Conformational Variability of the Glycine Receptor M2 Domain in Response to Activation by Different Agonists*

Models describing the structural changes mediating Cys loop receptor activation generally give little attention to the possibility that different agonists may promote activation via distinct M2 pore-lining domain structural rearrangements. We investigated this question by comparing the effects of different ligands on the conformation of the external portion of the homomeric α1 glycine receptor M2 domain. Conformational flexibility was assessed by tethering a rhodamine fluorophore to cysteines introduced at the 19′ or 22′ positions and monitoring fluorescence and current changes during channel activation. During glycine activation, fluorescence of the label attached to R19′C increased by ∼20%, and the emission peak shifted to lower wavelengths, consistent with a more hydrophobic fluorophore environment. In contrast, ivermectin activated the receptors without producing a fluorescence change. Although taurine and β-alanine were weak partial agonists at the α1R19′C glycine receptor, they induced large fluorescence changes. Propofol, which drastically enhanced these currents, did not induce a glycine-like blue shift in the spectral emission peak. The inhibitors strychnine and picrotoxin elicited fluorescence and current changes as expected for a competitive antagonist and an open channel blocker, respectively. Glycine and taurine (or β-alanine) also produced an increase and a decrease, respectively, in the fluorescence of a label attached to the nearby L22′C residue. Thus, results from two separate labeled residues support the conclusion that the glycine receptor M2 domain responds with distinct conformational changes to activation by different agonists.

Models describing the structural changes mediating Cys loop receptor activation generally give little attention to the possibility that different agonists may promote activation via distinct M2 pore-lining domain structural rearrangements. We investigated this question by comparing the effects of different ligands on the conformation of the external portion of the homomeric ␣1 glycine receptor M2 domain. Conformational flexibility was assessed by tethering a rhodamine fluorophore to cysteines introduced at the 19 or 22 positions and monitoring fluorescence and current changes during channel activation. During glycine activation, fluorescence of the label attached to R19C increased by ϳ20%, and the emission peak shifted to lower wavelengths, consistent with a more hydrophobic fluorophore environment. In contrast, ivermectin activated the receptors without producing a fluorescence change. Although taurine and ␤-alanine were weak partial agonists at the ␣1R19C glycine receptor, they induced large fluorescence changes. Propofol, which drastically enhanced these currents, did not induce a glycine-like blue shift in the spectral emission peak. The inhibitors strychnine and picrotoxin elicited fluorescence and current changes as expected for a competitive antagonist and an open channel blocker, respectively. Glycine and taurine (or ␤-alanine) also produced an increase and a decrease, respectively, in the fluorescence of a label attached to the nearby L22C residue. Thus, results from two separate labeled residues support the conclusion that the glycine receptor M2 domain responds with distinct conformational changes to activation by different agonists.
Glycine receptor (GlyR) 4 chloride channels mediate inhibitory neurotransmission in the central nervous system (1). They comprise an assembly of five subunits that are each composed of a large N-terminal extracellular ligand-binding domain and four transmembrane ␣-helices (M1-M4). Cryo-electron microscopy images of the homologous Torpedo electropax nicotinic acetylcholine receptor (nAChR) transmembrane region reveals that the pore-lining M2 domains are kinked radially inward to form a central constriction at the membrane midpoint (2). There is currently a great deal of interest in understanding how M2 domains move to open the channel. Although the original model proposed a drastic rotation of M2 domains about their long axes (3), more recent evidence suggests that there is little if any rotation during receptor activation (4 -6). The precise nature of the structural change remains a matter for debate.
The central pore kink is likely to introduce a degree of structural discontinuity because the hydrogen bonds responsible for maintaining ␣-helix rigidity are most likely broken. The kink may therefore act as a swivel enabling the outer half of M2 to move asynchronously with the inner half, where the gate is most likely positioned. Indeed, several lines of evidence suggest gating is mediated by a backbone rearrangement at this midpoint (7)(8)(9)(10). In addition, a rate equilibrium free energy state analysis shows that the M2-M3 domain is positioned midway along the agonist-induced "conformational wave" that extends from the ligand binding domain to the gate (11). These considerations imply that the outer part of M2 should be an informative place to investigate the structural basis of receptor activation, as the movements in this region should not reflect either those occurring at the ligand binding site or the activation gate. Thus, in this study we probe conformational changes at the 19Ј and 22Ј residues, near the external end of M2.
