Solution structure and backbone dynamics of the functional cytoplasmic subdomain of human ephrin B2, a cell-surface ligand with bidirectional signaling properties

The cytoplasmic domain of B ephrins plays a central role in bidirectional signal transduction processes controlling pattern formation and morphogenesis, such as axon guidance, cell migration, segmentation, and angiogensis. In particular, the extremely conserved last 33-residue cytoplasmic subdomain was shown to bind to both a PDZ domain for one signaling pathway [Lu et al. (2001) Cell 105, 69-79] and an SH2 domain from an alternative signaling network [Cowan and Henkemeyer (2001) Nature 413, 174-179]. To date, no structural information is available for the cytoplasmic domain of ephrin B proteins. We report here a detailed NMR study on the structural and dynamic properties of the cytoplasmic domain of human ephrin B2. Our results reveal the following: (1) the N-terminal region of the cytoplasmic domain from residues 253 to 300 lacks the ability for structure formation and is particularly prone to aggregation; and (2) the C-terminal functional subdomain from residues 301 to 333 assumes two distinctive structural elements with residues 301-322 adopting a well-packed hairpin structure followed by a flexible C-terminal tail. Furthermore, the backbone (15)N relaxation data demonstrate that the hairpin structure has significantly limited backbone motions, indicating a high conformational stability for the folded structure. Therefore, while the flexible C-terminal tail is suitable for binding to the PDZ domain, the folded hairpin may represent a latent structure requiring phosphorylation-induced conformational changes for high-affinity interactions with the SH2 domain.     

Local structural plasticity of the prion protein. Analysis of NMR relaxation dynamics

A template-assisted conformational change of the cellular prion protein (PrP(C)) from a predominantly helical structure to an amyloid-type structure with a higher proportion of beta-sheet is thought to be the causative factor in prion diseases. Since flexibility of the polypeptide is likely to contribute to the ability of PrP(C) to undergo the conformational change that leads to the infective state, we have undertaken a comprehensive examination of the dynamics of two recombinant Syrian hamster PrP fragments, PrP(29-231) and PrP(90-231), using (15)N NMR relaxation measurements. The molecular motions of these PrP fragments have been studied in solution using (15)N longitudinal (T(1)) and transverse relaxation (T(2)) measurements as well as [(1)H]-(15)N nuclear Overhauser effects (NOE). These data have been analyzed using both reduced spectral density mapping and the Lipari-Szabo model free formalism. The relaxation properties of the common regions of PrP(29-231) and PrP(90-231) are very similar; both have a relatively inflexible globular domain (residues 128-227) with a highly flexible and largely unstructured N-terminal domain. Residues 29-89 of PrP(29-231), which include the copper-binding octarepeat sequences, are also highly flexible. Analysis of the spectral densities at each residue indicates that even within the structured core of PrP(C), a markedly diverse range of motions is observed, consistent with the inherent plasticity of the protein. The central portions of helices B and C form a relatively rigid core, which is stabilized by the presence of an interhelix disulfide bond. Of the remainder of the globular domain, the parts that are not in direct contact with the rigid region, including helix A, are more flexible. Most significantly, slow conformational fluctuations on a millisecond to microsecond time scale are observed for the small beta-sheet. These results are consistent with the hypothesis that the infectious, scrapie form of the protein PrP(Sc) could contain a helical core consisting of helices B and C, similar in structure to the cellular form PrP(C). Our results indicate that residues 90-140, which are required for prion infectivity, are relatively flexible in PrP(C), consistent with a lowered thermodynamic barrier to a template-assisted conformational change to the infectious beta-sheet-rich scrapie isoform.     

Identification of a novel human islet amyloid polypeptide beta-sheet domain and factors influencing fibrillogenesis

Human islet amyloid polypeptide (hIAPP) accumulates as pancreatic amyloid in type 2 diabetes and readily forms fibrils in vitro. Investigations into the mechanism of hIAPP fibril formation have focused largely on residues 20 to 29, which are considered to comprise a primary amyloidogenic domain. In rodents, proline substitutions within this region and the subsequent beta-sheet disruption, prevents fibril formation. An additional amyloidogenic fragment within the C-terminal sequence, residues 30 to 37, has been identified recently. We have extended these observations by examining a series of overlapping peptide fragments from the human and rodent sequences. Using protein spectroscopy (CD/FTIR), electron microscopy and X-ray diffraction, a previously unrecognised amyloidogenic domain was localised within residues 8 to 20. Synthetic peptides corresponding to this region exhibited a transition from random coil to beta-sheet conformation and assembled into fibrils having a typical amyloid-like morphology. The comparable rat 8-20 sequence, which contains a single His18Arg substitution, was also capable of assembling into amyloid-like fibrils. Examination of peptide fragments corresponding to residues 1 to 13 revealed that the immediate N-terminal region is likely to have only a modulating influence on fibril formation or conformational conversion. The contributions of charged residues as they relate to the amyloid-forming 8-20 sequence were also investigated using IAPP fragments and by assessing the effects of pH and counterions. The identification of these principal amyloidogenic sequences and the effects of associated factors provide details on the IAPP aggregation pathway and structure of the peptide in its fibrillar state.     

