Molecular determinants of proton modulation of glycine receptors.

Extracellular pH regulates glycine receptors through an unknown mechanism. Here we demonstrate that acidic pH remarkably inhibited glycine-activated whole-cell currents in recombinant glycine alpha1 and alpha1beta receptors transiently expressed in human embryonic kidney 293 cells. The proton effect was voltage-independent and pharmacologically competed with glycine receptor agonist glycine and antagonist strychnine. Using site-directed mutagenesis, we have identified an N-terminal domain that is essential for proton-induced inhibition of glycine current. In alpha1 homomers, removal of the hydroxyl group by mutation of residue Thr-112 to Ala or Phe abolished inhibition of glycine currents by acidification. In contrast, mutation of Thr-112 to another hydroxylated residue (Tyr) produced receptors that retained partial proton sensitivity. In alpha1beta heteromers, a single mutation of the beta subunit T135A, which is homologous to alpha1 Thr-112, reduced proton sensitivity, whereas the double mutation alpha1(T112A)beta(T135A) almost completely eliminated the proton sensitivity. In addition, the mutation alpha1 H109A greatly reduced sensitivity to protons in homomeric alpha1 receptors. The results demonstrate that extracellular pH can regulate the function of glycine alpha1 and alpha1beta receptors. An extracellular domain consisting of Thr-112 and His-109 at the alpha1 subunit and Thr-135 at the beta subunit plays a critical role in determining proton modulation of glycine receptor function.

Glycine is a fast inhibitory neurotransmitter in the mammalian central nervous system. It acts by binding to Cl Ϫ conducting glycine receptors that belong to a superfamily of ligand-gated ion channels including nicotinic acetylcholine, ␥-aminobutyric acid type A, and 5-hydroxytryptamine type 3 receptors. Glycine receptors are pentameric membrane proteins composed of ␣1-4 and ␤ subunits. Only the ␣ subunits can form functional homomeric receptors that contain major determinants of agonist and antagonist binding. Homomeric glycine receptors (primarily ␣ 2 ) appear to be expressed during embryonic and early postnatal development and might serve as extrasynaptic receptors throughout maturity (1,2). Adult postsynaptic glycine receptors are composed of ␣1-4 and ␤ subunits that form functional heteromeric receptors (3␣:2␤) and are located in spinal cord, brainstem, and many regions in brain (hippocampus, amygdala, striatum, cortex, etc.) (3,4). Each subunit consists of a large extracellular N-terminal region, four transmembrane domains, and a large cytoplasmic domain. Transmembrane domain II forms the channel lumen (5), whereas the extracellular N-terminal region at ␣ subunits contains the glycine and strychnine binding sites (6 -11).
The brain is exposed to both transient changes in extracellular pH under physiological conditions, such as spontaneous neuronal firing and respiratory changes (12) and more sustained acidosis (up to a 1-unit drop in pH) after various pathophysiological conditions such as seizure, ischemia, and stroke (13,14). It is well known that protons modulate neuronal excitability, and this effect is partially mediated through pH modulation of activity of ion channels. Previous studies have shown that extracellular acidification decreases the opening of voltage-gated (Na ϩ , K ϩ , and Ca 2ϩ ) (15,16) and ligand-gated channels (acetylcholine, N-methyl-D-aspartate, ␥-aminobutyric acid, type A) (17)(18)(19)(20)(21). The pH modulation of glycine receptors has been only preliminarily assessed (22).
In the present study, we have characterized the sensitivity of homomeric ␣1 and heteromeric ␣1␤ receptors to extracellular pH and performed site-directed mutagenesis to investigate the structure-function relationship of glycine receptors in pH sensing. We have identified three residues at N terminus, Thr-112 and His-109 of the ␣1 subunit and Thr-135 of the ␤ subunit, which play a critical role in determining proton modulation of glycine receptor function.