Current models of Cys loop receptor activation consider only structural changes associated with transitions from the resting closed state to the agonist-induced open state (5,6,12,13). Relatively little attention has been given to the possibility that different agonists and pharmacological modulators may promote different structural conformations in the pore region. However, it has been suggested on the basis of substituted cysteine accessibility studies that the pore blocker, picrotoxin, can change the conformation of the outer GlyR pore region suggesting in turn that it may interact allosterically with agonistinduced conformational changes (14). Our aim is to investigate the conformational changes induced by agonists, antagonists, and allosteric modulators by covalently labeling different residues near the extracellular M2 boundary with the sulfhydrylreactive fluorophores, methanethiosulfonate-rhodamine (MTSR) and tetramethylrhodamine-maleimide (TMRM), and simultaneously measuring current changes and fluorescence changes (15,16). This approach has been employed so far in only a few studies on Cys loop receptors (15,17,18). A particular advantage of the GlyR is that it displays little or no desensitization thereby facilitating the quantitation of ligand-induced fluorescence spectral shifts.
Molecular Biology-The human GlyR ␣1 subunit cDNA was subcloned into the pGEMHE vector. All constructs used in this study contained the functionally silent C41A mutation, which eliminated the sole uncross-linked extracellular sulfhydryl group (19). Site-directed mutagenesis to generate the R19ЈC, S21ЈC, and L22ЈC mutants was performed with the QuickChange mutagenesis kit (Stratagene, La Jolla, CA). Successful incorporation of the mutations was confirmed through automated sequencing of the entire cDNA coding region. The mMessage mMachine kit (Ambion, Austin, TX) was used to generate capped mRNA for oocyte injection. mRNA concentrations were adjusted to 200 pg/nl and aliquots stored at Ϫ70°C.
Oocyte Preparation, Injection, and Labeling-Female Xenopus laevis (Xenopus Express, France) frogs were anesthetized, and stage VI oocytes were removed from ovaries and washed thoroughly in OR-2 (2.5 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 5 mM HEPES, pH 7.4). The oocytes were incubated in collagenase (Sigma) in OR-2 for 2 h at room temperature, rinsed, and stored at 18°C.
All oocytes were injected with 10 ng of mRNA into the cytosol. To achieve the high levels of expression required for the detection of the fluorescent signal over the background (because of oocyte autofluorescence and nonspecific binding of the dye), the oocytes were incubated for 3-10 days after injection at 18°C. The incubation solution contained 96 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 1.8 mM CaCl 2 , 5 mM HEPES, 0.6 mM theophylline, 2.5 mM pyruvic acid, 50 g/ml gentamycin (Cambrex Corp., East Rutherford, NJ), 5% horse serum (Hyclone, Logan, UT), pH 7.4.
To maximize surface expression of receptors prior to labeling, oocytes were incubated at room temperature for 2 h. The oocytes were then transferred into ND96 (96 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 1.8 mM CaCl 2 , 5 mM HEPES, pH 7.4) and stored on ice. For labeling, oocytes were transferred into the labeling solution containing 10 M MTSR in ND96 for 25 s or 10 M TMRM in ND96 for 60 min. The oocytes were then washed and stored in ND96 for up to 6 h before recording. All labeling steps were performed on ice.
Voltage Clamp Fluorometry (VCF)-For VCF experiments, an inverted Nikon Eclipse TE300 microscope (Nikon Instru- ments, Kawasaki, Japan) was equipped with a high-Q TRITC filter set (Chroma Technology, Rockingham, VT), a Plan Fluor ϫ40 objective lens (N.A. 0.6, WD 3.7-2.7 mm) (Nikon Instruments, Kawasaki, Japan), and a PhotoMax 200 photodiode detection system (Dagan Corp., Minneapolis, MN) attached to the side port of the microscope. An excitation filter wheel including a shutter and an emission filter wheel were controlled through a Lambda 10-2 unit (Sutter Instruments, Novato, CA). A Lambda LS 175-watt xenon arc lamp served as a light source and was coupled to the microscope via a liquid light guide (Sutter Instruments, Novato, CA). The design of the custom-made recording chamber is described in Ref. 15. An automated perfusion system operated by a Valvebank 8 module (AutoMate Scientific, San Francisco, CA) was used for perfusion of the recording chamber.
Electrodes for two-electrode voltage clamp recordings were filled with 3 M CsCl and moved by automated ROE-200 micromanipulators coupled to an MPC-200 controller (Sutter Instruments, Novarto, CA). Cells were voltageclamped at a holding potential of Ϫ40 mV in all experiments, and currents were recorded using a Gene Clamp 500B amplifier (Molecular Devices). Current and fluorescence traces were acquired at 200 Hz via a Digidata 1322A interface and Clampex 9.2 software (Molecular Devices). The fluorescence signal was further digitally filtered at 1-2 Hz with an eight-pole Bessel filter for analysis and display. The base line was corrected for bleaching where appropriate. All values for changes in fluorescence indicate an increase in fluorescence, unless stated otherwise. The empirical Hill equation, fitted with a nonlinear least squares algorithm (SigmaPlot 9.0, Systat Software, Point Richmond, CA), was used to obtain half-maximal concentrations (EC 50 or IC 50 ) and Hill coefficient (n H ) values for ligand-induced activation and inhibition. All results are expressed as means Ϯ S.E. of the mean of five or more independent experiments.
