Advertisement

Evidence that the TM1-TM2 Loop Contributes to the ρ1 GABA Receptor Pore*

  • Natalia Filippova
    Affiliations
    Department of Neurobiology, University of Alabama at Birmingham School of Medicine, Birmingham, Alabama 35294
    Search for articles by this author
  • Virginia E. Wotring
    Affiliations
    Department of Neurobiology, University of Alabama at Birmingham School of Medicine, Birmingham, Alabama 35294
    Search for articles by this author
  • David S. Weiss
    Correspondence
    To whom correspondence should be addressed: Dept. of Neurobiology, UAB School of Medicine, 1719 Sixth Ave So. CIRC410, Birmingham, AL 35294. Tel.: 205-975-5093; Fax: 205-934-4066;
    Affiliations
    Department of Neurobiology, University of Alabama at Birmingham School of Medicine, Birmingham, Alabama 35294
    Search for articles by this author
  • Author Footnotes
    * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
      Considerable evidence indicates the second transmembrane domain (TM2) of the γ-aminobutyric acid (GABA) receptor lines the integral ion pore. To further delineate the structures that constitute the ion pore and selectivity filter of the ρ1 GABA receptor, we used the substituted cysteine accessibility method with charged reagents to identify anion- and cation-accessible surfaces. Twenty-one consecutive residues were mutated to cysteine, one at a time, in the presumed intracellular end of the first transmembrane domain (TM1; Ala271-Met276), the entire linker connecting TM1 to TM2 (Leu277-Arg287), and the presumed intracellular end of TM2 (Ala288-Ala291). Positively (MTSEA+) and negatively (pCMBS) charged sulfhydryl reagents, as well as Cd2+, were added extracellularly to test accessibility of the engineered cysteines. Four of the mutants, all at the intracellular end of TM2 (R287C, V289C, P290C, A291C), were accessible to positively charged reagents, whereas seven mutants (A271C, T272C, L277C, W279C, V280C, P290C, A291C) were functionally modified by negatively charged pCMBS. These seven modified residues were at the intracellular end of TM2, in the TM1-TM2 linker, and at the intracellular end of TM1. In nearly all cases (excluding P290C), the rate and the degree of modification were state-dependent, with greater accessibility in the presence of agonist. Select cysteine mutants were combined with a point mutation (A291E) that converted the pore from chloride- to non-selective. In this case, positively charged reagents could modify residues in the TM1-TM2 linker (Leu277 and Val280), supporting the notion that the modifying reagents were reaching their target through the pore. Taken together, our results suggest that, up to its intracellular end, the TM2 domain is not charge selective. In addition, we propose that the TM1-TM2 linker and the intracellular end of TM1 are along the pathway of the permeating ion. These findings may lend new insights into the structure of the GABA receptor pore.
      Ligand-gated ion channels play a fundamental role in neuronal communication. The γ-aminobutyric acid (GABA)
      The abbreviations used are: GABA, γ-aminobutyric acid; nACh, nicotinic acetylcholine; TM1–TM4, transmembrane domains 1–4; MTSEA, 2-aminoethyl methanethiosulfonate hydrobromide; pCMBS, p-chloromercuribenzene sulfonic acid.
      1The abbreviations used are: GABA, γ-aminobutyric acid; nACh, nicotinic acetylcholine; TM1–TM4, transmembrane domains 1–4; MTSEA, 2-aminoethyl methanethiosulfonate hydrobromide; pCMBS, p-chloromercuribenzene sulfonic acid.
      receptor is the major inhibitory neurotransmitter-activated ion channel in mammalian brain and is an important target for a variety of clinically prescribed therapeutic compounds. Detailed knowledge of its structure and function can provide important information for understanding the fundamentals of neuronal communication, the treatment of neurological disorders, and the design of therapeutic compounds that target the GABA receptor.
      The family of ligand-activated ion channels that includes the GABA, nicotinic acetylcholine (nACh), glycine, and serotonin type 3 receptors (
      • Noda M.
      • Takahashi H.
      • Tanabe T.
      • Toyosato M.
      • Furutani Y.
      • Hirose T.
      • Asai M.
      • Inayama S.
      • Miyata T.
      • Numa S.
      ,
      • Grenningloh G.
      • Rienitz A.
      • Schmitt B.
      • Methfessel C.
      • Zensen M.
      • Beyreuther K.
      • Gundelfinger E.D.
      • Betz H.
      ,
      • Cutting G.R.
      • Lu L.
      • O'Hara B.F.
      • Kasch L.M.
      • Montrose-Rafizadeh C.
      • Donovan D.M.
      • Shimada S.
      • Antonarakis S.E.
      • Guggino W.B.
      • Uhl G.R.
      • Kazazian H.H.
      ,
      • Schofield P.R.
      • Darlison M.G.
      • Fujita N.
      • Burt D.R.
      • Stephenson F.A.
      • Rodriguez H.
      • Rhee L.M.
      • Ramachandran J.
      • Reale V.
      • Glencorse T.A.
      • Seeburg P.H.
      • Barnard E.A.
      ,
      • Maricq A.V.
      • Peterson A.S.
      • Brake A.J.
      • Myers R.M.
      • Julius D.
      ) is composed of a group of proteins with diverse functional properties but homologous structure. This structure includes four predicted transmembrane domains (TM1–TM4), a large extracellular amino-terminal domain containing the agonist-binding site, and a large intracellular linker between TM3 and TM4 that has been suggested to play a key role in receptor regulation and targeting. Evidence suggests that the TM2 domain of each of the five subunits forms the central ion pore (
      • Leonard R.J.
      • Labarca C.G.
      • Charnet P.
      • Davidson N.
      • Lester H.A.
      ,
      • Akabas M.H.
      • Stauffer D.A.
      • Xu M.
      • Karlin A.
      ,
      • Xu M.
      • Akabas M.H.
      ,
      • Hucho F.
      • Oberthur W.
      • Lottspeich F.
      ). The location of the charge selectivity filter, although still somewhat controversial, has been localized to the intracellular end of the TM2 domain because mutations in this region can reverse the charge selectivity of both cationic (nACh and 5-HT3) and anionic (GABA and glycine) receptors (
      • Galzi J.L.
      • Devillers-Thiery A.
      • Hussy N.
      • Bertrand S.
