Boger needed no lectures, least of all from Schreiber. For most of the past year he had concentrated on finding a corporate partner. He had sold Vertex’s story in a down market by projecting confidently what the company could and would do. Now, the goals were scientific and more stringent by nature. Immunophilins research had reached a steady boil. Besides Merck, Glaxo and Sandoz, the obvious front-runners, nearly every major U.S. and European drug company now had a program in the area, and dozens of top academic scientists from every field were stampeding either to collaborate or compete with Schreiber, Starzl, and the other early leaders. With a perspicacious Chugai closely marking the performance of its newest U.S. asset, and show-me pharmaceutical and financial critics fervently waiting for Boger to deliver—or better yet, as some at Merck hoped, fail—he needed to do competitive science. Vertex needed to make a drug.

All of the company’s senior scientists including Boger had worked on promising leads before, but not one of the molecules had, strictly speaking, become a drug. “A compound may be brilliantly designed—everything absolutely rational,” explains a former Merck vice president, showing something of the company’s bias toward pills, “but until that compound has been shown not only to do clinically what you want it to do, but to be safe, to be active orally, to stay around in the body, and not to give you nightmares, it’s not a drug.” New drugs are exceedingly rare; novel ones still rarer. Of the hundreds of thousands of new compounds tested each year by the pharmaceutical industry, only about thirty are eventually approved by the FDA. Only three or four of those, like cyclosporine, actually do something new by working through a new pathway or (like FK-506, which was still not approved) suggest new uses because of their potency. Most others are variants of existing molecules. Like sperm cells, which are launched with great fanfare by the hundreds of millions and of which only one or two, if any, reach their goal, potential drug molecules may be expected to swim mightily for a while, but that’s it. That no one at Vertex had ever “had a drug” wasn’t especially alarming—many people in the industry work their entire careers without one—though it worried some of the scientists, who wondered if they weren’t being overly presumptuous.

Boger, of course, not only intended to prove that Vertex could make drugs but drugs that were in all ways superior to those of the world’s pharmaceutical leaders. Besides working exquisitely, Vertex’s molecules were to be prototypes for a more rational system of drug discovery and a death knell for screening. The competition with Schreiber was thus for him neither the big picture nor the big prize, however much it might rankle him. Boger had set himself not against an icon like Woodward but against the best-run, most admired scientific company in the world. When Boger had left Merck twenty-two months earlier, it had perhaps forty scientists working on FK-506. In time, that number would probably reach one hundred. That, far more than Schreiber, impelled him now.

A year after the labs had opened, Vertex’s science by mid-fall 1990 was operating almost in full. There were groups making molecules and tearing molecules apart. There were researchers synthesizing genes and screening gene libraries, extracting proteins, growing them, crystallizing them, chasing their structures, simulating hypothetical structures on computers. There were people modeling molecular interactions and writing new software to try to design drug molecules from scratch. Each week the company tested dozens of new compounds to see how they bound to FKBP, whether they inhibited protein folding and whether they stopped, through any of a half-dozen mechanisms, T cells from proliferating in test tubes. There was an animal pharmacology lab where promising molecules were pumped into mice with skin grafts on their foot pads, a simple model for testing rejection.

Drug development is an iterative process, a backbreaking multiyear series of assays aimed at funneling molecules upward through an evolutionary gauntlet. At the lowest level, molecules are tested for chemical affinity to a target, usually a protein. Those that are successful are examined for biochemical activity. If they’re active—if they change the functioning of a target—they go into cells. Those that somehow affect cells without simply killing them in a toxic onslaught are tested in mice, then rats, then rabbits, and so on, through dogs and primates and up to humans. Each compound’s progress is adjudged, especially in the later stages, by its “therapeutic index”—a kind of cost-benefit review whereby medicinal rewards are weighed against toxicity. Eventually, Boger planned for Vertex to do all such testing itself. Now, however, the company was focused entirely on discovery. It was looking for new molecules that showed the potential—with perhaps a decade of testing, $200 million, endless tweaking, and the ever-present threat that it all might simply explode—to become a drug.

In August, Boger had promised Chugai that Vertex would have its own compounds that were as potent as cyclosporine in cell assays by the end of the year. It was an exhorbitant leap—a hundredfold increase in activity from Vertex’s best molecules—and Boger’s pledge made the chemists, whose job it was to make the compounds, shudder with misgiving. “We’ve got until Christmas to do what Sandoz hasn’t been able to do in twelve years,” Dave Armistead grumbled.

Armistead was the lead chemist on the project, a surprisingly thoughtful, gravel-voiced Virginian who had come to Vertex via Yale and Merck and whose soft-spokenness belied a macho sensibility toward science. Armistead likes to synthesize “ball-buster molecules.” At age thirty-four, he power-lifts several nights a week after work, a discipline that has paid off with a bulging chest and arms like dock ropes. With piercing blue eyes, high ruddy cheekbones, and a spike of brown hair, he combines the feral, imperturbable air of a Native American brave with the rowdy menace of a Brian Bosworth, the football player. And yet though generally fearless about work and as much a natural chemist perhaps as Schreiber, he found Boger’s arrogant predictions disturbing. Making new molecules is one thing; picking up two orders of magnitude of biological activity in four months is another. Armistead knew what Boger was asking, and he didn’t share Boger’s confidence that it could be done.

