Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Report of the Research Briefing Panel on Protein Structure and Biological Function
Research Briefing Panel on Protein Structure and Biological Function Frederic Richards (Chairman), Yale University, New Haven, Conn. Robert Baldwin, Stanford University Medical Center, Palo Alto, Calif. Gerald R. Galluppi, Monsanto Company, St. Louis, Mo. Robert Griffin, Massachusetts Institute of Technology, Cambridge, Mass. Emil Thomas Kaiser, Rockefeller University, New York, N.Y. Brian Matthews, University of Oregon, Eugene, Oreg. l. Andrew McCammon, University of Houston-University Park, Houston, Tex. Alfred Recifield, Brandeis University, Waltham, Mass. Brian Reid, University of Washington, Seattle, Wash. Robert Saner, Massachusetts Institute of Technology, Cambridge, Mass. 38 Alan Schechter, National Institutes of Health, Bethesda, Md. Paul Sigler, University of Chicago, Chicago, Ill. Peter van Hippel, University of Oregon, Eugene, Oreg. Don Wiley, Harvard University, Cambridge, Mass. Staff Barbara FiIner, Director, Division of Health Sciences Policy, Institute of Medicine Naomi Hudson, Administrative Secretary Allan R. Hoffman, Executive Director, Committee on Science, Engineering, and Public Policy
Report of the Research Briefing Panel on Protein Structure and Biological Function INTRODUCTION Proteins are involved in every biological function. As enzymes, they catalyze the chemical reactions of cells. As hormones and growth factors, they regulate the develop- ment of cells and coordinate the functions of distant organs in the body. In various filamentous forms, they control the shape of cells and the dramatic alterations that oc- cur during cell division. In muscle, proteins change chemical energy into mechanical en- ergy and cause movement. As components of membranes, they control the traffic of molecules anct information among the var- ious cellular compartments. Hemoglobin, a protein in blood, is specifically designed to transport oxygen between organs; other blood proteins, such as clotting factors and circulating antibodies, act as defenses against trauma and infection. In plants, a highly organized collection of membrane proteins is involved in the complex process of pho- tosynthesis, without which there would be no higher animal forms. Proteins are polymer molecules com- posed of amino acids that are connected by links known as peptide bonds. An individ- ual protein molecule may contain hundreds 39 or thousands of amino acids arranged in one, or several, polypeptide chains. Each chain folds into a particular three-dimensional configuration that is essential for its highly specific biological function. In this report, we focus on the experimental and theoret- ical investigation of the three-dimensional structure of proteins, at the level of reso- lution of individual atoms. The study of protein structures, fre- quently referred to as structural biology, is in a period of great excitement brought about by developments in several fields. Many proteins of special interest are now available in unprecedented amounts. The chemical synthesis of polypeptide chains of increas- ing size has steadily improved over the past 20 years, and recently a chain length of over 100 amino acids was achieved. In the past 5 years, the biological synthesis of proteins, through cloning, has been reduced to stan- dard practice, not only in the laboratory but also on a commercial scale. These comple- mentary procedures supply proteins of de- fined sequence in substantial amounts. At the same time, spectacular advances have been made in x-ray diffraction and nuclear magnetic resonance spectroscopy, two tech- niques for determining structure. New pro
cedures and refined equipment have ex- panded the range of application and the size of the proteins that can be studied. Chem- ical theory also has been developing at a rapid rate, especially those branches related to polymers. Furthermore, with the distri- bution of faster, smaller, and less expensive computer hardware, there has been con- stant improvement in access to more ad- vanced computing capability, a factor that has played an important role in theoretical and experimental studies. The coming to- gether of these separate developments makes structural biology ready for an explosive in- crease in the determination and under- standing of high-resolution structures of proteins and protein complexes. The new surge of structural information will dramatically improve our understand- ing of the processes of biological control and will guide the design of proteins or other products that will be developed either to cause purposeful malfunctions (e.g., insec- ticides) or to correct natural malfunctions (e.g., to improve human health). Guided by the three-dimensional structure, changes can be introduced into the sequence of an en- zyme that can alter the specificity of the cat- alyzed reaction and/or its catalytic rate. In addition, structural analysis may reveal spe- cific differences in essential enzymes that will enable geneticists to engineer protein sequences, or drug designers to produce re- agents, that will selectively counter harmful bacteria or insects without harming the host. RESEARCH GOALS THE FOLDING PROBLEM Inside a cell, amino acids are assembled into peptide chains by a complex system that translates the genetic message into spe- cific amino acid sequences. Following syn- thesis, the chains fold into compact protein molecules. For many isolated polypeptides, conditions can be provided in which this folding step will occur spontaneously, 40 yielding