      • Changeux J.P.
      • Bertrand D.
      ,
      • Keramidas A.
      • Moorhouse A.J.
      • French C.R.
      • Schofield P.R.
      • Barry P.H.
      ,
      • Keramidas A.
      • Moorhouse A.J.
      • Pierce K.D.
      • Schofield P.R.
      • Barry P.H.
      ,
      • Wotring V.E.
      • Miller T.S.
      • Weiss D.S.
      ,
      • Gunthorpe M.J.
      • Lummis S.C.
      ).
      In this study, we have used the substituted cysteine accessibility method (
      • Karlin A.
      • Akabas M.A.
      ) along with mutations that alter the ion selectivity (
      • Wotring V.E.
      • Miller T.S.
      • Weiss D.S.
      ) to test the potential role of the TM1-TM2 loop and adjacent regions in ion permeation through the ρ1 GABA receptor pore. Employing positively (MTSEA+, Cd2+) and negatively (pCMBS) charged reagents that interact with these cysteine residues, we determined that the intracellular TM2 domain, all the way to its intracellular end, is non-selective in terms of charge. Interestingly, the TM1-TM2 linker and the intracellular end of TM1 are only accessible to the negatively charged reagent, pCMBS. After converting the pore to be non-selective by incorporating the A291E mutation, we find three residues in the TM1-TM2 linker that are now accessible to positively charged reagents. The results are interpreted in terms of the pore architecture, location of the selectivity filter, and potential structural rearrangements during channel activation.

      EXPERIMENTAL PROCEDURES

      Mutagenesis of GABA ρ1 Receptor and in Vitro RNA Transcription— The human ρ1 subunit was obtained through PCR as described previously (
      • Amin J.
      • Weiss D.S.
      ) and cloned into the pGEMHE vector (
      • Liman E.R.
      • Tytgat J.
      • Hess P.
      ). The single A271C, T272C, L273C, M274C, V275C, M276C, L277C, S278C, W279C, V280C, S281C, F282C, W283C, I284C, D285C, R286C, R287C, A288C, V289C, P290C, A291C and double A271C/A291E, T272C/A291E, M276C/A291E, L277C/A291E, W279C/A291E, V280C/A291E, R287C/A291E mutations were accomplished with PCR mutagenesis using an overlap extension protocol (
      • Kammann M.
      • Laufs J.
      • Schell J.
      • Gronenborn B.
      ). All constructs were verified by cDNA sequencing.
      The cDNA of each clone was linearized with NheI, and run-off capped cRNA was transcribed from the linearized cDNA with standard in vitro transcription procedures (
      • Amin J.
      • Dickerson I.
      • Weiss D.S.
      ). The cRNA was purified using the RNeasy Mini Kit (Qiagen, Valencia, CA), and the integrity and yield were verified on a 1% agarose gel.
      Expression and Recording from Oocytes—Oocytes from Xenopus laevis were prepared and maintained as described previously (
      • Amin J.
      • Weiss D.S.
      ). Individual ρ1 wild-type or mutant cRNA were injected into each oocyte (total of 0.5–5 ng). A two-electrode voltage-clamp (GeneClamp 500, Axon Instruments, Foster City, CA) was used for current recording 3 or 4 days after cRNA injection at a holding potential from –70 to –50 mV. The current signal was low-pass filtered at 10 Hz and digitized at 50 Hz. Data were analyzed using Igor software (Wavemetrics, Lake Oswego, OR). Dose-response relations were fit with the following form of the Hill equation using a nonlinear least-squares method
      I=Imax1+(EC50/[A])n
      (Eq. 1)


      in which I is the peak current response at a given concentration of agonist (A), Imax is the maximum current response, EC50 is the concentration of the agonist yielding half-maximal activation, and n is the Hill coefficient. Data were compared statistically by a Student's t test. Statistical significance was determined at the 5% level. All results are presented as the mean ± S.E.
      The external recording solution contained 92.5 mm NaCl, 2.5 mm KCl, 1 mm CaCl2, 2.5 mm MgCl2, 5 mm Hepes, pH 7.5. The electrodes were filled with 3 m KCl and had resistances ranging from 1 to 3 mΩ. The ground electrode was connected to the recording bath by means of a 3 m KCl/agar bridge.
      Determination of Permeability Ratios—Details of the method for determining the relative permeability to sodium, potassium, and chloride have been provided previously (
      • Wotring V.E.
      • Chang Y.
      • Weiss D.S.
      ). In all cases, GABA (Calbiochem) was applied at the concentration required for half-maximal activation (EC50), and voltage ramps were used to determine reversal potentials under the various ionic conditions. The permeabilities (P) of ions relative to chloride were then calculated using the Goldman Hodgkin-Katz equation (
      • Hodgkin A.L.
      • Katz B.
      ) with internal ion activities used previously (
      • Wotring V.E.
      • Chang Y.
      • Weiss D.S.
      ) and the known activities for external chloride and replacement ions. Permeability coefficients were determined using Eq. 2, and reversal potentials (Vrev) were measured at various activities of extracellular chloride and sodium.
      Vrev=58log[(PNa[Na]i+PK[K]i+PCl[X]o)/(PNa[y]o+Pk[K]o+PCl[Cl]i)]
      (Eq. 2)


      Permeability coefficients for sodium and potassium were fit simultaneously (holding PCl at 1 for normalization purposes) with reversal potentials measured in OR2, low sodium, and low chloride. Data from five individual oocytes were fitted separately and then averaged to provide a standard error.
      Accessibility Tests of Mutated Residues to pCMBS, MTSEA+, and Cd2+We tested the accessibility of cysteines to sulfhydryl reagents applied extracellularly in the presence or in the absence of GABA. For each mutant, we used a GABA concentration at least 8-fold higher than the EC50. Before sulfhydryl reagent application, each oocyte was stabilized in the recording solution for 3–10 min, at which time three GABA-evoked currents were obtained. Experiments were continued if the amplitude of the control GABA-activated current varied by less than 3%. After sulfhydryl reagent application, the oocyte was washed for 3–4 min in the recording solution, and test GABA pulses were applied. The amplitude of the current after exposure to the modifying reagent was normalized to the control amplitude. Functional modification of wild-type and mutant receptors was compared using the Student's t test (p ≤ 0.05) or the analysis of variance Dunnet test (p < 0.5).
