Nanotechnology is currently faced with significant problems for general methods of construction of molecular devices, atom-by-atom. Solutions to this problem have however already been reduced to practice by nature: the natural proteins are molecular scale machines of various highly sophisticated functionality. They teach us that molecular machinery might be constructed not by placing an atom-at-a-time into a three dimensional assembly, but by "folding up" essentially linear polymers. The assembly method is that these polymers contain instructions along their length as to how chemical features, judiciously placed along the polymer, will be brought together in space. Moreover, this "bringing together in space" is a spontaneous folding process. In the protein world, therefore, machines are constructed according to a "string theory": these compact functional devices are on finer inspection seen as "balls of string", the string being wrapped up in various specific ways which determine the various corresponding specific mechanisms and functions.
In a series of TIBS and Nature News & Views articles in the early 1970s (see for example refs 1-2) I described the implications of understanding protein folding for the design not only of novel pharmaceuticals, but explicitly on non-biological, molecular scale machinery as novel catalysts, biosensors, and computer chips. In some cases such as catalysts (3), we judged that relatively tractable design routes from first principles lay not too far ahead, at least as relatively non-specific catalysis is concerned. In contrast, we have recently emphasized that ab initio design is difficult for more sophisticated, specific machinery, and that genomic information, representing some 4 billion years of design through natural selection, is required (4). These observations are further developed here.
In particular, studies with a number of teams (e.g. Ref. 5) of the conformational problems encountered when building peptide chains, plus experience in designing synthetic systems and testing stereochemistry by immunological means in a series of some eight inventions (e.g. Ref. 6), led us to thinking of improved synthetic and assembly methods. Thanks in particular to technology developed by Dr. Steven Kent at the Scripps Research institute, I was able to assist in the construction of large protein and protein-like systems. Since these were not based on ribosomal synthesis but new chemical methods, we were not confined to biological chemistries. That is, such methods "break free" of the constraints of the ribosome, yet, starting as they do from a protein concept, they can still borrow information from nature.
For example, functional superoxide dismutase, a protein of 153 D-amino acid residues, was made in mirror image form using D-amino acids (7). Proteins made of D-amino acids fold and function in a mirror image universe to normal proteins: except in cases such as superxoide dismutase where the substrate is achiral, they act on material of opposite chirality compared with those acted upon by the normal L-protein. A polypetide chain made wholly or largely of D-amino acid residues is essentially a non-biological and less biodegradable plastic (8). Despite the traditional interests of my laboratories in conformational design, some choices of polymeric chemical systems have special features which bypass design and enable direct experimental methods of encoding new functions into the molecules. For example, I and others such as Peter Kim of the Whitehead Institute have conceived powerful different experimental methods of programming recognition and potentially very sophisticated function into D-protein or "Doppelganger" protein systems (8). Such Doppleganger proteins have medical and nanotechnological applications (8).
Such special cases are but example glimpses of what is possible. Indeed, folding of a polymeric chain in a specific information-rich way is but one programmable self-assembly method. Further self-assembly methods also potentially yield complex macromolecular assemblies but are initially "chainless". Non-chain methods involve chemical assembly of a population of differing heteropolyvalent monomeric "residue" fragments in a specific manner precoded in their heteropolyvalent "latch" points, analogous to the assembly of the (e.g. p22) virus coat from protein building blocks.
Nonetheless, analysis of the underlying mathematics of spontaneous assembly (which occur in unexpectedly short-timescales in the real world) shows that such mathematics has a great deal in common with the mathematics underlying the protein folding problem itself, and that both chain and non-chain assembly systems depend on developing the same set of computational tools. Indeed, many embellishments on the implementation of the non-chain method hint at this by effectively reducing back to a chain approach (a protein chain is, in effect, a partially pre-linked assembly of tetravelent units employing both covalent and non-covalent interactions). Be that as it may, chemical and computer experiments help shed light on a somewhat larger, more general class of "code of codes". These codes relate programmable one dimensional information in chain, or equivalently in non-chain, form, to the three dimensional structure of complex, functional matter. This will be discussed.
2. "Designing Biologically Active Peptides", B. Robson (1980), Trends, Biochem. Sci. 2, 240-244.
3. "Artificial Enzymes", B. Robson and A. Marsden (1987) Biochem. Soc. Trans. 15, 119.
4. "Fast Track Routes from Genome to Peptidomimetic" B. Robson (1998) In "Peptidomimetics & Small Molecule Design", Eds. W. Hori, E. M. Nagle, and L. M. Savage, IBC Library Series.
5. "The Conformation of Peptides During Solid Phase Synthesis", C. Baris, A. Brass, B. Robson and G. Tomalin. (1990). Innovation Perspect. Solid Phase Synth. Collect. Paper. 1st International Symposium SPCC, Birmingham, UK. Ed. R Epton. 89, 441-445; "N-Alkoxy amid backbone protection in BOC chemistry : improved synthesis of a 'difficult sequence'", L. E. Canne, G. M. Figliozzi, B. Robson, M. A. Siani and R. J. Simon (1996), Protein Science, 5, Suppl.1, 72 [published poster 82-M] "Total Chemical Synthesis of Chemokines and their Analogues", M. A. Siani, L. E. Canne, G. M. Figliozzi, B. Robson, D. A. Thompson and R. J. Simon , to be submitted, synopsis in "Chemokines", International Business Communications (1996
6. PATENT: "Synthetic Peptides", R. V. Fishleigh, B. Robson. P.C.T. International Patent Application No. PCT/GB91/00392 International Publication No: WO91/13909, dated 19th September 1991
7. "Chemical synthesis and activity of D-superoxide dismutase", G. M. Figliozzi, M. A. Siani, L. E. Canne, B. Robson, and R. J. Simon (1996) Protein Science, 5, Suppl.1, 72, [published poster 87-S]
8. "Doppelganger Proteins as Drug Leads", B. Robson (1996), Nature Biotechnology, 14, 892-893
Barry Robson Ph.D., D.Sc.
Strategic Advisor, Distinguished Engineer, Computational Biology Center
IBM T.J. Watson Research Center
30 Saw Mill River Road, Hawthorne, NY 10523