a biologically active molecule iden- tical to the native, ceD-derived protein. Thus, all of the information required to produce the final structure is contained in the amino acid sequence. The prediction of the de- tailed three-dimensional structure of a pro- tein from a given sequence is known as the folding problem. It is the most fundamental problem at the chemistry-biology interface, and its solution has the highest long-range · - priority. The folding problem is not only a major intellectual challenge but also an urgent and immediate problem at the practical level in biotechnology. Successful industrial pro- duction of a biologically active protein fre- quently depends on the ability to induce a cloned polypeptide to fold correctly. PROTEIN STABILITY Protein stability Is a specific issue within the folding problem. Thermodynamically, protein structures are only marginally sta- ble, and small changes can substantially in- crease or decrease their effective stability. Not only is prediction of stability a strin- gent, but elusive, test of theoretical un- derstanding, but also direct practical appli- cations, through genetic engineering of pro- teins, are immediately at hand. Resistance to thermal destruction or to degradation by enzymes secreted by microorganisms, for example, are highly desirable properties for pharmacological agents and enzymes in in- dustrial use. L~GAND BINDING The specificity and strength with which ligands, either small or large molecules, bind to proteins is a central feature of biological function. The result of this binding may be simple sequestration for storage or removal, a catalytic event if the protein is an enzyme, the development and transmission of a sig- nal if the protein is a receptor, or the switch- ing off of a gene if it is a repressor. The
PROTEIN STRUCTURE AND BIOLOGICAL FUNCTION ability to predict the structure either of a protein ligand or an active site the part of the protein where the ligand binds is cen- tral to the rational design of new drugs or new enzymes. SIGNAL TRANSMISSION Signal transmission from one part of a molecule to another is essential for the reg- ulation of complex enzymes and for the ac- tivity of many receptor systems. Its mechanism, however, is poorly under- stood. Changes in the dynamic properties of an entire protein molecule can be caused by the binding of a ligand to a relatively small active site; this phenomenon has clear Implications for information transfer. In some proteins, however, similar interactions pro- duce only localized changes in structure or dynamics. Full understanding will only come through a detailed study of the relevant structures and their properties. RECENT ADVANCES IN KNOWLEDGE About 300 protein structures are known, and about 10 to 20 new structures are re- ported each year. In many cases, knowledge of these structures has brought us closer to the goals outlined above understanding folding, stability, ligand binding, signal transmission, and catalytic activity. A few examples will illustrate what has been gleaned from studies of structure and the potential applications of the knowledge. Ang~otensin is a chemical in human blood that is involved in regulation of blood pres- sure. When it is modified by an enzyme known as ang~otensin-converting enzyme (ACE), it causes a rapict increase in blood pressure. Control of high blood pressure seemed possible if an inhibitor of ACE could be found. At the time a drug development program was started, ACE had not been iso- lated in pure form from humans. But the structure of an enzyme from the pancreas of cows, which happens to catalyze a chem 41 ically similar reaction, was known at high resolution. Based on the detailed structure of the animal enzyme, especially the cata- lytic binding site, it was possible to make a mode! for the active site of human ACE. With this model, the drug captopril, a strong inhibitor of human ACE, was successfully designed and synthesized. Subsequently, protein chemistry was used to design ena- lapril, a new drug in which some of the unwanted side effects of captopri] have been eliminated. Influenza virus offers another example. This virus causes recurring epidemics (and the continuing need for development of new vaccines) because its surface proteins vary so much. Recently, the structure of hae- magglutinin, a surface protein, was deter- mined. Consequently, the regions that vary with the strain of virus have been located. Even more promising is the discovery of a region that does not vary; it provides a pocket in the protein and may be an excellent target for drug development. The full high-reso- lution structure was essential to the discov- ery of this region. Recent structural studies of the coIct and polio viruses and adenovi- ruses have opened up similar exciting op- portunities. The recent determination of the structure of a part of an enzyme called DNA poly- merase I, in conjunction with related kinetic studies, is beginning to lay a general foun- dation for understanding how "processive" enzyme reactions work. Such enzymes latch onto a long molecule (the DNA thread in the case of DNA polymerase) and then move rapidly along it without letting go, much like a train on a track. Such a mechanism is entirely new in the field of enzymology. Not only is this fascinating to biochemists study- ing the replication of DNA and RNA of central importance to life but the practical importance is considerable. The processive digestion of polysaccharides and other macromolecules is of great importance to food processing and pharmaceutical indus- tries, for example.