      To estimate the rate of modification by sulfydryl reagents, the amplitude of the test GABA pulse was plotted as a function of the cumulative time in the reagent (applied for 15 s every 3–5 min). The relationship between current amplitude and time was fitted (and well described) by a single exponential function. Data were compared statistically by the Student's t test. Statistical significance was determined at the 5% level. All results are presented as the mean ± S.E.

      RESULTS

      General Properties of the Cysteine Mutants—Twenty-one residues at the presumed intracellular ends of TM1 and TM2, as well as the linker connecting TM1 to TM2, were individually mutated to cysteine in the ρ1 GABA receptor (Fig. 1). The properties of the homomeric ρ1 mutants were probed by GABA application (from 0.1 μm to 100 mm) 2–3 days after cRNA injection into Xenopus oocytes. Nineteen mutants showed robust GABA-activated currents, whereas M274C and D285C exhibited no functional expression. We do not know whether these two mutant receptors were expressed on the cell surface and were non-functional, or whether the mutation impaired receptor assembly. Fig. 2A shows GABA-mediated responses for wild-type receptors and select mutants. L273C, M276C, L277C, W279C, S281C, I284C, R287C, A288C, V289C, P290C, and A291C exhibited minor (< 3-fold) shifts in the GABA EC50 (concentration required for half-maximal activation) compared with the wild-type receptor (Fig. 2B). The largest shift in EC50 (∼22-fold increase) was observed for V280C.
      Figure thumbnail gr1
      Fig. 1Schematic representation of ρ1 GABA receptor showing the mutated residues. *, residues in TM1, TM2, and the TM1-TM2 linker domains that were mutated to cysteine (one at a time). In the TM2 prime notation, Arg292 would correspond to 0′.
      Figure thumbnail gr2
      Fig. 2Effects of cysteine mutations on GABA receptor function. A, examples of GABA-activated currents from the wild-type receptor and select cysteine mutants. S281C exhibited a steady-state current in the absence of GABA that was antagonized by picrotoxin, suggesting the mutant was spontaneously active. Note that Met274 and Asp285 were non-functional. B, average of the EC50 for all cysteine mutants compared with that of the wild type. Mutants that exhibited a signficant difference from the wild-type receptor are indicated by asterisk. Only V280C exhibited a dramatic change in GABA sensitivity. C, average of the relative permeabilities for potassium (□) and sodium (▪) relative to that of Cl. All mutants retained their high selectivity for Cl over that of sodium and potassium. For comparison, the ΔP290/A291E and A291E mutations plotted to the right of the graph show substitutions that have been shown previously to convert the receptor to cationic and non-selective (
      • Wotring V.E.
      • Miller T.S.
      • Weiss D.S.
      ).
      Some of the cysteine mutations imparted other changes in receptor properties. For example, although wild-type ρ1 receptors show essentially no desensitization, M276C, F282C, I284C, R287C, V289C, and A291C exhibited a decay in the GABA-mediated current during prolonged agonist application (Fig. 2A). In addition, oocytes expressing S281C exhibited a greater than normal resting current in the absence of GABA that was blocked by the GABA receptor antagonist picrotoxin, suggesting the receptor was spontaneously active (Fig. 2A). Finally, W279C exhibited a dramatically prolonged current decay after GABA removal (τ = 200 ± 20 s, n = 6), compared with that of the wild-type ρ1 receptor (τ = 20 ± 3 s, n = 8).
      We also examined all the functional cysteine mutants for changes in the relative cation and anion permeability. Fig. 2C is a plot of the relative permeability of sodium (black squares) and potassium (white squares) compared with that of chloride. For comparison, the wild-type permeability ratio for Na:K:Cl was 0:0.03:1.0 (
      • Wotring V.E.
      • Chang Y.
      • Weiss D.S.
      ). The symbols to the right of the graph in Fig. 2C indicate the sodium and potassium permeability for the ΔP290/A291E mutant that has been previously shown to convert the receptor from anionto cation-selective and the A291E mutant that has been previously shown to convert the receptor to be essentially equally permeable for Na, K, and Cl (
      • Wotring V.E.
      • Miller T.S.
      • Weiss D.S.
      ). Note that all cysteine mutations retained their high degree of selectivity for anions over cations. Overall, the cysteine mutations induced minor changes in gating and selectivity when compared with the ρ1 wild-type receptor, suggesting the general structure, as well as the activation and permeation mechanisms, were similar to that of the wild type.
      Accessibility of the Cysteines for Positively Charged Reagents—In an effort to identify residues that line the cation-accessible surfaces of the pore, the accessibility of the cysteines were probed with the positively charged sulfhydryl reagent MTSEA (2 mm, 1-min application) or Cd2+ (100 μm, 1 min) added extracellularly both in the absence and presence of GABA. MTSEA increased the maximum current amplitude of R287C and V289C and decreased the maximum current amplitude of P290C compared with wild-type ρ1, in which no significant changes were observed (Fig. 3). The degree of modification of R287C and P290C was significantly higher in the presence of GABA, whereas V289C was the same in the absence or presence of agonist. Although three of the cysteine mutants exhibited a significant change in the maximum current with MTSEA modification, only P290C demonstrated a significant change in the EC50 for GABA (Fig. 3B). Finally, A291C, a mutant that did not demonstrate a change in the maximum current or EC50 with MTSEA exposure, exhibited a change in the rate of desensitization (Fig. 3C). In this case, the cysteine substitution on its own imparted an enhanced desensitization rate, and MTSEA reduced this rate by a factor of 2.5 ± 0.5 (n = 5). Thus, it seems that four of the examined residues (R287C, V289C, P290C, and A291C) were modified by the positively charged reagent, MTSEA.
      Figure thumbnail gr3
      Fig. 3Modification by MTSEA+ and Cd2+. Effects of MTSEA on the peak current amplitude (A) and EC50 (B). Mutants that exhibited a significant difference from the wild-type receptor are indicated by the asterisks at the bottom of A and B. In all cases, 2 mm MTSEA was applied for 1 min. The black symbols represent MTSEA in the absence of GABA, and the white symbols are in the presence of a saturating concentration of GABA (8-fold higher than the EC50 for that particular mutant). C, summary of the MTSEA modification results. In the case of A291C, although there was no significant change in the EC50 or maximum current, the rate of desensitization increased after MTSEA exposure, indicating modification had taken place. D, effects of Cd2+ (100 μm) on the peak current amplitude and on the EC50 of the wild type and mutants either in the absence (▪) or in the presence (□) of GABA.
