Richard J. Feldmann

Membrane proteins: Amino acid sequence and membrane penetration

Abstract A computer study shows that the membrane-penetrating portion of the erythrocyte surface MN-glycoprotein (Winzler, 1969; Marchesi et al. , 1972) is distinguishable by informal cluster analysis from other segments of globular proteins when sequence length is plotted against hydrophobicity This analysis further suggests the possibility that other membrane-penetrating segments of proteins can be identified in the same way.

Rhodopsin's protein and carbohydrate structure: Selected aspects

A topographic model for rhodopsin has been constructed based upon evaluation of rhodopsins sequence by a secondary structure prediction algorithm as well as chemical and enzymatic modification of rhodopsin in the membrane [Hargrave et al. (1983) Biophys. Struct. Mech. 9, 235-244]. The non-uniform distribution of several amino acids in the primary structure and within the topographic model is discussed. The seven predicted helices were evaluated and each helix was found to have one surface which is much more hydrophobic than the other. Stereoscopic views of a three dimensional model with a functional color-coding scheme incorporating these features are presented. The amino acid sequence of rhodopsin has been compared to other proteins in the Dayhoff Protein Data Bank. No obvious relationship to any other protein sequenced was found. High resolution proton magnetic resonance spectroscopy was used to reinvestigate the structure and relative proportions of rhodopsins major and minor oligosaccharide chains. One major (Man3GlcNAc3) and two minor (Man4GlcNAc3 and Man5GlcNAc3) were observed.

A hypothetical space-filling model of the V-regions of the galactan-binding myeloma immunoglobulin J539

Abstract J539 is an IgA myeloma protein whose V-region amino acid sequences and ligand binding for β-1.6-galactosyl oligosaccharides have been determined in the past. Two hypothetical, three-dimensional models of this protein have now been constructed on a computer display system by modifying the known structures of McPC603. The J539 Fv is 11 amino acids shorter than McPC603, and its χ-carbon backbone was formed by eliminating positions from McPC603 and rejoining the chains at three deletion sites in CDR-1 of V L and CDR-2 and -3 of V H . This maneuver was performed using a three-dimensional viewing system that displayed the amino acids with side chains in the regions around the break sites. The side chains of the amino acids in the model were oriented by a computer program that searched the available space for positions that did not make van der Waals contacts with neighboring atoms. β-1,6-Galactohexaose has been fitted to the J539 combining site using the computer display system. The terminal non-reducing sugar made close contacts with hydrogen bond accepting amino acids in CDR-2 of V H and oxygens of the third and fourth sugars were close to hydrogen bond accepting amino acids in CDR-1 and CDR-3 of V L . While the proposed structures are not unique solutions for J539 they present working models for further immunochemical experiments.

A COMPUTER‐GENERATED THREE‐DIMENSIONAL MODEL OF THE B CHAIN OF BOVINE α‐THROMBIN*

A computer graphic molecular display system has been used to construct a three-dimensional model of the B chain of bovine thrombin. The model is derived from the bovine alpha-chymotrypsin structure as determined by X-ray crystallographic studies. The amino acid sequence of bovine thrombin has been substituted for that of alpha-chymotrypsin, preserving the beta-barrel structure and maximizing homology of the amino acid sequence of the two proteins. With the exception of an area in the vicinity of the specificity binding pocket, most of the changes observed in thrombin occur on the surface of the molecule. The most notable changes observed in the model are the increases on the surface of positively charged (arginine and lysine) and negatively charged (glutamate and aspartate) residues. A glutamate replaces methionine 192 near the entrance to the specificity binding pocket. The nature of this site was further altered by the substitution of an aspartate for serine 189 and an alanine for serine 190. The structure of the resulting specificity binding pocket is consistent with that of serine proteases, which have trypsin-like substrate specificity. The computer graphics molecular display system has been used to insert models of synthetic thrombin inhibitors into the active site of the thrombin B chain model. With the model, it has been possible to correlate the interaction of thrombin with the observed binding constants of two inhibitors of trypsin-like serine proteases, p-amidinophenylmethylsulfonylfluoride (Ki = 1.27 x 10(-6) M) and m-[m-(trifluoromethyl)phenoxypropoxy]benzamidine (KD = 2.9 x 10(-6) M).

A refined model for the variable domains (Fv) of the J539 ß(1,6)-d-galactan-binding immunoglobulin

A refined protocol for building a hypothetical model of the J539 Fv is described. Computer programs for positioning amino acid side chains and structure energy minimization [CHARMM program of Brooks et al., J. comp. Chem. 4, 187-217 (1983)] were employed. Computer modeling was accomplished on an Evans and Sutherland picture system which permitted structure visualization in three dimensions. Peptide backbone breaksites were rejoined by monitoring for correct distances and torsion angles. A physical model was then constructed and used as a basis for further refinements such as aligning conformations around remodeled sites, adjusting proline substitutions and optimizing hydrogen-bond-forming potentials. This structure (J539-ADO) was energy minimized and the final coordinates were obtained from the energy-refined model. The resulting hypothetical J539 structure can be compared to the structure of J539 now being determined by X-ray crystallography. The procedures described can be used for other Fv fragments.

