Biomedical subjects
Publications and source records attributed to M G Rossmann.
Cauliflower mosaic virus has been crystallized in hanging and sitting drops. The hexagonal and octahedrally shaped crystals are up to 0.5 mm in mean diameter. The octahedrally shaped crystals diffract to about 27 A resolution. The results are discussed in relation to the lability and aggregation of the virions.
The structure of Mengo virus had been determined from crystals grown in the presence of 100 mM phosphate buffer at pH 7.4. It is shown that Mengo virus is poorly infectious at the phosphate concentration similar to that in which it was crystallized. Maximal infectivity is achieved at 10 mM phosphate or less in physiological saline. The phosphate effect is ameliorated when the pH is lowered to 4.6. Although it has not been possible to study the crystal structure of the virus at low phosphate concentrations, it is shown that increasing the Cl- concentration at pH 6.2 or decreasing the pH to 4.6 causes substantial conformational changes confined to the "pit," a deep surface depression. These structural changes involve a movement of the "FMDV loop" (GH loop) in VP1, an ordering of the "VP3 loop" (GH loop in VP3) between 3176 and 3182, the displacement of a bound phosphate near the "FMDV loop" (GH loop in VP1), and movement of the carboxy terminus of VP2. The changes in conformation are correlated with the dissociation of the virion into pentamers at pH 6.2 and 150 mM Cl-. The localization of the conformational changes and the correlated role of the phosphate in controlling infectivity support the hypothesis that the "pit" is the receptor attachment site.
A recombinant form of human immunodeficiency virus capsid protein, p24, expressed in Escherichia coli has been purified to homogeneity and separated into distinct isoelectric forms. A monoclonal antibody, mAb25.4, which recognizes an epitope in the amino-terminal region of p24, has been purified to homogeneity from ascites fluid and digested with papain to produce the respective antigen-binding fragment (Fab). The Fab25.4 was purified from the digestion mixture and separated into two distinct isoelectric forms. The two Fab species were each complexed with one isoelectric form of the recombinant p24 by incubating equimolar quantities of the two proteins. Two different crystal morphologies of the p24-Fab25.4 complex were obtained by the vapor-diffusion method with 12-24% PEG 3350 as the precipitant. One of these crystal forms has unit-cell parameters of a = 92.1 A, b = 85.4 A, c = 54.0 A, alpha = gamma = 90.0 degrees and beta = 90.4 degrees and belongs to the monoclinic space group P2(1), with one molecule of the complex per asymmetric unit. These crystals strongly diffracted x-rays to at least 2.7-A resolution.
The three-dimensional structure of human rhinovirus 14 has a deep surface depression or "canyon" encircling each of the twelve 5-fold vertices. The canyon's surface is inaccessible to the broad antigen binding region of antibodies, permitting conservation of residues that might be required for host cell receptor recognition without danger of attack by the host's immune system. In contrast, the exposed surface features, where neutralizing antibodies are known to bind, change rapidly under pressure from the host's immune system. It was, therefore, hypothesized that this depression was the site of receptor attachment. Similar, but smaller, depressions had been observed previously on both the hemagglutinin and neuraminidase spikes of influenza virus. These have also been shown to be the site of host cell interaction. Although support for the canyon hypothesis was only circumstantial in the first place, there are now extensive confirmatory data. These include site-specific mutations of residues in the canyon and conformational changes induced in the canyon by the binding of small organic molecules, all of which alter receptor attachment. The strategy used in human rhinovirus 14 to protect the viral receptor attachment site from immune surveillance may be utilized not only in other picornaviruses but also in many other types of viruses including human immunodeficiency virus.
The core protein of Sindbis virus has been crystallized. Three different crystal forms have been observed. They diffract variously from 2.5 A to 3.5 A resolution.
Mutants of human rhinovirus 14 were isolated and characterized by searching for resistance to compounds that inhibit viral uncoating. The portions of the RNA that code for amino acids that surround the antiviral compound binding site were sequenced. X-ray analysis of two of these mutants, 1188 Val----Leu and 1199 Cys----Tyr, shows that these were single-site substitutions which would sterically hinder drug binding. Differences in the resistance of mutant viruses to various antiviral compounds may be rationalized in terms of the three-dimensional structures of these mutants. Predictions of the structures of mutant rhinovirus 14 with the substitutions 1188 Val----Leu, 1199 Cys----Tyr and 1199 Cys----Trp in VP1 were made using a molecular dynamics technique. The predicted structure of the 1199 Cys----Tyr mutant was consistent with the electron density map, while the 1188 Val----Leu prediction was not. Large (up to 1.4 A) conformational differences between native rhinovirus 14 and the 1199 Cys----Tyr mutant occurred in main-chain atoms near the mutation site. These changes, as well as the orientation of the 1199 tyrosine side-chain, were correctly predicted by the molecular dynamics calculation. The structure of the predicted 1199 Cys----Trp mutation is consistent with the drug-resistant properties of this virus.
