Engineered pores have several advantages as potential sensor elements: sensitivity is in the nanomolar range; analyte binding is rapid (diffusion limited in some cases) and reversible; strictly selective binding is not required because single-channel recordings are rich in information; and for a particular analyte, the dissociation rate constant, the extent of channel block and the voltage-dependence of these parameters are distinguishing, while the frequency of partial channel block reflects the analyte concentration. A single sensor element might, therefore, be used to quantitate more than one analyte at once. The approach described here can be generalized for additional analytes.

Glutamate binds to AMPA receptors within a deep cleft between two globular protein domains (domains 1 and 2). Once glutamate binds, the cleft closes, and agonist-bound structures of the isolated ligand binding core suggest that closure of the binding cleft is sufficiently complete that it essentially prevents ligand dissociation. There is also considerable evidence supporting the view that cleft closure is the initial conformational change that triggers receptor activation and desensitization, and it has been clearly demonstrated that there is a correlation between the degree of cleft closure and agonist efficacy. It is unknown, however, whether the stability of binding cleft closure also influences receptor-channel properties. The crystallographic structures indicate that closed-cleft conformations are stabilized by the formation of hydrogen bonds that involve amino acid side chains of residues in domains 1 and 2. We show here that mutations that disrupt one such cross-cleft hydrogen bond (in the AMPA receptor subunit GluR2) decrease both agonist affinity and efficacy. The same mutations also hasten recovery from desensitization. We conclude that the stability of binding cleft closure has a significant impact on AMPA receptor function and is a major determinant of the apparent affinity of agonists. The results suggest that the stability of cleft closure has been tuned so that glutamate dissociates as rapidly as possible yet remains a full agonist.

Elucidation of the accurate subunit stoichiometry of oligomeric membrane proteins is fraught with complexities. The interpretations of chemical cross-linking, analytical ultracentrifugation, gel filtration, and low-resolution electron microscopy studies are often ambiguous. Staphylococcal a-hemolysin (crHL), a homooligomeric toxin that forms channels in cell membranes, was believed to possess six subunits arranged around a sixfold axis of symmetry. Here, we report that analysis of x-ray diffraction data and chemical modification experiments indicate that the aHlL oligomer is a heptamer. Self-rotation functions calculated using x-ray diffraction data from single crystals of acHL oligomers show a sevenfold axis of rotational symmetry. The aHiL pore formed on rabbit erythrocyte membranes was determined to be a heptamer by electrophoretic separation of aHL heteromers formed from subunits with the charge of wild-type aHL and subunits with additional negative charge generated by targeted chemical modification of a single-cysteine mutant. These data establish the heptameric oligomerization state of the aHlL transmembrane pore both in three-dimensional crystals and on a biological membrane.As exemplified by studies on cholera toxin (1), chemical cross-linking and electron microscopy may not provide precise and unequivocal information on the subunit stoichiometry of oligomeric proteins. Complete cross-linking of an oligomeric protein can give both cyclic and linear species, which may migrate differently on denaturing polyacrylamide gels, giving the false indication of an additional subunit (1). Dynamic or static rotational disorder about the axis of molecular symmetry in two-dimensional crystals can yield misleading diffraction data (1). Resolution of the subunit composition of cholera toxin was achieved by the collection and analysis of high-resolution x-ray diffraction data (2). A second example is aerolysin, a pore-forming protein secreted by Aeromonas hydrophila. Aerolysin was thought to form a pentamer or a hexamer (3) until a recent examination by rotational correlation analysis of particles in averaged electron micrographs suggested that it is a heptamer (4). However, this conclusion is being reevaluated (5). The gap junction connexon provides another example of an oligomeric membrane protein for which the number of subunits is still unclear. The connexon was believed to be a hexamer (6), but recent work suggests that it may be a pentamer (7).Reservations concerning the reliability of methods employed in determining the subunit stoichiometry of other membrane proteins have led us to reexamine the quaternary structure of a-hemolysin (aHL), a protein that is a model system for studying membrane protein assembly (8-11), and that has potential applications in biotechnology [e.g., as a component of immunotoxins (12) or an element in biosensors (13)]. The aHL polypeptide of 293 amino acids is secreted by Staphylococcus aureus as a water-soluble monomer and assembles upon contact with lipid bilayers or the deterg...