Spectral Analysis-For spectral analysis a MicroSpec 2150i (Acton Research Corp., Acton, MA) coupled to an ORCA-ER CCD camera (Hamamatsu, Hamamatsu City, Japan) replaced the photodiode detection system on the side port of the micro- Conformational Changes of the GlyR M2 Segment DECEMBER 7, 2007 • VOLUME 282 • NUMBER 49 scope. The spectrometer was operated using SpectraPro monochromator software (Acton Research Corp., Acton, MA). For excitation, a HQ535/ϫ50 filter was used in combination with a Q565LP dichroic mirror (Chroma Technology, Rockingham, VT); no emission filter was used. Prior to the experiment, the spectrometer was calibrated using the 434, 546, and 577 nm lines of a mercury lamp spectrum.
To analyze spectra, the image of the region of interest to be monitored was aligned with the slit of the spectrometer. Fluo-rescence from the region of interest was reflected onto the grating (300 g/mm; 500 nm blaze), and the extracted spectrum was imaged on the ORCA-ER CCD (chip size 1344 ϫ 1024 pixels) using Meta-Morph 6.2 (Universal Imaging Corp., Downington, PA). The x axis of the resulting "spectral image" represents the wavelength dimension, whereas the y axis represents the one-dimensional spatial dimension of the slit. The wavelength coverage for this combination of grating and camera was 170 nm. Adjacent averaging (15 points) was used to smooth recorded spectra (ORIGIN 6.0, Microcal Software Inc., Northampton, MA).  Table 1 shows the mean EC 50 and n H values, determined by averaging the parameters determined from curve fits to individual dose responses. Table 1 also shows the mean I max values. As reported previously (20), the ␣1R19ЈC mutation results in a dramatically decreased glycine sensitivity. The mean I max was not significantly different in ␣1R19ЈC relative to ␣1WT receptors (9.3 Ϯ 1.6 A, n ϭ 5, versus 10.0 Ϯ 0.6 A, n ϭ 9). The effect of MTSR labeling on glycine-evoked currents was also studied. As shown in the example in Fig.  1B (upper panel), our standard MTSR labeling protocol produced no significant change (ϩ11.7 Ϯ 8.1%, n ϭ 8) in the magnitude of currents activated by EC 10 (5 M) glycine in ␣1WT GlyRs (Fig. 1B, upper panel). However, EC 10 (1 mM) glycine currents were strongly potentiated in ␣1R19ЈC GlyRs following MTSR labeling with a mean current increase of 471 Ϯ 68% (n ϭ 6) (e.g. Fig. 1B, lower panel). As shown in Fig. 1C, this was caused by a leftward shift in the ␣1R19ЈC GlyR glycine EC 50 value,  although the n H remained unchanged ( Table 1). As expected, the mean glycine EC 50 of the ␣1WT GlyR was not significantly changed by the MTSR labeling procedure ( Fig. 1C and Table 1). Fig. 1D shows a model of three nAChR M2 helices (based on Protein Data Bank code 2BG9), one of which shows the approximate relative locations of the labeled rhodamine 5 and the 9Ј leucine residue thought to be part of the channel gate.

Specific MTSR Labeling of the
Glycine-evoked Changes in Fluorescence Intensity at R19ЈC-MTSR treatment of oocytes injected with ␣1R19ЈC mRNA resulted in an ϳ3.5-fold increase in resting fluorescence compared with that observed in identically treated uninjected oocytes or oocytes injected with ␣1WT mRNA (data not shown). VCF experiments showed that glycine application produced changes in fluorescence intensity (⌬F) in ␣1R19ЈC-injected oocytes that were coincident in time with the activated chloride current (⌬I) ( Fig.  2A), with relative changes in fluorescence (⌬F/F) regularly reaching 20% at saturating glycine concentrations. In contrast, no significant ⌬F was ever observed with uninjected or ␣1WT-injected oocytes. When normalized, the ⌬I and ⌬F glycine dose responses were virtually superimposed ( Fig. 2B and Table 1), strongly suggesting a tight coupling between conformational changes at the 19Ј residue and at the channel gate during glycine-induced activation.
