Craig M. Crews was sifting through back issues of the Journal of Antibiotics when he stumbled upon a molecule that piqued his curiosity. The compound—a microbial natural product called epoxomicin—was a tetrapeptide with an unusual epoxy ketone group at one end.
At the time, in the late 1990’s, Crews was an assistant professor at Yale University on the hunt for a good project. But last month when an analog of epoxomicin, called carfilzomib (Kyprolis), was approved by the Food & Drug Administration for the treatment of recurrent multiple myeloma, Crews was elevated to the elite club of academics whose work has led directly to a drug.
“I confess that we didn’t set out to make a drug,” Crews says. “My research interests are trying to understand how biologically active compounds work.” Epoxomicin fit that bill nicely: The researchers from Bristol-Myers Squibb Research Institute in Tokyo who discovered the compound knew it had potent in vivo antitumor activity, but they had no idea how it worked.
So the BMS team dropped the project, and, as Crews would later learn, that’s the only reason the initial work on epoxomicin was even published. BMS worried that without knowing epoxomicin’s mechanism of action, FDA wouldn’t let them move forward, despite the natural product’s apparent potency.
Because the goal of Crews’s lab is to figure out how such biologically active mystery molecules work, the researchers adopted epoxomicin. By that time, BMS had shut down its research institute in Tokyo, so there was none of the natural product to be had. Undaunted, Crews and his students devised the first total synthesis of the compound.
They also created a version of epoxomicin with a biotin tag that would help them identify the specific proteins the natural product was binding to. Because epoxomicin has an epoxide that’s prone to nucleophilic attack, Crews reasoned that it most likely covalently bound to its target protein. By seeing which proteins were newly biotinylated upon exposure to the tagged epoxomicin, they could fish out and identify the compound’s target.
They found that epoxomicin was binding to the proteasome—the biochemical machinery that slices and dices damaged or unwanted proteins. “If you can gum up that proteolysis, you can imagine the garbage piles up rather quickly,” Crews explains, and cells die.
That sounds like a pretty vital function for the cell, one that scientists might be reluctant to mess with. But, Crews continues, cells that undergo a high rate of growth, such as cancer cells, have a higher rate of metabolism and therefore need to turn over proteins more quickly than do normal cells. So there exists a therapeutic window in which you can selectively kill cancer cells via proteasome inhibition.
Crews’s lab tested epoxomicin against known proteasome inhibitors and found that it was extremely potent. They also learned that the compound did not have any cross-inhibition with proteases—a problem that plagued other proteasome inhibitors.
In collaboration with Robert Huber of Max Planck Institute for Biochemistry, in Martinsried, Germany, Crews was able to pinpoint the source of this selectivity by solving the crystal structure of epoxomicin bound to the proteasome. They published the work in 2000 (J. Am. Chem. Soc., DOI: 10.1021/ja993588m).
Crews and Huber found that a threonine residue, which happens to be the N-terminus of the proteasome catalytic subunits, covalently attaches to epoxomicin via formation of a morpholino ring. Normally, the alcohol side chain of the threonine attacks amide bonds, but when presented with epoxomicin, it goes after the ketone portion of the epoxy ketone group. This, the researchers believe, sets the epoxide group up for nucleophilic attack by the amino group of the N-terminus, leading to a morpholino ring adduct.
“Obviously, other proteases have nucleophilic side chains in their active sites, but their active sites aren’t really at the very beginning of the protein, so they can’t have this double nucleophilic attack across the epoxy ketone involving the side chain as well as the N-terminus. They can’t form that morpholino ring. That’s why we have this proteasome specificity,” Crews explains.
Crews had found the answer to his question of how epoxomicin worked. But the proteasome specificity got him thinking that he might be on his way to an actual cancer treatment. His group used the knowledge they had gleaned from their biological studies to make the epoxomicin analog YU-101. That compound had potent antitumor activity, and it turned out to be better at inhibiting the proteasome than epoxomicin.