      Two or more cysteines, if they are within 6–7 Å, can form a high affinity divalent cation-binding site. Therefore, we examined the actions of Cd2+ on select cysteine mutants. Fig. 3D is a plot of the maximum GABA-activated current for wild type and A291C, P290C, V289C, R287C mutant receptors after Cd2+ (in the presence (white symbol) and absence (black symbol) of GABA. Note, that Cd2+ significantly reduced the maximum current amplitude for P290C and A291C in the presence of GABA when compared with the wild-type receptor (no significant change, p > 0.20, n = 5). In addition, Cd2+ significantly decreased desensitization of the A291C mutant in the presence of GABA. Thus, Cd2+ significantly modified P290C and A291C mutants. That Cd2+ does not modify R287C may indicate these cysteines are not at the proper orientation to coordinate the divalent cation or, alternatively, the divalent cation cannot reach a site that deep in the pore. The observation that P290C and A291C are not blocked by Cd2+ in the absence of GABA may indicate that structural rearrangements occur in this region upon receptor activation that orient the cysteines in the proper configuration to coordinate the divalent cation.
      Accessibility of the Engineered Cysteines for Negatively Charged Sulfhydryl Reagents—Having evidence that positively charged reagents can modify cysteines near the intracellular end of TM2, we next tested the actions of the negatively charged reagent, pCMBS. Fig. 4A plots the maximum GABA-activated current after pCMBS exposure (1 mm, 1 min) in the presence (white symbol) and absence (black symbol) of GABA. pCMBS induced significant modification of the maximum current in five mutants (A271C, T272C, L277C, V280C, and P290C) when compared with the wild-type ρ1 receptor (no changes, n = 6). Fig. 4B presents the change in the EC50 after pCMBS exposure. In this case, the wild-type ρ1 receptor exhibited a small but significant decrease in the GABA EC50 (from 0.9 ± 0.1 to 0.51 ± 0.5 μm; n = 5). Three mutants (T272C, V280C, P290C) exhibited a further decrease, and one mutant (A271C) exhibited an increase in the EC50 for GABA after pCMBS modification.
      Figure thumbnail gr4
      Fig. 4Modification by pCMBS. Effects of MTSEA on the peak current amplitude (A) and EC50 (B). Mutants that exhibited a significant difference from the wild-type receptor are indicated by the asterisks at the bottom of A and B. In all cases, 1 mm pCMBS was applied for 1 min. The black symbols represent pCMBS in the absence of GABA, and the white symbols are in the presence of a saturating concentration of GABA (8-fold higher than the EC50 for that particular mutant). C, in addition to exhibiting an increase in the EC50, A271C was locked into the open state after modification. Although W279C and A291C did not show a change in the current amplitude or EC50 after pCMBS exposure, there were changes in the deactivation kinetics, confirming modification had occurred. D, summary of modification of cysteine mutants by pCMBS.
      Several mutants exhibited other changes in receptor properties after pCMBS treatment, indicating modification. For example, in addition to the increased EC50, A271C exhibited a steady current nearly equal in magnitude to the GABA-activated current before pCMBS exposure (Fig. 4C, panel 1). Picrotoxin partially antagonized this agonist-independent current. In the case of W279C, the deactivation kinetics changed from a non-exponential to a double-exponential decay (τ1 = 20 ± 5 s and τ2 = 300 ± 50 s; n = 6; Fig. 4C, panel 2). As was true for MTSEA exposure, A291C exhibited a significant decrease in the desensitization rate after modification (from τ = 3.5 ± 0.5 to 9 ± 2 s, n = 5; Fig. 4C, panel 3). Note that in all cases, the residues were modified in both the absence and presence of saturating concentrations of GABA. In sum, pCMBS induced a functional modification in seven mutants: A271C, T272C, L277C, W279C, V280C, P290C, and A291C (Fig. 4D). These data indicate that, in addition to modifying residues at the intracellular end of TM2, the negatively charged reagent pCMBS has access to residues in the TM1-TM2 linker and at the presumed intracellular end of TM1.
      Fig. 5 presents a summary of the residues that were modified by positively and negatively charged sulfhydryl reagents. Functional modification of A291C and P290C mutants by Cd2+ suggests that the TM2 domain is permeable for divalent cations up to P290C in the open state. Accessibility of A291C, P290C, V289C, and R287C for MTSEA+ extends the region accessible for positively charged reagents to R287C. Because V280C, W279C, L277C, T272C, and A271C were modified only by negatively charged pCMBS, these residues may line the anion-accessible surface of the pore.
      Figure thumbnail gr5
      Fig. 5Summary of modification by MTSEA+ and pCMBS. M274C and D285C are blank because these two mutants were non-functional.
      The Rate of Modification Is State-dependent—Data presented thus far demonstrates that modifying reagents can reach their respective targets in either the absence or presence of GABA. In those experiments, however, high concentrations of reagents were used, and sufficient time was allowed for the reactions to approach completion. Therefore, these data do not allow us to discern the relative state-dependence of the accessibility (e.g. absence or presence of GABA). For such information, one must compare the rates of modification. To check whether the rate of modification was state-dependent, sulfhydryl reagents (at low concentrations) were applied for 15-s intervals (total duration of 90 s) in either the presence or in the absence of GABA between GABA test pulses without the modifying reagent present. The amplitude of the GABA-mediated test pulse was plotted as a function of cumulative time of reagent exposure, and these data were described by a single exponential function. In some experiments, the antagonist picrotoxin (from 0.1 to 1 mm) was applied along with the sulfhydryl reagent to test for the protection of cysteine modification.
