Enzymes are important to nanotechnology in several ways. They are examples
of atomically precise functional structures. They may be useful as tools
to construct the building blocks for more rigid, polycyclic structures.
The papers described below report advances in the design and analysis of
A.D.Mesecar, B.L.Stoddard, and D.E.Koshland Jr., writing in [Science277:202-206 11Jul97--MEDLINE
Abstract] have experimentally demonstrated the sensitivity of
enzyme catalysis rates to small changes in the alignment of the substrates.
They studied isocitrate dehydrogenase (IDH). This enzyme transfers hydrogen
from isocitrate to nicotinamide adenine dinucleotide phosphate (NADP), performing
one of the reactions in the Krebs cycle. The geometry of the active site
was determined by x-ray diffraction of the Michaelis complex of IDH with
its substrates and a metal ion cofactor (Mg2+ in the normal case).
Three crystal structures were compared: the structure of a complex of the
normal substrates, the normal metal, and a mutant (Y160F) enzyme, the structure
of a complex of a modified substrate (nicotinamide hypoxanthine dinucleotide
phosphate NHDP), the normal metal, and the normal enzyme, and the structure
of a complex of the normal substrates, Ca2+, and the normal enzyme.
At the reaction center, a hydride is transferred from the isocitrate to
the nicotinamide ring in all three of these cases. None of the modifications
studied changes the identity of the atoms which participate directly in
the reaction. The modifications do alter the geometry of the reaction
In the experiment modifying NADP to NHDP, the "substitution occurs
at a position that is >12Å and 21 bond lengths away from the hydride
that is transferred and has no effect on the G0
of the reaction." The substitution does, however, shift the positions
of the substrate atoms in the complex, propagating to the nicotinamide ring,
where it "causes the distance of the hydride donor-acceptor pair to
increase by 1.55 Å (a covalent carbon-carbon bond distance is 1.54
Å) and the angles of approach to deviate between 10° to 20°
from the more in-line geometry observed for the NADP structures." The
effect of this shift in geometry is to reduce the catalysis rate by a factor
of 3.8 x 10-5.
In the experiment substituting Ca2+ for Mg2+ a cascade
of geometric changes starts with the enlargement of the coordination sphere
from the normal 6 ligands (for Mg2+) to 8 ligands (for Ca2+).
The net effect at the reaction center is "an adjustment of isocitrate,
NADP, and the side chains of the aspartate residues, thereby decreasing
the distance between the hydride donor-acceptor pair by 0.55 Å and
altering the attacking and dihedral angles." The effect of this shift
in geometry is to reduce the catalysis rate by a factor of 2.5 x 10-3.
The authors emphasize that the Ca2+ data rule out a simple dependence
of the catalytic speed on the distance between the hydride donor and acceptor.
They write that: "The results provide evidence that orbital overlap
produced by optimal orientation of reacting orbitals plays a major quantitative
role in the catalytic power of enzymes."
These experiments have a number of implications for nanotechnology. First,
they provide direct experimental evidence for sharp geometrical control
of reaction rates, even in the unfavorable case of hydride transfer (with
maximal tunneling effects). This strengthens the claim that controlling
the approach of a reagent to a workpiece can permit selective reaction at
just one of a number of nearby chemically equivalent sites. Second, this
adds to the desirability of very stiff, heavily crosslinked "diamondoid"
structures for controlling reaction state geometries. Third, these experiments
imply that purely x,y,z control of reactant positioning, as in current STM
and AFM piezoelectric actuators, is likely to be insufficient for efficient
mechanosynthesis. Control of orientation via tiltable stages or similar
mechanisms is likely to be necessary to steer reagent orbitals towards desired
Writing in [C&EN 35-36 30Jun97], S.Borman reports on recent work
from J.P.Caradonna's lab, which redesigned a noncatalytic host protein,
thioredoxin, to incorporate an active iron center from superoxide dismutase
(SOD). They were able to demonstrate catalytic activity analogous to that
of SOD (albeit four orders of magnitude slower) in the hybrid structure.
In a separate modification, they were also able to introduce an Fe4S4
group into thioredoxin and to demonstrate the redox activity of this group
in the modified protein. Other groups have added metal binding groups into
proteins previously, "but this is the first time metal-modified proteins
have exhibited functional activity such as catalysis." A program called
Dezymer (originally from H.W.Hellinga) was used to do the combinatorial
search to find feasible modifications to thioredoxin in order to accommodate
the SOD active site while preserving the folding of the host protein. Caradonna
said that they "are using Dezymer as a tool to systematically investigate
the effect of the protein matrix on reactivity at metal centers, just as
the effect of ligand substitutions on the reactivity of small metal complexes
have been studied." From the perspective of nanotechnology, our best
current technologies for building atomically precise 3D structures are currently
biomolecules such as proteins. This work helps to disentangle this geometrical
design work from the design of reactive tips, which are roughly analogous
to the metal centers in catalytic proteins.