Instrumentation clevelopments have been essential to these and numerous other ex- amples of major progress in structural bi- ology. New or improved instrumentation has led to tremendous savings of time and man- power, as well as to unique routes for solv- ing research problems. Accordingly, much of the remainder of this report will focus on research advances in instrumentation and data analysis. TECHNOLOGICAL ADVANCES X-RAY DIFFRACTION Since the early 1960s, x-ray crystallogra- phy has continued to provide us with the most detailed and comprehensive picture of three-dimensional protein structure. When x-rays are passed through a crystal, they are diffracted in many directions, and the ge- ometry and intensity of the many diffracted beams are directly related to the structure of the crystal. Improvements in x-ray sources, in data collection equipment, and in the power and availability of computers are ex- pected to continue to enhance the power of these structure studies. Area Detectors The diffraction pattern from a crystal can be recorded all at once (with a photographic fiIm) or one beam at a time (with an appro- priate counter). Data on film must be read optically and then must be converted to a digital form to determine the intensity of each beam. Area detectors are a major new innovation in such data collection. They have the advantage of direct counting while re- taining the multiple recording capability of film with a higher signal-to-noise ratio. Analyses that took weeks or months have been reduced to hours, with a considerable · . gain in accuracy. This time-saving device has had far- reaching effects on the kinds of experiments that are being planned. For example, it will 42 now be possible to take full advantage of the ability of molecular genetics to generate many different mutational changes in a sin- gle protein. The crystallographic examina- lion of the different forms of a Oven protein will be practical for a small group of inves- tigators, or even a single individual. The comparison of structures will be of inesti- mable value in elucidating determinants of structure and structure/function relation- ships. Synchrotron X-Ray Sources Synchrotron x-ray sources are becoming available at various national facilities, and their accessibility provides unique oppor- tunities. Because synchrotron radiation has a continuously varying wavelength, it is possible to collect data at two or more dif- ferent wavelengths. Proper combination of the data sets provides substantial help in overcoming the major stumbling block in solving an unknown structure. The high intensity of the synchrotron ra- diation also far exceeds that of any usual laboratory source. It is possible to collect enough data for structure definition in 10 to 100 milliseconds, and perhaps even faster in the future. With such high data rates, full structural studies of short-lived intermedi- ates in enzyme reactions are possible in principle. The determination of the struc- ture of such intermediates at physiological temperatures would lead to a dramatic im- provement of our understanding of enzyme catalysis. Currently, intermediate states can only be studied when stabilized under un- usual conditions, such as very low temper- ature, so there is always uncertainty about the relevance of any findings to the actual catalytic process. The full benefits of fast synchrotron data collection are not yet being realized because the diffracted intensities are recorded on photographic film. More effective use of synchrotron facilities for protein structure investigations will require a hich-flux area de O J
PROTEIN STRUCTURE AND BIOLOGICAL FUNCTION sector capable of recording perhaps 108 events per second. This is at least i,000 times faster than the commercial instruments now avail- able for laboratory use. Computing and Graphics X-ray studies require computers for data analysis and interpretation. Mode! refine- ment to get the "best" structure is particu- larly computer-intensive. Computer-aided molecular graphics plays an increasing role in the solution of structure and in subse- quent study of the structure. Color is rou- tine and is now central to the effective use of graphics as a laboratory tool. It is possible to examine very complex structures and to selectively "flag" special features or prop- erties by color-coding the atoms. Spatial re- lations that would be extremely difficult to detect by computation become very obvious to the human eye. The emphasis on graph- ics is likely to continue, even with marked improvements in automatic data analysis, which is itself highly computer-intensive. Neutron Diffraction Neutron diffraction has a special role in structure studies because of its unique abil- ity to reveal the position of hydrogen atoms. These atoms are frequently of central im- portance in enzyme-catalyzed reactions, but because hydrogen is so light, it is poorly detectec! if at all in x-ray structures of proteins. However, hydrogen