A publication of the Foresight Institute
An advance in understanding the folding and stability of proteins has been reported by the protein design group at DuPont Merck. Building on their previous work with a4 — a protein of about six dozen amino acids designed completely from scratch — these researchers are now studying how metal-binding sites affect the stability of protein structure.
a4 is a compact bundle of four very stable helices connected at the ends by flexible loops of five amino acids. NMR spectroscopy has shown that despite its thermodynamic stability (i.e., reluctance to spontaneously unfold and flop around randomly), a4 is not as well-ordered internally as many native proteins are. The suspected cause of this intramolecular fluidity (known as a “molten globule state”) was the lack of specific bonds between the helices making up the bundle. The introduction of metal-binding sites at appropriate places in the molecule might allow metal atoms to act as braces between the helices, reducing their jiggling and sliding
So these protein engineers replaced six of the amino acids of a4 by histidine, an amino acid that forms bonds with zinc atoms. The resulting protein, called H6–a4, behaved as anticipated: in the absence of zinc it was far less stable than a4, since the histidines tended to pull the helices into solution; but in the presence of zinc it folded into bundles, like a4, yet also showed the internal well-ordering seen in native proteins. [Tracy M. Handel, et al., Science 261:879:885, 13Aug93]
With a few minor design adjustments, a4 is likely to take on fully native — or even superior — thermodynamic properties. To give this protein a useful function beyond being an interesting research tool, however, will require severe redesign. The big challenge will be to find ways to add functionality without compromising the molecule’s stability.
My personal opinion is that learning to make proteins with improved properties will mainly be of value in the genetic engineering of living things rather than in the chemical or drug industries or in nanotechnology. Although industry today does use a few proteins as catalysts and as drugs, future industry will probably abandon these in favor of smaller molecules that will be more durable, easier to make, and more effective. But for such tasks as improving our crops or upgrading our bodies, we will probably prefer to introduce genes for redesigned proteins rather than to rely on special fertilizers, pesticides, and drugs. Eventually I suppose that nanomachines will reduce the need for protein engineering even in these areas.
A collaboration between researchers at UC Berkeley and Affymax Research Institute has come up with an unusual family of polypeptide analogs called oligocarbamates. Like polypeptides, these new molecules have a molecular “backbone” with various side-chains attached. But instead of the (-C-C-N-)n pattern of carbon and nitrogen atoms that make up the peptide backbone, the pattern in oligocarbamates is (-C-O-C-C-N-)n. In principle, each monomer could be furnished with two side-chains instead of just one, although the work so far has not made use of this feature.
After developing the individual chemical reactions needed for routine assembly of oligocarbamates, the researchers generated a library of these polymers using Affymax’s “spatially addressable” solid phase technique (see this column in Update 12), resulting in a 1-molecule thick layer of 256 different oligocarbamates on a glass plate, arranged as a 16 x 16 array of 800 µm-square patches. Treating the plate with a solution of fluorescent antibody provided a way to measure the affinity of each oligomer for this antibody. The highest affinity oligomers were then synthesized in larger quantities on an automated peptide synthesizer for further testing. Oligocarbamates turned out to be quite resistant to degradation by proteases, and significantly less water soluble than corresponding peptides. [Charles Y. Cho, et al., Science 261:1303-1305, 3Sep93]
In recent years chemists have come up with a variety of different analogs to proteins and nucleic acids, each characterized by a different molecular backbone. From the standpoint of nanotechnology they have shared the drawbacks of their biological counterparts: mushiness, and (in the case of nucleic acid analogs) poor functionality. Polycarbamates appear to avoid both of these problems and therefore hold some promise as materials for nanotechnology. Rigidity can be increased by putting two side-chains on each monomer — since they would be attached to adjacent backbone carbons, they could be engineered to restrict each others’ random motions, producing a stiffer, better-defined molecule.
The ability to predict the properties of materials from theoretical principles would greatly facilitate the development of useful new materials, whether they are to be produced by traditional methods or by nanotechnology. Any chosen property of any substance can in principle be calculated from quantum theory, but today’s computers are seldom adequate for the task, particularly if the material in question is a solid. For certain properties, however — hardness, for example — empirical formulas can give excellent results over a wide range of materials.
