Foresight Update 7
page 4
A publication of the Foresight Institute
 Recent
Progress: Steps Toward Nanotechnology
by Russell Mills
Of the various research paths leading toward nanotechnology,
protein engineering seems to be carrying the most traffic--a
situation stemming from the ability of protein engineers to draw
from a pre-existing treasure trove of protein designs: Earth's
multi-billion-year accumulation of molecular tinkering. The sheer
volume of exciting results causes this subject to dominate my
column; I expect that it will for some time also dominate the
"toolkit" of molecules and techniques that researchers
develop for working with atoms.
Protein engineering itself follows several parallel tracks
defined by the strategies used in designing and building new
protein molecules. Let us look at three of these strategies:
site-directed mutagenesis, transfer of structural cassettes, and
induction of antibodies.
Site-directed mutagenesis
This widely used technique is based on the alteration of
existing protein chains through addition, deletion, or
substitution of amino acids at particular points along the chain.
While in principle capable of producing any conceivable protein,
this method's usefulness is circumscribed by our limited ability
to make rational choices of sites and alterations. Modifying a
protein to achieve a given structure or function may involve
changes at dozens of sites. Experience can provide a rough guide
to the changes needed, but pinning down the details requires the
calculation of interactions between thousands of atoms--an
ability still well beyond current computational techniques.
The work of Roger Bone and colleagues at UC illustrates the use
of site-directed mutagenesis to explore protein function. The
enzyme alpha-lytic protease is a protein that selectively cuts
other protein chains having certain sequences of amino acids in
accessible positions. Bone's group substituted amino acids at
either of two locations in the enzyme's active site (the pocket
in the enzyme's surface where substrate molecules are bound and
transformed). A mutation at one location removed a bump in the
active site, causing the enzyme to select larger substrate
molecules for binding and modification. A mutation at the other
location enlarged the active site by the same amount, but gave it
a different geometry; the enzyme's overall activity declined
drastically. Both mutations made the enzyme less rigid,
broadening the class of substrates to which it could bind. [Nature
339:191-195,18May89]
Transfer of structural cassettes
This approach to protein engineering consists of transferring
segments of one protein to another protein--cutting and pasting
chains of amino acids, as it were. The rationale is that
evolution has already optimized the structure of such segments
for the functions and local environments in which they occur. The
protein engineer still must pay attention to how the components
join together--for example, rejecting candidate chains whose
overall geometry would disrupt adjacent parts of the target
molecule--but is spared the task of understanding and calculating
every molecular detail.
Thomas Hynes and his colleagues at Yale and Stanford recently
used this method to make a hybrid between two unrelated proteins.
A chain of 5 amino acids from one protein was replaced by a
6-amino acid chain from the other, yielding a fully functional
protein with somewhat reduced stability. In choosing the proteins
and the transferred segments, the researchers ignored the amino
acid sequences; they strove for similarity in the angles at which
the segments fused to adjacent protein chains. [Nature
339:73-76,4May89]
Structural cassette transfer is a major shortcut in the
development of the nanotechnological toolkit. Treating proteins
as modular devices whose parts can be selectively interchanged is
one way of accessing evolution's ancient but extensive
engineering experience.
Induction of antibodies
Antibodies are proteins produced by mammalian immune systems
in response to molecules that the immune system classifies as
foreign. A given antibody binds tightly to a specific pattern of
atoms (called an "antigen"); we say that it recognizes
that antigen.
When a mammal (typically, a rabbit) is injected with a substance
that provokes an immune reaction, the resulting antibody
molecules can be separated from the blood and tested for
reactivity with molecules resembling the original antigen. Often
some of the antibodies are able to recognize molecules chemically
distinct from the antigen but resembling it in shape or charge
distribution. This has made it possible to induce the production
of antibodies that catalyze chemical reactions, much as enzymes
do. The trick is to choose an antigen resembling a chemical
"transition state"--i.e., a transitory
configuration of molecules in mid-reaction, as the system passes
over the energy barrier separating the reactant configuration
from the product configuration. By binding the transition state,
the resultant "catalytic antibodies" make it somewhat
less unstable, lowering its energy and thus lowering the energy
barrier that hinders the progression of the chemical reaction.
In a new extension of this strategy, K. M. Shokat and others in
Berkeley and Zürich have developed a degree of control over the
microstructure of an antibody's binding site. Although it would
be possible (in principle at least) to achieve the same results
through site-directed mutagenesis or even by synthesizing the
relevant parts of the antibody and "pasting" them into
another antibody (as structural cassettes), the method used by
Shokat's group is indirect and elegant: they synthesized an
antigen that mimicked not the transition state of a chemical
reaction but the complementary shape and charge distribution they
hoped to realize in the binding site of an antibody. Using this
antigen, they induced the production of antibodies with desired
characteristics built into their binding sites. [Nature
338:269-271,16Mar89]
The procedure may be made clearer with this analogy: a mechanic
(protein engineer) who wants a new kind of wrench (the catalytic
antibody) makes a wax model of the wrench's jaws (the shape and
charge distribution of the binding site) and gives it to a
foundry (the rabbit). The foundry builds a plaster mold (the
complementary shape and charge distribution of the binding site)
around the wax model, melts out the wax, uses the mold to cast a
metal wrench head (the actual binding site), and attaches the
wrench head to a standard handle (the rest of the antibody). The
resulting wrench (catalytic antibody) can now be used by the
mechanic to carry out mechanical tasks (catalyze certain chemical
reactions).
