Mightier machines from tiny atoms may someday grow

Molecular engineering, using mechanisms like those of viruses and cells, will replace the ‘primitive’ tools of our industrial epoch

Coal and diamond, cancer and healthy tissue: throughout history, the arrangement of atoms has distinguished the cheap from the cherished, disease from health. Arranged one way, atoms make soil, air and water; arranged another, they make ripe strawberries. Arranged one way, atoms make air and steel fenders; arranged another, they make rusty scrap.

We have come far in our atom arranging, from chipping flint for arrows to machining aluminum for spaceships. We take pride in our technology, with its gleaming metal and desk-top computers. Yet for all our progress in arranging atoms, we still use primitive methods: we handle them not as individuals, but as unruly herds. Like it or not, the greatest technological revolution in history still lurks in the future.

Today’s technology builds on an ancient tradition. Chipped flint was the high technology of 20,000 B.C. Our ancestors grasped chunks of flint containing trillions of trillions of atoms and removed chips containing billions of trillions of atoms; they had skills difficult to imitate today, and they did fine work. They had other skills as well: they made patterns on cave walls in France with sprayed paint, using their hands as stencils. In tune, they made pots by heating clay, then bronze by cooking rocks. Tile) shaped bronze by pounding on it. They made iron, then steel, and shaped it by heating, pounding and removing chips.

Nowadays we cook up pure ceramics and strong steel, and shape them by pounding, chipping anti so forth. We cook tip silicon (using pure ingredients in a clean pot), saw it into slices and make patterns on its surface using special sprays and tiny stencils. We call time products ‘‘chips," and think that they are small.

Microelectronics has stuffed the monster computers of the 1950s onto a few silicon chips, and sent them speeding away in Japanese motorcycles. Engineers use big machines such as vacuum vapor-deposition chambers to make ever-smaller devices, slinging hordes of atoms at a surface to build up wires and components one-tenth the width of a fine hair. Although small by the standards of flint chippers, these microcircuits remain gargantuan: a single circuit element holds trillions of atoms, and a simple computer chip remains visible to the naked eye.

Yet today a technology is rising on a new foundation. Like microelectronics, it is a technology of the tiny and it also fills tile press with news. Unlike microelectronics, it builds devices from the bottom up, putting every atom in its place.

Biochemists use machines made of protein molecules These machines differ from ordinary machines in many ways, yet like them they have parts of various shapes and sizes that can move with respect to one another. All machines have parts made from clumps of atoms; protein machines simply use small clumps. A chance to build with such tin)’ devices would drive microcircuit designers mad with joy, yet difficulties remain; while engineers can use photographic reduction to turn drawings into circuit patterns, biochemists must build more indirectly, mixing chemicals in various sequences with uncertain results.

In time, however, advanced molecular machines will let engineers build molecular structures as easily and directly as a microcircuit, or a car. This will change the nature of technology. As the metal machines of tile Industrial Revolution spread wealth in a world of poverty, artificial molecular machines will reshape the world as we know it.

Many years will pass before engineers build such molecular machines, yet genetic engineering is already opening a path toward them. The new gene machines lie on this path, and illustrate what chemists can do even today.

Picture clumps of atoms linked by bonds as lumpy beads linked by snaps. To form ordinary polymers, chemists mix the right beads–molecules–in a vessel. These bounce and tumble in the molecular chaos of the liquid, bumping and snapping together. Ordinarily, one snaps onto another until all have joined chains.

In a gene machine, the beads are the letters of the genetic alphabet, the nucleotides–or segments–of DNA, and they are not all thrown in together. To extend a chain, the machine first puts one kind of nucleotide in the vessel to snap onto the chain’s end. Then, the machine washes excess nucleotides of the first kind from the vessel and adds other chemicals to prepare the chain ends for the next nucleotide in the chosen sequence. Plastic beads anchor the very first nucleotide to keep the chains from washing away with the chemical baths.

Unfortunately, blindly stirring chemicals in a pot leads to addition and deletion errors which become unacceptably frequent in chains more than about two dozen nucleotides long. Genetic engineers first sort good chains from bad, like quality inspectors discarding bad parts before assembling a car. Then, to splice these short chains into working genes (typically thousands of nucleotides long), they turn to molecular machines found in bacteria. Some protein machines, called restriction enzymes, "read" certain DNA words as "cut here." Other enzymes splice pieces together, reading matching parts as "glue here." Together with gene machines to type the "words," these cutting and pasting enzymes let genetic engineers write any DNA message they choose.