Functional dissection of the ETS transcription factor MEF

We previously indicated that myeloid elf-1-like factor (MEF) but not elf-1, specifically activated lysozyme gene expression in epithelial cells. MEF is highly homologous at the nucleotide and amino acid level, with elf-1 especially in the ETS domain. Here, we report the functional analysis of the nuclear localization and transactivation properties of MEF. To investigate the intracellular localization of MEF, we transiently transfected MEF-green fluorescence protein (GFP) fusion protein expression vector into HeLa cells. A region spanning residues 177-291 is required for nuclear localization. We produced deletion mutants of MEF to determine the transactivation domain. The data showed that the N-terminal region, encompassing amino acids 1-52 is a potent transactivation domain. The C-terminal region spanning residues 477-663 can also mediate transactivation but not as strongly as the N-terminal region. The activity of the amino acid residues 1-52 was confirmed by experiments with fused constructs of MEF to the DNA binding-domain of the yeast GAL4 protein. These results, which determined the localization of the functional domains of MEF, will provide us with new clues to its transactivation mechanisms to regulate lysozyme gene expression in epithelial cells.     

Characterization of the reversible conformational equilibrium of the cytoplasmic domain of erythrocyte membrane band 3

The cytoplasmic domain of the erythrocyte membrane protein, band 3, contains binding sites for hemoglobin, several glycolytic enzymes, and ankyrin, the linkage to the cytoskeleton. In an earlier study, we found evidence which suggested that band 3 might undergo a native conformational change. We demonstrate here that the cytoplasmic domain of band 3 does exist in a reversible, pH-dependent conformational equilibrium among 3 native states. At physiological salt concentrations this equilibrium is characterized by apparent pKa values of 7.2 and 9.2; however, these apparent pKa values change if the domain's sulfhydryl groups are modified. A major component of the structural change appears to involve the pivoting of two subdomains of the cytoplasmic domain at a central hinge, as evidenced by both hydrodynamic and fluorescence energy transfer measurements. The probable site of this hinge is between residues 176 and 191, a region highly accessible to proteases and also rich in proline. These structural rearrangements also apparently extend to the cluster of tryptophan residues near the N terminus, since the domain's intrinsic fluorescence more than doubles between pH 6.5 and 9.5. No measurable change in band 3 secondary or quaternary structure could be detected during the conformational transitions. A structural model of the cytoplasmic domain of band 3 is presented to show the possible spatial relationships between the regions of conformational change and the sites of peripheral protein binding.     

Site-directed mutagenesis and 1H NMR spectroscopy of an interdomain segment in the pyruvate dehydrogenase multienzyme complex of Escherichia coli

Deletion of two of the three homologous lipoyl domains that form the N-terminal half of each dihydrolipoamide acetyltransferase (E2p) polypeptide chain of the Escherichia coli pyruvate dehydrogenase complex can be achieved by in vitro deletion in the structural gene aceF. A site-directed mutagenesis of this shortened aceF gene was carried out to replace the glutamine residue at position 291 (wild-type numbering) with a histidine residue. Residue 291 is near the middle of a long segment (about 30 amino acid residues) of polypeptide chain, rich in alanine, proline, and charged amino acids, that links the remaining lipoyl domain to the dihydrolipoamide dehydrogenase (E3) binding domain in the E2p chain. A fully active enzyme complex was still assembled, and despite the enormous size of the particle (Mr approximately 4 x 10(6)), sharp resonances attributable to the single new histidine residue per E2p chain could be detected in the 400-MHz 1H NMR spectrum of the complex. The sharpness of these resonances, their chemical shifts (7.94 and 7.05 ppm), and the apparent pKa (6.4) of the side chain were all consistent with this histidine residue being exposed to solvent in a conformationally flexible region of the E2p polypeptide chain. These experiments provide direct proof for the conformational flexibility of this region of polypeptide chain, which is thought to play an important part in the movement of the lipoyl domain required for active site coupling in the enzyme complex.(ABSTRACT TRUNCATED AT 250 WORDS)     

An interdomain linker increases the thermostability and decreases the calcium affinity of the calmodulin N-domain.

