A publication of the Foresight Institute
Concepts in Protein Engineering and Design, ed. Paul
Wrede and Gisbert Schneider, Walter de Gruyter & Co., 1994,
378 pgs, some color illustrations.
This volume provides an excellent survey of the state of the art in protein engineering, covering topics ranging from protein analysis to the use of neural network techniques in protein sequence design. Taken as a whole, it offers a good picture both of the achievements to date and of the challenges that remain. As a bonus, the last chapter (by Nadrian Seeman, a speaker at the second Foresight Conference) describes achievements in the engineering of three-dimensional structures from nucleic acids.
Readers with a background in chemistry or molecular biology will find this book a valuable introduction to biomolecular engineering, which may prove to be a key step on the path to advanced nanotechnologies. Its focus on achievable steps that are on today's research agenda will suggest practical career moves for those with their eyes on the road ahead.
Prospects in Nanotechnology: Toward Molecular Manufacturing,
ed. Markus Krummenacker and James Lewis, Wiley, 1995, 297 pages,
hardcover. Proceedings of the First Foresight General Conference;
see announcement in this issue.
Experiment Zukunft: Die Nanotechnologische Revolution, by K. Eric Drexler, Chris Peterson, with Gayle Pergamit, Addison Wesley, 1995, 320 pages, hardcover. German edition of Unbounding the Future.
Technotrends by Daniel Burrus, HarperBusiness, 1994, paperback. Gives "24 technologies that will revolutionize our lives," but does not get nanotechnology quite right. Confuses the current ability to design and build molecules (e.g. new enzymes) and the coming ability to design and build molecularly-precise materials. Confuses micromechanics (mechanics) with quantum structures (electronics). Author is "one of the world's leading technology forecasters," and his basic point -- the importance of emerging technologies to business -- is excellent. Let's educate him further.
Foresight is pleased
to announce the publication of Prospects in Nanotechnology:
Toward Molecular Manufacturing (John Wiley & Sons, Inc.,
1995). When we held our first conference for non-researchers we
weren't planning a proceedings book, but thanks to long hours of
hard work by coeditors (and Senior Associates) Markus
Krummenacker and James Lewis, many of the papers presented are
now available in one volume. Because Wiley is a technical
publisher, the papers selected are the more technical ones
presented, but all are accessible to the general reader.
From the back cover of the book:
A fascinating journey through the microscopic world of nanotechnology and its macroscopic implications
Complex mechanical devices with feature sizes on the molecular scale and with all the power of today's supercomputers; diamond and other ultrastrong building materials-and all of this accessible through low-cost automated molecular manufacturing. We've all read about the vast potential of nanotechnology. Some theorists have hailed it as the most important technological breakthrough since steam power. But how far have we really come to realizing any of that potential? And how have recent developments in the field already begun to shape the world of the twenty-first century?
Now this discussion-oriented book takes you to the front lines of nanotechnology theory and practice to provide answers to these and other questions. Featuring contributions from a number of international top names in the field, it offers a provocative look into a future shaped by nanotechnological applications and also provides an overview of the enabling technologies that are in current use in a variety of industries, and which will be important in attaining this breakthrough technology.
a former researcher at the Institute for Molecular Manufacturing,
now working with Nanothinc, a San Francisco-based company.
James Lewis, Ph.D., is a scientist at Bristol-Myers Squibb Pharmaceutical Research Institute in Seattle. Dr. Lewis is author of more than forty scientific research publications in biochemistry, virology, and molecular biology.
Foreword by K. Eric Drexler
Most Foresight conferences have been highly technical, but
those of you who attended the 1992 General Conference will recall
that it was specifically designed for those not engaged in
research, covering the topic of nanotechnology for the
There was some confusion at the meeting: the press-expecting a research conference-did not know what to make of it, and the background level of knowledge of the participants varied widely. Still, many Foresight members enjoyed the event tremendously and asked for more.