is easily "seen" by neutrons, although such experiments can only be carried out at the national laboratory reactors. Together with the hoped-for up- grading of the high-flux reactors, the de- velopment and capabilities of the new pulse neutron source at Los Alamos win be watched with great interest. Crystallization Protein crystals are unusual examples of the solid state in that they contain a large 43 amount of liquid water. The structure of those proteins for which comparisons have been made is essentially identical in a crystal and in aqueous solution. Production of the highly ordered crystals required for x-ray diffraction studies re- mains an art rather than a science. None- theless, the number of crystallized proteins is increasing rapidly. Of special note is the recent successful crystallization of integral membrane proteins such as the photosyn- thetic reaction center and bacterial rhodop- s~n. NUCLEAR MAGNETIC RESONANCE With a nuclear magnetic resonance (NMR) spectrum, researchers measure the absorp- tion of radio-frequency energy by the nuclei of molecules placed in a magnetic field. The frequency at which an atomic nucleus ab- sorbs radiation is very sensitive to the chem- ical environment provided by the structure of the molecule. A basic problem, however, has been the identification of which peak in the spectrum belongs to which atomic nu- cleus in the structure. New data collection techniques have been developed to extract information about the distances between neighboring atoms, and these techniques have revolutionized the study of proteins up to approximately 12,000 in molecular weight. From these identified spectra, sci- entists can derive characteristic patterns of substructures in the protein. More detailed analysis with computer-intensive distance geometry algorithms can provide the full three-dimensional structure in favorable cases. The recent progress in research on small proteins has been directed toward the de- termination of average structures in solution for comparison with models from x-ray dif- fraction. This work has set the stage for the next fascinating phase of NMR, the study of changes in structure that are induced, for example, by the binding of ligands. While much of this work will be done in partner
ship with investigators using x-ray diffrac- tion, many problems are only accessible through NMR procedures notably those cases in which crystalline materials cannot be obtained. Unique opportunities exist to learn about partially or totally disordered molecules that are important both in equi- librium populations and as reaction inter- mediates. Defining procedures for the precise ma- nipulation of nuclear spin in a molecule- spin engineering will continue to play an important role in the development of op- erating procedures for NMR spectrometers, especially for macromolecules with their complex spin systems. An appropriate se- quence of radio-frequency pulses can dras- tically simplify a complex spectrum, reveal relations between spatially distant atoms, and greatly assist in the essential step of assigning peak signals to portions of the protein structure. Further developments re- quire that young investigators with a back- ground in quantum physics be attracted to this particular area of structural biology. Several other techniques designed for structures larger than those with molecular weights of 15,000 to 20,000 have also been developed. Solid-state NMR has no inher- ent size limit, and there are very interesting applications for membrane proteins or fi- brous materials, such as collagen, which are intrinsically insoluble. Another approach is the direct study of small substrates or in- hibitors interacting with active sites of large enzymes. A number of new developments are being intensively pursued in this area, such as the use of labeled, tightly bound substrates. A third approach is to simplify the spectra by preparing samples with sta- ble isotopes inserted in a limited number of known positions. By a combination of chemical and biological procedures, amino acids are prepared with the isotopes 2H, 13C, or i5N in appropriate positions. These are incorporated into proteins at known loca- tions. The syntheses are often difficult, but the rewards are great because the spectra of 44 the isotopically substituted proteins can be very simple and easy to interpret. More- over, this technique can produce very large signals in comparison with the low back- ground absorption. Even low concentra- tions of relatively unstable intermediates, such as are likely to be important in the protein folding problem, may be detectable in these enriched samples. And as ~ further benefit, data collection times are markedly shortened. SYNTHESIS OF PROTEINS Sensitivity and sample size continue to limit both x-ray crystallography and NMR, which require amounts of material in the 10- to lOO-mg range. An adequate quantity of highly purified proteins of specified