Staring at the skeleton sketch of BR, I imagine myself inside the Lilliputian columns when the sun breaks through the San Francisco fog. A photon zigzagging through the salty bay dives into the sensitive retinal A, causing it to shift shape, from straight to bent. As it kinks, the protein columns attached to retinal A are rattled as well. The amino-acid molecules studded throughout the rattled columns bump against one another, like passengers colliding in a lurching bus. The new proximity starts a handoff of a proton from amino acid to amino acid. In a nanoheartbeat, the positive charge moves from the inside of the membrane to the outside. A sunny morning can keep the handoff of protons working continuously.
What interests computer engineers is only the first part of this scene—the photon of light hits, and the molecule shifts. This flip-flop, from one state to another and back again, is automatic, even if the protein molecule is separated from its live host. “What most people don’t realize is that you can remove BR from Halobacterium and embed it in plastic and it will work quite beautifully,” says Hong. “In Russia, scientists have made a film of BR that they can still flip back and forth after fifteen years. That stability, we thought, would make it a good medium for storing information in computers.”
Another of BR’s talents is its knee-jerk reaction to certain frequencies of light—this means you can use one color of light to kink it (recording a one) and another color of light to unkink it (recording a zero). Here’s how it works: In its relaxed state, BR will absorb only green light. If you flood it with green light, it kinks to a red-absorbing state. Then, if it’s zapped by red, it unkinks, returning to the green-absorbing state. It’s an endless toggle controlled by light.
This mechanism reminded computer scientists of the system already in use for storing digitized information. The working surface of magnetic hard drives or floppy disks is covered with tiny iron-oxide crystals, and they are able to flip their poles like little magnets. As the read/write sensors make their passes over different parts of the disk, they turn electrical signals into magnetic energy and vice versa.
In the case of optical protein computing, the working surface of the disk would be covered with BR molecules (much smaller than iron-oxide crystals) packed shoulder to shoulder. The read/write heads would be red and green laser beams, which, when aimed at specific “addresses” on the drive, would kink and unkink molecules, storing ones and zeros and then reading them out. An optical detector would measure whether or not light has been absorbed at each site. To keep from erasing information during the reading process, a second pulse of light would follow the read light to reset the flipped BR.
The thought of using a protein this small to store information quickens the pulses of computer engineers. Robert R. Birge, director of the W. M. Keck Center for Molecular Electronics at Syracuse University, went beyond dreaming and teamed up with physicist Rick Lawrence of the Hughes Aircraft Corporation in Los Angeles to flight-test a BR storage device. They laminated a thousand layers of BR, each a molecule thick, onto a thumbnail-sized quartz plate. “It looked like a piece of glass with a clear, deep, rich red coating,” Birge said.
A laser was used to address not one molecule at a time (laser beams are still far too wide to do this) but a patch of about ten thousand molecules, flipping them all at once. Even in this configuration, says Birge, the device has a potential storage density of nearly ten megabytes per square centimeter, comparable to the storage density of elite magnetic devices available only in multimillion-dollar supercomputers. But that’s only a beginning. When we find a way to focus the beams to write to each molecule, says Birge, a single 5¼-inch floppy disk coated with BR could theoretically hold 200 million megabytes (compared to the 1.2 megabytes that a disk that size holds now). Access times would be cut way down, too. It takes BR only five trillionths of a second to change absorption states. Give it a nanosecond, and it’ll kink and unkink two thousand times, beating conventional magnetic devices by a factor of a thousand.
But in the speed-addicted world of computing, even this is not fast enough. Researchers at the Naval Research Center in Dahlgren, Virginia, are hoping to find or engineer a strain of Halobacterium with an even faster BR flip-flop. Ann Tate, manager of the Molecular Computing Group, explains, “When the BR molecule flexes from its unkinked to its kinked state, it goes through a continuum of shapes, each one with a different absorption spectrum. Right now, we concentrate on the ground state and the kinked state, and it takes five picoseconds to get from one to the other. What if we could speed up the flexing? Find a BR that kinks at three picoseconds instead of five?”
The scientists hope to find that speedier BR in one of the millions of Halobacterium offspring they are raising in laboratory tanks. Once they locate the winning microbe, they’ll want to put its BR in a storage medium that breaks all the records in terms of capacity. That means going beyond two-dimensional BR films. “What we are starting to do now is suspend BR in a Jell-O—like plastic that hardens into cubes. When we get BR memory in 3-D like this, storage capacity will really balloon.”
Imprisoning BR in a cube presented an opportunity but also a logistical problem—how to read the molecules in the center of the cube without the light beam triggering or destroying information on the way in. Once again, BR’s special qualities allowed engineers to jog around this problem. Researcher Dave Cullin was explaining this to me (with copious drawings) in the windowless belly of a Quonset hut at naval headquarters in Dahlgren. “BR actually uses two photons when it photosynthesizes—it adds up the energy in the two. This ability to absorb and combine two photons gave us an idea. We could penetrate the cube with two rays, each entering from a different face, each of a frequency that, by itself, didn’t affect BR molecules on the way in. At the point where the rays converged, however, their frequencies would combine, and this energy would be enough to write or read the data at that particular address.” Dave paused after this punch line, giving me time to admire the simple ingeniousness of the two-photon scheme. For the thousandth time I noticed my own tendency (a human tendency, I think) to be absolutely delighted by this sort of elegance. The same elegance that nature, of course, has been choosing for eons.
