing heteromeric α2/β GlyRs (Fig. 5H).

The Difference in ICS Sensitivity between Homomeric α1 and α2 GlyRs Does Not Result from Their Difference in Transmembrane M2 Segments.

In the case of GlyRs, as in the case of other ligand-gated channels of the same family, the second transmembrane segment (TM2) of each subunit, lining the ionic channel, is known to be involved in several types of modulations of the activity, in particular in the potentiation of Gly responses by alcohols and volatile anesthetics (see Discussion, below). Furthermore, the agents previously known to discriminate between different types of GlyRs, picrotoxinin and cyanotriphenylborate, were reported to act via TM2 segments (Pribilla et al., 1992; Rundström et al., 1994). The TM2 segments of α1 and α2 GlyR subunits differ only by one amino acid (the Gly-254 of α1, which is an alanine at the equivalent position in α2; Grenningloh et al., 1990b). We tested the effect of ICS on homomeric GlyRs obtained from the mutant subunit α1G254A (i.e., receptors having the same TM2 segments as homomeric α2 GlyRs). These GlyRs kept the ICS sensitivity of wild-type homomeric α1 GlyRs. Their response to 15 μM Gly was reversibly potentiated by supramicromolar concentrations of ICS (data not shown) and the mean percentages of potentiation [32.8% ± 2.4%, 66.1% ± 3.7%, and 74.2% ± 4.2% for 4, 10, and 20 μM ICS, respectively (mean ± S.E.M.,n = 3)] were not significantly different from those obtained with homomeric wild-type α1 GlyRs (see Fig. 2).

Discussion

We have shown that recombinant GlyRs can be differentially modulated by ICS, depending on their subunit composition. Homomeric GlyRs formed from human α1 (Figs. 2 and 3) or α2 (Fig. 4) subunits were both insensitive to submicromolar ICS concentrations and were modulated in opposite directions by 1 to 20 μM ICS: α1 GlyRs were potentiated whereas α2 GlyRs were inhibited. The modulation of α1 GlyRs by ICS was biphasic, with an inhibition being induced above 100 μM ICS (Fig. 2, C and D; see also Maksay et al., 1999); this suggests the presence of two binding sites for ICS, the site of higher affinity accounting for the potentiation. In contrast with the situation encountered with homomeric GlyRs, in oocytes expressing heteromeric GlyRs, a potentiation of Gly responses was evoked by submicromolar ICS concentrations; in addition, a reverse modulation developed for higher ICS concentrations (Fig. 5). The difference in ICS sensitivity between homomeric and heteromeric GlyRs cannot be attributed to possible slight differences in Gly sensitivity (see Fig. 1 and references below), nor to differences in experimental conditions (very low Gly concentrations were used, especially for homomeric GlyRs, which might even have favoured their potentiation). Thus, ICS seems to be a useful tool to discriminate α1 from α2 homomeric GlyRs, and homomeric GlyRs from heteromeric GlyRs. In our experiments, the successful expression of heteromeric α1/β and α2/β GlyRs was demonstrated by their low picrotoxin sensitivity and high ICS sensitivity. The inhibitory effect of supramicromolar ICS concentrations might also seem weaker for oocytes expressing α1 and β rather than α2 and β (Fig. 5, E and F). However, this is likely to result from the simultaneous expression of homomeric GlyRs (differentially modulated according to the α subunit type). This probably led to an underestimation of the potentiation of both types of heteromeric GlyRs by low ICS concentrations (which do not affect the fraction of the response carried by homomeric GlyRs). Note that difficulties in successfully expressing heteromeric GlyRs in X. laevis oocytes have already been encountered (Downie et al., 1996).

In the previous report concerning modulations by ICS of recombinant GlyRs in X. laevis oocytes (Maksay et al., 1999), only the inhibitory effect of high concentrations of ICS was observed; neither potentiation of homomeric α1 GlyRs by micromolar ICS concentrations nor potentiation of heteromeric GlyRs by submicromolar ICS concentrations were detected. Because the percentage of potentiation induced by a given ICS concentration clearly decreases with increasing Gly concentrations (Fig. 3; see also Fig. 7 in Chesnoy-Marchais, 1996), this discrepancy between the two studies can be explained by the difference in Gly concentrations used. We mainly used concentrations ≤20 μM, whereas Maksay et al. (1999) used half-saturating Gly concentrations that, in their experiments, were between 177 ± 19 μM and 807 ± 240 μM depending on the receptor type expressed. The pure inhibitory effect of ICS that we observed on homomeric α2 GlyRs was noncompetitive (Fig. 4E), in contrast to the previously reported inhibitory effect of atropine, another tropeine (Maksay et al., 1999).

Our results concerning heteromeric GlyRs agree qualitatively with previous results obtained from rat ventral spinal neurons (Chesnoy-Marchais, 1996) or purified motoneurons (Lévi et al., 1999). It was known that β subunits are involved in the interaction between GlyRs and gephyrin (Meyer et al., 1995), that they influence the elementary conductance states of the Gly-gated channels (Bormann et al., 1993) and lower their picrotoxinin sensitivity (Pribilla et al., 1992). However, all the amino acids affecting agonist binding to GlyRs belong to α subunits (Becker and Langosch, 1998), and the participation of β subunits to the gating process was never clearly demonstrated. There is no β subunit mutation that has been reported to affect GlyR function. The Gly EC₅₀ is usually similar for homomeric and heteromeric GlyRs (Fig. 1; Pribilla et al., 1992; Takagi et al., 1992; Bormann et al., 1993; Rundström et al., 1994; Pistis et al., 1997). Furthermore, β subunits did not significantly affect the sensitivity of α1-containing GlyRs to various anesthetics (propofol, pentobarbitone, etomidate, and trichloroethanol; Pistis et al., 1997). By showing that coexpression of β subunits with α1 subunits lowers by about 2 orders of magnitude the threshold concentration of ICS required for potentiation of Gly responses, we demonstrate the involvement of β subunits in the transduction process between ligand binding and channel opening. Furthermore, the requirement of β subunits for potentiation by ICS of α2-containing GlyRs suggests that these subunits contribute to the high-affinity site recognizing this new class of glycinergic potentiator. The influence of β subunits on the sensitivity of GlyRs to ICS, contrasting with their lack of influence on the sensitivity to ethanol (Valenzuela et al., 1998) and various anesthetics (Pistis et al., 1997), confirms the difference between the mechanisms of potentiation involved (Chesnoy-Marchais, 1999).