EXPERIMENTAL PROCEDURES
Cloned Receptors-Wild type human glycine ␣1 and ␤ cDNA were generous gifts from H. Betz. The wild type or mutant receptor cDNA was expressed in human embryonic kidney cell lines via the mammalian expression vector pCIS2. Cells were transfected using calcium phosphate precipitation technique to achieve transient expression. Briefly, human embryonic kidney cells were plated onto coverslips and transfected with wild type or mutant subunits. Typically, ␣1 or ␣1 plus ␤ (1:10) cDNA (24) was added to cells growing exponentially on one coverslip placed in a 35-mm culture dish. After 6 -8 h, cells were washed and placed in fresh culture medium. Transfected cells were used for electrophysiological analysis 24 -48 h after the transfection.
Mutagenesis-Mutations of receptor cDNA were performed using commercially available QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) with commercially produced mutagenic primers (Integrated DNA Technologies). All mutants were verified by DNA sequencing (Biotechnology Core Facility, Texas Tech University, Lubbock, TX).
Electrophysiology-Whole-cell patch recordings were made at room temperature (22-25°C). Patch pipettes of borosilicate glass (1B150F, World Precision Instruments, Inc., Sarasota, FL) were pulled (Flaming/ Brown, P-87/PC, Sutter Instrument Co., Novato, CA) to a tip resistance of 1-2.5 megaohms. The pipette solution contained 140 mM CsCl, 10 mM EGTA, 10 mM HEPES, 4 mM MgATP, pH 7.2. Coverslips containing cultured cells were placed in a small chamber (ϳ1.5 ml) on the stage of an inverted light microscope (Olympus IMT-2) and superfused continuously (5-8 ml/min) with the following external solution containing 125 mM NaCl, 5.5 mM KCl, 0.8 mM MgCl 2 , 3.0 mM CaCl 2 , 20 mM HEPES, 10 mM D-glucose, pH 7.3. Glycine-induced Cl Ϫ currents from the whole-cell patch clamp technique were obtained using an Axoclamp 200A amplifier (Axon Instruments, Foster City, CA) equipped with a CV-4 headstage. Glycine-induced Cl Ϫ currents were low pass-filtered at 5 kHz, monitored on an oscilloscope and a chart recorder (Gould TA240), and stored on a computer (pClamp 6.0, Axon Instruments) for subsequent analysis. 60 -80% series resistance compensation was applied at the amplifier. To monitor the possibility that access resistance changed over time or during different experimental conditions, at the initiation of each recording we measured and stored on our digital oscilloscope the current response to a 5-mV voltage pulse. This stored trace was continually referenced throughout the recording. If a change in access resistance was observed during the recording period, the patch was aborted, and the data were not included in the analysis. Except during acquisition of current-voltage (I-V) relationships, cells were voltage-clamped at Ϫ60 mV. I-V relationships were determined by subtracting responses to continuous voltage ramp (Ϫ50 to ϩ50 mV in 0.5 s) before EC 30 glycine concentration (15 M for ␣1 and 10 M for ␣1␤) application from the responses to the same voltage protocols during glycine application.
Heteromeric ␣1␤ glycine receptors are 17-fold less sensitive to picrotoxin than homomeric ␣1 receptors (25). Thus, the incorporation of ␤ subunits with ␣1 subunits was confirmed by assessing picrotoxin sensitivity. If 500 M picrotoxin caused 30% or less inhibition of the current activated by EC 30 glycine, it was assumed that heteromeric ␣1␤ receptors were expressed.
Experimental Protocol-pH of external solutions was altered by the addition of NaOH or HCl and routinely checked before and during experiments. Glycine was prepared in the extracellular solution and applied from independent reservoirs by gravity flow for 10 s to cells using a Y-shaped tube positioned within 100 m of the cells. With this system, the 10 -90% rise time of the junction potential at the open tip was 12-51 ms (21). Once a control glycine response was determined, the effect of pH on the response was examined. To assess the pH effect cells were first bathed in media that was set to the test pH, then glycine, dissolved in the same test pH solution, was applied to the cells. Because in the whole-cell recordings, external solution of low pH elicited, as in other studies (20,21,23), a transient whole-cell inward current through acid-sensing ion channels, glycine application at various pH test values was made after the transient current had recovered and stabilized. Glycine applications were separated by at least 3-min intervals to ensure both adequate washout of glycine from the bath and recovery of receptors from desensitization if present.