The annual Senior Associates Gathering, begun formally in 1994, provides the value of the former General Conference with the following improvements: (1) we do not look for press coverage; in fact, the meeting is regarded as off-the-record, to encourage freer discussion, and (2) the level of knowledge and commitment of participants is kept high by drawing from the Senior Associates groups of Foresight, IMM, and CCIT. This also keeps the size of the group down to a level that enables a more intense, interactive meeting. Participants are expected to have read Engines of Creation or Unbounding the Future.
The 1995 Gathering will take place on November 11-12 in the Bay Area. Those wishing to participate who are not already Senior Associates can request a Senior Associate information package from the Foresight office. Senior Associates make a five-year pledge of $250, $500, or $1000 annually.
For a Senior Associate information package, contact Foresight at tel 415-917-1122; fax 415-917-1123; or email email@example.com.
[Editor's Note: This page has been optimized for Netscape 2 and later. If you are using a browser, such as Netscape 1.1, that does not support the html tag for superscripts, please be aware that an number like "2x109" is meant to be scientific notation for "2 times ten raised to the 9th power," and that "e2" means "e squared," etc.]
Robertson, Dunlap, Brenner, Mintmire, and White at NRL
described simulations of atomically perfect fullerene gears. They
designed 5-8 sprocket gears with 290-464 atoms. A six-tooth gear
was extended with a fullerene shaft to simulate the shaft that
would conduct mechanical power in a complete machine. Simulations
of power transmission from one gear to another were done with a reactive
potential, so the simulation verified that the bonds in the
structure would withstand the forces applied. During one
simulation, one gear spun up the other from a standing start to
20 gigarev/sec in a quarter of a turn. In addition, the overall
binding energies of the gears were calculated. They were
uniformly found to be more stable (per atom) than C60.
[Novel Forms of Carbon II 283-288]
In experimental news on fullerene systems, Lüthi et al. measured unusually low friction (a shear strength of 0.05 to 0.1 megapascal) between C60 islands and an NaCl substrate using a scanning force microscope. They note that "These results could find use in the field of nanotechnology; for example, C60 islands could be developed into a sled-type transport system on the nanometer scale." [Science 266: 1979-1981, 23Dec94]
In more general news on friction, Singer wrote a review article on studies of friction at the level of molecular mechanics simulations. Oddly, there is almost no overlap with Drexler's analysis in Nanosystems. The investigations described are all of much more dissipative systems than would be desirable for mature nanotechnology. The dominant dissipation mechanism in evidence is irreversible merging of potential wells (ideally confined to erasure of bits, but occurring here whenever methyl groups pass each other on a surface). At the level of accuracy relevant in this paper, "Low friction, including zero friction, can be achieved at low loads, with weak surface interactions and with 'small' atoms at the interface." The extensions suggested for current work are towards pushing the length and time scales upwards to better predict current macroscopic systems, rather than to optimize microscopic systems. The analysis is interesting for contemporary lubrication systems, and may be important for analysis of early protein machines, but hopefully these mechanisms will be designed out of mature nanomachines. [J.Vac.Sci.Tech.A 12(5) 2605-2616, Sep/Oct 1994]
Friction is important once one has a means to move things at
all. One approach to supplying energy for movement is to use
chemical energy sources. An analysis paper by Magnasco provides a
general framework for calculating the properties of
"Molecular Combustion Motors." Magnasco describes
molecular motors with a potential energy surface on two
coordinates: the reaction coordinate for fuel burning, and the
mechanical movement coordinate. Given this potential surface,
this paper describes how to calculate fuel consumption rate,
speed of movement as a function of load, and so on. There is an
interesting analogy made for the losses due to thermally
activated slipping on the potential energy landscape. "It is
evident that thermal noise will allow the gears to slide over
each other, not unlike actual engine transmission coupling
through transmission fluid before the teeth actually lock."