Spectral analysis of the fluorescence signal from MTSR-labeled ␣1R19ЈC GlyRs revealed that the fluorescence increase was accompanied by an ϳ10 nm blue shift (n ϭ 6) in the MTSR emission peak during full receptor activation (Fig. 2, C and  D). The magnitude of this shift was concentration-dependent, with EC 50 (1 mM) glycine producing roughly half of the shift produced by a saturating (20 mM) glycine concentration (Fig. 2, C and D). The blue shift in the MTSR emission peak is characteristic of an increase in hydrophobicity in the fluorophore environment, most likely reflecting a movement of the fluorophore to a more hydrophobic environment in the open state (15). As expected, MTSR-incubated oocytes expressing ␣1WT GlyRs did not show ⌬F or spectral shifts (Fig. 2D).
Effects of Other Agonists on R19ЈC Fluorescence Changes-The antihelminthic compound, ivermectin, binds to an as yet unidentified binding site and activates the GlyR in an irreversible manner (21). To confirm the previous conclusion that it activates the GlyR via a different mechanism from glycine (21), we sought to compare the relationship between ⌬I and ⌬F produced by ivermectin with that produced by glycine. In MTSR-labeled ␣1R19ЈC GlyRs, 10 M ivermectin evoked slowly developing currents that were roughly double the size of those produced by 1 mM glycine (203 Ϯ 30%, n ϭ 10). Despite this large current, no significant increase in the fluorescence intensity was ever observed (e.g. Fig. 3A). Indeed, the mean fluorescence change in eight oocytes showing robust glycine currents, the ⌬F response was only 0.5 Ϯ 0.1%. As observed previously (21), the currents evoked by 10 M ivermectin were only weakly sensitive to strychnine, with 10 M strychnine reducing current magnitude by 18.9 Ϯ 0.7% (n ϭ 4, Fig. 3B). No ivermectin-gated currents were observed in uninjected oocytes (n ϭ 3). Taken together, these results confirm that ivermectin activates GlyRs by a mechanism different from glycine.
Taurine and ␤-alanine are low efficacy agonists of ␣1WT GlyRs, with taurine having a lower efficacy than ␤-alanine (22). Both compounds are converted into antagonists by a variety of R19Ј mutations, including R19ЈC (20,23,24). These actions suggest that taurine and ␤-alanine bind to the ␣1R19ЈC GlyR without activating the receptor and thereby they act as classical competitive antagonists of glycine binding. If this hypothesis is    DECEMBER 7, 2007 • VOLUME 282 • NUMBER 49 correct, their binding should not produce significant ⌬F. We sought to investigate whether these ligands produced a significant ⌬F change upon receptor binding and, if so, whether this was spectrally similar to that produced by glycine. As expected, a 50 mM concentration of taurine did not produce a significant ⌬I, although small, nA-sized currents were regularly recorded. Nevertheless, 50 mM taurine produced large changes in ⌬F, with a mean value near 35% that evoked by a saturating glycine concentration (Fig. 3C). Both ⌬I and ⌬F stimulated by taurine were completely blocked by 10 M strychnine (data not shown). Similar results were obtained with ␤-alanine. A 50 mM concentration of ␤-alanine evoked only a very small current (ϳ4% that produced by saturating glycine in the same oocyte), but it evoked a large ⌬F that was roughly 60% that produced by a saturating glycine concentration (Fig. 3D). Again, both ⌬I and ⌬F stimulated by ␤-alanine were potently blocked by 10 M strychnine (data not shown).

Conformational Changes of the GlyR M2 Segment
The taurine-activated current was too small to be fitted with confidence to the Hill equation (n ϭ 8). However, Hill equation fits to taurine ⌬F dose responses (Fig. 3E) from the same eight oocytes yielded a mean EC 50 of 2.4 Ϯ 0.4 mM and n H of 0.9 Ϯ 0.1. Applications of increasing ␤-alanine concentrations revealed a close relationship between ⌬F and ⌬I with no significant differences in their EC 50 and n H values (Fig. 3E). The ⌬F dose response was fitted with a mean EC 50 of 2.02 Ϯ 0.34 mM and an n H of 0.8 Ϯ 0.1, whereas the ⌬I dose response was fitted with an EC 50 of 2.61 Ϯ 0.93 mM and an n H of 1.1 Ϯ 0.4 (n ϭ 6 for each). To facilitate comparison with glycinemediated responses, Fig. 3E also shows the ⌬I and ⌬F dose response relationships for glycine activation. The fluorescence and current maxima for taurine and ␤-alanine were both normalized to those produced by glycine in the same oocyte.