      Fig. 6 presents examples of the time dependence of modification of P290C by MTSEA (Fig. 6A) and pCMBS (Fig. 6B) in the absence and presence of GABA as well as in the presence of GABA plus picrotoxin. In Fig. 6A, note the faster rate of modification of P290C in the presence of GABA. Also, picrotoxin slowed the modification significantly. For pCMBS, the rates were the same whether agonist was present or not (Fig. 6B). Picrotoxin (up to 0.5 mm) was unable to completely block the GABA-induced current for V289C, A271C, and A291C (not shown), so the protection from modification, although present, was likely compromised. In all other cases, picrotoxin protected the cysteine from modification. The data for the rate of modification are summarized in Fig. 6, C and D where the average time constants in the absence (black symbols) and presence (white symbols) of GABA are plotted. For MTSEA, a significantly faster rate was observed in the presence of GABA in all four mutants examined. For pCMBS, a significantly faster rate was observed in the presence of GABA in four of the five examined mutants. P290C showed a comparable rate in the absence or presence of GABA (Fig. 6B). For both MTSEA and pCMBS, there seemed to be a clear correlation between the amino acid position and the degree of modification in the absence of GABA, with the rate decreasing as one progresses from TM2 toward TM1. Regardless of the pattern, these data indicate that the accessibilities of these residues (excluding P290C) were dependent upon the receptor state. More specifically, the presence of GABA imparts structural changes in the receptor that facilitate modification.
      Figure thumbnail gr6
      Fig. 6Relative accessibility of select residues in the absence and presence of GABA. Examples of modification of P290C by MTSEA+ (A) and pCMBS (B). Traces are shown for MTSEA+, MTSEA+ plus GABA, and MTSEA+ plus 20 μm GABA and 50 μm picrotoxin. Sulfhydryl reagents were applied for 15-s intervals for a total duration of 90 s. Data from four cysteine mutants are summarized in C (MTSEA+) and D (pCMBS). Note that the modification in the presence of GABA was faster compared with the modification in the absence of GABA for all mutants, excluding P290C modification by pCMBS.
      The Accessibility of L277C, V280C, and R287C for Charged Reagents Correlates with Selectivity—As summarized in Fig. 5, the intracellular end of TM1 (A271C, T272C) and the TM1-TM2 linker (L277C, W279C, and V280C) were accessible for negatively charged pCMBS, but not positively charged MTSEA or Cd2+. One possible explanation for these findings is that a charge-selectivity filter lies somewhere between Val280 and Arg287 and thus allows only negatively charged molecules beyond that point (toward TM1). To test this possibility, we incorporated (along with select cysteine substitutions) the A291E mutation, which has been shown to convert the ρ1 anionic pore into a non-selective pore that is equally permeable to anions and cations (
      • Wotring V.E.
      • Miller T.S.
      • Weiss D.S.
      ). The double mutants, A271C/A291E, T272C/A291E, L277C/A291E, W279C/A291E, V280C/A291E, and R287C/A291E were then probed by positively and negatively charged sulfhydryl reagents.
      Fig. 7 shows modification of L277C, V280C, L277C/A291E, and V280C/A291E by positively charged MTSEA. MTSEA (2 mm, 1 min) did not modify the L277C mutant in either the absence or presence of agonist (Fig. 7A). However, MTSEA (2 mm, 1 min) induced a significant increase in the holding current of L277C/A291E (up to 3-fold; Fig. 7B), which was about equal to the level of the GABA-induced current before modification. In addition, MTSEA modification produced a significant decrease in the EC50 (Fig. 7C). MTSEA (2 mm, 2 min) did not significantly modify V280C in the absence or presence of agonist (Fig. 7, D and F). In contrast, MTSEA (2 mm, 2 min) induced a significant increase in the maximum current amplitude (41 ± 6%, n = 7) and decreased the EC50 (38 ± 8%, n = 5) of V280C/A291E in the presence of agonist (Fig. 7, E and F), although there was still no modification in the absence of GABA (not shown). Recall that V280C in the wild-type background was accessible to the negatively charged reagent pCMBS.
      Figure thumbnail gr7
      Fig. 7Change in the pattern of modification after conversion of the pore from chloride- to non-selective by a concomitant A291E mutation. A, L277C was not modified by MTSEA in the presence of GABA. B, when L277C was introduced into a receptor containing the A291E mutant, modification by MTSEA+ was observed. Note that modification produced a steady current, which was evident upon GABA removal, likely because of spontaneous receptor opening. C, comparison of the modification of V280C and V280C/A291E on the maximum current and EC. *, significant change of the maximum current or EC50. D, V280C was not modified by MTSEA in the presence of GABA. E, when V280C 50was introduced into a receptor containing the A291E mutant, modification by MTSEA+ was observed. F, comparison of the modification of V280C and V280C/A291E on the maximum current and EC50. *, significant change of the maximum current or EC50.
      We did not observe a significant modification of the maximum current amplitude for W279C/A291E, A271C/A291E, or T272C/A291E by MTSEA, nor did we observe a modification of the maximum current amplitude for A271C/A291E, T272C/A291E, L277C/A291E, V280C/A291E, or R287C/A291E by pCMBS (not shown). However, we did observe a significant potentiation of the current amplitude for W279C/A291E (45 ± 5%; n = 4) after pCMBS treatment, compared with the W279C mutant, where only the deactivation kinetics were altered after modification and not the maximum amplitude. In the case of R287C/A291E, there was a clear increase in the relative accessibility when compared with R287C. The maximum current was potentiated 12 ± 6% (n = 5) versus 50 ± 6% (n = 4) by 67 μm MTSEA for R287C and R298C/A291E, respectively. Nevertheless, the general finding is that converting the GABA receptor anionic pore to a non-selective pore allows positively charged reagents to modify residues beyond the intracellular end of TM2 in the TM1-TM2 linker, all the way to the putative edge of TM1.
