We have devised a generally applicable strategy for analysis of protein structure and have applied it to examine the structure of the transmembrane portion of the Tar receptor of Escherichia coli. The basis of our approach is the use of disulfide cross-linking to identify residues that are within close proximity. To generate and test large numbers of cysteine pairs, we used an unusual method of mutagenesis by which cysteine substitutions can be created randomly at a number of targeted codons. Cysteine-substituted proteins encoded by mutagenized genes may be screened directly for disulfide formation within oligomers or, alternatively, different pools of genes may be randomly recombined to generate gene populations with substitutions in multiple regions. Thus, it is possible to detect a variety of disulfide cross-links between and within individual protein molecules. Interactions between the four membrane-spanning stretches of the Tar dimer were probed by measuring the tendency of 48 cysteine substitutions throughout this region to form disulffide cross-links with one another. We have interpreted these data to suggest a helical-bundle structure for the transmembrane region. The four helices of this bundle are not structurally equivalent: the two TM1 helices interact closely, whereas the TM2 helices are more peripherally located.The Tar receptor of Escherichia coli is a transmembrane signal transducer protein involved in the detection of chemotactic effectors. It is a member of a family of related receptors, the methyl-accepting chemotaxis proteins, which share considerable amino acid sequence similarity (1). Tar appears to act as a dimer (2). It mediates repellent responses to nickel and cobalt ions and attractant responses to aspartate and maltose (3). Each 553-amino acid monomer of Tar is comprised of relatively large periplasmic and intracellular domains as well as two membrane-spanning domains that are thought to be a-helical (4). (The transmembrane topology of Tar is illustrated in Fig. 1.) Ligands bind to the periplasmic region of Tar (5-8) leading the receptor to modulate the activity of the soluble autophosphorylating kinase CheA (9, 10). This initial event leads, through a series of additional intracellular interactions, to altered bacterial swimming behavior.The nature of the protein conformational changes that mediate communication across lipid bilayers is unknown. The Tar receptor resembles the mammalian epidermal growth factor and insulin receptors in that each of these proteins consists of large soluble domains linked by a small number of transmembrane stretches. In contrast to the epidermal growth factor receptor, the oligomeric state of the Tar receptor does not change as a consequence of ligand binding (2); thus any structural changes involved in transmembrane signal transmission occur within the preformed receptor dimer. We presume that ligand binding results in a conformational change in the periplasmic domain and that this change is then propagated by the four transmembrane domains to t...

There is tremendous variability in the importance of individual amino acids in protein sequences. On the one hand, nonconservative residue substitutions can be tolerated with no loss of activity at many residue positions, especially those exposed on the protein surface. On the other hand, destabilizing mutations can occur at a large number of different sites in a protein, and for many proteins such mutations account for more than half of the randomly isolated missense mutations that confer a defective phenotype. At sites that are key determinants of stability or activity, even residue substitutions that are generally considered to be conservative (e.g., Glu in equilibrium Asp, Asn in equilibrium Asp, Ile in equilibrium Leu, Lys in equilibrium Arg and Ala in equilibrium Gly) can have severe phenotypic effects. Unfortunately, this means that there is no simple way to infer the likely effect of an amino acid substitution on the basis of sequence information alone. A nonconservative Gly----Arg substitution could be phenotypically silent at one position while a conservative Asn----Asp change could lead to complete loss of activity at another position. For proteins whose structures are known, it is often possible to predict whether particular residue substitutions will be destabilizing, as long as detailed estimates of the destabilization energy are not required. Substitutions that introduce polar groups, large cavities, or overly large side chains into the hydrophobic core are potentially the most destabilizing. Substitutions that disrupt hydrogen bonding or electrostatic interactions can also have significant effects, although the destabilization caused by these substitutions is smaller than that caused by severe core mutations. Destabilizing substitutions that involve replacing glycines in turns, or introducing prolines into alpha-helices and other disallowed positions are also reasonably common. Finally, most solvent exposed residues can apparently be freely substituted without serious effects on protein stability. Although exceptions may occur, these generalizations serve to summarize a large body of information and can be rationalized in physical and chemical terms. It is an especially encouraging result that proteins appear to tolerate most substitutions, even those that are destabilizing, without significant changes in the native structure. For proteins whose structures are known, this means that it is reasonable to interpret mutant phenotypes in terms of the wild-type structure. For proteins whose structures are not known, it is reasonable to infer that mutations that reduce activity without affecting stability are directly involved in function.(ABSTRACT TRUNCATED AT 400 WORDS)

It is rare for amino-acid substitutions on the surface of proteins to have large stabilizing or destabilizing effects. Nevertheless, one substitution of this type, the Tyr 26----Cys mutation in lambda Cro, increases the melting temperature of the protein by 11 degrees C and the stability by 2.2 kcal mol-1. Here we show that the stability of Cro can be increased by many different amino-acid substitutions at position 26, with increasing stability showing a good correlation with decreasing side-chain hydrophobicity. As Tyr 26 is hyper-exposed to solvent in the Cro crystal structure, we suggest that wild-type and variant proteins with other hydrophobic side chains at position 26 are destabilized as a result of a reverse hydrophobic effect caused by the side chain being more exposed to solvent in the native than in the denatured state.