The techniques described in the sections above all rely on some form of
self-assembly to contruct structures. An alternative strategy for constructing
and analyzing atomically precise structures is to modify one molecule at
a time. These techniques allow direct control of position with atomic precision.
Some type of replication technique needs to be used with this strategy in
order to produce large numbers of structures.
I. Amato, writing in [Science276:1982-1985 27Jun97] presents
a brief summary of the history of STM and AFM microscopies. Most of the
work that he touches on will be well known to the readers of this column,
but one group's work was unfamiliar to me, at least. R. Dunn's group at
the University of Kansas has been monitoring cellular pores with an AFM
probe. Amato quotes Dunn as saying: "In the open state you see a channel,
but after triggering the pore [with calcium ions] you see something like
a piston stick up and block the central part of the channel." This
experiment demonstrates techniques that should be useful in debugging nanoscale
machinery. First, it detected a nanometer scale mechanical event
rather than just a static image. Second, the mechanism for triggering the
event (Ca2+ addition) was independent of the AFM, so modification
of the AFM tip for more specialized imaging can be attempted without interfering
with the trigger mechanism. Third, since the experiment used an AFM rather
than an STM, it is applicable to insulating samples as well as to conducting
Writing in [Nature387:688-691 12Jun97], F. Kulzer et. al.
"report light-induced reversible frequency jumps [of the absorption
frequencies of] single molecules of the aromatic hydrocarbon terrylene embedded
in a particular site of a p-terphenyl host crystal at a temperature of around
2K." When p-terphenyl crystals lightly doped with terrylene are condensed
from vapor, the terrylene can be inserted into four different crystal sites,
called X1-X4, with different electronic dynamics.
The authors found that "although single molecules in X2
are very stable with respect to changes in absorption frequency, we find
reversible, light-induced frequency jumps in site X1." The
authors examined isolated molecules by keeping the doping sufficiently light
that there were typically fewer than five terrylene molecules in a 5-µm
More specifically, when a terrylene molecule in an X1 site is
illuminated at resonance with a laser field of 0.25 W cm-2 for
10-60 seconds, it jumps to a second state (denoted as XY) where its absorption
frequency is 843 GHz higher than in its original state. Illumination at
the new resonant frequency switches the molecule back to its original spectral
position. The frequency jump is a small fraction (0.16%) of the absorption
frequency, but it is quite distinct, "more than 10,000 times greater
than the homogenous line width of terrylene in p-terphenyl." It is
also quite consistent from sample to sample. The authors "have investigated
~50 molecules in seven different crystals taken from four sublimation runs."
They have "always found the same reversible jumps over 843 ± 2GHz,
with a variation of at most 500 MHZ for molecules in the same crystal."
Illumination at XY's resonant frequency doesn't always switch the molecule
back to the X1 state. About 10% of the time, the molecule makes
a transition to a series of other states, although it can be returned to
the X1 state by annealing at 40K. The states are sufficiently
stable that the authors followed the light-induced state transitions in
a "spectral diary" of one molecule over a period of 22 days.
The authors interpret these states as changes in the orientation of phenyl
groups in the local environment of the terrylene molecule. They analyzed
the vibronic structure of the fluorescence from molecules in the various
states and conclude that "a central ring flip of the adjacent p-terphenyl
molecule...alters the dipolar host-guest coupling and shifts the absorption
frequency of the chromophore."
This work affects nanotechnology is several ways. First, as the authors
write: "Although many obstacles still exist, this observation of reversible,
reproducible frequency jumps might pave the way towards the design of host-guest
systems suitable for optical switching and storage functions at the single-molecule
level." Second, this provides a new technique for predictably reorienting
single molecular environments, possibly applicable where STM or AFM techniques
are not feasible. Third, together with a localization technique (such as
near-field optics or possibly Stark effect shifting of resonant frequencies
in STM fields), this technique may provide a novel patterning mechanism
for selectively reorienting individual molecules.
