In a small, darkened room tucked away in the imposing Noyes chemistry building on the sunny Caltech campus, a research assistant mixes two liquids in a beaker. Suddenly there is light. It shines steadily from the mixture, bathing expectant faces in an eerie pink glow. Light also shines in another room across the hall, from tiny vials set before spectrographs that record every flicker and flare. Almost everywhere in Jacqueline Barton's chemistry lab, a neonlike glow emanates from the glassware.
The glow comes from small molecules Barton has invented. She makes them from the exotic metals ruthenium and rhodium and mates them with DNA--the huge molecule that carries an organism's genetic information. While floating about in solution, her molecules probe crevices and bumps on the DNA's surface. When they find the right spot they combine with the DNA and begin to glow. The pink light they give off has a garish hue worthy of Las Vegas.
Barton likes the Day-Glo effect, and she designs her molecules to produce it. When doing experiments, she keeps careful track of the light's intensity because it shows how well her chemistry is working. But the pink glow bears another message. It signals the birth of a new technology: tailor-made molecule-size tools for manipulating genes.
Over the past decade, Barton has designed scores of such molecules. Some are remarkably selective. They recognize bits of genetic code embedded in a string of DNA and flock to those targets with surgical precision, a talent that makes them look promising as drugs. Someday, for example, they might find and disable mutated, cancer-causing genes. In diagnostic tests, they might home in on alien bits of code inserted by a virus into human DNA--a glow would signal infection. Other molecules in Barton's collection are tiny, light-activated scalpels--they absorb light instead of emitting it and use its energy to cut the bonds that hold the DNA strand together.
Gene therapy is still new, and it's too soon to say just how Barton's molecules will be used. They might block defective genes in living cells, for example. Small and precise, they might also work like tugboats, guiding supertanker viruses that transport remedial genes to the proper place on the DNA. For now, Barton is still experimenting, still finding new things her molecules can do.
Her work is shaped by two formidable challenges. "The first conceptual hurdle," she says, "is how do you recognize a specific spot?" How does a molecule "know" where to latch onto DNA? "Then you've got to decide what to do when you get there." All the glory, you might think, lies in the second step: doing things to DNA. But it's not as easy as sending in the Marines to kick genes. Finding the right spot to attack is equally important, and it's probably the more difficult task.
Just think of the enormity of the search: if you extracted all the DNA from one of your cells and joined the pieces end to end, you would have an invisibly thin hair of a molecule three feet long. It would consist of two spiraling strands held together by cross-links, as the sides of a ladder are held together by rungs. The rungs that connect the two strands are made of chemical bases called A, C, G, or T, after the first letters of their chemical names: adenine, cytosine, guanine, thymine. There are 3 billion of these rungs in three feet of DNA.
If you think of those rungs as bits of code--CTAGGATC, and so on- -then there are 3 billion bits in your genetic library. How long would it take you to find a particular eight-base-long set of instructions? Imagine enlarging the DNA just enough to make each ladder rung visible: make the spacing between rungs a sixteenth of an inch. On that scale, your DNA would stretch across the continent--3,000 miles. To find a particular sequence of eight bases, you might need to examine the whole molecule, bit by bit, from coast to coast. You couldn't do it in a lifetime. Even if you assembled a small army to help you look--as in the cell, where many identical proteins search the DNA strand simultaneously--you'd have a difficult time of it.
Real molecules inside cells somehow do it routinely, however. Specialized proteins recognize particular segments of DNA, latch on, and read the encoded instructions to make whatever molecules the body needs. Other molecules then duplicate the DNA bases in that segment and transport the copies to factories inside a cell, where the bases become templates for proteins. Still other molecules can attach to DNA and prevent these processes from happening. Everything depends on molecules combining with DNA at exactly the right spot.
Barton hopes to learn the strategies of natural DNA-reading proteins and mimic them. "What we're trying to do," she says, "is tackle the question of how, chemically, the information that we talk about as being stored in DNA--how is that information read?"