As de facto head of Vertex’s chemistry effort, Armistead objected strongly to what he saw as two equally disturbing precedents. He thought that by setting expectations too high Boger was inviting Chugai to exercise an escape clause in its contract, a clause that tied its payments to Vertex’s making “adequate” scientific progress. Why incite disappointment? Armistead wondered. He was also discouraged for philosophical reasons about the demands of corporate oversight. “Things are going to change now that we have someone to answer to,” he said. “It’s going to be a lot more like working for Merck. We’re going to be making reports, and reports need results.” Like Boger, Armistead believed that research was distorted when scientists were made to account for their efforts. “People at big companies are of the mind-set that science is just turning the crank,” he says. “They want to be able to stand up at the end of the quarter and say, ‘We’ve turned out these 200 compounds, and though they’re all dead, we worked hard.’ They’re interested in generating data points. That’s what’s going to earn them a good rating.”

But Armistead was far more troubled by the scientific issues. Despite Boger’s statement to reporters on the day of the Chugai signing that Vertex “likes to go into projects where we are very sure of the biochemistry,” the company was still, eighteen months into the immunophilins program, sure of disturbingly little. The scientists still didn’t know how FK-506 worked and whether FKBP was its relevant target within the body. That no one else did either was scant consolation, as no one else claimed to need the information as badly or made such a point of having it. Immunosuppression was still a mystery with very few molecular clues. And though scientists at Vertex and elsewhere were trying to solve it, the chemists had nothing but their own hit-or-miss assay results to use in coming up with new compounds. They had no structural information to draw against: no lock, just other keys. Worse, they weren’t sure what they were trying to make. They didn’t know, for instance, whether blocking the protein folding action of FKPB had anything at all to do with suppressing the immune system or whether FK-506-like molecules weren’t simply all toxic.

Such uncertainies were dark clouds fulminating on Vertex’s horizon; indeed, one had already begun to break loose. There was now a gathering body of evidence that the protein folding hypothesis that Boger and everyone else had seized upon eighteen months earlier was incorrect, a red herring. Schreiber had raised the first public doubts a year earlier with his 506BD, which showed that, by itself, the binding portion of FK-506 didn’t block the enzyme from working: In cells, 506BD was a dead compound. He postulated that the business-end of FK-506 must therefore be the portion that sticks out into space—the “effector domain,” he called it. The implication was that those atoms mingled with something else, another “partner” protein, and that that was what accounted for the drug’s activity. Boger disputed Schreiber’s conclusion; he thought it simplistic. Indeed, Vertex’s own assay results had raised a similar specter sometime earlier. By the time of the Chugai signing, the company had made several molecules that were substantially smaller than FK-506 and inhibited the enzyme activity of FKBP just as well but that were feeble immunosuppressants. Either the molecules weren’t getting to their targets, or they were getting there but were irrelevant. In any case, they were far from being drugs. “We have more inherent potency than we have potency in cells,” Boger warned the scientists. “We’re at the point now where I would encourage chemistry to start being very careful of the questions it asks.”

Addressing those questions without benefit of knowing which atoms on the molecule bound to FKBP or even whether inhibiting the enzyme was a meaningful goal left Armistead and the other chemists to practice the very science they abhorred: raw medicinal chemistry. In other words, trial and error: Make a compound, assay it, adjust it subtly to make another molecule, test that second molecule, compare how they behave, make a third molecule that maximizes the best features of the first two, test that, and so on. To the chemists, few results could have been more frustrating—or embarrassing. Not only did they not have the information that they needed to design drugs and that Boger had promised them, they were having to compete against Merck and the other large companies in exactly that area where the traditional drugmakers were most dominant: their ability to generate and test thousands of compounds. If medicinal chemistry was essentially a prerational, “monkeys with typewriters” approach, as Boger liked to say, the chances of producing a Macbeth strongly favored the firm with the most, and most experienced, monkeys. Vertex had five chemists on the project, with a couple of decades of pharmaceutical experience and not a drug among them.

Armistead and the chemists wanted intensely to be careful about the kinds of questions they asked, but in truth they had no choice. Until Navia and Yamashita solved the structure of FKBP and could show them how their molecules bound to it, and until the biologists, mainly Harding and his group, identified the true molecular target, they were flying blind. Contrary to Boger’s drumbeating, they weren’t doing structure-based design. They were doing its opposite, perilously outflanked and outnumbered.