Given that the relationship between ⌬I and ⌬F produced by glycine activation was not preserved during receptor activation by taurine or ␤-alanine, we hypothesized that these compounds produced different conformational changes at the 19Ј residue. We investigated this by measuring the spectral changes produced by both compounds. As shown in Fig. 3F, taurine binding produced no detectable spectral shift, whereas the spectral shift produced by ␤-alanine (ϳ5 nm) was much smaller than that produced by gly-  (Fig. 3G). Together these results strongly suggest that the binding of these weak partial agonists results in different conformational changes from those produced by glycine. Fluorescence Responses of Other Labeled Residues in the Extracellular M2 Region-As the R19ЈC mutation impacts severely on receptor function, we assessed whether other cysteine mutants in the extracellular M2 region with less impact on receptor function show a similar pattern of differential responses to different agonists. A previous study (20) showed that cysteine substitutions at the 21Ј and 22Ј positions disrupt receptor function to a lesser degree than the R19ЈC mutation. Thus, we focused on these two residues. When injected into oocytes, the ␣1S21ЈC GlyR exhibited a WT-like glycine EC 50 , although the n H was significantly reduced ( Table 1). The ␣1S21ЈC-injected oocytes showed no evidence of labeling by either MTSR or TMRM. Incubation with MTSR or TMRM in the presence or absence of 100 M glycine produced no significant ⌬F in response to saturating (3 mM) glycine applications (n ϭ 3 each; data not shown), and fluorophore labeling produced no change in ␣1S21ЈC GlyR electrophysiological properties (Table 1). We conclude that ␣1S21ЈC is either not labeled by these fluorophores or that it experiences no change in microenvironment during receptor activation.
The ␣1L22ЈC GlyR showed robust expression with electrophysiological properties intermediate between those of ␣1WT and ␣1R19ЈC (Table 1). Importantly, taurine and ␤-alanine acted as strong partial agonists (see below), as at ␣1WT receptors. When receptors were labeled with MTSR, no change in ⌬F could be detected, even at saturating (3 mM) glycine concentrations (n ϭ 4) (Fig. 4A). However, following labeling by TMRM, ␣1L22ЈC-injected oocytes showed a small but significant ⌬F (1.5 Ϯ 0.3%, n ϭ 13) at saturating glycine concentrations (Fig.  4A). TMRM labeling also produced a significant increase in the glycine sensitivity of ␣1L22ЈC GlyRs (Table 1 and Fig. 4B). The small size of these signals precluded quantitation or spectral analysis of the ⌬F response. In contrast to the effect of glycine, application of saturating (10 mM) concentrations of taurine or ␤-alanine evoked a significant decrease in ⌬F (Fig. 4, C and D).
Taurine induced a maximal current of 1.7 Ϯ 0.4 A and a ⌬F of Ϫ1.1 Ϯ 0.2% (n ϭ 8). ␤-Alanine induced a maximal current of 3.1 Ϯ 0.3 A and a ⌬F of Ϫ1.1 Ϯ 0.3% (n ϭ 7). We then tested whether the sign of the fluorescence change simply depended on the size of the current or was indeed agonist-specific. Application of a low (100 M) glycine concentration evoked currents of 1.6 Ϯ 0.3 A (n ϭ 7), a value not significantly different from those evoked by 10 mM taurine (1.7 Ϯ 0.4 A). However, this concentration of glycine produced no significant ⌬F (0.0 Ϯ 0.1% compared with Ϫ1.1 Ϯ 0.2%) (n ϭ 7). A similar result was obtained for ␤-alanine, where currents induced by 300 M glycine were not significantly different from those evoked by 10 mM ␤-alanine (3.8 Ϯ 0.6 A and 3.1 Ϯ 0.3 A, respectively), but the respective fluorescence changes were significantly different in size and sign (0.3 Ϯ 0.2% compared with Ϫ1.1 Ϯ 0.3%; both n ϭ 7). Together these results demonstrate that glycine produces a different conformational change at L22ЈC relative to those produced by taurine and ␤-alanine.
Effects of GlyR Inhibitors on R19ЈC Fluorescence Changes-We next sought to compare the effects on the relationship between ⌬I and ⌬F of a putative classical competitive antagonist, strychnine, an allosteric inhibitor, picrotoxin, and a putative classical channel blocker, ginkgolide C. The molecular pharmacology of each of these compounds at the GlyR has been reviewed recently (25). When applied alone, strychnine did not evoke significant ⌬I or ⌬F changes, even at a concentration of 10 M (Fig. 5A). However, when co-applied with 1 mM glycine, 10 M strychnine simultaneously decreased ⌬I and ⌬F values (Fig. 5B). As shown in Fig. 5C, the sensitivities of ⌬I and ⌬F to strychnine were very similar. Analysis of the strychnine ⌬I dose-response relationships yielded IC 50 and n H values of 114.2 Ϯ 9.4 nM and 1.4 Ϯ 0.1 (n ϭ 5), whereas the ⌬F dose responses were fitted by IC 50 and n H values of 142.5 Ϯ 14.4 nM and 1.3 Ϯ 0.2 (n ϭ 5). Thus, strychnine itself cannot induce conformational changes that are detectable with a fluorophore attached to ␣1R19ЈC, although it can antagonize glycine-induced conformational changes. This response profile is expected for a classical competitive antagonist.