The crystal structure of Escherichia coli aspartate carbamoyltransferase complexed with adenosine 5'-triphosphate (ATP) has been solved by molecular replacement and has been refined to a crystallographic residual of 0.17 at 2.6-A resolution by using the computer program X-PLOR. The unit cell dimensions of this crystal form are a = b = 122.2 A and c = 143.3 A and the space group is P321. Although the c-axis unit cell dimension is approximately 1 A longer than the corresponding dimension of the CTP-ligated P321 crystal form (c = 142.2 A), the ATP-ligated enzyme adopts a T-like quaternary structure. The base moiety of ATP interacts with residues Glu10, Ile12, and Lys60 while the ribose is near Asp19 and Lys60; the triphosphate entity is bound to Lys94, although His20 and Arg96 are nearby. We observe a higher occupancy for ATP in the allosteric site of the R1 regulatory chain in comparison to the occupancy of the R6 allosteric site. These crystallographically independent sites are related by a molecular 2-fold axis. There are other violations of the noncrystallographic symmetry that are similar to those observed in the refined CTP-ligated aspartate carbamoyltransferase structure. These infringements on the molecular symmetry might be the result of intermolecular interactions in the crystal. To ensure the most meaningful comparison with the ATP-ligated structure, we refined the previously reported CTP-bound and unligated structures to crystallographic residuals between 0.17 and 0.18 using X-PLOR. These X-PLOR refined structures are not significantly different from the initial structures that had been crystallographically refined by a restrained least-squares method. After making all possible comparisons between the CTP- and ATP-ligated and the unligated T-state structures, we find that the most significant differences are located at the allosteric sites and in small changes in the quaternary structures. At the allosteric site, the binding of CTP and ATP successively enlarges the nucleotide binding cavity, particularly in the vicinity of the base. The changes in the quaternary structure can be characterized by an increase in the separation of the catalytic trimers by approximately 0.5 A as ATP binds to the unligated T structure. On the basis of these structural studies, we discuss the relationships between the conformational differences in the allosteric site and the small changes in the quaternary structure within the T form to the possible mechanisms for CTP inhibition and ATP activation.

The three-dimensional structure of the ternary complex of carbamoyl phosphate, succinate, and aspartate carbamoyltransferase (EC 2.1.3.2) has been determined to 2.6-resolution. The binding of the phosphate of carbamoyl phosphate is similar to the binding of the phosphonate of N-(phosphonoacetyl)-L-aspartate (PALA); interacting with the carboxylates of succinate are some of the same residues that interact with the carboxylates of PALA. The aminio group of carbamoyl phosphate donates hydrogen bonds to the mainchain carbonyls of residues Pro-266 and Leu-267 and the side-chain carbonyl of Gln-137. In comparing the structure of the active sites in the PALA-enzyme complex to the active sites in the carbamoyl phosphate-succinate-enzyme complex, we find that they are similar.Aspartate carbamoyltransferase (also called aspartate transcarbamylase, ATCase; EC 2.1.3.2) from Escherichia coli catalyzes the reaction between L-aspartate and carbamoyl phosphate to yield N-carbamoyl-L-aspartate and phosphate (1). The dodecameric enzyme consists of two catalytic trimers (2C3) and three regulatory dimers (3R2) that spontaneously assemble to form the holoenzyme (C6R6) (2,3). As the paradigmatic regulatory enzyme, ATCase has been studied for more than 30 years not only in an effort to understand its allosteric properties but also because it catalyzes a reaction that is requisite for cell division (4, 5). Consequently, ATCase has been a target for antineoplastic agents (6, 7). However, there remain many questions concerning the specific conformational changes that aspartate and carbamoyl phosphate produce, inducing the homotropic allosteric transition. One important answer to these questions came in the solution of the structure of the R state of ATCase complexed with the bisubstrate analogue N-(phosphonoacetyl)-L-aspartate (PALA; Ki = 2 x 10-8 M) (8,9). To address the numerous remaining structural questions, we have embarked upon a number of crystallographic studies in which the enzyme was complexed with various combinations of substrates and substrate analogues. Regrettably, the substrates carbamoyl phosphate and aspartate react too rapidly for x-ray diffraction studies under usual conditions (although it might be possible to study the reaction in real time by using the technique of Laue diffraction). Hence, we have chosen to determine the structure of the ternary complex of enzyme, carbamoyl phosphate, and succinate, an excellent competitive inhibitor with respect to aspartate (K1 = 4.4 x 10-4 M at pH 6.0) (10). MATERIALS AND METHODSThe half-life of carbamoyl phosphate is short [about 11 hr at 20°C in the crystallization buffer, as shown by 31P NMR (J.E.G., unpublished results)] compared to the time required to grow diffraction-quality crystals. Consequently, the carbamoyl phosphate-and succinate-ligated crystals were prepared by soaking crystals grown in the presence of 1.0 mM PALA (9) with saturating amounts of carbamoyl phosphate and succinate. The ligand exchange was accomplished by placing the crystals in a buffer cont...