This, Armistead believed, was the hoariest feature of Boger’s pledge to Chugai. Armistead admired Boger greatly. He hadn’t lost faith in him. But he was a realist, and Boger notwithstanding, Vertex’s chemistry effort was up against a sheer, potentially ruinous, and deeply humiliating wall. “Nobody’s going to buy an inhibitor of FKBP,” he said. “We’ve got to figure out what else you have to do that will throw the final switch and light up the cell assays. That’s what FK-506 and cyclosporine do.

“If we don’t have that, we’re just jerking off.”

•  •  •

What they had, what they’d always had, was the structure of FK-506.

Born of a soil fungus, isolated from a beery black broth, purified and deposited in a culture bank in Japan where it was sealed away from competitors, synthesized in a pair of internecine tong wars, deconstructed by X-ray crystallographers, then suddenly soaring to become best supporting actor in Science’s 1989 Molecule of the Year issue, the compound was self-championing. Whatever quarrels there were about the importance of FKBP, there was none about FK-506. It was a drug, and it was a drug because of its architecture, its shape.

From the moment Armistead saw the structure of FK-506 three years before, he knew he wanted to synthesize it. It was exactly the kind of “big, sexy, macho molecule” that quickens the blood of synthetic organic chemists and that he had just had his first major success with as a Yale postdoc. That was October 1987, near the time of the crash on Wall Street, and Armistead had just been hired by Boger to join his nascent structure-based design group at Merck. Boger had little interest in the race to be first to make the molecule, but to redesign it he needed to know how it was assembled. Armistead leapt at the opportunity: “You’ve got some guy that flashes you this very provocative molecule and you know this isn’t something you’re going to put together in two to three weeks. This is going to be a major challenge. You knew you were going to get press for it.”

There are various approaches for reproducing natural compounds, but the one that appealed most to Boger and Armistead, for different reasons, was a “convergent” synthesis. Says Armistead: “Boger thought—and I still believe this—that the inherent problems with toxicity and bioavailability [how efficiently a molecule performs within the body] with FK-506 were part of the structure of the molecule. Those problems aren’t going to be solved, I don’t think, by running a chemical reaction on FK-506 and taking that product and making a drug out of it. What he wanted to do—this was prior to the identification of FKBP, so there was no receptor to study—was a synthesis made of four major building blocks, so that you could take these four pieces and learn how to glue them together. Once you did that, you could take one of these pieces, make drastic changes to it, glue it all back together again, and see what came out of the process. The idea was to have a synthesis that literally converged on the total synthesis: We would build the components, glue them together, and have FK-506.”

To Armistead, who would be trying to make the molecule, a late-stage convergence was preferable for another reason. “From an efficiency standpoint, if you’ve got sixty steps, and you try to do them in a row, you’re never going to get there. A linear synthesis wouldn’t work. You needed a convergent synthesis. Coming together as late as possible was the key.”

From the outset, Boger’s plans ran counter to both Merck’s culture and Merck’s needs. His charter, supported by the company’s top executives, was to develop a new way to discover drugs, but Merck’s success was based on generating large numbers of new compounds. A lengthy, uncertain, total synthesis meant that Merck would be left to sit idly in a prime area where the stakes were likely to run into billions of dollars and where its competitors were all charging ahead. “Middle management hated it. Hated it!” Armistead says. “It was not a speedy way to make analogs. As soon as Boger left, they chopped the nuts off of this and started doing analog work.”

As Boger found out, there was another issue: His group wasn’t the only one at Merck trying to make FK-506. Within the drug industry, there are two rival groups of chemists: medicinal chemists, who try to invent new drugs, and process chemists, who work on devising new, cheaper ways of producing them. Max Tishler, patron saint of process chemists, elevated the art tremendously, but at Merck and elsewhere, process chemists still struggle for respect. Their work is thought to be dull, unimaginative, unglamorous. Merck’s process chemistry group, seeing in FK-506 the same opportunities that Armistead did, also had launched a major effort to synthesize it. A similar duel was meanwhile developing at Yale, where Schreiber and Sam Danishevsky, his department chairman (in whose lab Armistead did his postdoc), were also competing to make the molecule. Typically, if not surprisingly, the intramural competition at Merck was just as secretive and ruthless as Merck’s competition with the Yale groups and others outside the firm. “They wouldn’t tell us anything,” Armistead recalls. “I’m over here banging my head against the wall, trying to solve a problem that’s been solved in the building next door, and they’re getting the same paycheck that I am. It was an insanely competitive situation.”

Merck’s process group narrowly won the race to synthesize FK-506, beating Armistead and his collaborators, who immediately dropped the project. Completed in the fall of 1988, the group’s structure attracted international attention and dealt Schreiber, who eventually came in second, a small but significant blow. Then, in December, Boger announced he was leaving the company. Armistead, “very, very disappointed,” was also shocked. He’d never heard of anyone quitting a major drug company to start his own firm, much less a fast-tracker like Boger. Historically, biomedical start-ups were launched by molecular biologists interested in producing proteins and other large biomolecules. They competed with each other in markets that were small and untested. But Boger was a chemist. He was starting a drug company, not a biotech company, which meant he would be competing not with other small companies but behemoths like Merck and Glaxo, with their stables of billion-dollar drugs. Armistead echoed the skepticism that prevailed at Merck at the time. “You’re going to go out and start a drug company,” he said. “I thought it was ridiculous.”