Chemicals-All drugs were obtained from Sigma. Glycine, strychnine, and ZnCl 2 stocks were made in double distilled H 2 O.
Data Analysis-Glycine concentration-response profiles were fitted to the following equation I/I max ϭ [glycine] n /(EC 50 n ϩ [glycine] n ), where I and I max represent the normalized glycine-induced current at a given concentration and the maximum current induced by a saturating concentration of glycine, respectively, EC 50 is 50% effective glycine concentration, and n is the Hill coefficient.
Concentration-response curves for strychnine inhibitory effect were fitted to the following equation: I/I max ϭ [strychnine] n /([strychnine] n ϩ IC 50 ), where I is the Cl Ϫ current amplitude at the end of drug application normalized to control at a given strychnine concentration, IC 50 is the half-blocking concentration, and n is the Hill coefficient. Concentration-response profiles were evaluated using approximately the EC 30 glycine concentration. A minimum of three individual experiments was conducted for each paradigm. All data are presented as the means Ϯ S.E. Student's t test (paired or unpaired) or a one-way analysis of variance was used to determine statistical significance (p Ͻ 0.05).

Extracellular Protons Inhibit Glycine-activated Currents-
Based on the characterization of recombinant glycine receptors at the control condition pH 7.3, 15 and 10 M glycine are approximately EC 30 concentrations for ␣1 and ␣1␤ receptors, respectively (see Fig. 2, A and B). The EC 30 concentrations generate a stable current, elicit minimal receptor desensitization, and allow -fold modulator-induced potentiation of the glycine-evoked response. In assessing the effect of pH on the EC 30 glycine response, we varied the pH of the external medium between 8.4 and 5.4. The modulatory effect of glycineactivated current by extracellular pH in human glycine ␣1 and ␣1␤ receptors is illustrated in Fig. 1. The amplitude of current activated by EC 30 glycine was increased when the pH was increased from 7.3 to 7.8 and 8.4 and markedly attenuated when pH was decreased from 7.3 to 6.8 and 6.4 (Fig. 1, A and  B). The effect of pH on glycine current was rapid and completely reversible (Fig. 1, A and B). No adaptation to proton effect was observed during a prolonged perfusion (up to 10 min) with acidic or alkaline medium (data not shown). Fig. 1C shows the average sensitivity of glycine-activated current to protons over the range of pH 5.4 -8.4. In both receptors the acidic pH had a more profound effect than alkaline pH on glycine current. For homomeric ␣1 receptors, the glycine current activated by 15 M glycine was enhanced to 118 Ϯ 4.8% at pH 7.8 and to 129 Ϯ 10% of the control at pH 8.4 (n ϭ 3-5). The response to 15 M glycine was inhibited to 43 Ϯ 6.6% at pH 6.8, 22 Ϯ 3.4% at pH 6.4, and 12 Ϯ 1.8% of the control at pH 5.4 (n ϭ 3-4). For heteromeric ␣1␤ receptors, glycine-activated currents were also sensitive to acidic pH and hardly sensitive to alkaline pH. The current activated by 10 M glycine was inhibited to 39.1 Ϯ 5.4 and 1.3 Ϯ 1.3% of the control at pH 6.4 and 5.4, respectively M for ␣1␤) were remarkably inhibited by acidic pH and potentiated by alkaline pH. Note that the pH effects were reversible and dependent on the concentration of protons. C, summary of relative current activated by EC 30 glycine plotted as a function of pH. Glycine currents over the range of pH from 8.4 to 5.4 were normalized to the current recorded at control condition (pH 7.3, assigned as 100%, overlaps for ␣1 and ␣1␤). Data plotted are the means Ϯ S.E. Each data point is the average of at least 3 cells at a holding potential of Ϫ60 mV. The curve shown is the best fit of the data to the logistic equation, which yielded a transition point of pH 7.0 and 6.5 for ␣1 and ␣1␤ receptor, respectively.