A hydrophobic core is a widely accepted determinant of protein stability. However, regulatory proteins undergoing ligand-induced conformational switching may expose interior residues to solvent and cannot afford to be extremely rigid. Optimizing the energetic balance between stability and binding is challenging. The addition of five interdomain residues to rat and Paramecium calmodulin N-domain fragments (residues 1-75) increased their thermostability by 9 degrees C and lowered their calcium affinity by a factor of 4. This demonstrates that the flexible linker regulates functional properties as well as tethering the neighboring domains and that protein stability may be increased markedly by minor modifications of the C-terminus. The sensitivity of this domain to few and conservative variations in helices A and D (D2E, S17A, T70S and M71L) is demonstrated by the rat CaM fragments having lower stability and higher calcium affinity than fragments of the same length derived from Paramecium CaM.     

Disruption of a calmodulin central helix-like region of 10-formyltetrahydrofolate dehydrogenase impairs its dehydrogenase activity by uncoupling the functional domains

10-Formyltetrahydrofolate dehydrogenase (FDH) is composed of three domains and possesses three catalytic activities but has only two catalytic centers. The amino-terminal domain (residue 1-310) bears 10-formyltetrahydrofolate hydrolase activity, the carboxyl-terminal domain (residue 420-902) bears an aldehyde dehydrogenase activity, and the full-length FDH produces 10-formyltetrahydrofolate dehydrogenase activity. The intermediate linker (residues 311-419) connecting the two catalytic domains does not contribute directly to the enzyme catalytic centers but is crucial for 10-formyltetrahydrofolate dehydrogenase activity. We have identified a region within the intermediate domain (residues 384-405) that shows sequence similarity to the central helix of calmodulin. Deletion of either the entire putative helix or the central part of the helix or replacement of the six residues within the central part with alanines resulted in total loss of the 10-formyltetrahydrofolate dehydrogenase activity, whereas the full hydrolase and aldehyde dehydrogenase activities were retained. Alanine-scanning mutagenesis revealed that neither of the six residues alone is required for FDH activity. Analysis of the predicted secondary structures and circular dichroic and fluorescence spectroscopy studies of the intermediate domain expressed as a separate protein showed that this region is likely to consist of two alpha-helices connected by a flexible loop. Our results suggest that flexibility within the putative helix is important for FDH function and could be a point for regulation of the enzyme.     

Mapping of the functional domains in the amino-terminal region of calponin.

Three chymotryptic fragments accounting for almost the entire amino acid sequence of gizzard calponin (Takahashi, K., and Nadal-Ginard, B. (1991) J. Biol. Chem. 266, 13284-13288) were isolated and characterized. They encompass the segments of residues 7-144 (NH2-terminal 13-kDa peptide), 7-182 (NH2-terminal 22-kDa peptide), and 183-292 (COOH-terminal 13-kDa peptide). They arise from the sequential hydrolysis of the peptide bonds at Tyr182-Gly183 and Tyr144-Ala145 which were protected by the binding of F-actin to calponin. Only the NH2-terminal 13- and 22-kDa fragments were retained by immobilized Ca(2+)-calmodulin, but only the larger 22 kDa entity cosedimented with F-actin and inhibited, in the absence of Ca(2+)-calmodulin, the skeletal actomyosin subfragment-1 ATPase activity as the intact calponin. Since the latter peptide differs from the NH2-terminal 13-kDa fragment by a COOH-terminal 38-residue extension, this difference segment appears to contain the actin-binding domain of calponin. Zero-length cross-linked complexes of F-actin and either calponin or its 22-kDa peptide were produced. The total CNBr digest of the F-actin-calponin conjugate was fractionated over immobilized calmodulin. The EGTA-eluted pair of cross-linked actin-calponin peptides was composed of the COOH-terminal actin segment of residues 326-355 joined to the NH2-terminal calponin region of residues 52-168 which seems to contain the major determinants for F-actin and Ca(2+)-calmodulin binding.     