If it were not for molecular machines, genetic engineers would never bother to make DNA. In itself, DNA is (heresy!) a worthless molecule. It is neither strong like Kevlar (a plastic fiber stronger and lighter than steel), nor colorful like a dye, nor active like an enzyme. Industry spends millions of dollars making DNA because DNA can provide instructions to molecular machines called ribosomes. First, a machine transcribes genetic information in the nucleus to make RNA "tapes." Then, much as automated machines read tapes and shape metal, the ribosomes in tile cytoplasm "read" the RNA tapes and build proteins. This gives genetic engineering its purpose beLause proteins are useful.

Like DNA, proteins are strings of lumpy beads. Unlike DNA, protein molecules coil up to form small objects able to do things. Some are enzymes, machines that build and tear down molecules (and copy DNA, transcribe it and build other proteins in the cycle of life). Others are hormones, proteins that bind to other proteins in cells and signal them to change behavior.

Absorbing and rearranging with quiet care

Scientists are now trying to program bacteria to make human insulin, growth hormone and interferon (an antiviral agent) on an industrial scale. Researchers are tackling the problem of engineering plants that are able to fix nitrogen from the air, thereby making their own fertilizer. Since human DNA is responsible for genetic defects, there is even interest in patching tip mutations in human cells, such as those that produce defective hemoglobin in the blood.

Biochemists have reason to study the workings of molecular machines, both to learn to build them and to learn to wreck them. Around the world (and especially the Third World) a great variety of viruses, bacteria, protozoa, fungi and worms parasitize people. Drugs fight them by jamming their molecular machinery without greatly affecting the molecular machinery of humans. Penicillin, for example, jams machines thiat build bacterial cell walls, thus killing bacteria while leaving people unharmed Dr. Seymour S. Cohen, Distinguished Professor of Pharmacological Sciences at the State University of New York at Stony Brook, argues that biochemists should systematically study thie molecular machinery of parasites. Once the shape and function of a protein machine vital to a parasite is determined, chemists could design a molecule shaped to jam it and ruin it. Such drugs could free humanity from such horrors as leprosy and herpes.

In the long run, however, progress will rest not on jamming existing machines with drugs, nor on moving them from cell to cell with genetic engineering, but rather on building new machines to do our bidding. Existing molecular machinery points the way.

In the cell, proteins serve basic mechanical functions. Some push and pull; some act as cords or struts. Parts of some molecules make excellent bearings. A reversible, variable-speed motor drives bacteria through the water by turning a corkscrew-shaped propeller. If someone were foolish enough to build tiny cars around such motors, several billions of billions would fit in a pocket, andi150-lane freeways could be built through our finest capillaries.

Simple protein machines combine to form complex systems The tape-programmed ribosome is one example; viruses provide others. One virus acts as a spring-loaded syringe, first attaching itself to a bacterium, then punching a hole in its wall and injecting viral DNA. Like a conqueror seizing factories to build mom-c tanks, this DNA directs the cell’s machines to build more viral DNA and syringes.

What is more, protein machines assemble themselves, even in the test tube, driven by thermal agitation and thermodynamics. Imagine protein molecules dissolved in water. Each is a lump covered with bumps, hollows and patterns of oiliness, wetness and electric charge. Picture them wandering and tumbling, jostled by the thermal vibrations of the surrounding water molecules. From time to time they bounce together, then bounce apart. Sometimes, though, two bounce together that fit, bumps in hollows, with sticky patches matching–then they pull together and stick. Proteins fit like pieces in a jigsaw puzzle, but form machines. This ability is surprising: imagine putting the parts for a car in a large box, shaking it up, and then finding an assembled automobile inside!

With motors, devices that push and pull, and parts that move on bearings, engineers build complex machinery. Today they use plastic and steel, and build robot arms to handle car parts. Tomorrow, when biochemists master the art of protein design, engineers will build robot arms to handle individual molecules.

How far off is such an ability? In 1959, many geneticists called engineering impossible; today, it is an industry. With biochemistry and computer-aided design exploding fields, protein design may come sooner than one might think. Already, biochemists have mapped the structures of many proteins. With gene machines to help write DNA tapes, they can direct cells to build any protein they can design. The real difficulty lies in knowing how to design chains that will fold to make proteins of the right shape and function. Because the forces that fold proteins are weak and the number of plausible ways a protein might fold is astronomical, designing a large protein from scratch remains beyond reach.