But Armistead was intrigued enough to want to hear more—an attitude, he recalls, not shared by others in Boger’s group. “In a big bureaucratic situation, there’s a lot of rewards just for being there, for tenure,” Armistead says. “Boger rewarded for performance. A lot of people thought his leaving was a big opportunity for them. He gathered us all together to tell us he was leaving and I remember one person saying, ‘There is a God after all,’ and walking out.”

Now, at Vertex, a different sort of internal competition drove the FK-506 chemistry effort. When Armistead was weighing whether to leave Merck for Vertex, he enlisted his closest friend from graduate school, Jeff Saunders, then a chemist at Squibb. “I figured if I was going to jump off this diving board, I was going to take somebody with me,” he says. Saunders, the son and grandson of chemists, was born to the trade. His careful thinking and commitment to doing his own benchwork impressed Boger, who, taking him to dinner, instantly offered him a job. (Says Saunders: “I wasn’t unhappy until Boger told me I was.”) To Armistead and Saunders it seemed ideal: Ever since they were students, they’d discussed starting a business, a custom chemical company or, reflecting another shared passion, a wine store. This was better: It had all the upside—the opportunity to work together and get rich—with little of the risk. Saunders, keener, wirier, with a diamond stud earring, was quieter than Armistead, less blustery and intuitive, more modest, but with a similar air of freelance bravado. Like brothers, the two were inveterate rivals.

Armistead and Saunders’s friendly competition set the tone for Vertex’s chemistry program. Working across from each other at the bench and at adjoining hoods, they attacked neighboring sections of the molecule—Saunders, an unusual pairing of oxygen atoms on one side; Armistead, a sugarlike ring adjacent to it. They weren’t trying to design drugs, but to identify those portions of the molecule most involved in binding and inhibiting enzyme activity. They wanted to see how small a molecule they could engineer that would retain FK-506’s potency in those areas, while changing the structure enough to make it patentable. As with Boger’s original notion of a “scaffold redesign,” each hoped not just to make a better FK-506, but a new one, one that nature itself might have made if it had intended the molecule to block the action of FKBP inside the human body instead of performing some unexplained role in the life of a fungus.

Chemists “run reactions.” Because molecules are collections of atoms arranged according to physical principles that are well understood, knowing which reactions to run and how to run them define the chemist’s craft. Even as recently as Woodward’s time, a synthetic organic chemist often had to discern the optimal conditions and make the reagents himself before he (the overwhelming majority of organic chemists are men) attempted a new reaction. Now, however, after 150 years in which millions of facts have been amassed about hundreds of thousands of individual compounds, chemists tend to believe they can make a molecule into any shape they want from standardized parts. They’re more like builders than architects, drawing their ideas from catalogs the size of big city phone books.

As Vertex began making molecules, almost all its early successes belonged to Armistead, who quickly found a substitute for the sugarlike ring. Marveled Saunders: “It was probably a half hour in Aldrich [a well-known chemical supply book]: ‘What’s available? What can I buy? What will it do for me?’ ” Saunders, meanwhile, was becalmed. Nothing he did seemed to work. He was running as many reactions as Armistead but couldn’t make the molecules he wanted. In almost a year of twelve-hour days—of coming in on weekends when only he, Thomson, Yamashita, and Murcko were routinely in the labs—he submitted only four new compounds. None of them was successful.

“It bothered me a lot,” he recalled, “not only because of the image it was presenting, but I wasn’t producing. . . . It was annoying, and it got to be frustrating and embarrassing. It threw me off the pace. After six or seven months, I got so caught up with it that I was ready to drop it, but Boger said, ‘No, don’t drop this. This is worth doing.’ ”

Armistead also defended Saunders. They shared ideas, hung out together after work, drank, got into noisy arguments. In the lab, they razzed each other like mechanics at adjacent bays. But Armistead was no solace; he was cruising. In the company’s formative days, he was establishing himself as one of its leaders. To the extent that Vertex had a fast track or any track at all within the swirling atmosphere of Boger’s social experiment, Armistead was on it, traveling with Boger to Japan, representing chemistry at key meetings, speaking for the project. Saunders reacted painfully. “Dave’s one of the best chemists I’ve known,” he said. “Not the smartest. Not the best read. Not the easiest to get along with. But as far as being productive, there’s no question. Unfortunately, I’m compared to him.”

Saunders’s confidence was shaken and he had sharp pangs of envy over Armistead’s success, but the two never had a falling out as Saunders and Thomson had. They were rivals over status and position, but tied by an unspoken lingua familiaris. Armistead, bigger and stronger, seemed the competent and protective older sibling; Saunders, more sensitive, struggled erratically, quietly, trying to find himself in Armistead’s wake. Indeed, it may have been as much for Saunders as for himself that Armistead was now angered by the impositions of the Chugai deal: Vertex was about to become a less forgiving place, particularly in chemistry. He feared what impact the new production demands would have on his unsettled friend.