On the basis of a substituted cysteine accessibility analysis, it was recently proposed that picrotoxin can induce a conformational change in the M2-M3 loop that is not produced by glycine (14). We thus hypothesized that picrotoxin may produce a different relationship between ⌬I and ⌬F than was produced by glycine. Application of picrotoxin alone did not produce any changes in ⌬I or ⌬F, even at a concentration of 50 M (Fig. 5D). Although co-application of 50 M picrotoxin with 1 mM glycine potently blocked the current, the ⌬F remained intact (Fig. 5E). Dose-response analysis of ⌬I yielded an IC 50 of 626.1 Ϯ 11.4 nM and an n H of 1.0 Ϯ 0.2 (both n ϭ 6). However, in the same six oocytes, ⌬F showed full amplitude over a broad range of picrotoxin concentrations (Fig. 5F). This response profile is expected for a classical open state blocker that plugs the pore without  lower panels), 5% ⌬F. J, spectral emission from MTSR-labeled ␣1R19ЈC oocytes before, after, and during the application of 1 mM glycine and 1 mM glycine ϩ 100 M ginkgolide C (average of five cells each). K, difference emission spectra from MTSR-labeled ␣1R19ЈC oocytes recorded during application of glycine and glycine ϩ ginkgolide C, respectively. Both the gray trace (spectral emission before application of agonist) and the blue trace are normalized to the 1 mM glycine emission peak. Difference emission spectra were calculated as described in Fig. 2D.
altering the normal open state conformation. Although it provides no evidence to support the previous conclusion for picrotoxin imposing a distinct conformational change in the M2-M3 loop (14), we cannot eliminate the possibility that such a conformational change may occur without a change in hydrophobicity at R19ЈC.
We next investigated the effects of the GlyR pore blocker ginkgolide C. As with strychnine and picrotoxin, high concentrations (100 M) of ginkgolide C applied alone did not evoke detectable ⌬I or ⌬F (Fig. 5G). When co-applied with 1 mM glycine, ginkgolide C decreased ⌬I but dramatically increased ⌬F (Fig. 5H). Dose responses averaged from six oocytes revealed an averaged ⌬I IC 50 of 4.6 Ϯ 2.4 M and an n H of 0.5 Ϯ 0.1 and an averaged ⌬F EC 50 of 19.6 Ϯ 4.1 M and an n H of 1.0 Ϯ 0.1 (Fig. 5I). Spectral analysis revealed that the effect of ginkgolide C on the MTSR emission spectrum was similar to that produced by saturating glycine concentrations (compare Fig. 5K with Fig. 2D, inset). An exception is the ginkgolide C-induced hump near 625 nm for which we have no explanation. Overall, the results indicate that pore block by ginkgolide C induces a conformational change in M2-M3 similar to that produced by saturating concentrations of glycine.
Analysis of the Effects of the GlyR Potentiating Agent, Propofol-Propofol is an intravenous anesthetic that potentiates GlyRs by binding to the alcohol and anesthetic binding site located between TM2 and TM3 (26,27). Its potentiation, particularly of taurine-and ␤-alanine-gated currents, is facilitated by mutations to R19Ј in ␣1 GlyRs (28). As propofol is hydrophobic and its putative binding site lies in close proximity to the 19Ј-labeled MTSR, it may directly affect fluorescence by modulating the hydrophobicity of the environment surrounding the fluorophore. However, there was no significant change in the basal fluorescence level when MTSR-labeled ␣1R19ЈC GlyRs were incubated with 500 M propofol for 2 min. Similarly, preincubation with propofol and subsequent MTSR labeling in the presence of propofol (500 M each) did not change ⌬F max or the resting fluorescence intensity (data not shown). Following a 2-min incubation with 500 M propofol, MTSR-labeled ␣1R19ЈC GlyRs retained the ability to produce ∆I α19'C tau + Pro ∆F α19'C tau + Pro ∆I α19'C tau ∆F α19'C tau ∆I α19'C gly ∆F α19'C gly ∆I α19'C β-ala + Pro ∆F α19'C β-ala + Pro ∆I α19'C β-ala ∆F α19'C β-ala ∆I α19'C gly ∆F α19'C gly robust current and fluorescence changes in response to increasing glycine concentrations (Fig. 6A). Propofol pretreatment significantly increased the ⌬I max but not the ⌬F max value and significantly reduced the mean glycine EC 50 values for both ⌬I and ⌬F ( Fig. 6B and Table 2). The same propofol pretreatment drastically increased the taurine-and ␤-alanine-mediated ⌬I responses although their ⌬F values were less affected (Fig. 6, C and D). The effects of propofol on averaged ⌬I and ⌬F dose responses for taurine and ␤-alanine are summarized in Fig. 6, E and F, respectively, with mean parameters of best fit to the dose-response curves summarized in Table 2. As with glycine-mediated responses, the taurine-and ␤-alanine-mediated ⌬F EC 50 values were shifted to significantly lower values. The ⌬F max value for taurine was significantly increased by propofol, whereas the corresponding value for ␤-alanine was not. The ⌬I max values for both taurine and ␤-alanine were also significantly increased by propofol. Considering that propofol converted ␤-alanine and taurine into highly efficacious agonists, we tested whether their ⌬F characteristics may have changed to more closely resemble those produced by glycine. Surprisingly, however, propofol had no effect on the spectral maxima of the ⌬F increases induced by either ␤-alanine or taurine (Fig. 6, G and H).