se- quence, and, where required, with specific isotopic substitutions, is essential to bio- physical study of structure and function. Within different but overlapping size ranges, quantities of proteins can be produced to- day either by chemical or biological proce- dures. Chemical Synthesis Solid-phase chemical synthesis has been effectively automated, and peptides 30 to 40 amino acids long are readily produced in good yield. Substantially longer peptides also have been synthesized, and continuing im- provement can be expected. The next step toward the synthesis of longer chains is the condensation of preformed fragments. Con- densation is possible by enzymatic as well as chemical methods, but general proce- dures are not as well worked out and de- serve considerably more study. Chemical synthesis allows the insertion of an isotopically labeled amino acid, an amino acid derivative, or even a nonnatural amino acid in any single position in the chain. Multiple-site "mutations" at any selected group of sites thus become easy to produce. In the synthesis of drugs that mimic pep
PROTEIN STRUCTURE AND BIOLOGICAL FUNCTION tides, even the peptide bond may be cir- cumvented in specified locations, leading to resistance to degradation. Limitations in chemical synthesis at this time derive from the problem of optical pu- rity, the yield of the correct sequence, and the roughly linear relationship between the amount of the product and the cost of pro- ducing it. However, these limitations, as wed as the limitation on chain length, are over- come in some biotechnology processes. BioZogicaZ Synthesis The procedures that are used to produce proteins in high yield by cloning in bacteria are well developed. The companion proce- dures for producing single amino acid changes by site-directed mutagenesis are simple, fast, and reliable. In many cases, the fusion of a special "signal sequence" to the protein will result in its secretion, which as- sists in proper folding. Nonetheless, clon- ing is not always successful. Degradation of the product can severely reduce the yield; inside the bacterial cell, the chain may not fold to yield the desired, biologically active molecule; and modification of certain amino acids after polymerization of the protein as required for various eukaryotic proteins ("post-translational" modification) does not occur in bacteria. Cloning in eukaryotic systems, particu- larly human cell culture, is not yet as well developed and is very expensive for pro- ducing proteins in commercial quantities. Improved expression vectors to overpro- duce the desired protein, pro/ease-deficient strains, rapid lysis methods, and better bio- chemical separation procedures are all re- quired. Two important research opportunities are development of better methods for muZtipZe site-directed changes at positions widely sepa- rated in the sequence, and the deveZopment of high expression systems for eukaryotic proteins. it is equally important that these new pro- tein products be characterized rapidly. Pre- liminary structure evaluations by standard 45 biophysical procedures, particularly the var- ious forms of optical spectroscopy, enable screening of protein products and selection of the most interesting for detailed study by x-ray and NMR techniques. THEORY Theoretical studies of proteins are only beginning to have a real impact in relation to biological function. The field has been stimulated by advances in computer tech- nology, theoretical chemistry, and experi- mental biochemistry. It is clear that theoretical studies will play a major role in the design of new proteins and of molecules with which proteins interact. The most highly developed theoretical methods involve molecular dynamics sim- ulations, in which computers are used to simulate atomic motions in a protein and its surroundings. When combined with a new approach giving thermodynamic parame- ters for reactions, dynamic simulations can be used to make predictions concerning rec- ognition and binding among proteins and other molecules. This method has recently been used successfully to calculate the af- finity in an enzyme-inhibitor interaction; it has promise for studies of protein folding, stability, covalent reactivity, and noncova- lent association. Practical applications in- clude the design of drugs, enzymes, antibodies, and other molecules. The rates and mechanisms of enzyme-cat- alyzed reactions and ligand binding are po- tentially accessible through molecular dynamics. A modification known as Brown- ian dynamics is useful for extending calcu- lations into the time range of somewhat slower biological processes. A number of other theoretical approaches, clearly on the horizon, may be useful for predicting the structures of short peptides in solution. Continuing attention should be directed to improving the basic mathematical func- tions and input parameters that are needed both in molecular dynamics and energy