Saunders viewed Boger’s promise to have cyclosporinelike molecules in cells by the end of the year with an added dread. Not only did he, like Armistead, think it would be extremely hard to do, he felt he had the burden of a lost year to account for. He had barely made any contribution and now he was being expected to emerge suddenly from his slump and help make an almost mystical gain in activity.

And standing next to him, shoulder to shoulder, overshadowing him, was the ubiquitous, conspicuously more muscular, and occasionally goading specter of his closest friend.

“We’re a lot alike,” Saunders said of himself and Armistead one night after work. “I’ll never deny that he’s my nemesis, though.

“Dave’s . . . Dave’s always there.”

•  •  •

Jon Moore sat—his beetle brow furrowed, square Lebanese-American face flushed, and football player’s shoulders heaved forward—like a New Age keyboardist at a pale-green and beige seven-foot console, alternately adjusting dials and typing instructions into an unseen computer. An auditorium full of such instruments and it would look like Mission Control in Houston, but Moore sat alone, in half-light, in a narrow room dominated at the far end by an enormous, gleaming stainless-steel cylinder lit by a single overhead spotlight. The cylinder housed an electromagnet so powerful that it tugged at the electrons on Yamashita’s computer screen across the hall, blurring line after line of electronic type. Gangs of thick gray cables snaked across the carpet, connecting the device to Moore’s panel and crossing a swath of yellow and black tape that marked where it was safe to walk without wiping out one’s magnetically encoded credit cards or destroying one’s watch. Moore habitually removed his wallet and timepiece and tossed them on the console before going to work.

Like Yamashita and Navia, Moore is a biophysicist—“kind of a weird undergraduate major because you don’t really know what it is until you’re a senior,” he says. Like them, he was hired by Boger, at age thirty-three, to solve protein structures by a method neither as evolved nor as fashionable as X-ray crystallography. Moore is a proton nuclear magnetic resonance (NMR) spectroscoper—to crystallographers like Navia and Yamashita what a radiologist using magnet resonance imaging (MRI) was, perhaps five years ago, to those favoring more established computerized axial tomographic (CAT) scans. NMR spectroscopers and crystallographers both produce pictures of an unseen world by analyzing subatomic activity, and the competition between them is fierce, with NMR threatening to overtake crystallography, if not entirely, enough to make the rivalry uncomfortable. Like CAT scanners and MRI jocks, NMR experts and crystallographers outdo each other in pointing out the flaws in one another’s work.

NMR is based on the theory of resonance. Like electrons, the nuclei of certain atoms, particularly hydrogen atoms, are considered to spin like tops. Put them in an electromagnetic field and they align themselves like so many bar magnets, resonating, or spinning, all at about the same frequency. Thus, for instance, there are about 1600 hydrogen atoms in FKBP, each with a positively charged nucleus called a proton, which, when placed in a glass sample tube and dropped into the maw of the electromagnet in Vertex’s NMR lab, spin at approximately 500 megahertz, or 500 million cycles per second—the same frequency as the machine.

If all the protons resonated at exactly the same rate, it would be impossible to distinguish them. However, their spins vary ever so slightly, by about one five-hundredth of 1 percent. A hydrogen atom may, for instance, share an electron cloud with a larger, more muscular carbon atom, which exerts a drag on its spin. Or it may be near, in the roller-coaster structure of a protein, other protons that speed it up. By plotting the subtle shifts between protons, bio-physicists like Moore try to figure out the location of each hydrogen atom in a molecule—a thousand points of light, so to speak, or a thousand distant stars. Knowing the chemical structure of the molecule, they then can map out its overall architecture by using the points as landmarks. (MRI, NMR’s med-tech cousin, operates much the same way. Patients are immobilized and slid, like hot dogs on buns, into giant electromagnets that set every proton in their bodies whirring. Because water molecules, which have two protons each, resonate differently in different cellular environments, the technology makes it possible to identify, say, incipient cancer cells deep within tissue.)

Moore had gotten his first Thomson Unit of pure FKBP in August, several months after Navia and Yamashita. It was a month after he had arrived at Vertex, and the delay was unsurprising: Crystallography, having solved far more protein structures than NMR, is traditionally accorded such priority. Indeed, using NMR to solve the structures of large, hydrogen-rich molecules like proteins was still a new art, and Boger’s decision to pursue Moore, one of its rising stars, something of a long shot. FKBP, with more than one hundred amino acids, was just beyond the point at which proteins were thought to be too large and complex to be solved by NMR. On the other hand, Boger knew the limitations of crystallography—it was useless without hard-to-grow crystals—and Moore had once solved the structure of a protein with more than ninety amino acids, the largest protein ever to be solved by NMR up to that time. Intense, brash, competitive, with the determination of a nose tackle, the squat, powerfully built Moore represented to Boger not only the opportunity to exploit a wide open and extraordinarily promising new field, but a backstop should Navia and Yamashita fail.