DISCUSSION
Structural Basis for Fluorescence Change-Given the distance of R19ЈC and L22ЈC from the glycine-binding site (12) and their position midway along the conformational wave that extends from the binding site to the activation gate upon agonist binding (11), we propose that fluorescence signals arising from rhodamine-labeled R19ЈC and L22ЈC are not likely to simply reflect ligand binding. They are more likely to result from a rotation or tilting of the M2 helix during the conformational change that ultimately leads to channel opening. During this global conformational rearrangement, the MTSR fluorophore at the 19Ј position may move to a more hydrophobic environment. TMRM attached to L22ЈC moves into a more hydropho-bic environment when activated by glycine, but it moves into a more hydrophilic environment when activated by taurine or ␤-alanine. However, we cannot rule out the possibility that a quenching group moving toward or away from the fluorophores may contribute to the observed fluorescence changes. Our experiments do not allow us to discriminate among the existing structural models of Cys loop receptor activation, although they do permit us to compare M2 domain conformational changes produced by different ligands.
Agonist-induced Conformational Changes-Perhaps our most dramatic result is that ivermectin activates the pore without producing a significant ⌬F in the 19Ј-attached MTSR. Although it has been suggested previously that ivermectin and glycine activate the GlyR pore via structurally different mechanisms (21), the present result provides a much stronger case for this conclusion. The Cys loop receptor ivermectin-binding site is yet to be identified. Being lipophilic, it is possible that ivermectin binds either to a transmembrane or an intracellular site. Regardless of where it binds or how it activates the receptor, it is surprising that ivermectin opens the pore with no detectable change in the conformation at the top of M2. This raises the intriguing possibility that ivermectin produces conformational changes at the internal part of the M2 domain only, perhaps via an interaction with the M1-M2 linker.
Taurine and ␤-alanine are low efficacy glycinergic agonists that are converted into antagonists by R19Ј mutations (20,23,24). Although their efficacy at activating GlyR currents was very weak, both ligands evoked large ⌬F responses that were not accompanied by blue shifts in the spectral emission peak (Fig. 3,  F and G). This response profile differs drastically from that produced by glycine, which was characterized by a close coupling between ⌬I and ⌬F and a substantial blue shift in the emission spectral peak. Thus, the conformational change at R19Ј produced by these ligands is distinct from that produced by glycine. This is not surprising because it was already evident from their low efficacy that taurine and ␤-alanine must produce Dashed lines indicate interrupted illumination to prevent photobleaching between applications. G and H, difference emission spectra from MTSRlabeled ␣1R19ЈC oocytes recorded during application of saturating [taurine] (G) and [␤-alanine] (H) before and after a 2-min preincubation with 500 M propofol (ϩPro), respectively. Emission peaks after propofol treatment were normalized to those recorded before propofol treatment. Difference emission spectra were calculated as described in Fig. 2D.

TABLE 2
Summary of results for glycine-, taurine-, and ␤-alanine-evoked current and fluorescence recordings from MTSR-labeled ␣1R19C oocytes different conformational changes to glycine. However, propofol, which drastically increased the efficacy with which taurineand ␤-alanine were able to activate ⌬I, had no effect on the spectral properties of the concomitant ⌬F (Fig. 6, G and H). This result is more significant as it shows that taurine and ␤-alanine produce no ⌬F spectral change even when they strongly activate the GlyR. This provides a stronger case for taurine and ␤-alanine activating the GlyR pore via a different conformational change to glycine. However, a weakness in this argument is that the R19ЈC mutation may have altered structure and function to the extent that the results may not apply to the WT GlyR.