Nevertheless, biochemists have already designed short chains of a dozen or so pieces that fold and nestle onto the surfaces of other molecules as planned, and have rebuilt existing enzymes, slightly. James H. McAlear of EMV Associates inc., Forest L. Carter at tile U.S. Naval Research Laboratory, Kevin M. Ulmer of Genex Corporation, and researchers in university laboratories around the world are already beginning theoretical work on molecular switches and memory devices that might be incorporated into a protein-based computer.

Whether used to make gears or computers, however, protein machines have shortcomings. They quit when dried, freeze when chilled, and go to pieces when heated. Engineers do not build from flesh, hair and gelatin; long ago, people learned to use their hands of flesh and bone to build machines of wood, ceramic and steel. Similarly, we will use protein machines to build machines of tougher stuff.

Using tapes to direct them, like ribosomes or automated machine tools, protein machines could build nonprotein machines of more solid structures like metals, glass and diamond.

Machines of such stuff, the machines of the second generation, could be rugged. Able to tolerate acid or vacuum, freezing or baking, machines with such parts could use as "tools" almost any of the reactive groups used by chemists, and the)’ could use them with tile precision of programmed machines. With such machines, we will gain a basic ability lacked! by all past history: the ability to place atoms in any reasonable arrangement, the ability to build all that the laws of nature permit.

And why not? Artificial machines already improve on evolved machines. Cars outspeed cheetahs, jets out-fly falcons and computers already outcalculate head-scratching humans, though they are still made by the chipping, cooking and stencils of bulk technology.

Today, engineers make irregular patterns on the surface of silicon chips by throwing atoms and photons at them, yet the direction of progress clearly shows tile value of miniaturization. Tomorrow, engineers will build circuits to atomic precision, and they will build in three dimensions. The limits of electronic technology remain uncertain (the quantum mechanics of electrons in networks of tiny structures present complex problems) but, whatever they are, they will be reached.

Though some computers of time future will still use electronic effects, the smallest computers may not. This may seem odd. but the essence of computation has nothing to do with electronics. A digital computer is a collection of switches able to turn each other on and off. Its switches start out in some pattern (perhaps representing 2 + 2), and then switch each other into a new pattern (representiug 4), and so on–such patterns can represent almost anything. We use tin)’ electrical switches connected by wires simply because mechanical switches connected by rodis or strings would be big, slow and expensive, using bulk technology.

Time idea of a purely mechanical computer is hardly new. In England in 1834, Charles Babbage invented a mechanical computer built of brass gears. With components a few atoms wide, however, a simple mechanical computer would fit within 1/100th of a cubic micron, many billions of times smaller than today's so-called microelectronics. Even with a billion bytes of added memory, such computers could fit into a box a micron wide. Small size would make them fast. Although mechanical signals move about 100,000 times slower than the electrical signals in today’s machines, because they need travel only 1/1,000,000th as far they face less delay. Thus, a mere mechanical computer could work faster than the electronic whirlwinds of today.

Will such intricately-patterned matter cost a great deal? If each atom cost even 1/100th of a cent to position, then a block the size of a sugar cube would cost more than $10 billion billion to build; if this calculation were to make any sense at all, the whole idea might safely be ignored.

Once again living systems point the way. Growing bacteria show that molecular systems can build copies of themselves if given room and common materials (garbage, for example). Similarly, a device able to pick up atoms and assemble them according to instructions would be able to build a copy of itself. It would need a computer to direct it and a tape to store the instructions, but it could build a duplicate computer and copy the tape, thus replicating the entire system.

Replicators double with each replication, multiplying exponentially. In mere hours or days one microscopic replicator could turn a ton of raw materials and some fuel into a ton of general-assembly machines, able to make almost anything.

Picture a general-assembly machine as a stubby robot arm a few hundred atoms long and a hundred atoms wide. Behind it sits a computer reading a tape, and in front sits an unfinished structure. Tubes or conveyor belts bring molecules to the machine. Some molecules supply energy to motors driving the computer and arm; others supply groups of atoms for the arm to pick up and place where the computer directs. Atom by atom (or group by group) the arm moves pieces into place. Chemical reactions bond them to the structure on contact.

This may seem a tedious process, but tiny arms could move quickly. We are used to thinking of the way human arms wave or human fingers flutter; of the way hummingbird wings hum and mosquito wings whine. But for arms so small, to handle a million molecules per second would be to move in slow motion. Ordinary enzymes do as well using motions driven by thermal jostling. An assembly-machine system might contain billions of atoms, yet it could copy itself in a matter of minutes, like a bacterium.