For Moore, the decision to work for a company was equally venturesome and uncertain. After getting his Ph.D. from the University of Pennsylvania, where he also went as an undergraduate, and doing two postdocs, he had expected to start out as an assistant professor, launching an independent research career. He’d recently received two prestigious offers—one from Florida State University, a leading center of magnetic research, the other from the Brookhaven National Laboratory on Long Island. But looking at the landscape of academic research, Moore recoiled. The more he thought about being an assistant professor and what it entailed—uncertain funding, chronic job insecurity, the need to sacrifice research in order to teach, dependence on better-known collaborators—the more appealing Vertex looked. He was particularly attracted by the chance, practically impossible for someone at his level outside of industry, to plunge directly into a hot area.

“In academia, you can’t tread on the big guys’ toes,” he said. “It would be difficult for me to take on someone like Stuart Schreiber without the resources of a Vertex. I mean, think how long it would have taken to isolate protein on my own or with a graduate student, or even in collaboration with someone who might not have the expertise of a Thomson. You hear this scenario all the time: The guy you’re collaborating with has one graduate student making protein, but she isn’t showing up much in the lab. Stuff like that frustrates academic people to no end. You end up trying to find some little project where there’s no competition and you hope you can squeak by without getting crushed.”

If Navia and Yamashita could only guess at their competition, Moore had known even before he accepted Boger’s job offer that his main rivals would be Schreiber’s group. NMR was one of the developing technologies that Schreiber had also brought early on into his lab, and as early as mid-March, a graduate student named Mark Rosen had begun a first gamut of experiments with a batch of Schreiber’s enzyme. Rosen had never solved a protein structure before, and Boger had made much of his inexperience in recruiting Moore to Vertex and convincing him that Schreiber could be beaten. Still, with Schreiber’s luck as a talisman, Rosen and a colleague had advanced quickly. By September, just as Moore was beginning his work in earnest, they had already located about 700 of FKBP’s 1600 protons—more than enough to deduce a structure. In August, by assembling snatches of structural information, they had seen on their computer screens the first rough images of what researchers call the gross folding topology of the protein—its outer shape. Like the reconstruction of an ancient piece of pottery from smashed and buried shards, NMR structures are assembled bit by bit, and Schreiber’s group was still months away at the least from having a refined structure. But they were several months ahead of Moore, who, having no idea of their progress, now set out alone to catch up with them.

“If you’ve never done a structure before, getting from assignments to a structure is a ton of work. If you have done a structure before and have the right software, you can cut a lot of corners and do it much faster,” he said.

“Joshua did a very good job of making me think that ‘Ah, Schreiber’s never done a structure before. They’ll never get anywhere on it.’ Whether he knew that or not, I don’t know, but his job was to sell me. I took it as a challenge. I motivated myself thinking about Schreiber.”

•  •  •

Navia and Yamashita had no idea that Schreiber, along with Clardy, was also now competing with them directly, though even if they had it wouldn’t have propelled them any faster. They already had their demon: Merck.

Scientists, unlike, say, athletes, conjure their own competition. They seldom know precisely whom they’re competing with or where they are in the race. They hear things—rumors, reports—but the information comes at a distance and with a price: It may be exaggerated, deliberately misleading, or simply false. The atmosphere in a lab in the throes of a heated project is insular, secretive, xenophobic, superheated, and paranoid. Like the crew of a submarine in enemy waters, its scientists grimly hunker down at their stations, sweating anxiously, listening, awaiting the next depth charge. Even Boger, normally cavalier, now picked up each week’s Science and Nature with trepidation, expecting to read something that would suddenly and irremediably derail him.

With crystallography, he had reason for concern. Merck had been rumored to have crystals as much as four or five months before Vertex. More, the scientist who grew them, a young former associate of Navia’s named Brian McKeever, was well known to be one of the field’s true wizards. It was McKeever who’d produced the crystals of HIV protease that Navia had used to solve its structure, the one scientist Boger had failed to recruit in his tong war with Rahway. Now, throughout the fall, as Navia and Yamashita tried to grow better crystals of, FKBP than the ones that had first diffracted marginally in July, and as Yamashita sought to derive from an enormous mass of data the first glimmerings of a structure, McKeever’s specter was constantly with them.

Once scientists crystallize a protein they can eventually solve its structure, but getting from one to the other requires several translations in data. Yamashita by now had collected millions of pieces of data about FKBP, skeins of numbers that gave him the map coordinates for each atom on each molecule of protein as it was frozen in space. However, by itself such information was useless. None of the atoms was prominent enough to stand out against the others. He couldn’t tell where one molecule ended and another began. It was all one endless, undifferentiated mass. To interpret his numbers, Yamashita needed some identifying mark, a solid, stationary point from which to proceed.