Inhibition of Glycine Currents by Protons Is
Pharmacologically Competitive-Protons might inhibit glycine-activated current by decreasing the affinity of the receptors for glycine or decreasing the efficacy of glycine at receptors or both. To distinguish between these possibilities, concentration-response curves for glycine-activated current were determined at pH 7.3, 8.4, and 6.4 for ␣1 and ␣1␤ receptors. As shown in Fig. 2 Fig. 2B). In contrast, the maximal glycine-activated current and Hill coefficient values were not significantly affected by changes in pH in either receptor (p Ͼ 0.05). The lack of shift in EC 50 at pH 8.4 is comparable with minimal effect of alkaline pH on the glycine receptors (Fig. 1).
Proton Inhibitory Effect Is Voltage-independent-To evaluate whether protons inhibit the glycine response by altering the Cl Ϫ ion driving force, the effect of protons on the reversal potential of glycine-activated current was examined. Fig. 2, C and D, shows the current-voltage relationship for current activated by EC 30 glycine at pH 7.3 and 6.4 for ␣1 and ␣1␤ receptors recorded from the same cell. The glycine-induced currents reversed at the calculated equilibrium potential for Cl Ϫ ions (E Cl ϭ Ϫ3.14 mV). For ␣1 receptors the reversal potential was Ϫ4.9 Ϯ 1.6 mV at pH 8.4, Ϫ6.9 Ϯ 2.5 mV at pH 7.3, and Ϫ3.9 Ϯ 1.8 mV at 6.4 (n ϭ 4 -6). These values are not significantly different (p Ͼ 0.05, one-way analysis of variance). In addition, the average percentage inhibition by protons (at pH 6.4) did not significantly differ at membrane holding potentials between Ϫ50 and ϩ50 mV (71 Ϯ 7.4% at Ϫ50 mV, 73 Ϯ 11% at Ϫ30 mV, 63 Ϯ 11% at ϩ30 mV, 62 Ϯ 12% at ϩ50 mV, n ϭ 4 -6; p Ͼ 0.05, one-way analysis of variance). The proton effect on I-V relationship for ␣1␤ receptors was similar to that for ␣1 receptors For both receptors, the EC 50 value at pH 6.4 was significantly different from pH 7.3 (p Ͻ 0.05, unpaired t test). Changes in pH shifted the glycine concentration-response curve without significantly affecting the Hill coefficient or efficacy. C and D, current-voltage relationships (I-V curve) showing that protons did not alter the reversal potential in glycine ␣1 (C) and ␣1␤ receptors (D). I-V relationships were determined by subtracting responses to continuous voltage ramp (Ϫ50 to ϩ 50 mV in 0.5 s) before EC 30 glycine application from the responses to the same voltage protocols during glycine application.
(n ϭ 3, Fig. 2D). Therefore, the data suggest that protons neither alter the ion permeance ratio of the channel nor act on a site influenced by the transmembrane electric field.
Protons Attenuate Strychnine-induced Inhibition of Glycine Current-Proton sensitivity characterized above suggests that protons modulate homomeric and heteromeric glycine receptors in a qualitatively similar manner. The competitive inhibition of both glycine receptors by protons lured us to propose that protons may inhibit glycine receptors at ligand binding sites. Because ␣ subunits contain binding sites for agonists and competitive antagonists (24, 25), we tested this hypothesis in homomeric ␣1 receptors. We examined the proton effect on the apparent affinity of strychnine, a competitive antagonist.
Strychnine is a weak base with a pK a value of 8.26 (26). Based on the Henderson-Hasselbalch equation, ϳ90% of the strychnine molecules are estimated to remain in the zwitterionic form within the pH range from 6.4 to 7.3. As shown in Fig. 3A, a change of pH from 7.3 to 6.4 significantly shifted strychnine inhibition of currents activated by equal potency of glycine (EC 30 ). IC 50 for strychnine was increased from 43.9 Ϯ 6.1 nM at pH 7.3 to 102 Ϯ 32 nM at pH 6.4 (n ϭ 6, p Ͻ 0.05, paired t test). The maximal inhibition of strychnine was not affected by the change in pH.
