The Antidote: Inside the World of New Pharma (3 page)

Such good fortune and adulation might have gone to Boger’s head if not for the unending pressure, in the face of minimal resources, to get as many projects off the ground and find partners for them as soon as possible—an endeavor that, like drug discovery, resulted in failure more than 90 percent of the time. Right now AIDS and the virus that caused it, HIV, gave the company its best hope for getting a drug to doctors and patients.

Following Boger’s speech to the executives, senior scientist Mark Murcko delivered a spirited recap of Vertex’s design strategies for a new class of direct-acting antiviral drugs—HIV protease inhibitors—that the company was relying on to catapult it into the ranks of the world’s elite drugmakers. Murcko, thirty-three, was the last of a half dozen major Merck defectors to join Vertex and the one Boger had fought hardest to get, platooning the scientists to call him every other night for three months until he agreed to move. More and more, he’d become Boger’s technological seer, sidekick, everyman, and commentator: a stocky, precocious, deep-voiced, aside-tossing Sancho Panza. At first Boger refused to go into AIDS, but Murcko and a few others changed his mind.

Murcko peppered his slide presentation of elegant computer-generated three-dimensional molecular models with striking new data and throwaway gibes at traditional discovery methods. He told the audience about a drug hunt of rare effectiveness and speed. Proteases are enzymes that cleave other sections of protein: in the case of HIV,
enabling it to multiply. Theoretically, if a small molecule could latch onto the active site, it would block replication. HIV protease offered a promising drug target and Merck and several other companies had compounds in the clinic, although none had yet shown that their molecules were effective in patients. More to the point, the target site was only Vertex’s second project; the first had crashed spectacularly when the mechanism for which it was churning out compounds proved to be biologically irrelevant. Merck was spending more on HIV than on any other research effort in its history, throwing hundreds of chemists at the project. Vertex, Murcko said, had just five.

Vertex needed to speed up the process of getting its own compounds to patients, something it had never done before and that would not be possible at this stage without another partner. What the company had was Boger, hubris, and a passionate group of mostly ex-Merck researchers as determined as he was to “do it right”—“Merck’s first team making drugs for the hardest diseases with computers,” as Aldrich liked to tell potential partners and investors. Structure-based drug design is premised on having a reliable 3-D computer model of the target molecule in the body, so as to be able to visualize the inmost workings of living matter. Vertex had industrialized its processes for isolating and purifying proteins to the point where it could reliably solve complex structures showing how different prospective drug molecules bound to the business end of a target. Murcko’s group digitally modeled new atomic interactions and suggested improvements based on predicted changes in activity. This “feedback loop,” he told the audience, enabled Vertex’s small group of chemists to make much more informed choices than Merck’s legions about what molecules to make next.

Not everyone was sold. After lunch, the keynote speaker, Dr. Edward Scolnick, president of Merck laboratories, unfurled himself from a too-small chair on the dais and rose to deliver a talk entitled “Molecular Approaches to Drug Design.” Murcko, sitting just to his right, smiled, amused. Scolnick was notably hard charging. He had driven the billion-dollar drug Mevacor, the first of the cholesterol-reducing statins, to market, devoting a significant portion of his toxicology budget to overcoming concerns about statin toxicity. Merck’s findings convinced the
US Food and Drug Administration that what seemed to be cancerous changes in animals weren’t really cancers and that a resumption of human testing of statins, after a three-year moratorium, was warranted. Now he was driving Merck’s AIDS effort. Though the structure of HIV protease had initially been solved at Merck—by yet another of those who would soon grow frustrated and follow Boger to Vertex—Scolnick considered the information it generated just moderately useful for drug discovery.

“I gave you that title,” Scolnick said, “but that’s not what I’m going to talk about.” He spoke instead about the industry’s R&D mission and its impact on prices: “what we do compared with what we charge.” Merck CEO Dr. Roy Vagelos had frozen some prices as early as 1990, and the company was urging the industry to adopt voluntary controls in an effort to stave off federal regulation under the Clintons. Scolnick touted the industry’s case for not controlling high prices: namely, that for fifty years, since the introduction of penicillin and cortisone, both of which Merck played a key role in developing, America’s drug companies had produced incomparable value for patients, doctors, society, and shareholders. Their products were thus worth the cost. He finished with a staunch and—given the audience—surprising defense of Proscar, Merck’s new drug for shrinking prostates.

With Merck’s market capitalization off a staggering $20 billion in the past year due in no small part to Proscar’s disappointing launch, Boger thought that Scolnick’s road show–style talk indicated how far even the industry’s titans had been driven to stoop. Scolnick had enthusiastically backed Boger’s efforts at Merck and was put out by his departure, once asking a reporter, “Does Vertex have one single project that wasn’t first formulated here at Merck?” even though Merck itself never challenged Boger on intellectual property. Early defectors to Vertex were given up to a month to leave—and perhaps reconsider—but by the time Murcko left, he at first was told to get out in four days.

Boger called Scolnick’s talk “bizarre” and then beamed a few days later when he received a handwritten note of congratulations from Scolnick: “Dear Josh, It was good to see you again. Clearly your group is thriving, and the talk by Mark was excellent. I look forward to great things from your group.” However much he rebelled against
Merck’s organization, the
ideal
of Merck still fired Boger’s imagination, and he wasn’t above gloating when its senior people had to give Vertex its due.

With targets king, the one thing Vertex needed more than anything else were 3-D models of the protein receptors for its medicines. Drugs are molecules. Once they enter the body, they attach at critical junctures in the pathway of a disease, by finding affinities within the folds of the working units in and among cells and binding to them chemically in a way that alters their activity. Since the 1940s, nearly all small-molecule drugs had been discovered serendipitously (brutishly) by screening large libraries of compounds against presumed biological targets and searching for “hits” based on properties such as shape, electrical charge, and either a fondness for, or revulsion toward, water—properties medicinal chemists then tried to tweak through modification. “Monkeys with typewriters,” Boger called this approach.