Commonly, scientists overcome this hurdle by growing crystals that incorporate bigger, bulkier atoms—so called heavy-atom derivatives. Imagine, as author Horace Freeland Judson has proposed, the crystal lattice as “vulgar patterned wallpaper.” Heavy atoms are distinguishing marks—the “tip of a rosebud, eye of a bird”—that show up at repeated intervals. Two or three such marks, overlaid, make it possible to discern the simple repeating pattern within the overall matrix, in other words, the outer dimension—the global shape—of a single molecule.

Navia had predicted that he and Yamashita would have heavy-atom derivatives of FKBP within two months after they first crystallized the protein; structure, a month later. However, there had been mounting delays. The first heavy atom they used—platinum—acted like a bowling ball dropped into a fissure of ice: The stress around the crack forced the crystals to break apart. With other atoms, it had been impossible to get them to bind to the molecules, or if they bound, not to change the shape of the protein. In an extraordinary and provocative piece of science, Thomson had managed to make FKBP unfold—so that it was just a linear chain of amino acids with all its atoms, including those normally buried deep inside, laid bare—then snap back into its original conformation, without losing activity. It was like a child’s party blower: Thomson uncoiled the protein, laid in a heavy atom, then rolled it back up. Discouragingly, it still didn’t provide Yamashita the help he needed.

Each time a heavy atom didn’t work out, Yamashita had to try to grow new crystals under new conditions. He then had to collect an entire data set, which took about a week, before he could generate computer maps showing whether a derivative would work. The entire turnaround took about three weeks, and each failure plunged Yamashita into a wider desperation. He was working all the time, staying at Vertex most nights, napping for forty-five minutes or an hour on the floor by the X-ray generator before rousing himself with a cigarette and a Pepsi and hurling himself back into the numbing cycle of growing new crystals, collecting new data, crunching new numbers, making new maps, encountering new failures. As it had for Thomson before him, the world now receded, disappearing completely behind his obsession with his work. Like Thomson, he believed Vertex was depending on him utterly and that he couldn’t fail. Feeling isolated, he also now began to see the world as increasingly conspiratorial.

Everywhere Yamashita saw rivals, and everywhere he fought them privately by subsuming himself in plots, morality plays, agonies. He was sure his principle opponent, Brian McKeever, would know how to overcome the problems with heavy atoms and that Merck was therefore unbeatable. This made him distrust Boger, who had told him Merck couldn’t win, and Navia, whom he blamed for much of his misery. As Yamashita saw it, Navia was gravely ambitious. Certainly, he was impatient. Pushing hard for results, he often was too agressive with crystals, destroying them in the process of his experiments and leaving Yamashita to repeat the work. He used a lot of protein and, when they got in his way, could be careless with people, especially John Thomson. Yamashita believed Navia was determined to take credit for all crystal structures solved at Vertex, regardless of his role, and thus he—Yamashita—would never be able to advance within the company or his career no matter how slavishly he worked or brilliantly he performed. The months of failure with heavy atoms—months during which Navia was often away, attending NIH site visits and shuttle launches and business meetings, or was focused on other projects—left Yamashita in charge of a struggling program, but with no ultimate authority or hope of reward. Meanwhile, he was eyeing Jon Moore, whose progress he also feared. Losing to Moore, though not as ruinous as being beaten by McKeever or oppressed by Navia, would still deprive Yamashita of his great opportunity to be the first to solve the structure of FKBP. Swept up in this competitive torrent, Yamashita lashed himself more furiously.

For Yamashita, in this state, emotions became another kind of enemy, and he fought them just as hard. Ever since his earlier disappointments with growing the first protein crystals, he had affected a steely resolve. “The highest level of feeling I’m coping with right now is hunger,” he’d said during that period. Now, throughout the fall, whenever someone asked how he was doing, he answered in a low, dulcet voice: “I’m stable.” Thomson, who saw perhaps more of him than anyone else, questioned Yamashita’s notion of stability and assessed him rather as metastable, the scientific term for that last moment of tenuous equilibrium before a system spins out of control and crashes.

When he wasn’t acting dysfunctional, Yamashita was affecting and warm, laughing easily at his predicament and going out of his way to bolster others. He was particularly solicitous of Navia, who, despite his resenting and fearing, he also, in the Japanese tradition, honored and revered. Once, having to adjust the X-ray beam, he said: “I wouldn’t want Manuel to do it. He’s at an age where it would be unwise for him to spend too much more time at the machine.” (Noting that experimental crystallographers risk leukemia as a result of heavy X-ray exposure, he later remarked: “You have to die sometime.”)