Structure‐Function Relationships of Thrombin Based on the Computer‐Generated Three‐Dimensional Model of the B Chain of Bovine Thrombina

The advent of sophisticated computer graphics systems that permit the representation of macromolecular structure has made it possible to examine protein structure in detail. We have used one aspect of this technology to develop a model of thrombin. The model is based on structural and functional similarities this enzyme exhibits with respect to proteins found in the family of serine proteinases. This review has covered interpretations of the structure of the model based on analyses of data that had been collected before and after the model was developed. On one hand, the conceptualization of primary and secondary features in the model of the active site of thrombin has for the most part been preceded by data from experiments on the interaction of thrombin with naturally occurring substrates and inhibitors. The features of the model explain these data adequately. On the other hand, the model has been more recently used in an interactive way to derive information about the bioregulatory aspects of thrombin. The realization that the amino-terminus portion of the cyanogen-bromide fragment was probably not part of the chemotactic activity, because it was probably internalized in the native protein, has suggested that synthetic analogs should focus more on the carboxyterminus of the peptide. It is hoped that in the future the model will continue to serve more in this function and that it can be used to explore further other aspects about the structural and functional relationships of this enzyme.

PART II. COMPUTER‐ASSISTED MACROMOLECULAR STRUCTURE GENERATION: EXTENSION OF EXISTING INFORMATION: On the Construction of Computer Models of Proteins by the Extension of Crystallographic Structures

One of the central beliefs of molecular biology is that the sequence of amino acids determines the structure of a p r ~ t e i n . ~ ~ Yet our scientific culture has still to discover many of the rules that specify and govern this relationship. The results of X-ray crystallographic determination of specific proteins have led to an understanding of the taxonomy of protein architecture^.^ The protein sequence fileS which had been growing rather slowly because each amino acid sequence had to be determined by chemical techniques is now growing exponentially because amino acid sequences can be derived from nucleic acid sequencese6 Protein extension attempts to shorten the delay be-

Object-oriented, dataflow visualization systems: a paradigm shift?

Discusses the breadth and the effectiveness of application visualization systems (AVSs). The current and future research areas involving AVSs, drawbacks and limitations of certain application areas, possible improvements to AVSs, and alternative analysis and visualization approaches are discussed.<<ETX>>

The carboxyl-terminal one-third of bovine rhodopsin: Its structure and function

Abstract The covalent sequence of the carboxyl-terminal one-third of bovine rhodopsin has been determined. It is 108 amino acids in length and consists of three hydrophilic regions exposed to the aqueous environment at membrane surfaces and two hydrophobic regions which penetrate the disc membrane lipid bilayer. This carboxyl-terminal third of rhodopsin contains the retinal binding site, sites of phosphorylation, the fast-reacting sulfhydryl group and the thermolytic fragment F2. Hydrophobic helices in the sequence have been predicted by a new method developed for application to membrane proteins and a hypothetical “helical hairpin” space-filling working model has been constructed.

Topographic Antigenic Determinants Detected by Monoclonal Antibodies to Myoglobin

We have been interested in studying the structure of immunological receptors on T and B lymphocytes for natural protein antigens, and the nature of the antigenic determinants recognized by these. As a model protein antigens, we have used mammalian myoglobins, as we have also been studying the H-2-linked Ir gene control of the T and B lymphocyte responses to these antigens in mice (Berzofsky, 1978a, b; Berzofsky et al., 1979; Richman et al., 1980; Berzofsky and Richman, 1981; Kohno and Berzofsky, 1982; Berzofsky, submitted for publication). As a model of B lymphocyte receptors for sperm whale myoglobin, we prepared monoclonal hybridoma antibodies in A.SW mice, which have an antigen-specific H-2linked Ir gene for high responsiveness to sperm whale myoglobin, (Berzofsky, 1978), as well an a non-H-2linked gene which results in higher serum antibody responses to many antigens (Dorf et al., 1974; Berzofsky et al., 1977; Pisetsky et al.,1978. We hoped that these antibodies would allow us to extend the knowledge of antigenic determinants on globular proteins which had been obtained with heterogeneous immune serum antibodies (Crumpton, 1974; Reichlin, 1975; Atassi, 1975; Hurrell et al., 1977a; East et al., 1980), and also the knowledge of antibody combining sites, which, because of the requirement for homogeneous antibody preparations, was limited to myeloma proteins which bind small molecules (haptens) and carbohydrates (Potter, 1977; Kabat, 1978).