DNA-induced increase in the alpha-helical content of C/EBP and GCN4

Leucine zipper proteins comprise a recently identified class of DNA binding proteins that contain a bipartite structural motif consisting of a "leucine zipper" dimerization domain and a segment rich in basic residues responsible for DNA interaction. Protein fragments encompassing the zipper plus basic region domains (bZip) have previously been used to determine the conformational and dynamic properties of this motif. In the absence of DNA, the coiled-coil portion is alpha-helical and dimeric, whereas the basic region is flexible and partially disordered. Addition of DNA containing a specific recognition sequence induces a fully helical conformation in the basic regions of these fragments. However, the question remained whether the same conformational change would be observed in native bZip proteins where the basic regions might be stabilized in an alpha-helical conformation even in the absence of DNA, through interactions with portions of the protein not included in the bZip motif. We have now examined the DNA-induced conformational transition for an intact bZip protein, GCN4, and for the bZip fragment of C/EBP with two enhancers that are differentially symmetric. Our results are consistent with the induced helical fork model wherein the basic regions are largely flexible in the absence of DNA and become fully helical in the presence of the specific DNA recognition sequence.     

The lipid-associated conformation of the low density lipoprotein receptor binding domain of human apolipoprotein E

Apolipoprotein E (apoE) is a 34-kDa exchangeable apolipoprotein that regulates metabolism of plasma lipoproteins by functioning as a ligand for members of the LDL receptor family. The receptor-binding region localizes to the vicinity of residues 130-150 within its independently folded 22-kDa N-terminal domain. In the absence of lipid, this domain exists as a receptor-inactive, globular four-helix bundle. Receptor recognition properties of this domain are manifest upon lipid association, which is accompanied by a conformational change in the protein. Fluorescence resonance energy transfer has been used to monitor helix repositioning, which accompanies lipid association of the apoE N-terminal domain. Site-directed mutagenesis was used to replace naturally occurring Trp residues with phenylalanine, creating a Trp-null apoE3 N-terminal domain (residues 1-183). Subsequently, tyrosine residues in helix 2, helix 3, or helix 4 were converted to Trp, generating single Trp mutant proteins. The lone cysteine at position 112 was covalently modified with N-iodoacetyl-N'-(5-sulfo-1-naphthyl)ethylenediamine, which serves as an energy acceptor from excited tryptophan residues. Fluorescence resonance energy transfer analysis of apoE N-terminal domain variants in phospholipid disc complexes suggests that the helix bundle opens to adopt a partially extended conformation. A model is presented that depicts a tandem arrangement of the receptor-binding region of the protein in the disc complex, corresponding to its low density lipoprotein receptor-active conformation.     

Identification of the bile acid-binding region in the soy glycinin A1aB1b subunit

Soy glycinin has five major subunits which are classified into two groups according to their homology in amino acid sequences (group I, A1aB1b, A1bB2 and A2B1a; group II, A3B4 and A5A4B3). It has been reported that the peptide fragments derived from the A1a and A2 chains of the A1aB1b and A2B1a subunits had bile acid-binding ability and that the region of 114-161 residues of the A1a chain was responsible for this bile acid-binding ability. In this study, we constructed A1a, A3 and 9 deletion mutants of A1a lacking various numbers of residues at the C-terminus, and evaluated their bile acid-binding ability by a cholic acid-conjugated column and fluorescence analysis. The bile acid-binding ability of A1a was higher than that of A3 and there was a remarkable decrease in the bile acid-binding ability between the delta[138-291] and delta[130-291] mutants. The 130-138 region is rich in hydrophobic residues. In this regard, when we constructed the delta[129-134] mutant lacking six contiguous hydrophobic residues (VAWWMY) and evaluated its bile acid-binding ability, a similar remarkable decrease in the bile acid-binding ability was observed. These results indicate that the 129-134 residue region (VAWWMY) with high hydrophobicity was important for bile acid-binding of A1a.     

Nicked apomyoglobin: a noncovalent complex of two polypeptide fragments comprising the entire protein chain

