Conformational Proteomics of Macromolecular Architecture

The following sections are included:

• PREFACE

• CONTENTS

This paper traces the beginnings of structural virology, from the early 1950's to the presentation of the Caspar-Klug theory of virus structure in 1962. It focuses primarily on the virus research of Francis Crick, James Watson, Rosalind Franklin, Aaron Klug, and Donald Caspar. Collaborative efforts in X-ray crystallography and electron microscopy in combination with intellectual triggers from the Art world provided the soil from which the early theories of virus structure grew and matured.

The quest for relations between form and function is the substance of structural molecular biology. Biological systems combine a high degree of organization with a capacity for dynamic behavior. These features seem to originate from the specific bonding properties of protein molecules that can appear in more than one structural configuration. An attempt to explain how multiple copies of a protein can self-assemble into a cage with a strict symmetry lattice was given in the 1962 theory of quasi-equivalence. The same concept of protein adaptability may be used to approach dynamic behavior of regular biological assemblies as a general. We here describe the principle with two examples that did not originally apply to the theory.

Viral coat proteins have a variety of functions in addition to form the protecting shell around the viral nucleic acid. In many cases, the protein has disordered arms. These arms are used to regulate the assembly of coat proteins into symmetric shells, to interact with the viral nucleic acid and to control the release of the viral genome.

New findings challenge our understanding of the alpha virus structure and fusion mechanism. It is evident from recent work in electron cryomicroscopy, cryoEM, that the external domains of the membrane-anchored glycoproteins, E1 and E2, form a shell at some distance above the membrane. From there, the glycoproteins protrude further outwards as three-lobed spikes. They present a receptor-binding site residing in E2 at their outermost domains, distal to the center of the spike. The ectodomain of the fusion protein, the E1, has an elongated shape, as revealed by X-ray crystallography. Fitted in the cryoEM structure of the virus, the C-terminal and central parts of the E1 ectodomain fill the major portion of the shell, while the fusion peptide loop hides under the receptor-binding domain in the spike. With this structural background, the alphaviruses represent an intriguing new fusion principle, differing in many aspects from the established influenza model. This mechanism is now on its way to be revealed.

The membrane proteins, which have key roles in various biological processes such as photosynthesis, respiration, ion pump, and so on, are estimated for 25–30% of all the species of proteins. The membrane proteins whose tertiary structures are known are, however, less than one % of those of water-soluble proteins. It is because both purification and crystallization of membrane proteins are more difficult than other proteins. The procedures of purification and crystallization were summarized for the membrane proteins whose structures were determined by X-ray method. Each membrane protein has suitable detergents for respective purification and crystallization. There are two types of crystals of membrane proteins diffracting at high resolution. In the type II crystal hydrophilic surfaces of molecules contribute mainly to inter-molecular interactions, while in the type I crystals transmembrane surfaces with hydrophobic nature contact with each other. The quality of membrane protein crystals can be improved by adjusting solvent contents as that of water-soluble proteins.

We consider the possibility that X-ray Free Electron Lasers might be useful for structure determination of membrane proteins in 2-dimensional crystals. In particular, we consider whether it might be possible to collect useful diffraction data from thin sheets of membrane proteins in single-shot experiments using ultrashort, ultrabright X-ray pulses, where the diffraction is expected to be faster than much of the damage. Two things are needed from the X-ray source: high intensity, and short pulses. Data could be collected via grazing-incidence diffraction on separate samples in different orientations. Advances in ‘nanotechnology’ for ordering samples on surfaces will be important for this endeavor.

Water is the most abundant molecule in cells on the earth. The cell membranes are therefore required to provide an effective water channel function. By electron crystallography using electron cryo-microscopy, with a super cooled stage, the structure of the water channel protein; aquaporin-1 is analyzed. The resolved structure reveals a unique folding structure allowing an intriguing mechanism for water channeling. This exemplifies how improvement in technical design, based on theoretical considerations, has successfully advanced scientific insight.

The present understanding of Clathrin and its functions in cellular contexts comes from the combined efforts in different fields. These include strikingly diverse examples: the developing chicken oocyte as the yolk proteins are internalized, human placenta, nerve synapses and unfortunately, viral infection. Clathrin cage assemblies with adaptor protein AP2 and other components are here described, based on studies by electron cryomicroscopy and X-ray crystallography.