Like the structural cassette strategy, this method draws upon an
ancient engineering legacy--that which produced the antibody
factories of the immune system. By delegating much of the
molecular engineering and fabrication work to these agencies,
researchers stay within the practical limits of computational and
experimental complexity.
Protein engineering in perspective
The discussion above portrays protein engineering as having
begun to assemble a few very simple tools. It is mainly
preoccupied with making very small changes in existing protein
structures to test the relationship between structure and
function, or to develop variations on useful functions already
found in native proteins.
There are exceptions to this picture. The work of de Grado's
group at Du Pont, discussed in previous issues of Update,
belongs to yet another branch of protein engineering: de novo
protein design, in which moderate-sized proteins are designed and
built from scratch. These workers are confronted by the same
basic problem plaguing all molecular engineering today: that of
computing the structure and dynamics of large collections of
atoms. A future generation of computers should make such
computations feasible; meanwhile, protein designers simplify the
problem: using short chains of amino acids (for which useful
calculations can be done) they build larger structures from
multiple copies of these chains by arranging them in simple
patterns.
Short subjects
Molecular motors appeal to the mechanic in all of us, and
evolution has managed to come up with several kinds--some drive
rotary devices (like flagella), others drive muscle contraction,
while still others haul loads along fibers (called
"microtubules") inside cells. The last category
includes dynein, which generates movement away from the growing
end of microtubules, and kinesin, which moves in the opposite
direction. Jonathan Scholey in Denver and his colleagues have now
shown that kinesin consists of a pair of globular
"heads" about 10 nm in diameter, a 45 nm stalk, and a
fan-shaped "tail" about 20 nm long. The heads
apparently are motor domains that bind to microtubules and
generate force and motion. They contain a distinct ATP-binding
site for intake of energy. The tail probably binds to cell
organelles, which are then hauled along the microtubules to their
destinations. The researchers speculate that the two heads take
turns attaching and detaching as they track along a microtubule
so that at least one head is holding on at any given time. [Nature
338:291-292,23Mar89]. (Chapter 4 of Engines of
Creation describes an array of assemblers backed up by
conveyors carrying reactive molecules to the assembly area. Rows
of molecular conveyors--an engaging idea, but one with no
prospect of fulfillment for several decades, right? ... Wrong.
Prototypes exist already inside our every cell! Modifying
microtubules and kinesin motors for use in other contexts should
prove far easier than designing a whole new conveyor system from
scratch.)
Some fascinating work by Seth Stern and colleagues (UC and
Univ. Wis.) elucidates ribosome assembly and function.
Ribosomes--those intracellular particles that read the genetic
code and use the information to assemble proteins--are the
archetype of all molecular assemblers. We already use them to
produce proteins specified by recombinant DNA; when we learn to
redesign them we should be able to manufacture almost any
polymeric material with atomic precision. Ribosomes consist of
three RNA molecules, and dozens of different proteins whose role
has been obscure. Stern's group measured the accessibility and
reactivity of all 1500+ nucleotides making up one of the three
RNA molecules, doing this at each stage as this RNA folded into a
functional molecule. The work revealed a folding process
controlled and stabilized by the ribosomal proteins. The
researchers constructed a 3-dimensional map of this RNA molecule,
and found sites on the ribosomal surface for the binding of
antibiotics and of molecules that assist in protein synthesis.
The most significant findings, however, involved the accuracy
with which ribosomes read the genetic code. The error-rate is
affected by the conformation of the RNA chain, which in turn is
modulated by external factors, such as the binding of
streptomycin. Intriguingly, the authors mention a mutation that
causes ribosomes to read hyper-accurately, and suggest that
"translational accuracy is somehow held in balance at a low
level of misreading." [Science 244:783-790,19May89]
Protein motions--vibrations, rotations, folding, unfolding,
and other motions--unfortunately appear to be crucial for the
function of biological macromolecules. The time-scale of these
motions ranges from about 100 femtoseconds to more than a second.
The motions treatable by computer models lie between 100
femtoseconds and about 300 picoseconds, but experimental
verification has been lacking. Now Hans Frauenfelder of the Univ.
of Illinois suggests that recent measurements of myoglobin
dynamics not only permit comparisons to be made between theory
and experiment, but also hint at a simple underlying unity
between some protein motions and the dynamics of glassy
materials. He believes that progress in protein dynamics will be
rapid. (With 9 more orders of magnitude to cover, we'll be in
trouble if it isn't!) [Nature 338:623-624,20Apr89]
Chemists Nadrian
Seeman and colleagues at New York University say they aim to
build three-dimensional structures out of DNA segments, then hook
proteins or other catalytic molecules to the resulting framework.
[Science News, 136:126,19Aug89]
[Webmaster's note: See Update 23 for a report on the
award of the Feynman Prize in Nanotechnology to Dr. Seeman in
1995.]
Materials fabrication has reached nanometer dimensions (in one
dimension) in experiments at Simon Fraser University in Burnaby,
British Columbia. A paper by Anthony Arrott and others describes
the use of molecular-beam epitaxy to lay down alternating layers
of metals, each only a few atoms thick. The resulting materials
exhibit such properties as magnetic fields of unprecedented
strength and magnetic moments that can be switched from one
direction to another by an electric current. [IEEE Spectrum,
April 1989:12]
Dr. Mills has a degree in biophysics and assists in the
production of Update.
From Foresight Update 7, originally
published 15 December 1989.
Foresight thanks Dave Kilbridge for converting Update 7 to
html for this web page.
|