      The Accessibility of P290C to Cd2+ from the Inside—The data in Fig. 3 indicate that Cd2+ was unable to access residues beyond P290C. Furthermore, this accessibility was observed only in the presence of agonist. One possibility is that, in the closed state (absence of agonist), there is a barrier to this point that slows access of Cd2+ to its site. A second possibility is that agonist-mediated structural changes in this region (e.g. channel opening) alter the accessibility. For example, the residue might be partially buried in the closed state and become more exposed and accessible in the presence of agonist. In an attempt to discern between these two possibilities, the accessibility of P290C was probed from the inside by injecting 27.6 nl of 1 mm Cd2+ into the oocyte, resulting in an approximate maximum final Cd2+ intracellular concentration of 10–14 μm (assuming an oocyte diameter of 0.5–1 mm). Although this approach is clearly not accurate enough to allow a comparison between the rates of accessibility from the inside versus the outside, it does allow a comparison of the accessibility from the inside in the absence and presence of GABA. Fig. 8A, top traces show GABA-mediated currents before and after Cd2+ injection, and the amplitudes are plotted in Fig. 8C, white circles, yielding a time constant of 10.5 ± 2.0 min (n = 5). Fig. 8A, lower traces show the current decline in the presence of agonist. Note that at 11 min, the first time point taken after the Cd2+ injection, the current already declined completely, indicating a much faster rate than in the absence of GABA (Fig. 8C, black circles). No such decline in current was observed in the P290C mutant without Cd2+ (Fig. 8C, black circles) or in the wild-type ρ1 receptor (Fig. 8, B and C, black squares). Thus, P290C exhibits a state-dependent decline in current by Cd2+ from both the outside and inside, suggesting an agonist-mediated structural change that alters the degree of exposure of this residue to Cd2+.
      Figure thumbnail gr8
      Fig. 8Modification of P290C by Cd2+ from the intracellular side of membrane. A, after an initial test of the current amplitude, 1 mm Cd2+ was injected into a volume of 27.6 nl which, after diffusion, would result in a final intracellular concentration (negating buffering, extrusion, etc.) of 10–14 μm, assuming an oocyte diameter of 0.5–1 mm. Examples of modification of P290C after Cd2+ injection (upper traces) and after Cd2+ injection plus an 8-min incubation in GABA (lower traces) are shown. B, effects of intracellular Cd2+ injection on the wild-type receptor. No modification was observed. C, time course of modification by intracellular Cd2+ for wild type and mutant under various conditions. Note that modification of P290C mutant was significantly faster in the presence of GABA. In each experiment, the GABA-evoked current was normalized to the current amplitude before Cd2+ injection.

      DISCUSSION

      Position of the Selectivity Filter—Using the substituted cysteine accessibility method in conjunction with negatively and positively charged reagents, we attempted to identify anion- and cation-accessible surfaces of the ρ1 GABA receptor pore. These studies should be considered an extension of previous work that demonstrated the TM2 domain in this family of receptors lines the ion pore (
      • Leonard R.J.
      • Labarca C.G.
      • Charnet P.
      • Davidson N.
      • Lester H.A.
      ,
      • Akabas M.H.
      • Stauffer D.A.
      • Xu M.
      • Karlin A.
      ,
      • Xu M.
      • Akabas M.H.
      ). Earlier studies on the α1β2γ2 GABA receptor confirmed this role for TM2 and, based upon the alternating pattern of TM2 cysteine accessibility, proposed that the TM2 domain adopts an α-helical secondary structure (
      • Xu M.
      • Akabas M.H.
      ,
      • Xu M.
      • Akabas M.H.
      ). Finally, based on the accessibility of positively and negatively charged cysteine modifying reagents, they proposed the charge selectivity filter to be at least as cytoplasmic as α1Thr261, a position that is approximately at the putative midpoint of TM2.
      In the present studies on the ρ1 GABA receptor, positively charged reagents were able to access as far as R287C. Comparing the aligned sequences, Arg287 of ρ1 is 11 residues more toward the amino-terminal domain (deeper in the pore) than α1Thr261 of the α1β2γ2 study (
      • Xu M.
      • Akabas M.H.
      ). In fact, ρ1Arg287 is at the interface between TM2 and the linker connecting TM1 to TM2. Our results for negatively charged reagents were even more striking when compared with α1β2γ2 (
      • Xu M.
      • Akabas M.H.
      ). In ρ1, pCMBS modified cysteine residues in the TM1-TM2 linker as far as six residues up into the putative TM1 domain from the intracellular end. It should be mentioned that these TM1 and TM2 assignments are from the original ρ1 cloning report (
      • Cutting G.R.
      • Lu L.
      • O'Hara B.F.
      • Kasch L.M.
      • Montrose-Rafizadeh C.
      • Donovan D.M.
      • Shimada S.
      • Antonarakis S.E.
      • Guggino W.B.
      • Uhl G.R.
      • Kazazian H.H.
      ). Studies from other members of this receptor-operated superfamily have these domains shifted by as many as four residues (
      • Bertaccini E.
      • Trudell J.R.
      ). This slight disagreement in topology does not negate the differential patterns of accessibility.
      At face value, our data place the selectivity filter at or beyond (in the TM1 direction) R287C. One question that arises is whether or not MTSEA can only access residues up to R287C because of its positive charge or rather because of other physical constraints such as size. We propose it is the former for three reasons. First, dimensions of MTSEA are presumed to be 4.1 × 4.6 × 7.3 Å, whereas estimates of the pore diameter are ∼6.1 Å (
      • Wotring V.E.
      • Chang Y.
      • Weiss D.S.
      ). The second piece of evidence is that pCMBS can access residues beyond R287C, and it is presumed to be somewhat larger than MTSEA (
      • Xu M.
      • Akabas M.H.
      ). And third, the concomitant mutation (A291E) that results in a pore that does not select between anions or cations, permits MTSEA to assess residues past R287C, well into the TM1-TM2 linker. Mutagenesis studies of the ρ1 GABA receptor place the selectivity filter somewhat more extracellular then R287C, around Pro290 through Arg292, a region that aligns with the intermediate ring in the nACh receptor (
      • Imoto K.
      • Busch C.
      • Sakmann B.
      • Mishina M.
      • Konno T.
      • Nakai J.
      • Bujo H.
      • Mori Y.
      • Fukuda K.
      • Numa S.
      ). One possible explanation is that in the present SCAM analysis, the cylindrically shaped MTSEA may be able to modify sidechains somewhat distant from the location of the reagent charged group in the pore. Also, one must consider that different mutations employed in the two studies can have differential effects on local structure. Finally, it should be cautioned that we are examining accessibility of these domains to cysteine-modifying reagents. The extent to which these results can be extrapolated to ion accessibility remains to be determined.
      Pore Topology—Pioneering studies on the nACh receptor, a member of the same family as that of the GABA receptor (
      • Noda M.
      • Takahashi H.
      • Tanabe T.
      • Toyosato M.
      • Furutani Y.
      • Hirose T.
      • Asai M.
      • Inayama S.
      • Miyata T.
      • Numa S.
      ,
      • Schofield P.R.
      • Darlison M.G.