[Phys. Rev. Let. 72: 2656-2659, 18Apr9 4]
In more concrete work on a specific motor system, a paper by Spudich describes work on the myosin motor in biological systems. He describes the current understanding of the chemical kinetics of this molecule, how some structural features of the molecule affect reaction rate, and where in the reaction cycle the rate limiting effects appear. For nanotech applications the coupling between chemical energy and mechanical energy will continue to be useful, but some more precise mechanism to single-step the motor will be necessary in order to use it for atomically precise operations. Thermally activated ADP desorbtion won't be sufficiently precise for that application. He also describes a laser-trap system which allowed observation of single steps of 10-20nm, and forces of 7 pN.
Spudich also points out that "laser-trap technology allows many types of measurements that could be applied to virtually any protein. For example, the elasticity of a protein molecule can now be measured directly, and conformational changes associated with transitions between states are within reach of being explored." This capability sounds promising; however, there may be considerable duplication between this technique and AFM, which probes similar ranges of parameters. [Nature 372: 515-518, 8Dec94]
Proteins offer us a great deal of synthetic flexibility today.
We have well-tested methods for building custom proteins, both
biologically based and chemically based. These methods, however,
let us select the primary structure of a protein: the sequence of
amino acid residues within it. In order to use a protein as part
of a structure, or to place groups within it in some pattern in
space in order to control a reaction or bind to some substrate,
we must control the secondary and tertiary structure of the
protein: how it folds and assembles itself in space. We must go
"from structure to sequence."
Steven Brenner and Alan Berry have written a program to help systematically select amino acid sequences designed to fold in a prespecified way. Their program relies primarily on the statistics of the structures of known natural proteins. It takes a desired structure as an input, which specifies which parts of the amino acid sequence are in what types of secondary structures (alpha-helices, beta sheets, and so on) and how much each residue is exposed to the solvent. The program uses simulated annealing to pick sequences of amino acids which match the conformational preferences, the neighbor preferences, and the solvent accessibilities of the protein database. Another bias-term in the process selects for "diverse" sequences, those where the overall frequencies of residues in the sequence match the frequencies in the database. The justification for this nonstructural term is that it biases the designed sequences against designs that would fold in too many ways. One needs to avoid designing proteins that fold in the way that one wants...but fold even more stably in some other way. The sequences that were designed were tested by computer analysis in an independent program, and their predicted structures compared with the intended structure. [Protein Science 3: 1871-1882, 1994]
There have been a number of experimental protein design papers recently:
A group at Lausanne, Switzerland, assembled and demonstrated the function of an ion channel built from 4 copies of a natural peptide covalently linked to a 10 residue "template" peptide. In this design the peptide bonds to NH2 side chains of lysine residues in the template provide the structure that is typically provided by protein folding in ordinary proteins. The group was able to measure the conductances from single channels. One somewhat disturbing note is that "Three different channel states (called 01, 02 and 03) can be clearly distinguished." The channel states seem to imply that several active states are thermodynamically accessible, which would be unfortunate if we wish to exploit stable, unique structures for protein designs. [Protein Science 3: 1788-1805, 1994]
Munson et al. redesigned the 4-helix-bundle protein Rop. They repacked the hydrophobic core, achieving better thermal stability than in the original protein. The redesign was done by examining packing effects in space-filling molecular models. The protein consists of two copies of a 63-residue chain. Several modified versions of this protein were designed, with 7 and 9 residues changed in two versions. The revised protein was produced biologically. The 50% denaturing temperature was raised from 74.6C to 87.2C and 95.4C in two variants. Energy minimization of the modified proteins gave backbones which were "superimposable on the wild-type backbone." This showed that the energetics of the hydrophobic core were well explained by space filling considerations. [Protein Science 3: 2015-2022, 1994]
Fezoui, Weaver, and Osterhout described how they designed a 38-residue peptide in 1990. Their peptide was designed to have a simple, predictable tertiary structure. It consists of two helices covalently linked by a hairpin turn. The two helices were designed to have mutually stabilizing hydrophobic regions where they touched. A noteworthy constraint on the design was avoiding undesired aggregation between copies of the peptides. This set an upper bound on how large the hydrophobic region of the peptides could be. This is somewhat discouraging for designing multi-peptide structures, since the desirable structures for bonding the peptides together include hydrophobic regions, which will tend to destabilize the individual peptides and tend to make them form undesired aggregates before they are mixed to form the desired ones. After the design of the hydrophobic regions in this design, charged residues were selected to form salt bridges between the two helices to further stabilize the structure. In the last step of the design, 12 remaining residues were selected, including a fluorescent donor and acceptor. These last residues were constrained to be good helix formers. Three residues met this contraint, so roughly 312 possibilities remain for this design after all of the constraints are met.