Real proteins don't read, of course. Nor is DNA anything like the kind of text we're familiar with. We shouldn't even be thinking about DNA as if it were a linear code, Barton says. "It has three-dimensional form. And so we should be sculpting molecules to have shapes that complement the shapes of DNA." Rather than scan the DNA strand from end to end, protein molecules in the body dance around in three-dimensional space, jostling and groping. A small protein molecule "reads" DNA by nudging against the big molecule's flanks; it "recognizes" code, researchers think, by being able to snuggle up close when its shape fits into the grooves it finds.
"If we think about the spiral staircase of DNA," Barton says, we find that "the steps aren't always flat." Depending on the sequence of bases, "some of them twist up, others twist down. They're a little bit open, a little bit twisted, a little bit tilted." DNA strands twist and writhe to form hairpin loops, left-handed as opposed to right-handed helices, and supercoils (extra twists superimposed on the basic double helix). Stressed segments of DNA pop out of the helix as single strands. There are also triple strands and cruciforms. "So you could imagine going along in a Braille-like fashion and finding the variations in shape," Barton explains.
To design a molecule that locks onto one part of the DNA ladder, Barton needs to know just what that target shape is. To find out, she must isolate and study short stretches of DNA. This used to be a catch-22: you needed to recognize code in order to isolate it so that you could study recognition. Barton has a simple way out. In the hallway outside her lab a bland white machine the size of a photocopier churns out synthetic DNA. Attached to one side of the machine are four liquid-filled tubes labeled A, C, G, and T, which feed in the chemical bases. Perfectly predictable DNA strands come out, around 30 bases long, in any desired sequence.
"In the past 15 years," Barton notes--almost gloats, really-- "there's been a revolution in our ability to make DNA." Before then, biologists had to isolate it from a soup that they squeezed out of cells. Its lengths and sequences were all different. Today, she says, "you know exactly what it is, exactly what atoms are in it, and what bonds connect what to what. I can make these molecules, look at their structures, analyze and manipulate them. We can do that on tiny bits of material, use all sorts of new techniques. It's not biology; it's really chemistry."
You get the feeling that Barton has little use for biology. "What we are is chemicals," she flatly asserts. "DNA is a molecule, and that molecule defines everything that you are. The biological world has now become the realm of chemists."
At 42, Barton remembers the bad old days when big organic molecules like DNA defied close scrutiny. As an undergraduate she was suitably modest: she concentrated on inorganic chemistry. The focus there is mostly on metals, mostly on small molecules. But a wonderful discovery diverted her career.
In the mid-1960s Barnett Rosenberg, a biophysicist at Michigan State University, noticed that bacteria he was culturing in his lab were growing into strange shapes--a strong hint that their DNA was being damaged. He wasn't sure what was causing the distorted growth, so he began looking for contaminants in his growth medium. Eventually he learned that platinum was leaching from an electrical component hooked up to the medium and reacting with ammonium chloride to form cis-dichlorodiammineplatinum (cis-platinum, for short). If this molecule had such a strong effect on those rapidly growing cells, he wondered, maybe it could also disable rapidly growing cancer cells in tumors. It turned out to be a powerful anticancer drug, and today cis-platinum is routinely used to treat testicular and ovarian cancer.
In 1975 Stephen Lippard, then a chemist at Columbia, wanted to know why cis-platinum had such a strong effect. Barton was just entering graduate school. "She was clearly a very bright student," he says, "quite strong mathematically. Pure algebra, quantum mechanics--she was really good at that." She planned a career in physical chemistry, and she would need those skills to analyze, say, how electrons in metals absorb or give off light. Barton walked into Lippard's office one day, and they began to talk about what he was doing. Lippard convinced her she could do physical chemistry on cis-platinum, and she signed on with him as her Ph.D. adviser.
Cis-platinum was thought to react in the cell with uracil--one of the building blocks of RNA and a close relative of thymine, a DNA building block. Lippard asked Barton to determine the structure of the cis-platinum- uracil compound, known as "platinum blue" because of its color. It was no easy task. First she had to produce a pure crystal large enough to study; then she had to figure out the location of atoms in the crystal by analyzing the patterns produced by X-rays shot through the crystal. The bottom line, Lippard says, is that she succeeded. "It was a major breakthrough."