So now that we have trillions of BR molecules in a device the size of a sugar cube, what can we store? We could use BR just to store zeros and ones, of course, but Robert Birge has a more ambitious plan. He and his company, Biological Components Corporation, want to use the 3-D memory device to store analogue holographic images in the BR, imprinting patterns of light and dark instead of strings of zeros and ones.
A hologram is created by superimposing two beams of light onto a piece of film. One beam of light contains the image, and the other is plain light, called a reference beam. Where the light waves interfere on the film, they create a unique signature. The deconstructive interference (where there is no image) causes dark areas and the constructive interference (the image) is registered as areas of light. When you want to recall the original image, you simply flood the hologram with plain light—the reference beam—and it regenerates the original recorded pattern. In Birge’s device, the film would be BR, and the light and dark patterns of the light waves would be recorded in kinked or unkinked molecules.
Holographic memory is especially suited to what’s called correlation, or matching of images. You can take a picture of an airplane wing, for instance, and then fly the plane and take another picture. Comparing the two holograms would instantly show you where stress or strain has occurred. To make the holograms even more versatile, you can pass the images through a Fourier lens as you record them, which basically turns the image into a “frequency picture” so that the holographic correlator can recognize and match an object even if it is tilted at a different angle from the way it was when originally recorded. For instance, a pen would be recognizable whether it is held horizontally, vertically, or anywhere in between. (Our eyes have even more flexibility. We can recognize someone if they are close or far, or if their image is tilted side to side, forward or
back. “Nature is ahead of us here,” Tate admits, “but it gives us something to strive for.”)
Fourier transforms made with conventional film can be layered like transparencies and held up to the light—when light shines through two of the transforms in the exact same spot, you have your match. What holographic BR memory can do, with mirrors and lenses, is place hundreds of BR-embodied Fouriers on top of one another to simultaneously find a match. This puts it streets ahead of digital techniques.
You could store pictures of all the customers in your bank, for instance, and when someone walked up to a teller, a camera would see the face and quickly match it to the hologram database, bringing up the customer’s file. Even if the camera caught only an eye or the corner of a smile, it could recall the whole thing, because a hologram stores the whole in each and every part. If you wanted to do the same thing with a conventional silicon-matching device, you would first have to digitize the person’s image into zeros and ones and then comb pixel by pixel for a string of numbers in your database that matched that person’s numbers. In the holographic correlator, numbers are eliminated. You essentially put the entire stack of customer pictures on top of one another and look for the spot of light that shines through—signifying a match. This simultaneous search can be done so quickly that you could use a TV camera as an input device and identify people as they stroll through the lobby.
Information storage isn’t a problem, either. If you figuratively sliced the cube into “sheets,” you could store up to four hundred images per sheet, and then “pull up” a whole sheet at a time by slicing plain light through the cube to illuminate a cross-sectional slice. Even more images could be stored per page with a technique called angular multiplexing. By changing the angle at which the reference beam hits the cube, you could burn hundreds of holograms on the exact same spot and read them back with a tiltable laser.
If the system proves practical, Birge believes holographic memory could play an important role in robot vision, artificial intelligence, optical correlators, and other areas starved for complex pattern-processing capabilities. “This is an area where we could completely blow away semiconductors,” he says. “We’re going to be able to have the equivalent of twenty million characters of associative memory on a single film. You simply couldn’t build a semiconductor associative memory with that many connections.” And yet, I think to myself, an associative memory with many, many more connections has already been designed, and it’s balanced on the stalk of my neck at this very minute.
After Conrad’s compelling visions of self-assembling shapes bouncing in a maelstrom of motion, Birge’s BR, as fantastic as it is, feels a little too confined—too on-and-off digital. To get back into more open spaces, I book a flight to the University of Arizona, Tucson, where I’m told I’ll meet another biomimic who’s determined to climb his own peak in the range of computing possibilities. In their ascents toward natural computing, Stuart Hameroff and Michael Conrad could easily run into each other on the trail.
According to Hameroff, the ultimate computer is not chemicals dancing in neurons, or light kinking proteins in a membrane, but rather the net of the spidery strands (cytoskeleton) assembling and disassembling in your cells as you read this. My survey of nature-based biological computing would not be complete without a visit to the man who sees the roots of consciousness in a microtubule. Buckle your quantum belts for this one.
THE SCAFFOLDS OF CONSCIOUSNESS?