      • Fujita N.
      • Burt D.R.
      • Stephenson F.A.
      • Rodriguez H.
      • Rhee L.M.
      • Ramachandran J.
      • Reale V.
      • Glencorse T.A.
      • Seeburg P.H.
      • Barnard E.A.
      ), implicated TM2 as a major domain forming the ion pore. Site-directed mutagenesis studies, as well as cysteine scanning in all member of this receptor-operated family, have further supported this role for TM2 (
      • Leonard R.J.
      • Labarca C.G.
      • Charnet P.
      • Davidson N.
      • Lester H.A.
      ,
      • Akabas M.H.
      • Stauffer D.A.
      • Xu M.
      • Karlin A.
      ,
      • Xu M.
      • Akabas M.H.
      ). A cysteine scan of TM1 in the homologous nACh receptor demonstrated a stretch of accessible residues on the extracellular half, leading the authors to conclude that the extracellular third of TM1 is accessible by means of the pore and thus may contribute to the pore structure (
      • Akabas M.H.
      • Karlin A.
      ). A cysteine scan of TM3 of the GABA receptor α subunit revealed two accessible residues near the extracellular end in the absence of GABA and three additional residues that were accessible in the presence of agonist. In this study, however, the authors proposed that access was through a water-filled crevice rather than the ion pore (
      • Williams D.B.
      • Akabas M.H.
      ). Finally, a chimeric study in the homologous 5-HT3 receptor identified a domain in the large cytoplasmic loop between TM3 and TM4, the H-A stretch, that controls single-channel conductance (
      • Kelley S.P.
      • Dunlop J.I.
      • Kirkness E.F.
      • Lambert J.J.
      • Peters J.A.
      ). The authors proposed that this domain may contribute to the narrow openings in the channel wall through which ions are filtered that was previously proposed from electron microscopic analysis at a 4.6 Å resolution (
      • Unwin N.
      ).
      We have identified residues in the intracellular third of TM1 and in the TM1-TM2 linker that are accessible by negatively, but not positively, charged reagents. We propose that this accessibility is through the pore, rather than through a hydrophilic crevice, because a mutation that changed the channel from anion-permeable to non-selective allowed positively charged reagents to reach residues at the interface of the TM1-TM2 linker and in TM1. Our present working hypothesis is that this region of the TM1-TM2 linker, as well as the intracellular third of TM1, are along the pathway of ion permeation. Perhaps this region, in addition to the TM3-TM4 H-A stretch (
      • Kelley S.P.
      • Dunlop J.I.
      • Kirkness E.F.
      • Lambert J.J.
      • Peters J.A.
      ), helps form the previously identified transverse tunnels in the channel wall (
      • Unwin N.
      ). It is also possible, although highly speculative, that the TM1-TM2 linker may reinsert into the membrane and contribute to the lining of the ion pore. In fact, using proteolysis in conjunction with mass spectrometry, an alternate model for the glycine receptor structure has been proposed in which TM1 is folded into several β strands, and TM3 may make as many as three passes through the membrane (
      • Leite J.F.
      • Cascio M.
      ,
      • Leite J.F.
      • Cascio M.
      ).
      Location of the Gate—From the earlier SCAM analysis in ligand-gated ion channels, studies have compared accessibility of TM2 residues in the presence and absence of agonist in an attempt to delimit (although not define) the location of the gate. Initial studies in the nACh and GABA receptors have demonstrated accessibility in the absence of agonist near the cytoplasmic end of TM2 (
      • Akabas M.H.
      • Stauffer D.A.
      • Xu M.
      • Karlin A.
      ,
      • Akabas M.H.
      • Karlin A.
      ). More recent studies using reagents added to either side of the membrane identified a barrier for accessibility in the absence of agonist in the α subunit of the nACh receptor between Gly240 and Thr244 (
      • Wilson G.G.
      • Karlin A.
      ). In the presence of agonist, this region became accessible in the nACh receptor (
      • Wilson G.G.
      • Karlin A.
      ). Although this sequence is not strictly conserved in the ρ1 GABA receptor, it would correspond to the region Pro290 to Pro294, positioned right at the putative intracellular end of TM2 (Fig. 1). A recent atomic model obtained from electron microscopy of crystalline postsynaptic membranes (
      • Miyazawa A.
      • Fujiyoshi Y.
      • Unwin N.
      ) revealed a tapering pore when viewed from the synaptic cleft. The maximal constriction of the nACh receptor pore was observed in the middle of the membrane, and this was presumably due to the hydrophobic side chains forming a girdle around the pore (
      • Miyazawa A.
      • Fujiyoshi Y.
      • Unwin N.
      ). As this is the only constriction in the entire ion pathway, the authors suggest this is likely to be the gate. This supports previous photolabeling studies (
      • White B.H.
      • Howard S.
      • Cohen S.G.
      • Cohen J.B.
      ,
      • White B.
      • Cohen J.
      ) as well as mutagenesis studies that demonstrate mutations in this region increase receptor sensitivity and stabilize the open state of the pore (
      • Filatov G.
      • White M.
      ,
      • Labarca C.
      • Nowak M.
      • Zhang H.
      • Tang L.
      • Deshpande P.
      • Lester H.A.
      ,
      • Chang Y.
      • Weiss D.S.
      ,
      • Chang Y.
      • Weiss D.S.
      ). Clearly, previous substituted cysteine accessibility method studies and the data we have presented here favor a gate located closer to the cytoplasmic boundary of TM2 (
      • Xu M.
      • Akabas M.H.
      ,
      • Akabas M.H.
      • Karlin A.
      ). We are a bit skeptical about assigning a gate location, even an approximate one, from a substituted cysteine accessibility method analysis, because many mutations in the pore can destabilize the closed state relative to the open state and produce agonist-independent openings. Even rare spontaneous openings that do not yield a detectable GABA-independent current could be sufficient to promote labeling beyond the gate.
      Modification rates were faster for both MTSEA and pCMBS in the presence of GABA. One possible explanation is that GABA induces a conformational change that now allows access of the reagents to the residues beyond this barrier. In its simplest form, assuming a comparable level of exposure, this would suggest a comparable rate of modification for residues beyond this barrier. This, in fact, is what was observed for all residues other than R287C, which demonstrated a slow rate of modification even in the presence of GABA.