In a sense, the discovery of the structure of platinum blue was similar to the discovery of the structure of DNA. The ingredients of these molecules were well known. But no one knew how the atoms in the molecule were actually put together. And without knowing the three-dimensional form, no one could understand how the molecule worked. As it turned out, Barton found that the molecule contains several exposed platinum atoms, which Lippard and other researchers guessed could combine with bases in the DNA of living cells--and probably at a gene responsible for cell division. That would explain cis-platinum's strong effect on rapidly dividing cells: it poisons them by crippling a gene they must use to replicate. Once researchers knew the structure of platinum blue, they could try all kinds of variations. So Barton's discovery opened up a whole new field of anticancer compounds.
Barton was now hooked on metallo-organic compounds, intrigued by their interactions with DNA. She branched out into other metals, such as cobalt. A burst of achievement and early recognition--a Waterman award for outstanding young scientists, a MacArthur genius award, and many others-- propelled Barton from graduate work at Columbia to a full professorship at Caltech. There she got funding, she says, to study the general principles behind "applied things that might be particularly amenable to industrial applications," such as synthetic drugs or diagnostic tools. In particular, she wanted to "rationally think about tailor-making molecules from scratch."
At this early stage in her career, Barton was taking a huge gamble. It wasn't obvious at the time how important metals are in the chemistry of life. They barely register as trace ingredients; the human body contains, for example, less than a hundredth of an ounce of zinc. Yet nature seems fond of metals: they make up three-fourths of the elements in the periodic table. It was natural for Barton, with her background in inorganic chemistry, to favor them, too.
The way Barton saw it, metals would make ideal building blocks for her molecules. Metallic atoms have gaps in their outer electron shells that welcome other electrons. The platinum in cis-platinum, for example, readily receives electrons from ammonia and chloride. The three atoms then effectively "share" these electrons, and the sharing holds the cis-platinum molecule together. Similarly, the metal atoms in Barton's molecules bond tightly to whatever pieces she decides to add.
Today Barton makes her molecules from exotic metals not naturally found in living organisms--rhodium and ruthenium. The atoms of these metallic elements are large, with six potential bonding sites; you could imagine them as fat round Legos with six little snap-in pegs. She starts with a single metallic atom and bulks it up with atoms typically found in proteins--rings of carbon and nitrogen atoms, for instance. These extensions give her molecules the shapes they need to latch onto particular bits of DNA.
The first things she bonded to her metallic atoms were flat, nonmetallic "blades," fashioned from hydrogen, carbon, and nitrogen, that were shaped like stop signs and that made her molecules resemble propellers. Like a ship's propeller, the molecules even had a twist, matching the twist in the double helix. "Two simple metal complexes," Barton says, "are the first ones we studied. They have three propeller blades around them. What we imagine is that one of the blades of the propeller can partially insert between the base pairs of DNA." Technically speaking, the blade can intercalate--that is, get its tip between the rungs of the DNA ladder.
Barton got the propeller-blade idea from work done by Karen Winterhahn, a chemist now at Dartmouth who also trained with Lippard. Winterhahn discovered that flat platinum compounds--single metallic blades- -could slip between the ladder rungs of DNA. Winterhahn's blades were neat and flat; the ones Barton tried were three-dimensional and gnarly with hydrogens. "I was sure that this bulky little molecule that had all these different blades sticking out couldn't possibly bind to anything," Barton recalls. "It was just going to sit there. But then I saw that it actually did." Other intercalating compounds had simple flat blades that could slip into DNA anywhere, but they had no specificity. Barton's three-dimensional molecules could be used to tell one step on the DNA ladder from another.
Barton's designs from scratch were working, but she also sometimes got lucky. One day, for example, an assistant noticed that in the presence of ultraviolet light, a ruthenium molecule would glow when it slipped a blade into DNA. This was an extra-long propeller blade--Barton called it a tongue. An electron in the tongue released the absorbed ultraviolet light energy as visible pink light. Normally, such a tongued propeller blade wouldn't glow when surrounded by water; chemical bonds in water molecules vibrate in tune with bonds in the tongue and dissipate the absorbed energy as heat. But lodged in the rungs of the DNA ladder, the tongue can't vibrate in the same way, and sympathetic vibrations are quenched. "So it doesn't glow at all in water," Barton says, unless you add some DNA that it can recognize. Then the tongue goes in where it's protected from water, "and it glows a whole lot. This is our molecular light switch."