The hardness of carbon-nitrogen compounds has been the focus of numerous studies, both theoretical and experimental. One particular compound, b-C3N4, was predicted to be at least as hard as diamond. The desire to test this prediction has led researchers to attempt by various methods to synthesize the substance, but to no avail — until now. A group at Harvard University announced this summer that they have prepared thin films of a material that appears to be b-C3N4. The films were made by vaporizing graphite with a laser and introducing atomic nitrogen into the carbon plume where it impinged upon a target surface. Spectroscopic and electron diffraction analyses of the product strongly suggest that a portion of it consists of b-C3N4. The film was quite hard: dragging a metal needle over it did no damage. Further twiddling of synthetic conditions should lead to a purer product that can be tested to see how well its properties were quantitatively predicted by current theoretical methods. [Chunming Niu, et al., Science 261: 334-337, 16July93]
In Update 16 I mentioned a new family of carbon-boron compounds called “carborods” that are reported to have exceptional stiffness and stability. These properties and the hardness of b-C3N4 derive from simple characteristics of the chemical bonds between boron, carbon, and nitrogen atoms. Hardness varies inversely with bond length; this makes sense intuitively if you think of bonds as little springs. Bond length, in turn, varies directly with the radii of the atoms forming the bond. Therefore one would expect to find the hardest materials made up of elements having small atomic radii — namely atoms near the top of the periodic table. Hardness varies directly with the average number of bonds made by atoms in the material; this, too, makes sense if you think of bonds as struts that hold the atoms in place. Therefore one would expect to find the hardest materials made up of elements in the center of the periodic table, not at the right or left. A glance at the periodic table confirms it all: the constituent atoms of carborods and b-C3N4 — boron, carbon and nitrogen — are found front row center.
Of the two parameters — bond length and bond number — bond length is the more important. Because the C–N bond is shorter than the C–C bond, we should anticipate that in a contest of hardness diamond will lose to b-C3N4. For construction of nanodevices, where stiffness enhances the accuracy of positioning objects, carbon-nitrogen bonds may be used extensively. [For a good brief discussion of theoretical approaches to hardness, see Marvin L. Cohen’s article in Science 261:307-308, 16July93. For a more technical treatment and invaluable tables and graphs, see chapter 3 of K. Eric Drexler’s book Nanosystems.]
This schematic illustration shows a series of hydrogen deposition tools (above) placing hydrogen atoms on a series of cylindrical work pieces (below). The two belts move from left to right, carried and brought together by two rollers (atomic detail not shown). Hydrogen atoms are shown in white, except for those undergoing transfer, highlighted with stripes. The deposition tool is a diamondoid cage structure ending in a hydrogen-tipped silicon atom. The cylindrical workpieces initially have four carbon radical sites, one of which receives the added hydrogen. Because carbon binds hydrogen more tightly than does silicon, the hydrogen atoms transfer reliably. ©1993 K. Eric Drexler. All rights reserved.
A landmark event in biology occurred this past summer when researchers at the University of Bath published a map, at atomic resolution, of the molecular motor responsible for muscle action. [Ivan Rayment, et al., Science 261:50-57; and Science 261:58-65, 2July93]. This motor, called “myosin subfragment-1” or “S1”, is the portion of the myosin molecule that converts chemical energy into force and motion.
S1 contains more than 50,000 atoms and is shaped like a whale about 5 nm wide at the head and 19 nm long. The head contains two active sites: a nucleotide binding pocket where ATP molecules are trapped and tapped for energy to run the motor; and an actin-binding site where the myosin head interacts with a protein filament called actin. The myosin tail is permanently connected to a myosin filament. Myosin filaments are densely studded with S1. In intact muscle, arrays of myosin and actin filaments interdigitate with each other. Where actin and myosin filaments overlap, they are bridged by the S1 motors.
During muscle contraction the motors travel along actin filaments pulling their myosin filament along in 5 nm steps — like a line of whales joined to a thick rope by their tails, grabbing and releasing another rope with their mouths, flexing and unflexing their bodies to reposition themselves. Decades of microscopy and biochemical studies had already provided a picture of muscle mechanics at this level of detail. The amino acid sequences of the proteins have long been known, and the actin molecule was mapped in detail several years ago. A mechanical understanding of the system, however, has waited for a map of the myosin motor.
Mapping S1 was not a straightforward task. For 30 years the molecule has resisted efforts to crystallize it for X-ray diffraction mapping. The Bath researchers circumvented this problem by chemically modifying some of the amino acids on the molecule’s surface. Other technical problems were overcome only after prodigious chemical and computational efforts.
The 3-D map of the myosin motor, although it is a view at only one moment in the contractile cycle, gives a rough indication of how the motor works:
An energy-bearing ATP molecule is captured at the nucleotide binding pocket when the pocket is open — this happens to be when the motor is tightly bound to its actin filament. The ATP capture causes a movement in molecular domains beneath the site, opening a cleft at the front end of the head — like the opening of a whale’s mouth — and causing S1 to release its hold on the actin filament. (It may remain weakly attached by a flexible peptide loop.) The pocket now closes upon the nucleotide, and a bend develops in the myosin tail. Crunch! — the terminal phosphate is severed from the ATP. The products of this hydrolysis reaction, ADP and phosphate, remain stuck in the reaction pocket for now. The myosin head again approaches the actin filament and weakly binds with it, altering intramolecular stresses. Domains shift, closing the cleft at the actin-binding site and partially opening the nucleotide binding pocket; the severed phosphate is released. Now the myosin head can bind more tightly to actin. When it does, the tight binding state triggers the power stroke in which the bent myosin tail is allowed to straighten, converting its stored energy into force and motion. Finally, in the straightened configuration, the binding pocket fully opens, releasing ADP.
Five years from now, this rough and still rather speculative verbal description of the actin-myosin contractile cycle will probably have become a hundred times more detailed, and correspondingly hard to follow. But by then I hope someone will have created an animated stereoscopic computer model that will fully reveal the workings of this motor in terms we can understand intuitively.