Mutation ␣1 T112A Abolishes Protons Sensitivity-It has been reported that the Zn 2ϩ -induced inhibition of glycine current was greatly attenuated in acidic pH (22,25), suggesting that Zn 2ϩ may interact with protons at the same site(s). Previous studies indicate that the residues Thr-112 on the ␣1 subunits are critical to the inhibition by Zn 2ϩ (25,27). Furthermore, Thr-112 is also important for inhibition by the competitive antagonist strychnine (34). Inspection of the amino acid sequence for ligand-gated anionic channels revealed that Thr is conserved at the equivalent position across all the subunits among ␥-aminobutyric acid type A, glycine, and ␥-aminobutyric acid type C receptors (residues marked with an asterisk in Fig.  4). In addition, both Zn 2ϩ and protons exert an inhibitory effect on these receptors (21,22,28). Therefore, the Thr-112 residue is a candidate for proton action sites on glycine ␣1 receptors.
To investigate the functional role of Thr-112 in modulation of glycine-activated currents, site-directed mutagenesis was employed to replace the threonine residue at 112 with alanine, which has similar residue volume to threonine but lacks a hydroxyl group (-OH). The mutation T112A did not cause a significant shift in glycine EC 50 (from 23 Ϯ 2.1 M in wild type to 42 Ϯ 5.6 M (n ϭ 4) in mutant receptors at pH 7.3, p Ͼ 0.05, unpaired t test), which is consistent with previous investigations (11,34). However, mutation T112A greatly reduced the inhibition of glycine currents by strychnine at pH 7.3 compared with wild type receptors. Fig. 3B shows that receptors expressing T112A were about 3-fold less sensitive to strychnine (IC 50 value of 120 Ϯ 30 nM, n ϭ 4) than the wild type (IC 50 value of 44 Ϯ 6.1 nM, n ϭ 6, p Ͻ 0.05, unpaired t test). The Hill coefficient and maximal inhibition were not altered by the mutation. As shown in Fig. 5A  in wild type receptors (Fig. 5B). As shown in Fig. 5C, a shift of pH from 7.3 to 6.4 was unable to affect EC 50 values for glycine or maximal available glycine-induced current for T112A mutant receptors (EC 50 ϭ 27.6 Ϯ 3.8 M at pH 7.3 and 28.7 Ϯ 5.78 M at pH 6.4, respectively, n ϭ 5) compared with a right shift of concentration-response curve observed in wild type receptors ( Fig. 2A). These data suggest Thr-112 is sufficient to account for the effect of protons on glycine ␣1 receptors.
To further explore whether the volume of amino acid or hydroxyl group is responsible for action of proton, Thr-112 was mutated into tyrosine (Tyr), which has a hydroxyl group but is of larger volume. As shown in Fig. 3B, the strychnine potency was reduced in mutant T112Y beyond that observed in T112A (IC 50 value of 333 Ϯ 78 nM, n ϭ 4, p Ͻ 0.05, compared with T112A or wild type, unpaired t test). In contrast, the T112Y mutation only partially attenuated the response to acidic pH compared with T112A mutant receptors. The current activated by EC 30 glycine was 40 Ϯ 3.8% of the control at pH 5.4 and 63 Ϯ 9.1% of the control at pH 6.4 for mutant T112Y receptors (n ϭ 5). These values were significantly different from the wild type or T112A receptors (Fig. 5B). Next, we mutated Thr-112 to another aromatic amino acid, Phe, which lacks hydroxyl group but has a volume similar to Tyr. As shown in Fig. 5B, compared with T112Y mutation, the T112F mutant receptor completely lost its sensitivity to pH 6.4 and had reduced proton sensitivity at pH 5.4 (glycine current was 107 Ϯ 9.2% at pH 6.4 and 63 Ϯ 6.1% of the control at pH 5.4, n ϭ 5). These data indicate that hydroxyl group at Thr-112 of the glycine receptors is a key element in determining the receptor sensitivity to protons.
Thr-113 is another conserved residue in all glycine ␣ subunits (Fig. 4). To determine whether Thr-113 also contributes to proton sensitivity in conjunction with Thr-112, we mutated Thr-113 to Ala. The T113A mutation caused a left shift of the concentration-response curve for glycine (EC 50 of 6.5 Ϯ 1 M, n ϭ 4, p Ͻ 0.05, unpaired t test, compared with wild type). The sensitivity to protons in T113A, tested with glycine EC 30 (5 M), was similar to wild type receptors (n ϭ 5, data not shown). Therefore, the role of the threonine residue in modulation of glycine receptors seems site-specific.