Limited proteolysis of the 153-residue chain of horse apomyoglobin (apoMb) by thermolysin results in the selective cleavage of the peptide bond Pro88-Leu89. The N-terminal (residues 1-88) and C-terminal (residues 89-153) fragments of apoMb were isolated to homogeneity and their conformational and association properties investigated in detail. Far-UV circular dichroism (CD) measurements revealed that both fragments in isolation acquire a high content of helical secondary structure, while near-UV CD indicated the absence of tertiary structure. A 1:1 mixture of the fragments leads to a tight noncovalent protein complex (1-88/89-153, nicked apoMb), characterized by secondary and tertiary structures similar to those of intact apoMb. The apoMb complex binds heme in a nativelike manner, as given by CD measurements in the Soret region. Second-derivative absorption spectra in the 250-300 nm region provided evidence that the degree of exposure of Tyr residues in the nicked species is similar to that of the intact protein at neutral pH. Also, the microenvironment of Trp residues, located in positions 7 and 14 of the 153-residue chain of the protein, is similar in both protein species, as given by fluorescence emission data. Moreover, in analogy to intact apoMb, the nicked protein binds the hydrophobic dye 1-anilinonaphthalene-8-sulfonate (ANS). Taken together, our results indicate that the two proteolytic fragments 1-88 and 89-153 of apoMb adopt partly folded states characterized by sufficiently nativelike conformational features that promote their specific association and mutual stabilization into a nicked protein species much resembling in its structural features intact apoMb. It is suggested that the formation of a noncovalent complex upon fragment complementation can mimic the protein folding process of the entire protein chain, with the difference that the folding of the complementary fragments is an intermolecular process. In particular, this study emphasizes the importance of interactions between marginally stable elements of secondary structure in promoting the tertiary contacts of a native protein. Considering that apoMb has been extensively used as a paradigm in protein folding studies for the past few decades, the novel fragment complementing system of apoMb here described appears to be very useful for investigating the initial as well as late events in protein folding.     

Importance of the D and E helices of the molecular chaperone DnaK for ATP binding and substrate release

The C-terminal domain of the molecular chaperone DnaK is a compact lid-like structure made up of five alpha-helices (alphaA-alphaE) (residues 508-608) that is followed by a 30-residue disordered, flexible region (609-638). The lid encapsulates the peptide molecule bound in the substrate-binding domain, whereas the function of the 30-residue disordered region is not known. By sequentially deleting the flexible subdomain and the individual lid helices, we deduced the importance of each structural unit to creating long-lived DnaK-peptide complexes. Here we report that (i) the alphaD helix is essential for long-lived DnaK-peptide complexes. For example, ATP triggers the dissociation of a acrylodan-labeled p5 peptide (ap5, a-CLLLSAPRR) from wtDnaK and DnaK595(A-D) with k(off) equal to 7.6 and 8.9 s(-1), respectively, whereas when the D-helix is deleted, creating DnaK578(A-C), k(off) jumps to 207 s(-1). (ii) The presence of the alphaB helix impacts the rate of the ATP-induced high-to-low affinity conformational change. For example, ATP induces this conformational change in a lidless variant, DnaK517(1/2A), with a rate constant of 442 s(-1), whereas, after adding back the B-helix (residues 518-554), ATP induces this conformational change in DnaK554(A-B) with a rate constant of 2.5 s(-1). Our interpretation is that this large decrease occurs because the B-helix of the DnaK554(A-B) is bound in the substrate-binding site. (iii) The deletion analysis also revealed that residues 596-638, which comprise the alphaE helix and the flexible subdomain, affect ATP binding. Our results are consistent with this part of the lid producing conformational heterogeneity, perhaps by binding to the ATPase domain.     

On the structure of the stator of the mitochondrial ATP synthase

The structure of most of the peripheral stalk, or stator, of the F-ATPase from bovine mitochondria, determined at 2.8 A resolution, contains residues 79-183, 3-123 and 5-70 of subunits b, d and F6, respectively. It consists of a continuous curved alpha-helix about 160 A long in the single b-subunit, augmented by the predominantly alpha-helical d- and F6-subunits. The structure occupies most of the peripheral stalk in a low-resolution structure of the F-ATPase. The long helix in subunit b extends from near to the top of the F1 domain to the surface of the membrane domain, and it probably continues unbroken across the membrane. Its uppermost region interacts with the oligomycin sensitivity conferral protein, bound to the N-terminal region of one alpha-subunit in the F1 domain. Various features suggest that the peripheral stalk is probably rigid rather than resembling a flexible rope. It remains unclear whether the transient storage of energy required by the rotary mechanism takes place in the central stalk or in the peripheral stalk or in both domains.     