Following random mutagenesis of the bacteriophage X cro gene, we have isolated missense mutations that affect approximately half of the 66 residue positions of Cro. About two-thirds of the mutations change residues involved in the maintenance of Cro structure and stability. The corresponding mutant proteins are severely degraded in the cell but often have specific activities near that of wild-type Cro. The remaining mutations affect residues involved in DNA binding. These mutant proteins are present at moderately reduced intracellular levels, but their specific activities are much lower than that of wild type.The study of missense mutants provides a means of testing and adding to our understanding of the determinants of protein structure, stability, and function. In this paper, we describe mutations that affect the intracellular level and activity of bacteriophage X Cro, a small repressor with a known three-dimensional structure (1-3). We find that a Crophenotype results from replacing any 1 of at least half of the 66 residues of Cro; some of the mutations change residues on the DNA binding surface, some change solvent-exposed residues distant from the proposed binding surface, and some affect residues in the hydrophobic core. Mutations that would be expected to destabilize the folded structure of Cro generally result in extreme sensitivity to intracellular proteolysis. In many cases, the resulting reduction in concentration is severe enough to account fully for the observed mutant phenotype. MATERIALS AND METHODSPlasmids, Phage, and Bacteria. Three related Cro-producing plasmids-pAP100, pAP101, and pAP104-were constructed for this work. Each plasmid is a pBR322 derivative, and each bears the same fusion of the lac UV5 promoter to the phage X cro gene. This fusion was derived from plasmid pTR214 (4). Plasmid pAP100 was constructed by inserting the 353-base-pair EcoRI/Nae I fragment from pTR214 between the EcoRI and filled-in Cla I sites of pBR322 (5). Plasmid pAP101 was constructed by replacing the 391-base-pair Cla I/BamHI fragment of pAP100 with the 375-base-pair EcoRI/BamHI fragment from pBR322. pAP101 has an intact tet promoter and, thus, confers a significantly higher level of tetracycline resistance than does pAP100. pAP104 is similar to pAP101 but contains the phage M13 origin of replication. pAP104 was constructed by replacing the small EcoRI/Bam-HI fragment of pZ150 (6) with that from plasmid pAP101. Strains containing pAP100, pAP101, or pAP104 express the same level of phage X Cro, as judged by the resistance of the strain to phage infection and by competitive radioimmunoassay. Thus, Cro-mutations isolated in any of the three plasmid backgrounds can be directly compared.X200 is an immunity 21 bacteriophage that carries the lacZ gene under the control of the phage X PR promoter (7). Phage M13 appl was constructed by deletion in vitro of the lac promoter and operator sequences of M13 mp8 (8). DNA between the unique Ava II and EcoRI sites was deleted, the DNA was ligated? and candidates were scree...

The crystal structure of the periplasmic domain of the aspartate receptor from Escherichia coli has been solved and refined to an R-factor of 0.203 at 2.3 A, resolution. The dimeric protein is largely helical, with four helices from each monomer forming a four-helix bundle. The dimer interface is constructed from four helices, two from each subunit, also packed together in a four-helix bundle arrangement. A sulfate ion occupies the aspartate-binding site. All hydrogen bonds made to aspartate are substituted by direct or water-mediated hydrogen bonds to the sulfate. Comparison of the Escherichia coli aspartate-receptor structure with that of Salmonella typhimurium [Milburn, Prive, Milligan, Scott, Yeh, Jancarik, Koshland & Kim (1991). Science, 254, 1342-1347; Scott, Milligan, Milburn, Prive, Yeh, Koshland & Kim (1993). J. Mol. Biol. 232, 555-573] reveals strong conservation in the structure of the monomer, but more divergence in the orientation of the subunits with respect to one another. Mutations that render the Escherichia coli receptor incapable of responding to maltose are either located in spatially conserved sites or in regions of the structures that have high temperature factors and are therefore likely to be quite flexible. The inability of the receptor from Salmonella typhimurium to respond to maltose may, therefore, be because of differences in amino acids located on the binding surface rather than structural differences.