minimization procedures; to improving the treatment of electrostatic interactions; and to the detailed treatment of the water-pro- tein interface where biological activity is ex- pressed. More ad hoc approaches in applying other aspects of basic chemical theory may also be useful in attacking the protein fold- ing problem in which the proper application of first principles is still elusive. BOTTLENECKS AND RECOMMENDATIONS Progress in solving the scientific problems of structural biology will require both per- sonne} qualified in this multidisciplinary area and sophisticated equipment in individual laboratories and national centers. For struc- tural biology to achieve its full potential in contributing to fundamental science, med- icine, agriculture, and the chemical indus- tries, a number of policy concerns should be addressed. I. Basic Research Support Of absolutely central concern in this area, as in others, is maintenance of support for basic research pro- grams at the level of the individual investi- gator and small consortia. 2. Professional Personnel Of comparable concern is the continuing supply and training of professional personnel. Scientific opportun- ities will not be realizect, and the equipment initiatives suggested below will have little effect, if trained personnel are not available. In the recent past, there has been a relatively small number of individuals entering bio- physics, biophysical chemistry, and the general field of structural biology. The ma- jor attraction during that period was clearly molecular genetics. During the past year or two, there has been an increasing number of entering graduate students interested in quantitative structural studies. This student interest has coincided with a dramatic upsurge in activity in the indus- trial sector. A substantial number of bio- technology firms have set up structural units, 46 including x-ray crystallography, high-reso- lution NMR, and theoretical modeling. Re- searchers experienced in one or more aspects of structural biology and general protein chemistry are in high demand at this time, and corporations have attracted much of the presently available talent. Currently available predoctoral training programs may be adequate to provide the necessary graduate student input to this field provided these predoctoral training programs are maintained at a level at least equal to their present levels. A particularly effective source of highly qualified personnel is provided by the Med- ical Scientist Training Program. If there are further cuts in any of the programs, the structure area, which is poised scientifically for substantial progress, will be nipped in the bud, and will suffer proportionally more than the well-populated areas of molecular and cellular biology. The most serious concern for personnel is postdoctoral training. The new genera- tion of structural biologists must be familiar with molecular biology as well as biophys- ics, and it would be highly desirable that molecular biologists with an interest in structure learn at least the rudiments of the biophysical methods. Similarly, physicists and instrumentation engineers whose ex- pertise could be shifted rapidly to structural biology must learn some molecular biology. This interdisciplinary training is difficult to accomplish properly in the time period of a normal doctoral program; postdoctoral training thus becomes even more essential than usual under these circumstances. We urge that the postdoctoral fellowship programs be maintained and, if possible, expanded to cover the present and anticipated needs in structural biology. 3. Supply and Development of Major Instruments The entire field of structural biology is heavily dependent on major equipment items and on unique facilities available at certain national centers. Contin- uing progress on the biological problems in this field will be closely correlated with the
PROTEIN STRUCTURE AND BIOLOGICAL FUNCTION improvement and availability of advanced instrumentation. X-Ray Diffraction Currently available area detectors are having an enormous impact on the efficiency of data collection and on the types of research programs that it is re- alistic to plan. Agencies should! be prepared to fund acquisition of area detectors, and anciliar~u equipment for efficient utilization, widely throughout the structural bioZo~u community over the next few years. cam ~ The effective use of the synchrotron x-ray sources at the national laboratories will depend on the development of high-flux area detectors capable of recording at least 108 events per second. Ef- forts are uncler way for other scientific fields; the needs of structural biology should be consiclered as part of this general develop- ment effort. Nuclear Magnetic Resonance The distri- bution of 500-MHz instruments, or their fu- ture replacements, will continue to present a policy problem. Laboratories devotee! to the development of NMR techniques, or to major long-term protein-structure