The structure of Mengo virus was determined to 3.0 A resolution using human rhinovirus 14 as an initial phasing model at 8.0 A resolution. Oscillation diffraction photographs were collected at the Cornell High Energy Synchrotron Source using orthorhombic Mengo virus crystals. The crystal space group was P2(1)2(1)2(1), a = 441.4, b = 427.3 and c = 421.9 A, with one icosahedral particle per asymmetric unit, giving 60-fold noncrystallographic redundancy. The orientations of the four viral particles in the unit cell were determined with a rotation function. Their positions relative to the crystallographic symmetry axes were found by a combination of Patterson-function analysis and a subsequent R-factor search using human rhinovirus 14 atomic coordinates as a model. The initial phases to 8.0 A resolution were then computed by placing human rhinovirus 14 particles in the orientations and positions of Mengo virus particles. These phases were improved by ten cycles of real-space molecular replacement averaging. Phases between 8.0 and 3.0 A resolution were obtained by molecular replacement phase extension. One or two reciprocal-space lattice points were used for each extension followed by two cycles of averaging.
X-Ray diffraction data have been obtained for nine related antiviral agents ("WIN compounds") while bound to human rhinovirus 14 (HRV14). These compounds can inhibit both viral attachment to host cells and uncoating. To calculate interpretable electron density maps it was necessary to account for (1) the low (approximately 60%) occupancies of these compounds in the crystal, (2) the large (up to 7.9 A) conformational changes induced at the attachment site, and (3) the incomplete diffraction data. Application of a density difference map technique, which exploits the 20-fold noncrystallographic redundancy in HRV14, resulted in clear images of the HRV14:WIN complexes. A real-space refinement procedure was used to fit atomic models to these maps. The binding site of WIN compounds in HRV14 is a hydrophobic pocket composed mainly from residues that form the beta-barrel of VP1. Among rhinoviruses, the residues associated with the binding pocket are far more conserved than external residues and are mostly contained within regular secondary structural elements. Molecular dynamics simulations of three HRV14:WIN complexes suggest that portions of the WIN compounds and viral protein near the entrance of the binding pocket are more flexible than portions deeper within the beta-barrel.
Explore the source record for details and available documents.
The tertiary structure of most icosahedral viral capsid proteins consists of an eight-stranded antiparallel beta-barrel with a hydrophobic interior. In a group of picornaviruses, this hydrophobic pocket can be filled by suitable organic molecules, which stop viral uncoating after attachment and penetration into the host cell. The antiviral activity of these agents is probably due to increased rigidity of the capsid protein, thereby inhibiting disassembly. The hydrophobic pocket may be an essential functional component of the protein and may have been conserved in the evolution of many viruses from a common precursor. Since eight-stranded anti-parallel beta-barrels, with a topology as in viral capsid proteins, are not generally found in other proteins involved in cell metabolism, antiviral agents that bind in the interior of viral capsid proteins are likely to be more virus-specific and less cytotoxic. Furthermore, the greatest conservation of viral capsid proteins occurs within this pocket, whereas the least conserved part is the antigenic exterior. Thus, compounds that bind to such a pocket are likely to be effective against a broader group of serologically distinct viruses. Discovery of antiviral agents of this type depends on designing compounds that can enter and fit snugly into the hydrophobic pocket of a particular viral capsid protein.
The three-dimensional structure of human rhinovirus 14 has a deep surface depression or "canyon" encircling each of the twelve fivefold vertices. The canyon's surface is inaccessible to the broad antigen binding region of antibodies, permitting conservation of residues that might be required for host cell receptor recognition without danger of attack by the host's immune system. In contrast, the exposed surface features, where neutralizing antibodies are known to bind, change rapidly under pressure from the host's immune system. It was, therefore, hypothesized that this depression was the site of receptor attachment. Similar, but smaller, depressions had been observed previously on both the hemagglutinin and neuraminidase spikes of influenza virus. These have also been shown to be the site of host cell interaction. Although support for the canyon hypothesis was only circumstantial in the first place, there are now extensive confirmatory data. These include site-specific mutations of residues in the canyon and conformational changes induced in the canyon by the binding of small organic molecules, all of which alter receptor attachment. The strategy used in human rhinovirus 14 to protect the viral receptor attachment site from immune surveillance may be utilized not only in other picornaviruses but also in many other types of viruses including human immunodeficiency virus.
The 3-dimensional structures of a number of small animal RNA viruses are now known to near-atomic resolution. All these viruses have similar structures and have, in all probability, diverged from a common ancestral virus able to infect a variety of organisms. The structures have elucidated the site of attachment to host cell receptors and the mode of protection of this site against host immune pressure. An internal hydrophobic pocket in rhinoviruses is the site for binding of antiviral drugs that inhibit uncoating and can inhibit attachment in some viruses. The pocket is probably a functional necessity, and thus is a suitable target for well-designed antiviral agents in many viruses.