Effect of Charged Residues Neighboring Thr-112 on pH Sensitivity-In addition to the polar residue Thr-112, charged residues neighboring Thr-112 were tested for potential proton coordination. By examining residues in the vicinity of ␣1 Thr-112, we found that two acidic amino acid residues, Glu-110 and Asp-114, are conserved in all glycine subunits (marked with dots in Fig. 4), and these flank residue Thr-112. Both Glu-110 and Asp-114 on glycine ␣1 receptors contain a negatively charged carboxyl group, which may provide an electrical requirement for interaction with protons. To test this hypothesis, Glu-110 and Asp-114 were mutated individually to the neutral amino acid alanine. As shown in Fig. 6A, the mutation of E110A or D114A caused an approximate 4-fold shift of EC 50 for glycine. The glycine EC 50 was significantly increased from 23 M for wild type to 95.5 Ϯ 10.5 M for E110A (n ϭ 3) and 76 Ϯ 16 M for D114A (n ϭ 4, p Ͻ 0.05, unpaired t test, compared with wild type). Inhibition by protons was not impaired by these mutations, suggesting that Glu-110 and Asp-114 residues are unlikely involved in the inhibitory effect of protons on glycine receptors (Fig. 6B).
The titratable residue histidine has been reported to be involved in determining proton sensitivity in a number of ion channels (29 -32). His-109 is highly conserved across all subunits of the ligand-gated anionic channels (Fig. 4) and has been identified as a Zn 2ϩ binding site for Zn 2ϩ -induced inhibition of glycine receptors (22,27). We, thus, examined the role of the His-109 residue in proton sensitivity in glycine ␣1 receptors. Fig. 6A shows that the replacement of histidine with alanine did not change the EC 50 value for glycine (EC 50 , 24 Ϯ 2.4 M; Hill coefficient, 1.64 Ϯ 0.12, n ϭ 3, p Ͼ 0.05, unpaired t test, compared with wild type). As shown in Fig. 6B, the mutation H109A resulted in significant reduction of proton sensitivity. The mutation H109A eliminated 38 -54% of the inhibition caused by protons as the percentage of current inhibition was 54 Ϯ 4.5% at pH 5.4 and 39 Ϯ 6.4% at pH 6.4 compared with 88 Ϯ 1.4% inhibition at pH 5.4 and 78 Ϯ 3.4% inhibition at pH 6.4 observed in wild type (n ϭ 4 -5, p Ͻ 0.01, unpaired t test, compared with wild type) (Fig. 6B). The potentiation of glycine currents by alkaline pH (pH 8.4) was also abolished (Fig. 6B). These data indicate that His-109 also takes part in proton modulatory effect of glycine ␣1 receptors, although His-109 cannot fully account for the proton effect.