The light switch gave Barton a convenient way to measure how well her molecules were finding their targets. But they still weren't good enough. When mixed with DNA at low concentrations, their shapes weren't recognizing code as well as natural proteins could.
At this point Barton would have welcomed another lucky breakthrough, but instead, she says, she got small, incremental steps. For example, she started building "very fancy molecules that would have all sorts of little groups on them for what are called hydrogen bonding interactions, essentially points of glue that might be useful in binding to DNA. From the very beginning, in the sixties and seventies, people always thought that putting hydrogen bonding groups on molecules would be important for recognition." Natural proteins, everyone thought, have hydrogen bonding groups, so they assumed that synthetic molecules should have them, too.
In hydrogen bonding, a hydrogen atom in a molecule loses its lone electron to some other atom in the same molecule: a nitrogen, say, in Barton's propeller blade yanks out a nearby hydrogen's electron, leaving that part of the blade positively charged. (A hydrogen atom without its electron is simply a positively charged proton.) DNA as a whole is negatively charged, so the positively charged protons on a propeller blade are bonding points--strong bits of glue that stick to the DNA.
So much for theory. "We kept on making molecules that didn't have any hydrogen bonding groups on them," Barton says, "and they were quite specific"--that is, they recognized their targets precisely and stuck well to DNA. And for a while, Barton kept setting those promising molecules aside. "We kept on going in the lab and making molecules that had a hydrogen bonding group here, a hydrogen bonding group there"--and those theoretically correct molecules didn't work as well.
"Even though we got quite sophisticated in terms of where we put the glue on the molecule," Barton continues, "what ended up being most important was the shape. Try as we might and get fancy as we might, it was really the shape of the molecule that determined its recognition characteristics with DNA. That was a lesson we've learned and then relearned and then relearned again."
When Barton's molecules have shapes that match their targets precisely, they get extremely close to DNA. The slightly positive atoms on her molecules then attract negative atoms on DNA. The close fit between her molecules and DNA makes these weak attractions stronger, in effect, than hydrogen bonds, just as the moon's closeness to Earth makes its weak gravity more powerful on Earth than the sun's gravity. (The moon's pull on Earth reveals its strength as twice-daily tides; the sun's contribution to our tides is comparatively negligible.)
To refine the shape of her molecules, Barton now adds small methyl groups. These are as simple as organic compounds can be: one carbon with two or three hydrogens attached. They don't form hydrogen bonds, but methyls in strategic places--on the propeller blades, for example--let Barton's molecules snuggle right up against specific sequences of DNA bases.
Barton is now convinced that having many weak bonds, rather than one strong bond, is the answer to the first mystery she set out to solve: How do molecules "read" DNA? It makes sense that nature would have hit on this strategy eons ago. If any passing big lug of a protein--with a single strong site for hydrogen bonding--could stick tightly to DNA, the works would quickly gum up. No code would ever be read, and no proteins ever constructed. On the other hand, if only intricate, interlocking shapes can combine with DNA, its code is well protected from random molecular jostling.
Assuming Barton can design molecules that latch onto DNA at just the right places, the next question, she says, is: What do you do when you get there? If you want to block genes from being read, as cis-platinum seems to do, "do you bind at the beginning of the spot that you're interested in, or the middle?" If you want to disable some bad DNA, just how do you do it? "There's repair machinery in the cell," she notes, and the cell might rebuild the defective stretch. "There are lots of issues," Barton says, not the least of which is whether your designer molecule goes after targets other than the intended one.
Barton can't research all the issues herself, but she's aware of them. Right now she focuses on bombing--breaking DNA's backbone at the spot her molecules recognize. It's something she knows how to do with her rhodium compounds, and it could be useful. "One potential application of these things," Barton says, "would be artificial restriction enzymes-- molecules that recognize a certain site and cut there."
Unfortunately, she admits, "the chemistry that we're doing now is much less subtle than what nature does. We literally shine a light and have a bomb go off, then pick up the pieces." That's too clumsy for genetic engineering, but good enough to help Barton with her own research. She can tell where her molecules latched onto DNA from the fragments produced after bombing.