I can think of no more effective way to allay the fears many people have about bioscience and nanotechnology than to demystify the workings of molecular motors. If the subject of automobile engines were as arcane and impenetrable for the average person as molecular biology is today, we would have jeremy rifkins trying to outlaw the use of cars. The mechanical aspects of molecular systems are the keys to public understanding and enthusiasm, and we are finally coming close to having those keys in hand.
Dr. Russell Mills is research director at a company in California.
The Third Foresight Research Conference on Molecular Nanotechnology: Computer-Aided Design of Molecular Systems attracted about two hundred participants to Palo Alto on October 14-16, 1993. The other academic/nonprofit sponsors, besides Foresight Institute, were the Stanford University Department of Materials Science & Engineering, the Molecular Graphics Society, and the Institute for Molecular Manufacturing. (See list of conference speakers.)
Besides the listed speakers, attendees had the opportunity to look at and discuss posters; attend demonstrations and exhibits; and hear panel discussions on “Can Today’s Computational Chemistry Model Tomorrow’s Molecular Machines?”, “Molecular Nanotechnology: Status and Next Steps”, and “General Purpose Molecular Manufacturing: How Long?”
Conference proceedings are being published as a special issue of the journal Nanotechnology, published by the Institute of Physics. Watch this publication for details on ordering the issue.
A day-long pre-conference tutorial was held October 13 on the subject “Introduction to Modeling Molecular Systems for the Computer Scientist”. The thirty participants attended the following sessions:
K. Eric Drexler Computational Methods for Molecular Manufacturing
Ralph C. Merkle Design Issues in Self Replicating Systems
William A. Goddard III An Introduction to Computational Chemistry
Biosym Methods & Applications in Density Functional Theory
Peter Kollman An Introduction to Molecular Mechanics
Ralph C. Merkle Computational Nanotechnology
The conference received corporate sponsorship support from Apple Computer, Beckman Instruments, Xerox PARC, ARCO, Biosym Technologies, Digital Instruments, Fenwick & West, JEOL, Nanoscale Progress, and Niehaus Ryan Haller Public Relations.
Introduction to the Design of Molecular Systems
Eric Drexler, Institute for Molecular Manufacturing
Ralph Merkle, Xerox Palo Alto Research Center
Modeling the Design of Proteins
Ken Dill, University of California at San Francisco
William Goddard III, California Institute of Technology
Crystal-Based Molecular CAD
Geoff Leach,Royal Melbourne Institute of Technology
Packing Molecular Building Blocks
Markus Krummenacker, Institute for Molecular Manufacturing
Visualization with Molecular Graphics
Michael Pique, Scripps Research Institute
Mechanical Engineering CAD
Joel Orr, Orr Associates, Inc.
Molecular Building Blocks
Ted Kaehler, Apple Computer, Inc.
Modeling Mechanochemical Processes
Charles Musgrave, California Institute of Technology
Modeling Diamond CVD with Density Functional Theory
Warren Pickett, Naval Research Laboratory
Atom Manipulation by Proximal Probes: Experiment and Theory
Makoto Sawamura, Aono Atomcraft Project, Japan
Scientific Visualization for Scanning Tunneling Microscopy (STM)
Russell Taylor, Univ. of North Carolina at Chapel Hill
Ab Initio Methods and Software
Charles Bauschlicher, NASA Ames
Nanocomputers and Reversible Computation
J. Storrs Hall, Rutgers University
Computational Chemistry, Parallel Supercomputers & Nanotechnology:
Current Capabilities and Future Progress
Ian Foster and Rick Stevens, Argonne National Laboratory
INVENTON: An Automated Molecular Invention System
Applied to Problems in Nanotechnology
Michael Pitman, UC Santa Cruz
Prof. Ken Dill of the University of California at San Francisco explains the state of the art in designing and modeling proteins, the chief building blocks of natural molecular machinery.
Geof Leach and Ralph Merkle present the results of their collaboration on Crystal Clear, the first nanotechnology CAD program.
Markus Krummenacker, Makoto Sawamura, Martin Edelstein, Eric Drexler, and Ted Kaehler debate development strategies on Friday’s panel “Molecular Nanotechnology: Status and Next Steps.”
Russ Taylor of North Carolina is mobbed with questions after his talk on virtual reality for interacting with the nanoscale world.
Jim Lewis, Tom McKendree, Josh Hall, and Neil Jacobstein discuss the critical question: “General Purpose Molecular Manufacturing: How Long?”
Makoto Sawamura describes progress toward atom manipulation in Japan’s Aono Atomcraft Project.
Rick Stevens of Argonne National Lab sketches his and colleague Ian Foster’s work on the intersection between computational chemistry, parallel supercomputing, and nanotechnology.
J. Storrs Hall of Rutgers University’s Laboratory for Computer Science Research describes energy-efficient computation based on nanotechnology. In addition to his research, Hall moderates the Internet nanotechnology discussion group sci.nanotech.