␤ Subunits Contribute to Proton Sensitivity-Finally, we examined the role of ␤ subunits in proton modulation of heteromeric glycine ␣1␤ receptors. If ␤ subunits do not contribute to pH modulation, heterologous expression of mutant ␣1(T112A) with ␤ subunits would produce receptors that remain insensitive to protons, as observed in ␣1(T112A) homomers. In contrast to complete loss of proton sensitivity observed in mutant homomeric ␣1(T112A) (see Fig. 5), proton sensitivity was reduced but still maintained in ␣1(T112A)␤ heterooligomers (Fig.  7A), suggesting that ␤ subunits may participate in proton modulation in heteromeric receptors. To further confirm this, Thr-135 of the ␤ subunit, which is the homologue of ␣1 Thr-112 (Fig.  4), was mutated to Ala. Compared with wild type ␣1␤ receptors, mutation ␤ T135A caused a small (ϳ10%) but significant reduction of proton-induced inhibition of EC 30 glycine currents (n ϭ 8, p Ͻ 0.05 at pH 5.4, unpaired t test). As expected, double mutation ␣1(T112A) and ␤(T135A) showed further reduction of the sensitivity to acidic pH (Fig. 7A). The EC 30 glycine currents were 105 Ϯ 3.3% at pH 8.4, 93 Ϯ 0.9% at pH 6.4, and 82 Ϯ 4.5% of the control at pH 5.4 (n ϭ 4). On the average, acidic pH inhibited EC 30 glycine currents in a rank order of ␣1␤ Ͼ ␣1␤(T135A) Ͼ ␣1(T112A)␤ Ͼ ␣1(T112A)␤(T135A) (Fig. 7A). Fig. 7B shows the mutations caused a modest shift in glycine EC 50 values (2-2.5-fold). The results indicate that the ␤ subunits contribute to proton sensitivity through Thr-135, in conjunction with the corresponding residue ␣ Thr-112. DISCUSSION Protons could potentially modulate glycine receptors through several sites of action. Protons could change the charge(s) of the glycine molecule itself, thus altering its interaction with the agonist binding site. This alternative is doubtful, as it is estimated from the Henderson-Hasselbach equation that glycine molecules (pK a ϭ 2.35 for -COOH and 9.60 for -NH 2 , pI ϭ 5.97) are almost exclusively in the zwitterionic form within the pH range used in this study. The possibility that protons may be inhibiting the glycine receptor by affecting electrostatic forces at the channel pore also seems unlikely, as the proton effect was voltage-independent and did not change reversal potential. We also considered the possibility that the target of protons could be intracellular, either the receptor itself or a regulatory protein that subsequently modified glycine receptor function. However, considering that proton action is rapid and reversible and that perfusing human embryonic kidney cells with HEPES-buffered solution at pH 5.4 and pH 9.4 induced negligible changes in intracellular pH (23), this possibility is also unlikely. The fact that protons competitively inhibited the effects of glycine and the antagonist strychnine presented the possibility that this extracellular site could be the ligand binding domain. Consistent with this hypothesis are the early findings of Young and Snyder (33) demonstrating that 3. Note that sensitivity to acidic or alkaline pH was not significantly different between ␣1(E110A) or ␣1(D114A) and wild type receptors. In contrast, the sensitivity to acidic or alkaline pH was significantly reduced in ␣1(H109A) receptors. Each data point represents the mean of at least four cells (**, p Ͻ 0.01, unpaired t test compared with proton sensitivity in wild type).
[ 3 H]strychnine binding is decreased in acidic pH. We, thus, studied the pH sensitivity of glycine receptors expressing Nterminal mutations that have been reported to be important for receptor binding and modulation (22,25,27,34).
Complete loss of proton sensitivity in homomeric ␣1 receptors with mutation of T112A indicates that Thr-112 is a key residue in determining pH sensitivity. Because threonine is not titratable within the pH range we tested here (pK a ϭ 15), Thr-112 is likely not to be a binding site per se for protons. It is worth noting that the residues involved in proton sensitivity may not be necessarily ionizable. Non-ionizable residues have been reported to play a major role in proton sensitivity of N-methyl-D-aspartate receptors (alanine residues) (35), Kir6.2 channels (threonine) (36), and Kir2.3 channels (threonine) (37). The role of the threonine residue in the modulation of protons is site-specific because mutation of another highly conserved threonine residue at position 113 did not affect receptor sensitivity to protons.
The mechanism by which Thr-112 influences pH sensitivity is complex. Mutation of Thr-112 to Ala converts the partial agonists taurine and ␤-aminoisobutyric acid into full agonists and dramatically reduces sensitivity to inhibition by strychnine and Zn 2ϩ (11,34), indicating that Thr-112 plays a fundamental role in actions of several modulators. T112A does not, however, affect glycine sensitivity (11,34). Consistent with these studies, we found that mutation of Thr-112 to Ala, Tyr, or Phe decreased sensitivity (i.e. increased IC 50 ) to inhibition by strychnine or Zn 2ϩ (data not shown) but had no effect on apparent glycine affinity, indicating that residue Thr-112 might participate in an antagonistic effect but not glycine affinity. With regard to the role of Thr-112 in proton-mediated inhibition, the presence of a hydroxyl group at this position appears to be critical. Receptors in which the Thr-112 residue was mutated to another hydroxylated residue (Tyr) retained sensitivity to protons, whereas the presence at the 112 position of Phe, which is similar in molecular volume to Tyr but does not have an hydroxyl group, conferred resistance to protons. Receptors with the T112Y mutation did not, however, retain sensitivity to inhibition by strychnine or Zn 2ϩ . Our results, coupled with those of others (11, 34) define a central role for Thr-112 in mediating inhibitory actions of protons, strychnine, and Zn 2ϩ , although the molecular determinants at this site for the three modulators are somewhat distinct.