projects, win need fully dedicated instruments of their own. Shared facilities should still be dedi- cated to the study of macromolecules and should not be expected to provide small molecule spectra as an acIditional service component. Improvements in resolution and sensitiv- ity will depend on the development of stronger magnets. Sensitivity and resolu- tion both become increasingly important as the size of the protein increases, and both are improved as the magnetic field strength of the spectrometer is increased. We suggest that a major effort be launched to interest and encourage instrument companies to produce spectrometers with a 17.5 T (750-MHz) magnet. This appears to be feasible with currently available superconducting wire technology, although some engineering difficulties re- main to be solved. A magnet at 20-25 T (~l,000 MHz) is not out of the question, although an intensive investigation in the materials science area may be involved. Suc 47 cess, however, would have a dramatic im- pact on biophysical NMR studies. Availability of precursor chemicals la- beled with stable isotopes is another NMR concern. The very high price of labeled ma- terial is a serious general problem for sci- entists. The Stable Isotope Facility at Los Alamos is bound by regulation to stop any activity that is taken up in the private sector. In contrast to many other examples of sci- entific resource supply, this particular reg- ulation has worked to the clisadvantage of the research community. There is not a suf- ficiently large market for labeled com- pounds to reduce the price through volume and competition. Even the development of clinical applications would produce a mar- ket for only a relatively small number of compounds and would not cover the broad range neecled for the research proposed above. It is essential that a mechanism be found for providing amino acids and nucleotides with a variety of different stable isotope labeling pat- terns for the research community. Computers The instrumentation needs for theoretical and experimental studies of pro- teins include increased and predictable access to supercomputers; improved communications between these machines anct remote users through networking; and additional access to high-quality graphics devices, scientific workstations, and special-purpose com- puters. X-ray crystallography, NMR, and theo- retical studies all have very heavy com- puting requirements. The larger the molecules and the faster the data collec- tion, the larger will be the computational requirements in the immediate future. Even for structures of modest size, the current iterative x-ray refinement procedures cre- ate major inconvenience to other users of a VAX. (Refinement cycles can occupy tens of hours per cycle.) Protein data process- ing even from current NMR spectrometers requires levels of computation not nor- mally available in the laboratories of in- dividual investigators. Many theoretical
problems are not at all practical on inter- mediate-leve} machines. We estimate that the needs of the current scientific community in structural biology nationwide may already exceed the com- puting power represented by two ad- vanced-leve! Cray machines. The present National Science Foundation Supercompu- ter Center Program is useful in providing access to these machines. This initiative, however, covers all areas of science and may wed become saturated. The necessary phased expansion of this program should include the present and anticipated requirements of the structural biologists. A library of pro- grams specifically written to take advantage of the architecture of these machines also will be essential for their effective use. The efficient use of the supercomputers will depend on the ease and convenience of access. The latter will depend on the speed and effectiveness of the networking that is available to make this possible. Networks suitable for connection of the lower-level 48 computers in the various structure labora- tories also will become increasingly useful. Experience over the next year or two with the currently developing general purpose networks will show whether or not they are adequate for this purpose. A large volume of protein sequence and structure data can be expected in the near future as a result of the many new metho- dologies. Therefore, it is time to plan for the accession and use of the ciata in a comput- erized central data bank. Computer searches of genetic structure data banks have pro- vided significant new insights into biologi- cal phenomena, and a similar outcome can be expected from the protein structure data. Finally, there will be increased need for dedicated microcomputers and various lev- els of graphics workstations in individual laboratories. Although these are not ex- pected to be particularly expensive, the need for their wide distribution is clearly visible now and should have a prominent place in the planning for future equipment support.