If Yamashita dealt with his frustration by repressing it, Navia was sharp edged, volatile. More than his work with either FKBP or HIV, his first love now was a method he had devised for using enzyme crystals to speed up and improve chemical reactions. Early trials had been so promising that Boger had begun talking about spinning off a new company just to market the technique, which had tantalizing applications for both the chemical and drug industries. Having already solved a significant number of crystal structures, Navia was looking, like Schreiber and Boger, to widen his mark on science, and he saw himself potentially fathering a new technology with his observations about using enzyme crystals as biological agents. Others were less persuaded. Two consultants flown by Boger from England to review the work found the work commercially unfeasible. After one particularly brutal session, one of them concluded, “I wouldn’t say it’s been blown out of the water at this stage, but it’s been in for a pretty serious dismemberment.” Navia was distraught. Publicly, he welcomed the criticism, though others, especially Yamashita, began maintaining a respectful distance, knowing his tendency to show his disappointment in sudden and unpredictable ways.

Driven yet inseparable, Navia and Yamashita were drawn together more and more with comic intensity. Wary of each other’s feelings and perhaps their own, they were gushingly polite toward each other, even as privately they seethed. “After you, Doctor,” Yamashita would say, passing through a doorway, Alphonse to Navia’s Gaston. “No, no, Doctor, after you,” Navia insisted. Beyond the conflicting clichés of their birth—Navia, the tempestuous, hot-blooded, Cuban-American; Yamashita, the inscrutable, stone-faced young Japanese-American—they were proving to be deeply incompatible. Despite their often genuine pleasure at being together and sincere mutual respect, something like hatred also now ran between them. Thomson, who liked them both but was becoming increasing annoyed with their relentless calls for more protein, now began referring to them in private as “Abbott and Costello.”

In September, a rumor raced through Vertex that sent shock waves through the labs. It was that Merck, in combination with researchers at Yale, had produced the ultimate experimental result for determining FKBP’s biological relevance, a so-called transgenic deletion, or gene knockout. Boger and the immunologists had long thought that the only way to determine absolutely whether blocking FKBP with FK-506 caused immunosuppression was to give the drug to an animal that didn’t have the protein. If the animal’s immunities were suppressed, it would prove conclusively that another receptor than FKBP was the drug’s true target. Now, the scientists began hearing from several sources that the Merck/Yale team had engineered such an animal. By deleting the gene for FKBP from mice embryos, the researchers reportedly had been able to “invent” an FKBP-less species. The shock was that the animals were said to be just as susceptible to FK-506 as normal mice.

Though the rumor was unconfirmed—no such experiment was ever reported in the literature—the implications were thunderous. If FKBP wasn’t the right target, then all Vertex’s work was suddenly voided. Everything. What good were better inhibitors of a protein that was biologically irrelevant? Of what value was Thomson’s Herculean extraction of the protein or Yamashita’s dogged search for heavy atoms? Yamashita heard that Merck was shutting down its program in FK-506 and implicitly believed it. Boger, forced to quell Yamashita’s distress along with everyone else’s, correctly pointed out that without published evidence the rumor had to be treated as false although he, too, was concerned. “You wouldn’t just hear that there was a paper out,” he told a small group consisting of Navia, Yamashita, and Murcko. “You’d hear the sky tearing.” Echoed Murcko: “The psychic cry of fifty anguished Ph.D.’s over the ether. It’s very unlikely that anyone in the world has that kind of evidence and anything else is speculative.”

Confronting the possibility that FKBP was irrelevant, beset by the problems with heavy atoms, assailed by the consultants over his immobilized enzymes, needing to go gingerly around Yamashita lest he snap and be gone, Navia was combustible. Witty and gracious one minute, he was explosive the next. On a day in early October, he slammed down the phone in the lunchroom after an escalating disagreement with a software vendor and launched into a tirade that even some of the others who’d grown accustomed to such outbursts found eye-opening.

“I’m stuck here until Friday with a pike up my ass while this marketing asshole is going to take until the fucking end of the week to tell me he can’t give me what I need,” he yelled, his Jesuit training squelched. “If it’s a marketing problem, that’s his fucking problem. . . . A bunch of goddamned used-car salesmen.” He stamped the floor, kicked two chairs, and jerked a thirty-five-pound plastic jug into the water cooler. Then, just as suddenly, he caught himself, apologized, and, straightening his tie, headed for the men’s room. “I’m going to stick my head under cold water so I don’t have a complete fit,” he announced contritely.

Yamashita, by comparison, seemed calm during these periods. He maintained the demeanor of the stringently controlled rational man. It was only after work, going out with Thomson and Laura Engle to drink, that his desperation poured out. He was working as hard as anyone at Vertex, as hard as Thomson had worked, and was making no headway. Returning to Vertex to sit through the night at his computer or at the hookup on his kitchen table, he still had no key for unlocking the numbers that would reveal the structure of the protein. He was bereft, exhausted. He decided to go to Hawaii to visit his parents for two weeks at Christmas, but as the date of his departure drew close, he still wasn’t sure he could leave. On the day before, he worked through the night, then, coming home and showering, fell asleep for five minutes on his couch.

He recalled: “In my dream, I thought, ‘I can’t go away. Brian will beat me. I’ve got to cancel my reservations. I can’t go. Brian will beat me.’ “