Titrable histidine residues are major determinants of pH modulation in many ion channels (29 -32). His-109, which is highly conserved among several ligand-gated ion channels, is likely one of the binding sites for protons in glycine ␣1 receptors. His-109 has been identified as the site of Zn 2ϩ coordination. Harvey et al. (22) show that Zn 2ϩ inhibition of glycine ␣1 receptors was greatly attenuated by acidification, suggesting that protons may compete with the Zn 2ϩ action. In good agreement with previous investigations (22,27), we found that mutation of His-109 into a non-charged residue abolished Zn 2ϩ inhibition (data not shown). Meanwhile, mutation H109A also led to partial loss of proton sensitivity, suggesting His-109 is involved in proton-induced inhibition. It seems that both Zn 2ϩ and H ϩ modulate glycine receptors through His-109. In addition, partial loss of proton sensitivity in mutation H109A indicates that protons might bind to other ionizable amino acid residues in addition to His-109. We have excluded two negatively charged residues, Glu-110 and Asp-114, in a neighboring position to Thr-112, from potential binding sites for protons because mutation of either of the residues failed to affect proton sensitivity.
We demonstrate that heteromeric glycine ␣1␤ receptors are sensitive to pH change in a manner similar to that observed in homomeric ␣1 receptors. Unlike in homomeric ␣1 receptors, mutation of ␣1(T112A) in ␣1␤ receptors did not completely abolish pH sensitivity. Effects of protons were only dramatically reduced when the equivalent ␤ subunit T residue was also mutated. This indicates the ␤ subunit contributes to pH sensitivity in ␣1␤ receptors. Although it is usually believed that the ␤ subunits alone neither form a functional channel nor form a ligand binding pocket (25), incorporation of ␤ subunits changes the pharmacological and functional properties of glycine receptors (for review, see Ref. 4). ␣1␤ glycine receptors have a stoichiometry of three ␣1 and two ␤ subunits (5). The presence of only two Ala residues in ␣1␤ receptors (␣1␤(T135A)) elicited only a small reduction of proton sensitivity, whereas the presence of three (␣1(T112A)␤) or five (␣1(T112A)␤(T135A) Ala residues had a significantly larger impact on proton inhibition. This indicates that multiple residues in both ␣ and ␤ subunits are required for proton modulation of heteromeric ␣1␤ receptors. Given the widespread distribution of glycine receptors in spinal cord, brainstem, and other regions of the brain (38), modulation of glycine receptors by protons is of significant physiological importance. Although we only studied ␣1 and ␣1␤ glycine receptor subtypes, the pH modulatory effect may be universal for other glycine subtypes because the critical residues identified here (Thr-112 and His-109) are conserved across the different ␣ subunits. Indeed, we have observed that pH modulates recombinant glycine ␣2 receptors in a way similar to ␣1 receptors. 1 The postsynaptic glycine receptors (predominantly ␣␤) might be saturated in the synaptic cleft at the moment of release of glycine (39). Considering that protons have potent effect on glycine response at submaximal glycine concentration, protons may modify the kinetics of glycinergic inhibitory postsynaptic potentials during the clearance of the glycine from the synaptic cleft. In addition, extrasynaptic glycine receptors (predominantly homomeric ␣ receptors) seem to be most effectively activated by sub-saturating concentrations of glycine spillover from the adjacent synaptic cleft (40). Proton-mediated modulation of extrasynaptic glycine receptors would result in alternation of tonic inhibition of neurons (4).