The revolution he intended to lead at Vertex, structure-based design, called for overthrowing this method by vastly increasing the value of precise atom-by-atom information in the hunt for new leads. The most common analogy is a lock and key; know the minute inner contours of a lock and you can design complementary features to trigger it. Proteases, however, are large, heavily folded
active
molecules that work like scissors, with a savage tendency to chew up everything around them. The kinds of small molecules that best inhibit them—gumming up the blades, as it were—mimic short chains of amino acids called peptides. But peptide-like compounds get chewed up by digestive enzymes in the gut, making them unusable in orally taken drugs. A lot of people in the industry thought it was impossible to block all but the scarcest proteases.

From the day Vertex opened its labs, the challenge of producing large enough quantities of ultrapure, active protein to supply the rest of the fast-expanding company fell to a dauntless Australian biophysical chemist named John Thomson. On its ill-fated first project, Thomson had slaved to provide the enzyme that enabled Vertex, from a standing start and many months behind the leaders, to solve its structure in a dead heat with a group at Harvard and ahead of Merck’s. He worked hours at a time in a 40-degree cold-room wearing only jeans, running shoes, and
a T-shirt. The more difficult the isolation and purification of a target, the longer and harder Thomson went at it. Once, because he’d stuck his neck out, he remained at the bench for eight days straight, isolating protein from calf thymus, on his swollen feet past dawn night after night, washing his own glassware for hours, his hands raw and eyes burning from solvents until he blanked out, sitting upright, Ray-Bans perched on his nose, on a bench in the lunchroom. He went home only to shower.

“One of our chief goals for the next year,” Aldrich had said, “is to keep John alive.” Proud, hard work—doing whatever it took—was Thomson’s key to everything and he exalted the purity of protein science. The wonder of biology is how chains of inert atoms receive, via other inert atoms, the operating instructions for how and when to fold precisely into bits of living matter, then interact with other bits to produce life. Every cell is an infinitesimal cosmos, a vast automated forge where proteins are made and interact, performing thousands of precise chemical operations, billions of times per second. Throughout the life sciences demonstrating biological activity is paramount and scientists prize above all else scarce reagents that they can use to mount further experiments. On the day Thomson delivered his first batch of active protein to the biophysicist who would try to solve its structure, Murcko, who shared his extreme commitment to structure-based discovery and his blue-collar work ethic, coined a new measurement: the Thomson Unit, 100 milligrams of ultrapure enzyme.

With the AIDS virus the problem was the production method. If native enzyme can’t be isolated from tissue, scientists turn to recombining DNA, putting microbes to work as infinitesimal automated “printers.” Genes that carry the instructions to make specific functional molecules are inserted into bacteria and as the microbes reproduce they churn out mature protein, which can be harvested. Not HIV protease. “The more you make the bugs make it,” Thomson says, “the more noxious it is to the bug. If you ask a cell to make too many scissors, it punctures itself or makes itself sick.” He continues:

So you’ve got to find a trick. One of the tricks—relatively new at the time—is to get the bugs to make and make and make the stuff with not very high fidelity, so that it doesn’t fold properly and falls
into the garbage dump of the cell. It’s inactive material. You have to then be able to take the inactive material and do something with it to refold it. So you basically brew it up in a harsh solvent—what you call a denaturant—and then you take the denaturant away, so that somehow the thing folds into a nice happy state.

I said, “I don’t want to see anyone trying to make this native protein anymore. I want to make it on an industrial scale and rely on ourselves to chemically process it later into an active form.” And it worked. It enabled us to jump from making a few milligrams, enough for a handful of crystallization experiments at a time, to batches that took us through a whole program.

A few weeks after the Boston biotech meeting, Boger sat impatiently across from Murcko, Thomson, and several others in a windowless conference room, at an urgent session of the HIV project council. The councils were an innovation of his. He wanted scientists from different disciplines exchanging ideas and sharing what they knew, and he thought the best way to organize them was to give them the power to organize and direct themselves.

The room was scarcely bigger than the narrow conference table, a dozen chairs, and a credenza, above which hung seasonal photos of Mount Fuji, a gift from Kissei. Crystallographer Eunice Kim presented a new structure of a strikingly potent inhibitor wedged deep inside a folded protein. She had determined the architecture, essentially, by taking Thomson’s enzyme, coaxing it to crystallize, firing X-ray beams at the crystals, then reverse-engineering a set of atomic blueprints from the resulting patterns when the beams bounced off the lattice and diffracted on a screen. For months, as the chemists made better and better molecules, Kim had been reducing her turnaround times for presenting them with detailed images of how the compounds bound with the active site of the protease. This latest, VX-328, had taken her five days.

“What took you so long?” Boger asked.

“Where’s 330?” Vicki Sato joked, as if Kim’s accelerating pace might already have caused her to lap herself. Sato was chief of research, accountable for moving compounds from the lab into clinical development.
She was referring to a molecule submitted in the last couple of days that was five times less potent than VX-328 but that had looked superior in its ability to survive the gut and remain intact in the blood. It was the ever-lethal difficulty of lasting in the body long enough to get to the virus and deactivate it that doomed nearly all protease blockers, and features to improve VX-330’s bioavailability had been deliberately engineered into the molecule based on Kim’s structures. The design team
knew
it was sacrificing a little potency in exchange for better properties. The compound differed from an earlier one by adding a single nitrogen atom within a binding pocket measured in billionths of meters.