The authors also describe the degree of structural success that they had. "CD experiments indicate that the peptide is approximately 60% helical at room temperature (89% of the residues are in the helical regions of the peptide), suggesting that the helices are slightly frayed and/or that the peptide is in equilibrium with unfolded, nonhelical conformations." The authors goals for further work in the field are consistent with the requirements of nanotechnology in that they are looking for more rigid structures. "...a major focus of the protein design field must be upon how to achieve protein designs with more rigid protein-like interiors." This is a requirement even for improving the quality of feedback about designs: "It is hoped that relatively simple notions of space filling can be used to design molecules that will exhibit enough structure to characterize by high-resolution techniques (NMR and X-ray diffraction)." [Protein Science 4: 286-295, 1995]
In supramolecular chemistry, Jin and Wells have shown that attractiveness is only about one residue deep, yet it is hard to graft. They studied the affinities of a series of antibodies to a series of mutated antigens and found that only a small number of residues on the antigens account for the bulk of the binding energy. They found that "...on average only 3-5 side chains could account for more than 80% of the binding..." There are x-ray studies to contrast these with, and they show far more residues physically in contact, "14-21 residues on each side." Jin and Wells were able to mutate as many as 16 neighboring residues to alanine without lowering the affinity by more than a factor of 10, while changing a single one of the 5 primary residues "caused a 6- to >500-fold reduction in affinity." While this appears to indicate that intermolecular affinity should be comparatively easy to produce, grafting the critical residues on to another antigen only worked when it was homologous to the original antigen. [Protein Science 3: 2351-2357, 1994]
In classical organic chemistry, Nicolaou and co-workers have synthesized Brevetoxin B. The molecule consists of 11 fused rings (all of which are cyclic ethers) with 23 stereocenters. The synthesis required 33 man-years by 25 graduate students and postdoctoral fellows spread across 12 calendar years. There are 83 steps in the synthesis, with an average yield for each step of 91%, and an overall yield of 0.043%. "The synthetic strategy was a convergent one, in which large parts of the molecule were preassembled and then combined." [C&EN 32- 33, 30Jan95]
In another synthetic note, Lagow et al. synthesized carbyne rods "with chain lengths in excess of 300 carbon atoms." The longest chains were produced by alternately condensing CF3 radicals from a C2F6 RF discharge and laser vaporized graphite on the walls of a glass reactor. The rods are not particularly reactive. They were dissolved in THF and toluene during the course of analysis. Related model compounds could be crystallized and heated to 130C before polymerizing. Unfortunately, attempts to separate some related reaction mixtures on the types of columns (Al2O3) used to separate fullerenes did not separate them, but rather they reacted with the column. I wish them better luck in future separation efforts. If we can separate 235U from 238U, surely there must be some way to separate F3C(CC)150CF3 from F3C(CC)151CF3. Nonetheless, it is encouraging to see that the rods used in the original rod logic proposal are in fact quite stable at room temperature, even without any solid matrix to prevent collisions. Useful mechanical components may be available right down to the limit of chemically plausible structures. [Science 267: 362-367, 20Jan95]
Jeffrey Soreff is a researcher at IBM with an interest in nanotechnology.
From Foresight Update 21, originally published 1 June 1995.