The sequences of the human α GlyRs used here are almost identical in the extracellular domain to the corresponding rat α variants (Grenningloh et al., 1990b; Kuhse et al., 1991; Malosio et al., 1991). Thus, comparison of the present results with those obtained from rat embryonic spinal neurons (Chesnoy-Marchais, 1996; Lévi et al., 1999) confirms that these neurons express heteromeric GlyRs after 1 to 2 weeks in culture (Tapia and Aguayo, 1998; Lévi et al., 1999).

The potentiating effect of ICS on recombinant GlyRs reported here is significant but its magnitude is small (maximum a doubling of the response to 15–20 μM Gly); it is relatively smaller than in neurons, because the Gly concentrations used induced a smaller fraction of the maximum response in oocytes than in neurons. Quantitative differences in pharmacological properties between heterologous GlyRs in X. laevis oocytes and spinal GlyRs were already known. For example, the Gly EC₅₀ are higher in oocytes (Fig. 1;Grenningloh et al., 1990b; Pistis et al., 1997; Mascia et al., 1998;Maksay et al., 1999) than in neurons (EC₅₀ ≤ 40 μM; Chesnoy-Marchais, 1996; Downie et al., 1996; Tapia and Aguayo, 1998; Lévi et al., 1999). The receptor density might partly account for this difference (Taleb and Betz, 1994). We expressed the receptors at a very high density and tried to superfuse the oocyte as rapidly and homogenously as possible. However, the size of the oocytes and their deeply folded membrane remain inappropriate to the study of responses to high, desensitizing, Gly concentrations. Other factors may also explain differences between results obtained from oocytes or neurons. For example, phosphorylation of GlyRs by protein kinase C controls their modulation by ethanol (Mascia et al., 1998) and the degree of modulation by ethanol of a given GlyR type depends on the expression system (Valenzuela et al., 1998). Differences in phosphorylation might account for differences between modulations observed in different cell types.

Identification of the amino acids involved in the potentiation of heteromeric GlyRs by low ICS concentrations could help in the design of more specific glycinergic potentiators. The modulations induced by other glycinergic potentiators, alcohols, and volatile anesthetics, are known to involve a serine conserved in TM2 segments of Gly and GABA_A receptors (Mihic et al., 1997; seeKrasowski and Harrison, 1999 for review of the role of TM2 segments in the potentiations induced by various anesthetics). This serine might participate in the binding site of these potentiators or influence transduction processes [it also affects agonists EC₅₀ (Ueno et al., 1999) and modulations by molecules unlikely to all bind to the same site (Belelli et al., 1999)]. TM2 segments affect transduction processes for all the ligand-gated channels of the GlyRs superfamily [see Galzi et al. (1996) for nicotinic receptors TM2 mutants showing altered isomerization constants between inactive and active states without modification of binding constants; see Moorhouse et al. (1999) for an α1 GlyR TM2 mutant showing a decreased function and normal ligand binding; see also Boileau and Czajkowski (1999) for GABA_A receptors TM2 mutations affecting modulation by benzodiazepines]. Up to now, the only agents known to discriminate between different GlyR types, picrotoxinin, and cyanotriphenylborate, recognize differences between TM2 segments of these receptors (Pribilla et al., 1992; Rundström et al., 1994). In contrast, we have shown that TM2 segments are not responsible for the striking difference in ICS sensitivity between homomeric α1 and α2 GlyRs (α1G254A GlyRs, having α2 TM2, showed the ICS sensitivity of wild-type α1 GlyRs). The binding site responsible for the potentiating effect of ICS on GlyRs (of much higher apparent affinity for heteromeric GlyRs than for α1 GlyRs, undetected for α2 GlyRs) is likely to involve some residues outside TM2, different on each subunit.

The binding of 5-HT₃ antagonists on 5-HT₃ receptors is known to involve a glutamate (E106 in mouse 5-HT₃ A_Lreceptors; Boess et al., 1997) and several tryptophan residues, probably involved in cation-Π interactions with the positive amine in the tropane of these antagonists (Venkataraman et al., 1999;Yan et al., 1999; Spier and Lummis, 2000). Interestingly, in the GlyR α1, α2, and β subunits, a glutamate is aligned with the E106 of 5-HT₃ A_L, and most residues aligned with the critical tryptophans are either tryptophan or phenylalanine, also able to participate in cation-Π interactions. Thus, the high-affinity binding site of ICS on GlyRs is likely to be extracellular and may resemble its site on 5-HT₃receptors. Sequence comparisons and chimera analysis should facilitate identification of the residues responsible for the differential ICS sensitivity of the different GlyR types. Potentiation by ICS of α1 homomeric and heteromeric GlyRs could both be explained by binding of ICS at the interface between subunits, as described for benzodiazepines on GABA_A receptors (Sigel and Buhr, 1997).