Stuart Hameroff and I are in a stale canteen just off the operating room, waiting for him to be summoned, as he puts it, “to pass gas.” At this moment, he looks more like the sax player on the cover of last month’s Downbeat than an anesthesiologist. He’s tilted chairback against the wall, feet up on the table, his scrub-green shower cap pulled low over a bushy ridge of salt-and-pepper eyebrows. At the back of his cap a ponytail struggles to break loose. He’s staring at a green wall and talking a blue streak.
As my tape spools, his thoughts bank like swallows over a wide landscape: quantum physics, philosophy, computer science, mathematics, neurobiology (another person who needs a Dewey decimal system for his personal library). But he keeps circling back to the same subject, one that has tangled many a fine mind over centuries: the brain/mind debate. That is, does the mind float above and separate from the brain, or does it sprout from the gray goo itself? If it sprouts, by what biological mechanism does it emerge? And then, most mysteriously, how do these biological interactions inside the brain converge to afford us a “unified sense of self”—a single identifiable I?
In a few months, Hameroff will host an international think tank at the University of Arizona in Tucson on consciousness, a conference that already has several hundred registrants pawing at the chance to reenact the old debate. Hameroff has stepped into the consciousness fray in a public way lately, appearing in glossy magazine spreads with Roger Penrose, a mathematical prodigy known for his theories about wormholes, black holes, and geometric tiling. With his latest book, Shadows of the Mind, Penrose appears to have headed down a new wormhole, into the quantum world of biology-based consciousness. For a journey like this, Penrose decided, it’s good to have a doctor along.
“I take away and revive people’s consciousness every day,” says Hameroff. “So I’ve thought about this in a very practical, nonabstract way. A biological way. We know, for instance, that certain structures in the brain physically change in the presence of anesthesia. That is, they stop moving when consciousness slips away. Wouldn’t it follow that those same structures, and their movements, are tied to consciousness? Maybe they’re the root of consciousness. I say they are.”
The physical structures that Hameroff refers to are protein polymer tubes called microtubules, and amazingly, though they are thoroughly ubiquitous structures, appearing in every cell of our body, they were incognito until 1970. It seems we had been inadvertently dissolving them with a fixative (osmium tetroxide) used to prepare specimens for the electron microscope. (Don’t you wonder what else we may be dissolving?) Once we realized how to prepare cells without destroying microtubules, we began to see them everywhere we looked, and it dawned on us how important they are.
Cells are not the droopy “bags of watery enzymes” that scientists once imagined. They are given their shape by the cytoskeleton—a Tinkertoy scaffolding of protein tubes and connectors that organize the interiors of all living cells. The protein tubes in this cytoskeleton are called microtubules, cylindrical fibers that can be anywhere from tens of nanometers long during early assembly to meters long in the nerve axons of large animals.
Microtubules are one of those examples of nature’s geometric mantra repeated over and over. The building blocks of the microtubule are proteins called tubulin. Two varieties of tubulin, alpha and beta, self-assemble into dimers, which self-assemble end to end into long protein chains. These strands always group together in bundles of thirteen, forming a hollow cylinder made of protein. The cylinder’s strands are twisted clockwise like twine in a rope, so that when the microtubules are viewed in cross section, they look like a child’s pinwheel.
Each cylinder sports protrusions along its length called microtubule associated proteins, or MAPS. Some MAPS are bridges connecting the tubules to one another, forming the 3-D lattice that gives the cell its shape. Other MAPS, such as dynein and kinesin, are sidearm proteins (contractile spurs) that can extend and contract. Moving like the legs of a centipede, they act in a coordinated way to pass cytoplasm (cell fluid) along the tubule in bucket-brigade style, or to move organelles from one part of the cell to another. The cell’s workers—chromosomes, nuclei, mitochondria, neurotransmitter synaptic vesicles, liposomes, phagosomes, granules, ribosomes, and the like—all ride the microtubule conveyor belt, meaning that microtubules are in on every just about every important cellular function you can think of.
Including reproduction. Remember those spindles forming and disappearing inside dividing cells in high school biology filmstrips? (I’m dating myself.) Those were microtu
bules helping to pull apart the doubling sets of chromosomes so that one cell could become two. Microtubules are also at work in cilia, the ubiquitous hairlike filaments that bacteria used to row themselves around your microscope slide. Cilia also line our mucous passages, and with the help of microtubules, they push materials up and down our body’s smallest corridors. It’s not an exaggeration to say that without microtubules, we wouldn’t be able to sense the world, swallow, grow, or, says Hameroff, remember our names.
That’s because brain cells are also full of these microtubular nets. Here they are not only conveyor belt and scaffolding, but also the builders and regulators of synaptic connections called dendritic spines. (The same spines that Donald O. Hebb said are responsible for opening a “dialogue” between two neurons so that learning can occur.) Microtubule assemblies are also present along the entire length of the spindly axon, and their branches are plugged directly into the neuron’s all-important membrane and into organelles such as the soccer-ball-shaped clathrins at the end of the axon. These clathrins control the release of neurotransmitter chemicals, which swim across the synapse, delivering the neuron’s signals. (In this last function, the microtubule has its finger in the very important pie of thought and feeling.)
Biomimicry Page 28