For this reason, we compared the agonist-specific conformational changes experienced by a TMRM attached to the nearby L22ЈC residue. Again, we found that glycine produced a different ⌬F response to that elicited by taurine and ␤-alanine (Fig. 4,  C and D). This shows that the agonist-specific conformational changes do not pertain only to the functionally impaired ␣1R19ЈC GlyRs but also to a nearby mutant with functional characteristics that more closely resemble the ␣WT GlyR.
Together, these results lead us to conclude that different agonists activate the GlyR by producing different conformational changes to the external region of the M2 domain. Thus, the top of M2 seems to display a conformational mobility that is not necessarily coupled to movements of the channel gate. Conversely, opening of the channel gate does not necessarily propagate back to the top of M2.
Antagonists-Strychnine has long been considered a classical competitive antagonist of the GlyR (1). Our finding that strychnine exhibits very similar inhibitory dose response profiles for ⌬I and ⌬F is consistent with this notion. By displacing glycine from its site, strychnine prevents glycine from binding and hence from inducing its ⌬F change.
Picrotoxin is thought to bind in the GlyR pore and to produce inhibition via an allosteric mechanism (25). We recently concluded on the basis of a substituted cystine accessibility study that it changes the M2-M3 loop conformation in a manner that cannot be achieved by glycine (14). If so, then picrotoxin might be expected to alter ⌬F in a way that cannot be achieved by altering glycine concentration alone. However, we found that ⌬F produced by picrotoxin plus glycine equals that produced by glycine alone (Fig. 5, D-F). This response profile is expected for a classical pore blocker that blocks current flow without altering the open state receptor conformation. However, we cannot rule out the possibility that picrotoxin changes the conformation of the M2-M3 loop in a way that does not involve a ⌬F change.
Several lines of evidence suggest that ginkgolides bind in the GlyR pore (29 -31), but there is no evidence to date that they produce a conformational change upon binding. Ginkgolides thus bind close to the covalently linked fluorophore. The results of Fig. 5, G-K, demonstrate that ginkgolide C-induced inhibition is accompanied by both an increase in ⌬F and a blue shift in the emission spectra maxima. Unexpectedly, ginkgolide C affects ⌬F at higher concentrations than those affecting ⌬I (Fig.  5I). This is difficult to reconcile with a model that assumes only one ginkgolide molecule binding per receptor. We propose that the ⌬F changes are mediated by ginkgolide C binding to a dis-crete, lower affinity site distinct from the well characterized pore blocker site at the 6Ј level. In support of this, a molecular docking study identified a putative ginkgolide-binding site near the M2-M3 loop (31). An alternate explanation is that the relatively large size of ginkgolides may force the pore to become stabilized in the activated state. However, the observed discrepancy between the ⌬F and ⌬I ginkgolide dose responses is not readily explainable by this model.
Comparison with Previous VCF Studies-Chang and Weiss (17) employed VCF to investigate molecular rearrangements at three positions in and around the recombinant 1 ␥-aminobutyric acid type A receptor ligand binding pocket. They found that two GABAergic (where GABA is ␥-aminobutyric acid) agonists produced a similar pattern of ⌬F responses at the three labeled positions. Competitive antagonists prevented these ⌬F changes but also produced distinct ⌬F responses in the absence of agonist. Perhaps most interestingly, picrotoxin produced a large ⌬F at one of the labeled positions, providing strong evidence that by binding in the pore it produces global conformational change (17). A more recent study on ␣1␤2 and ␣1␤2␥2 ␥-aminobutyric acid type A receptors showed that the ⌬F magnitude monitored at two binding domain residues exhibited subunit dependence (18). The results from these two studies provide strong evidence for ligand-and subunit-specific conformational changes in the ligand binding domain. However, they provide no information about whether these differential movements translate into different movements at the pore.
Finally, Dahan et al. (15) investigated acetylcholine-and epibatidine-mediated ⌬F and ⌬I changes at muscle (␣␤␥␦) nAChRs via a rhodamine label attached to the ␤ subunit 19Ј position. They concluded that ⌬F reported a conformational change at the ␣␦ subunit interface. The ⌬F signal was also modulated by the receptor desensitization status. Together with the present study, these results indicate that the M2 conformation reflects a wide variety of influences, including the identity of the bound agonist, the molecular identity of the labeled subunit, and the desensitization status of the receptor.
Conclusion-VCF recordings reveal that the GlyR agonists, glycine, taurine, and ivermectin, produce different ⌬F changes at fluorescently labeled 19Ј and 22Ј residues. The inhibitors, strychnine and picrotoxin, produced ⌬F and ⌬I changes as expected for a competitive antagonist and a channel blocker, respectively. On the other hand, the putative channel blocker, ginkgolide C, produced an increase in ⌬F that may have signaled a direct interaction between MTSR and a low affinity ginkgolide site. Taken together, our results suggest that the GlyR M2 domain responds with distinct conformational changes to the binding of different agonists.