Sites in TM2 and 3 are critical for alcohol-induced conformational changes in GABA receptors

Abstract gamma-Aminobutyric acid type A (GABA(A)) receptors are molecular targets for alcohols. Previous work suggests that S270 and A291 residues in the transmembrane (TM) 2 and 3 domains of the GABA(A) receptor alpha subunit are components of an alcohol-binding pocket, and S270I and A291W mutants abolished ethanol potentiation. Our results showed that A295C and F296C residues in the TM3 of the GABA(A) receptor alpha1 subunit are accessible to hexylmethanethiosulfonate (HMTS) in the alcohol-bound state, but not in the resting state. Thus, the A295C and F296C sites become water-accessible as a result of alcohol-induced conformational changes. If S270 or A291 residues are sites of alcohol binding, then S270I or A291W mutations should prevent alcohol-induced conformational movements within the TM3 domain. To investigate this question, the accessibility of HMTS reagent to double mutants (A291W/A295C, A291W/F296C, S270I/A295C or S270I/F296C) in the presence of ethanol or hexanol was tested. The A291W or S270I mutations markedly reduced the accessibility of HMTS to all the double mutants in the ethanol-bound state, and to S270I/F296C, A291W/A295C and A291W/F296C double mutants in the hexanol-bound state, suggesting that the A291 or S270 residues are critical sites for alcohol binding and alcohol-induced conformational changes.     

The solution structure of the periplasmic domain of the TonB system ExbD protein reveals an unexpected structural homology with siderophore-binding proteins

The transport of iron complexes through outer membrane transporters from Gram-negative bacteria is highly dependent on the TonB system. Together, the three components of the system, TonB, ExbB and ExbD, energize the transport of iron complexes through the outer membrane by utilizing the proton motive force across the cytoplasmic membrane. The three-dimensional (3D) structure of the periplasmic domain of TonB has previously been determined. However, no detailed structural information for the other two components of the TonB system is currently available and their role in the iron-uptake process is not yet clearly understood. ExbD from Escherichia coli contains 141 residues distributed in three domains: a small N-terminal cytoplasmic region, a single transmembrane helix and a C-terminal periplasmic domain. Here we describe the first well-defined solution structure of the periplasmic domain of ExbD (residues 44-141) solved by multidimensional nuclear magnetic resonance (NMR) spectroscopy. The monomeric structure presents three clearly distinct regions: an N-terminal flexible tail (residues 44-63), a well-defined folded region (residues 64-133) followed by a small C-terminal flexible region (residues 134-141). The folded region is formed by two alpha-helices that are located on one side of a single beta-sheet. The central beta-sheet is composed of five beta-strands, with a mixed parallel and antiparallel arrangement. Unexpectedly, this fold closely resembles that found in the C-terminal lobe of the siderophore-binding proteins FhuD and CeuE. The ExbD periplasmic domain has a strong tendency to aggregate in vitro and 3D-TROSY (transverse relaxation optimized spectroscopy) NMR experiments of the deuterated protein indicate that the multimeric protein has nearly identical secondary structure to that of the monomeric form. Chemical shift perturbation studies suggest that the Glu-Pro region (residues 70-83) of TonB can bind weakly to the surface and the flexible C-terminal region of ExbD. At the same time the Lys-Pro region (residues 84-102) and the folded C-terminal domain (residues 150-239) of TonB do not show significant binding to ExbD, suggesting that the main interactions forming the TonB complex occur in the cytoplasmic membrane.     

gamma-aminobutyric acid increases the water accessibility of M3 membrane-spanning segment residues in gamma-aminobutyric acid type A receptors

gamma-Aminobutyric acid type A (GABA(A)) receptors are members of the ligand-gated ion channel gene superfamily. Using the substituted cysteine accessibility method, we investigated whether residues in the alpha(1)M3 membrane-spanning segment are water-accessible. Cysteine was substituted, one at a time, for each M3 residue from alpha(1)Ala(291) to alpha(1)Val(307). The ability of these mutants to react with the water-soluble, sulfhydryl-specific reagent pCMBS(-) was assayed electrophysiologically. Cysteines substituted for alpha(1)Ala(291) and alpha(1)Tyr(294) reacted with pCMBS(-) applied both in the presence and in the absence of GABA. Cysteines substituted for alpha(1)Phe(298), alpha(1)Ala(300), alpha(1)Leu(301), and alpha(1)Glu(303) only reacted with pCMBS(-) applied in the presence of GABA. We infer that the pCMBS(-) reactive residues are on the water-accessible surface of the protein and that GABA induces a conformational change that increases the water accessibility of the four M3 residues, possibly by inducing the formation of water-filled crevices that extend into the interior of the protein. Others have shown that mutations of alpha(1)Ala(291), a water-accessible residue, alter volatile anesthetic and ethanol potentiation of GABA-induced currents. Water-filled crevices penetrating into the interior of the membrane-spanning domain may allow anesthetics and alcohol to reach their binding sites and thus may have implications for the mechanisms of action of these agents.