A series of eight antiviral compounds complexed with human rhinovirus 14 (HRV-14) were previously shown to displace segments of polypeptide chains in the floor of the "canyon" by as much as 0.45 nm in C-alpha positions from the native conformation (J. Badger, I. Minor, M. J. Kremer, M. A. Oliveira, T. J. Smith, J. P. Griffith, D. M. A. Guerin, S. Krishnaswamy, M. Luo, M. G. Rossman, M. A. McKinlay, G. D. Diana, F. J. Dutko, M. Fancher, R. R. Rueckert, and B. A. Heinz, Proc. Natl. Acad. Sci. USA 85:3304-3308, 1988). Because the canyon is thought to serve as the viral receptor-binding site (M. G. Rossmann, E. Arnold, J. W. Erickson, E. A. Frankenberger, J. P. Griffith, H. J. Hecht, J. E. Johnson, G. Kamer, M. Luo, A. G. Mosser, R. R. Rueckert, B. Sherry, and G. Vriend, Nature [London] 317:145-153, 1985; M. G. Rossmann and R. R. Rueckert, Microbiol. Sci. 4:206-214, 1987), these compounds were assessed for their ability to block adsorption of HRV-14 to HeLa cell membrane receptors. In parallel experiments, the compounds were assessed directly for antiviral activity in an in vitro plaque reduction assay in intact HeLa cells. All eight compounds blocked the adsorption of 50% of HRV-14 at approximately the same concentration required to reduce the number of visible plaques by 50% (MIC). A structurally related compound which was inactive in the plaque reduction assay had no effect on HRV-14 binding. A drug-resistant mutant of HRV-14 (Leu-1188), which was less sensitive to the eight compounds in plaque reduction assays was similarly less sensitive in the adsorption assay. We propose that the conformational changes in the floor of the HRV-14 canyon induced by these compounds substantially decrease adsorption of the virion to its receptor. These results provide further evidence for the role of the HRV canyon in receptor binding.
Spontaneous mutants of human rhinovirus 14 resistant to WIN 52084, an antiviral compound that inhibits attachment to cells, were isolated by selecting plaques that developed when wild-type virus was plated in the presence of high (2 micrograms/ml) or low (0.1 to 0.4 micrograms/ml) concentrations of the compound. Two classes of drug resistance were observed: a high-resistance (HR) class with a frequency of about 4 x 10(-5), and a low-resistance (LR) class with a 10- to 30-fold-higher frequency. The RNA genomes of 56 HR mutants and 13 LR mutants were sequenced in regions encoding the drug-binding site. The HR mutations mapped to only 2 of the 16 amino acid residues that form the walls of the drug-binding pocket. The side chains of these two residues point directly into the pocket and were invariably replaced by bulkier groups. These findings, and patterns of resistance to related WIN compounds, support the concept that HR mutations may hinder the entry or seating of drug within the binding pocket. In contrast, all of the LR mutations mapped to portions of the polypeptide chain near the canyon floor that move when the drug is inserted. Because several LR mutations partially reverse the attachment-inhibiting effect of WIN compounds, these mutants provide useful tools for studying the regions of the capsid structure involved in attachment. This paper shows that the method of escape mutant analysis, previously used to identify antibody binding sites on human rhinovirus 14, is also applicable to analysis of antiviral drug activity.
Explore the source record for details and available documents.
Examining the change in the state of the art of antiviral drug discovery, from the discovery nearly 40 years ago of a smallpox drug (methisazone) of unknown mechanism of action to the analysis of an antiviral agent bound to a virus at atomic resolution, gives the reader an appreciation for the magnitude of the advances in the science of antiviral chemotherapy. The clinical success of amantadine, acyclovir, and azidothymidine have proved that antiviral chemotherapy is a reality. The next step will be to apply the new tools of x-ray crystallography and computational chemistry to the antiviral challenges that lie before us.
Explore the source record for details and available documents.
The structure of human rhinovirus 14 has been refined, by the method of restrained least squares, to an R factor of 0.16 for various random samples between 6 and 3 A resolution with F greater than 3 sigma (F). As a first step the non-crystallographic symmetry parameters were optimized using the initial atomic model in a rigid-body refinement procedure. Phase determination by the molecular-replacement phase extension and refinement procedure was continued to 2.94 A resolution, employing the improved non-crystallographic symmetry operators. The resultant structure-factor phases and weights, together with the measured amplitudes, constituted the X-ray observations used in the restrained refinement. The Hendrickson-Konnert program system [Konnert & Hendrickson (1980). Acta Cryst. A36, 344-350] was modified to incorporate non-crystallographic symmetry constrains and structure-factor phases as observations. The non-bonded contacts between subunits related by non-crystallographic symmetry were also restrained.