Multifunctional enzyme complexes can reach molecular masses of MDa and above, and exceed the ribosome in size. Covalently attached swinging arms, such as lipoyl groups, biotinyl groups and phosphopantetheinyl groups, are essential to their reaction mechanisms. Unexpectedly, it turns out that protein domains contribute to the processes of molecular recognition that define, channel and protect the substrates and catalytic intermediates. The crucial part played by the mechanical motion of protein domains and the role of the molecular architecture that underlies the interactions of the component enzymes can now be identified and assessed. Such enzymes are better now regarded as sophisticated biological nanornachines.

The protein shells of the bifunctional Lumazine/Riboflavin synthase complex found in bacteria, archaea and plants show some similarity to the assembly of small spherical viruses. Sixty lumazine synthase subunits form a T = 1 icosahedral capsid, which instead of nucleic acids in the central core, contains a trimer of riboflavin synthase. Lumazine synthases from fungi, yeasts and some bacteria, however, exist only in pentameric form. Capsid formation in icosahedral lumazine synthases is dependent on the presence of certain substrate-analogous ligands, on pH and phosphate concentration. The experimental background from X-ray crystallography, X-ray small angle scattering and electron microscopy will be discussed. Different active assemblies of the enzyme are observed in vivo and in vitro. There is experimental evidence for the formation of large capsids, obtained spontaneously or after certain mutations to the sequence of the lumazine synthase subunit. Those presumably metastable T = 3 capsids can be reassembled into T = 1 capsids by ligand-driven reassembly in vitro. The active site of lumazine synthase is relatively resilient to point mutations. One lethal mutation to the binding site for the phosphate-substrate, however, has a strong influence on both, capsid stability and enzymatic activity.

The crystal structures of homooligomeric ATP sulfurylase from three different organisms have recently been determined. The enzymes are (a) the allosteric hexamer from the filamentous fungus, Penicillium ckrysogenum, (b) the non-allosteric hexamer from the yeast, Succharomyces cereviseae (both of which function in vivo to produce adenylyl-sulfate and PP_i from MgATP and inorganic sulfate), and (c) the dimeric enzyme from the Riftia (tubeworm) symbiont (a sulfur chemolithotroph, in which the enzyme runs in the reverse mode in vivo). The subunits of all three enzymes have the same overall fold and active site residues. The structures suggest that the different kinetic behavior of these enzymes relies not on differences in active site residues per se, but rather, on “second layer” residues that “tune” the properties of the active site. One important “second layer” is a mobile loop (“switch”) modulating the charge of an active site arginyl residue.

High-resolution crystal structures of functionally active ribosomal particles provide unique tools for understanding key questions concerning ribosomal function, mobility, dynamics, and involvement in cellular regulation. Structure analysis of complexes of ribosomal particles with substrate analogs and universal drugs indicated that ribosomes provide the structural frame for precise positioning of the tRNA molecules rather than participate in the catalytic event, and that the peptide bond is being formed by a nucleophilic attack of the amino moiety of the residue bound to A-site tRNA on the carbonyl carbon at the P-site. Clinically relevant antibiotics interact almost exclusively with rRNA. They interfere with substrate binding, limit the conformational mobility, block the nascent chain exit tunnel or hinder the progression of growing peptide chains.

Even though its structure is known to atomic resolution, the ways in which the ribosome accomplishes its tasks in synthesizing proteins are still unknown. The key to an understanding of its dynamics might be found in cryo-electron microscopy of trapped states, and an interpretation of the resultant density maps by “molding” the X-ray structures into them. First results, obtained by application of real-space refinement techniques to cryo-EM maps of complexes in different conformations, indicate a complicated internal reorganization. The question arises as to whether the observed conformational changes accompanying ribosomal function might be predictable based on the architecture of the macromolecular complex. It has indeed been possible to derive one of the principal motions (the ratchet motion) by normal mode analysis of the ribosome represented as a simplified mechanical system.