Now that her molecules can hit their targets, it would be nice to know which targets to pick. "The Human Genome Project is important," Barton says, "because we need to know the road map. There are 100,000-odd genes in the human genome, and we don't know where most of them are or what's responsible for what.
"But science does go so rapidly," she reflects. A few months after Barton was born, Watson and Crick proposed that DNA was a double helix. "The discovery of the double helix is only 40 years old. That's my lifetime. So it's hard to conceive of what will be in a lifetime from now."
There are strong hints, though, across the hall from Barton's synthetic-DNA maker, where she has begun a new stage of her work. She wants to find out if her best molecules can tamper with real genes inside living cells, as opposed to synthetic DNA in a test tube. So she has been feeding her metal complexes to bacteria cells growing in petri dishes. From the very first try, it was clear that the cells didn't like their doctored DNA. Within a few hours they started to die. "There's no question," says Barton, "the complexes don't make the cells very happy." But they didn't die off as readily as she would have liked. And she suspects the problem is the cell's ability to repair itself.
If you're doing chemotherapy or gene therapy, you want the genes you alter to stay altered. You want to make permanent changes in the cells' DNA. Yet our cells have evolved efficient mechanisms to repair the million or so bits of damaged DNA that accumulate each day in the human body.
That's the problem Lippard, now at MIT, is solving with cis- platinum--which he believes he's finally beginning to understand, after nearly two decades of intense research. Lippard found a group of human proteins that happen to have the right shape to attach to cis-platinum and shield it from a cell's repairing molecules. With these proteins in hand, he's set his sights on inoperable brain tumors. "What we'd like to do," he says, "is go in with a virus that carries the gene for the proteins that protect the cis-platinum." The virus prefers to infect rapidly dividing cells, and the extra gene would amplify the amount of this shielding protein that cancerous brain cells produce. "And then we go in with a combination of gene therapy and chemotherapy," Lippard says, "to try to knock out the cancer."
Brain tumors would be the ideal target because normal brain cells multiply very slowly. A combination of therapies that do the most damage to rapidly growing cells would be less likely to harm normal brain cells.
Barton hasn't pushed so relentlessly to develop any one molecule like cis-platinum. She has gone after broader, more basic knowledge instead. In fact, her most recent piece of research has taken her in a rather different direction. And it's already become something of a classic. Last year Barton and colleagues at Columbia and Caltech electrified the field of DNA research when they attached two of her propeller molecules to opposite ends of a DNA double helix. One molecule was made from ruthenium and the other from rhodium, two different metals like the copper and zinc in some batteries. When Barton attached these molecules simultaneously to DNA and shone a light on them, electricity flowed between them, zipping through the DNA like current through a wire.
The ruthenium molecule glowed when attached alone to the DNA, but it stopped glowing when the rhodium molecule was added to the other end. Contrary to everyday experience with electric currents, completing the circuit in this case caused the glow to disappear because the rhodium drained electrons from the ruthenium, much as an appliance that drains a lot of current can dim a light bulb. The rhodium, in effect, created a short circuit.
It was an elegant experiment with surprising results. First, Barton demonstrated that DNA is indeed an electrical conductor. "Since DNA was first proposed to be a double helix, in 1953," she explains, researchers have tried to nail down its electrical behavior. "People in the late sixties, I think, had a little pellet of DNA, put leads across it, and said DNA is actually a semiconductor." Barton's exotic molecules, attaching precisely to the double helix, allowed her to perform the same test on individual DNA molecules, and that's what made her experiment definitive. Electrons traveled through the DNA "in an incredibly short time," she says. It's a surprisingly good conductor.
"This wire characteristic," she adds, "ends up being a terrific new biosensor." Just as electricians send signals through a circuit to diagnose trouble, biologists might use Barton's molecules as electrical probes to test any part of a DNA strand. Breaks in the helix might show up as a glow. Missing bits of code, extra bits, or other abnormalities might show up as spikes or dips in the flow of current.
Barton calls DNA's conductivity a "crazy and exciting" discovery. Given her recent history, it won't be surprising if she produces a string of such surprises. In time, her marriage of metals and DNA will undoubtedly add a great number of useful molecular inventions to the toolbox of genetic engineers.