Editor's note: Dr. Lewis, a Senior Scientist at the genetic engineering
firm Oncogen in Seattle, has prepared a survey of the protein engineering
and related fields from a nanotechnology perspective. Excerpts follow:
A protein engineering meeting entitled "Protein domains: molecular
insights into structure/function relationships" was held at the Waksman
Institute at Rutgers on Nov. 2-3, 1987. It was cosponsored by the newly
created Center for Advanced Biotechnology and Medicine. The twelve talks
brought together topics ranging from pure structure determination to molecular
genetics.
A recently developed alternative approach to protein structure determination
was introduced by Prof.
Kurt Wuthrich of Switzerland. He explained how new techniques in nuclear
magnetic resonance spectroscopy enable the determination of three-dimensional
protein structures in solution so that it is not necessary to obtain
a crystal to know the structure. In addition, you can directly compare how
relevant the crystal structure is to the structure in solution. The results
he presented were impressive. The limitations of the technique at the moment
are (1) you can only do medium-size proteins, up to 25,000 daltons (about
230 amino acid units), (2) you need a 500 MHz machine, which costs over
$500,000, and (3) you need several days of CPU time on a Cray supercomputer.
The significance to nanotechnology of [this approach] is that our knowledge
of the rules governing the relationship between protein sequence and protein
structure, and between structure and function, is limited by the number
of structures that we know. At the moment, this number is less than 200,
but [new] techniques promise a rapid increase in this number.
[The current list of known protein structures is available in the Brookhaven
Protein Data Bank.]
A novel, and controversial, presentation on protein evolution by Prof.
Russell Doolittle of UCSD is directly relevant to the question of how
many basic protein structures have been invented by nature. If this number
is very large, then we have a long way to go to understand the structure/function
motifs that nature has invented, and to figure out what novel motifs nature
missed that we might use. Doolittle's work is not concerned directly with
three-dimensional structure, but instead with the comparison of protein
sequence information (about 30 times as many sequences as 3-D structures
are known). He has shown that a surprisingly large number of these sequences
are evolutionarily related, implying that most proteins are derived from
a small number of basic themes. Recognizing that our current databases contain
proteins chosen because someone wanted to work on them, rather than because
an attempt was made to be representative, Doolittle estimates that there
are fewer than a hundred fundamentally different protein structures in nature,
including the ones that have not yet been described. If he is right, a solid
understanding of sequence/structure/function relationships may not be too
far off. The question then becomes: is this small number all that you can
do with proteins; that is, did nature only make this many because the zillions
of other possibilities are redundant or useless--or did nature just get
lazy when she came up with a small number that worked so that these became
"locked in" by evolution, while other good possibilities were
ignored? Correspondingly, can we design additional structures that will
have novel and useful properties, rather than being more minor variations
on a well-studied theme?
Another meeting, "Protein Engineering '87", held at Oxford, was
reviewed by M. Gait, J. Thornton, and R. Wetzel in Protein Engineering
1:267-270. The major topics appear to have been the usual attempts
to use site-directed mutagenesis to improve commercially important enzymes.
These experiments may be a bit dull compared to the ultimate use of protein
engineering to develop nanotechnology, but we should remember that they
are important for two reasons: (1) they help develop the knowledge of structure/function
relationships that we need to design protein assemblers, and (2) they have
immediate commercial applications, which will provide the near-term market
force to drive nanotechnology development. Other topics included new algorithms
for predicting structures from sequences. This effort is a bit academic
from the nanotechnology perspective; however, there was also discussion
of a new algorithm to do the reverse: predict a sequence that should
produce a desired structure. Now this is something to follow! It is expected
to be easier to do than predicting structure from sequence, and will be
the heart of designing assemblers.
Although the paper "The design of a biochip: a self-assembling molecular-scale
memory device," by B.H. Robinson and N.C.
Seeman (Protein Engineering 1:295-300) was published
in a journal about protein engineering, it actually has nothing to do with
proteins. It describes the design of a computer to be constructed using
DNA, organic polymers, and metal ions. If the construction of such a computer
were actually realized, it could utilize a readable bit of 3x104
nm3 , and would operate at electronic speeds over short distances.
A novel feature of the design is the use of DNA for its structural--rather
than informational--content. Oligonucleotides would be synthesized with
sequences designed to give a particular pattern of complementarity so that
they could be ligated together and would self-assemble into "nucleic
acid junctions," a stiff scaffolding of DNA to be used to hold the
working parts of the circuits into place. Molecular architecture and computer
graphics programs to design these scaffolds are apparently available, and
some experimental work has been done. The computer would use conducting
organic polymers, such as trans-polyacetylene (tPA), in the preliminary
design. The use of bundles of ten to twenty tPA "wires" is contemplated
to ensure that the electronic properties are characteristic of bulk material.
The problem of how to attach tPA to DNA has not been solved, but plausible
solutions are presented. Metal ions chelated between adjacent tPA ends would
serve as redox bits that could be manipulated by a connected voltage source
through use of a suitable dopant in the polymer. A major feature is that
the conductance would be electronic rather than through phonon-limited nuclear
motion, so that speeds would be fast. The article explores several problems
involved, along with suggested solutions, and an estimate of a decade for
development. Although molecular in scale, the individual parts would be
considerably larger than those of Drexler's suggested mechanical nanocomputer--but
they might be faster, with estimated access times of picoseconds. Even with
a relatively large bit size (by molecular standards), a cubic cm array would
hold 3.3x1016 bits, or several hundred times as much as would
be needed to store each of the estimated ten million books in existence.
Current thinking (Update 1, p. 5) recognizes three technological
pathways to nanotechnology: (1) protein engineering, (2) atomic manipulation
through use of the scanning tunneling microscope (STM) and related technologies,
and (3) other approaches to chemical synthesis... The third path was briefly
considered in our previous installment by way of noting a short article
in Science News that reported novel "Cages, Cavities,
and Clefts" produced by organic chemists that are capable of recognizing
atoms and small molecules. Although these developments are currently seen
in terms of bulk-scale technology, they embody the principles of designing
specific molecular recognition, until now exclusive to the realm of biology.
These capabilities will surely become useful in designing molecular-scale
machines for manipulating individual atoms. This area of research has been
recognized by the 1987 Nobel Prize in Chemistry, showing once again that
science at large is realizing the enormous potential of recognition and
manipulation at the atomic scale. A short summary of the award-winning research,
entitled "Chemistry in the image of biology" appeared in Science
(238:611-612). It briefly describes how the work of the three Nobel
laureates mimics enzyme systems in function by the creation of small organic
"host molecules," typically one-tenth the size of enzymes, that
will recognize "guest molecules," such as metal ions, and bind
them to specially designed cavities, clefts, or cages. This host-guest chemistry
is essentially trying to understand in atomic detail the general principles
of how molecules recognize and react to each other. This knowledge will
be an essential aspect of nanotechnological manipulations. It will enable
our assemblers to feel, recognize, pick up, and put in place the atoms with
which our technological dreams will be built.
Readers who would like full copies of Dr. Lewis's Protein Engineering
Survey may obtain them through FI. Please include a donation of at
least $1 to cover our costs.