The analysis of protein synthesis has taken a leap due to the dramatic progress in crystallographic and cryo-EM studies of ribosomes and their subunits. To be able to understand the mechanisms involved in protein synthesis one has to consider the essential roles of the translation factors that catalyze different steps of the process. Most steps of protein synthesis are catalyzed by these factors. Of specific interest are the translational GTPases that undergo significant conformational changes associated with GTP hydrolysis. Two major questions are how is the GTPase activity induced and how does the conformational change induce the desired action in translation.

The bacterial flagellum is a dynamic molecular system made of a rotary motor, a universal joint, and a long helical propeller, by means of which bacteria swim. The helical propeller, for example, is made of a single protein flagellin, and yet its curved and twisted tubular structure can switch between left- and right-handed helical forms in response to the twisting force produced by quick reversal of the motor rotation, allowing bacteria to alternate their swimming pattern between run and tumble. Other parts also exert mechanical functions by their dynamic behaviors, and all these structures are constructed by a self-assembly process. Some of these dynamic aspects have been revealed by structural studies.

The cyclical interaction between myosin and the actin filament is responsible for muscle contraction. The myosin cross-bridge, which is the ATPase, binds to actin and then undergoes a conformational change (the power stroke) that “rows” the actin filament along. Protein crystallography of myosin has yielded high-resolution models of the beginning and end of the power stroke, which is driven by ATP hydrolysis. ATP also controls the cross bridge affinity for the actin filament. This is low in the presence of ATP and much higher without nucleotide. Recent high-resolution electron microscopy of the actomyosin complex has yielded atomic models of the actin myosin interaction that show two new myosin conformations. These explain the reciprocal link between actin affinity and ATP affinity. Thus there are four states of the myosin cross bridge. The function of the myosin cross bridge is carried out by regulated interactions between these four states.

Crystallization is recognized among structural biologists as a necessary process before three-dimensional structure can be solved at an atomic level. Crystallization has a dose of mysticism among protein chemist. Some treats it as an “art” and others as “black magic”. These concepts aroused from a limited knowledge in the physical chemistry of proteins in solution. Crystallization appears only in a metastable state. To define crystallization conditions the experiments are guided either by a chance search or by dedicated factorial design. Here we will briefly describe a factorial design method to rationally approach the metastable state. In summary, there is nothing mysterious in crystallization of biological macromolecules, and the success can often be achieved within a limited number of experiments.

Macromolecular assemblies, like viruses, are often built by multiple copies of a few components. These may have similar or diverse functions. The multivalency of the assembly allows ligand recognition with high avidity. Nevertheless, affinity is linked to the monovalent ligand interaction, related to the nature of the interactive surface. Such interactions can be followed in real time by the aid of surface plasmon resonance. Thus a sensor surface may be prepared with either the assembly or the ligand immobilized at the sensor and their interaction studied. Kinetic and thermodynamic properties of ligand binding to the macro-molecular assembly can be determined. Variations in the structure of the assembly, like those occurring during virus infection may also be revealed by this technique.

The development of the proteomics field in the post-genomics era has led to an accumulation of data concerning parameters affecting protein folding and assembly. Rationally implemented search-and-evaluation engines would provide a valuable tool for structural and functional predictions directly from the sequence of a protein. This would have an impact on many fields, such as folding, regulation of protein function and in other biotechnical applications.

Virus structures represent mega-molecular nucleoprotein complexes. The three dimensional structures of 74 unique virus capsids, from 21 families and 30 different genera, have been determined at near atomic resolution. We have devised a website and a database of high-resolution virus structures namely VIrus Particle ExploreR (VIPER: http://mmtsb.scripps.edu/viper/) as a repository of virus structures, where all the structures are stored in a single (standard) icosahedral convention. Each capsid is shown pictorially along with a list of the physical properties. Furthermore, the derived results of structural and computational analyses on each capsid are provided highlighting the inter-subunit residue-residue contacts, binding energies, quasi-equivalence and assembly pathways. The structural and analysis tools developed to analyze virus structures can be accessed through the VIPER web site. Efforts are ongoing to include cryo-EM reconstructions as well as the models fitted into these densities.

The following sections are included:

• INDEX