      In summary, we have provided evidence for a more “intracellular” selectivity filter than previously proposed for the GABA receptor. In addition, our data suggest that the linker between TM1 and TM2, as well as the intracellular third of TM1 itself, are along the pathway for ion permeation. These data should certainly be included when considering models for the structure of the pore. Finally, as is evident from differential accessibility studies, there are clear structural changes occurring in the TM1-TM2 linker that may have implications for the structure of the gate and the gating mechanism.

      REFERENCES

        • Noda M.
        • Takahashi H.
        • Tanabe T.
        • Toyosato M.
        • Furutani Y.
        • Hirose T.
        • Asai M.
        • Inayama S.
        • Miyata T.
        • Numa S.
        Nature. 1982; 299: 793-797
        • Grenningloh G.
        • Rienitz A.
        • Schmitt B.
        • Methfessel C.
        • Zensen M.
        • Beyreuther K.
        • Gundelfinger E.D.
        • Betz H.
        Nature. 1987; 328: 215-220
        • Cutting G.R.
        • Lu L.
        • O'Hara B.F.
        • Kasch L.M.
        • Montrose-Rafizadeh C.
        • Donovan D.M.
        • Shimada S.
        • Antonarakis S.E.
        • Guggino W.B.
        • Uhl G.R.
        • Kazazian H.H.
        Proc. Natl. Acad. Sci. U. S. A. 1991; 88: 2673-2677
        • Schofield P.R.
        • Darlison M.G.
        • Fujita N.
        • Burt D.R.
        • Stephenson F.A.
        • Rodriguez H.
        • Rhee L.M.
        • Ramachandran J.
        • Reale V.
        • Glencorse T.A.
        • Seeburg P.H.
        • Barnard E.A.
        Nature. 1987; 328: 221-227
        • Maricq A.V.
        • Peterson A.S.
        • Brake A.J.
        • Myers R.M.
        • Julius D.
        Science. 1991; 254: 432-436
        • Leonard R.J.
        • Labarca C.G.
        • Charnet P.
        • Davidson N.
        • Lester H.A.
        Science. 1988; 242: 1578-1581
        • Akabas M.H.
        • Stauffer D.A.
        • Xu M.
        • Karlin A.
        Science. 1992; 258: 307-310
        • Xu M.
        • Akabas M.H.
        J. Gen. Physiol. 1996; 107: 195-205
        • Hucho F.
        • Oberthur W.
        • Lottspeich F.
        FEBS Lett. 1986; 205: 137-142
        • Galzi J.L.
        • Devillers-Thiery A.
        • Hussy N.
        • Bertrand S.
        • Changeux J.P.
        • Bertrand D.
        Nature. 1992; 359: 500-505
        • Keramidas A.
        • Moorhouse A.J.
        • French C.R.
        • Schofield P.R.
        • Barry P.H.
        Biophys. J. 2000; 79: 247-259
        • Keramidas A.
        • Moorhouse A.J.
        • Pierce K.D.
        • Schofield P.R.
        • Barry P.H.
        J. Gen. Physiol. 2002; 119: 393-410
        • Wotring V.E.
        • Miller T.S.
        • Weiss D.S.
        J. Physiol. 2003; 548: 527-540
        • Gunthorpe M.J.
        • Lummis S.C.
        J. Biol. Chem. 2001; 276: 10977-10983
        • Karlin A.
        • Akabas M.A.
        Methods Enzymol. 1998; 293: 123-145
        • Amin J.
        • Weiss D.S.
        Recept. Channels. 1994; 2: 227-236
        • Liman E.R.
        • Tytgat J.
        • Hess P.
        Neuron. 1992; 9: 861-871
        • Kammann M.
        • Laufs J.
        • Schell J.
        • Gronenborn B.
        Nucleic Acids Res. 1989; 12: 4445-4452
        • Amin J.
        • Dickerson I.
        • Weiss D.S.
        Mol. Pharmacol. 1994; 45: 317-323
        • Wotring V.E.
        • Chang Y.
        • Weiss D.S.
        J. Physiol. (London). 1999; 521.2: 327-336
        • Hodgkin A.L.
        • Katz B.
        J. Physiol. 1949; 108: 37-77
        • Xu M.
        • Akabas M.H.
        J. Biol. Chem. 1993; 268: 21505-21508
        • Bertaccini E.
        • Trudell J.R.
        Protein Eng. 2002; 15: 443-454
        • Imoto K.
        • Busch C.
        • Sakmann B.
        • Mishina M.
        • Konno T.
        • Nakai J.
        • Bujo H.
        • Mori Y.
        • Fukuda K.
        • Numa S.
        Nature. 1988; 335: 645-648
        • Akabas M.H.
        • Karlin A.
        Biochemistry. 1995; 34: 12496-12500
        • Williams D.B.
        • Akabas M.H.
        Biophys. J. 1999; 77: 2563-2574
        • Kelley S.P.
        • Dunlop J.I.
        • Kirkness E.F.
        • Lambert J.J.
        • Peters J.A.
        Nature. 2003; 424: 321-324
        • Unwin N.
        Nature. 1995; 373: 37-43
        • Leite J.F.
        • Cascio M.
        Mol. Cell. Neurosci. 2001; 17: 777-792
        • Leite J.F.
        • Cascio M.
        Biochemistry. 2002; 41: 6140-6148
        • Filatov G.
        • White M.
        Mol. Pharmacol. 1995; 48: 379-384
        • Wilson G.G.
        • Karlin A.
        Neuron. 1998; 20: 1269-1281
        • Miyazawa A.
        • Fujiyoshi Y.
        • Unwin N.
        Nature. 2003; 423: 949-955
        • White B.H.
        • Howard S.
        • Cohen S.G.
        • Cohen J.B.
        J. Biol. Chem. 1991; 266: 21595-21607
        • White B.
        • Cohen J.
        J. Biol. Chem. 1992; 267: 15770-15783
        • Labarca C.
        • Nowak M.
        • Zhang H.
        • Tang L.
        • Deshpande P.
        • Lester H.A.
        Nature. 1995; 376: 514-516
        • Chang Y.
        • Weiss D.S.
        Mol. Pharmacol. 1998; 53: 511-523
        • Chang Y.
        • Weiss D.S.
        Biophys. J. 1999; 77: 2542-2551