These machines could then be turned to making other products, such as small computers, strong diamond fiber, industrial-grade synthetic muscle, molecular machines able to break down rock and separate pure metals, and more. Though tiny, molecular ma-chines can build big. Molecular machines working together build whales; seeds replicate their machinery and organize atoms into vast structures of cellulose, building redwoods. In time, we could build ‘‘seeds" able to soak up air, water, soil and sunlight to build simpler things–houses or spaceships. Yet building, even without labor or pollution. ma seem trivial once medicine makes use of molecular machines.
 

Moving beyond handling atoms in bulk

With continuing advances in drug design, however, medicine is moving toward sophisticated molecular technologies. Biologists already use molecular devices to study cells; enzymes help identify chemicals, antibodies tag proteins, and viral syringes inject DNA into bacteria.

As molecular machines and computers grow more advanced, biologists will move beyond simply smashing cells, peering at them through microscopes, and probing them; they will instead use advanced machines to probe and repair cells swiftly and directly, learning both structure and function. In time, biologists will catalog the tens of thousands of kinds of molecules in the body. and will map the structure of the hundreds of kinds of cells.

Cell-repair machines will take decades to develop, but they lie straight ahead. The medical need for new drugs and proteins already leads researchers, step by step, along a path to advanced molecular technology.

At this point, the idea of a technology that promises both a cornucopia and a panacea may well seem too good to be true. But should this news be surprising? These possibilities are part of nature, built into time structure of natural law. Biochemical systems already build with atoms; designing new machines will demand much engineering and detailed research, but no novel laws of nature. In a world made of atoms, time ability to handle atoms is basic; important, new and therefore startling consequences follow as summer follows spring.

Nevertheless, time picture painted above should seem suspicious because it is incomplete. Long ages ago, perhaps to guard against buying magic charms from shifty-eyed strangers, people learned to disbelieve tales of easy riches. They have since learned a similar lesson about technology. Chipped flint killed both game and people. Rockets both opened space to development and London to bombardment. Molecular technology, too, brings dangers on time scale of its benefits.

Consider one example: the danger of uncontrolled replicators. Recombinant DNA technology brought fears of the creation of a monster bacterium able to infect people or upset the ecology of Earth. These fears lessened as people realized that genes have been recombining for ages in nature, and that laboratory bacteria are pampered creatures unable to compete with bacteria adapted to the outside world. Artificial self-replicating molecular machines, however, might have no effective predators or competitors to keep

them in check. Solar-powered replicators could operate like plants or bacteria, but with greater energy efficiency, no vulnerability to disease, and little sensitivity to moisture, temperature, or soil composition. Replicators could act like bacteria, but with omnivorous tastes and few weaknesses. Any ecologist should see the changer: uncontrolled replicators could destroy time biosphere, and all we value. The danger is distant, but real.

We can avoid such dangers, for a time, through simple safeguards. It will be easy enough to build all replicators with a need for artificial "vitamins" made only in time laboratory thus tying them to human control. We must, however, use the time gained to develop means of preventing the deliberate redesign and misuse of replicators. Although possible, this poses a challenge of a different order; meeting it while preserving our liberty may one day be our greatest task.

Our world holds more than a hundred competing countries, and industrial companies beyond count. Most work today aims at such short-term goals as better drugs and better understanding of the processes of life. Nevertheless, governments, universities and chemical, electronic and pharmaceutical companies from Europe to Asia have set out on a path that leads ultimately to molecular machinery.

Time incentive to continue will only increase. In a competitive world, if something is possible, shows long-range promise and offers short-term payoff. it will l)e developed. We must count on a future with molecular technology, like it or not.

Fortunately molecular technology can bring more than an era of health, wealth and tiny computers: it can help resolve the conflicts that divide humanity. Today, industrialists and environmentalists are at odds not because the former like pollution or tile latter Scarcity, but because industry cannot avoid polluting while producing. Molecular technology can reduce this conflict by making manufacturing as clean as an apple orchard, and transportation far cleaner than horse-drawn carriages.

Expectations shape actions, and expectations rest on a view of time future. Tile idea of molecular technology brings a new view’ of the future, and with it a new view’ of time world and a new framework for action. If we stumble forward blindly, as if molecular technology were a mere idea about some stuff in a test tube, we will reap the benefits late, while risking destruction. We have little choice hut to open our eyes, to think hard on what we see, and to act as wisely as we are able.

Engineering with atoms and molecules