Not all work that creates a nanoscale structure directly aids the development
of molecular manufacturing. Writing in [Science276:1401-1404
30May97], M. Park et. al. describe a novel method for constructing a pattern
with a 30-40 nm pitch. Their technique relies on phase separation in diblock
copolymers. They synthesized diblock copolymers from polystyrene-polybutadiene
(PS-PB) and also from polystyrene-polyisoprene (PS-PI). Two different block
length combinations were used, SB with 36 kilodalton PS blocks and 11 kilodalton
PB blocks, and SI with 68 kilodalton PS blocks and 12 kilodalton PI blocks.
The basic ordering mechanism is evident even "in bulk, [where] the
SB 36/11 microphase separates into a cylindrical morphology and produces
hexagonally ordered PB cylinders embedded in a PS matrix; SI 68/12 adopts
a spherical morphology and produces PI spheres in a PS matrix with body-centered-cubic
order." The diameters of these structures is set by the chain lengths
of the blocks. In a 50 nm layer on Si3N4, SB 36/11
produces a sandwich structure, with a hexagonal array of PB spheres in a
PS matrix, bounded above and below by PB "wetting layers", which
coat the Si3N4 substrate and the air interface.
Since both PS and PB etch at similar rates under reactive ion etching conditions,
the authors needed to use special chemistry to distinguish them. They used
two techniques, which changed the effective thickness of the PB in opposite
ways. When they wished to make the PB volume act like a smaller volume of
polymer, they ozonated the wafer. This "attacks the carbon-carbon double
bonds in the PB backbone" and converted the PB (but not the PS) into
fragments that can be removed with water. When they wished to make the PB
more resistant to etching than the PS, they exposed it to OsO4,
which "adds across the carbon-carbon double bonds in the PB backbone,"
loading the PB with etch-resistant osmium. When these two modified films
are exposed to CF4/O2 reactive ion etching, they yield
holes and dots respectively. The holes, for instance, are approximately
15 nm deep, and have a period of 30 nm.
While this technique does generate nanometer scale patterns, and it does
build the patterns up from molecular patterning, it does not appear to directly
aid the development of molecular manufacturing. At the molecular scale,
the patterns of polymer blocks are too imprecise. The blocks have approximately
the stated molecular weights, but they do not all have precisely the same
length (unlike specific proteins, for instance) nor can they be expected
to pack into regions with atomically precise boundaries. At the other end
of the length scale, the authors do not appear to have attempted to synchronize
their patterns with the arbitrary long range patterns made possible by photolithography.
In fact, the patterns formed in this experiment have "a polygrain structure
that has an average grain size of 10 by 10," so it loses coherence
after about 300 nm. If this work is extended in both of these directions,
then it may provide a mechanism for exploiting conventional photolithography
to order atomically precise structures, but it does not currently do this.
Jeffrey Soreff is a researcher at IBM with an interest in nanotechnology.
Special thanks this issue go to Ka-Ping Yee and Terry Stanley for their work on Web Enhancement: Ping for coding the Backlink Mediator, and Terry for programming the LinkMap. See the "Inside Foresight" column in this issue for more details. Thanks also to our many informal advisors on the project, who include Mark Miller (who designed the backlink approach along with Ping), Norm Hardy, Marc Stiegler, and Dean Tribble. Also of help have been Dave Forrest, Wayne Gramlich, Ralph Merkle, and Russell Whitaker, and for PR guidance, Lew Phelps and Ed Niehaus.
For arranging for Foresight to have a London office, we vigorously (or should we say, vigourously?) thank Chris Portman and Philippe Van Nedervelde.
Continual thanks go to Gayle Pergamit, co-author of Unbounding the Future, for frequent assistance in briefing the media on nanotechnology and Foresight.
Ongoing thanks as well to the '97 nanotechnology conference team, including volunteer co-chairs Ralph Merkle and Al Globus, tutorial chair Deepak Srivastava, and planner Marcia Seidler.
For assistance to our summer policy intern, Franklin Van Ardoy, we thank Prof. Pat Parker (U.S. Naval Postgraduate School), Duncan Forbes, and Margaret Jordan.
For editorial contributions to this issue, thanks to Richard H. Smith, II of Georgetown University and Chris Worth of Singapore.
For sending information, we thank Jon Alexandr, Richard Counihan, Dave Forrest, Chris Fry, Richard Kluckhorn, Markus Krummenacker, Anthony Napier, Chris Portman, Salvatore Santoli, Nadrian Seeman, Richard Smith, Steve Vetter, and Russell Whitaker.
Chris Peterson, Executive Director, Foresight Institute