YU-101 also had better inhibitory activity than bortezomib, a proteasome inhibitor first invented by the small company ProScript, which was later acquired by Millennium Pharmaceuticals. During the time that Crews’s lab was developing YU-101, bortezomib was wending its way through clinical trials for the treatment of multiple myeloma, a blood disorder wherein antibody-producing cells go into overdrive. “Because the cells need to turn over protein very rapidly, if you block the proteasome, you get a buildup of these antibodies the cells produce and they die,” Crews explains.
Crews believed that YU-101 had promise, but he now faced the arduous development and clinical work that needs to be done to turn an interesting discovery into an actual drug. So many discoveries never make it through that process, largely because of a lack of funds, that it has come to be known among researchers as “the valley of death.”
For an academic, making it through the valley of death usually means getting your compound to a stage where a deep-pocketed pharmaceutical company will take an interest. “In days of yore, the pharmaceutical industry would consider taking a drug candidate when you showed it had activity in animals,” says David Newman, chief of the National Cancer Institute’s Natural Products Branch. “Companies now want to see drug candidates, and ideally, post-Phase I.”
Even so, academics say they need big pharma’s interest to advance a promising compound.
“You’re going to need a lot of good friends and a lot of different expertise to make it through the valley of death,” adds Paul A. Wender, a Stanford University professor who has been involved in founding three companies, starting with CellGate in 1997, to develop discoveries made in his lab. You need to have not only the skills of a good chemist but those of a good biologist and a good clinician, he says. “You have to worry about investment, intellectual property. It’s stuff that exceeds any one person’s area of expertise.”
Taking his first steps across the valley, Crews sought to start up a company to develop YU-101. He spent months pitching the project to venture capital firms trying to secure funding. But it wasn’t until May 2003 that his efforts met with success.
That’s when bortezomib (sold under the trade name Velcade) received FDA approval, becoming the first proteasome inhibitor therapeutic. But bortezomib was not without its problems. Patients complained of a painful condition known as peripheral neuropathy. It wasn’t known if that side effect would result from all proteasome inhibitors or whether it was an off-target effect.
Crews had a hunch that bortezomib’s boronate moiety was responsible. He reasoned that YU-101’s epoxy ketone wouldn’t have the same problems. He recalls telling venture capitalists that his group had everything they needed to make a better bortezomib, and the regulatory pathway had already been worked out. “By following bortezomib’s example, we were able to move very quickly,” Crews says.
Along with Caltech professor Raymond J. Deshaies, Crews started South San Francisco-based Proteolix. The company, Crews says, made more than 100 different analogs of epoxy ketone-containing peptides, ultimately coming back to YU-101 as the best of the bunch in terms of specificity and biochemistry. The company did make one change: They added a morpholine ring to the compound to boost its solubility, thereby creating carfilzomib.
“Within 18 months of securing the money and hiring the first team, Proteolix had filed a New Drug Application to start clinical trials,” Crews recalls. Proteolix pushed carfilzomib through Phase I and II clinical trials. As Crews predicted, the peripheral neuropathy that was problematic for bortezomib was not found to be a side effect for carfilzomib. In 2009, Onyx Pharmaceuticals acquired Proteolix and advanced the compound through a Phase IIb trial that led to the drug’s approval last month. Carfilzomib is also going through a Phase III trial as well as clinical trials to explore its efficacy in solid tumors.
“It’s fantastic for patients that there’s a second proteasome inhibitor that can offer relief from their disease,” says Julian Adams, president of R&D at Infinity Pharmaceuticals and the chemist behind the development of bortezomib at both ProScript and Millennium.
Adams calls Crews’s work a tour de force. “It’s a very elegant piece of science and directly led to the formation of a company that could actualize that science,” he adds.
Adams doesn’t, however, think Crews has made it successfully through the dreaded valley. “Getting acquired is not exactly making it through the valley of death,” he says. In his definition, “making it through the valley of death is being able to raise enough capital to stay independent.”
But others disagree. NCI’s Newman cites Crews’s research as a rare case of work from an academic lab crossing the valley successfully. “It’s a beautiful example of what can be done if all the gods are smiling brightly and you have a lot of luck,” he says.
“It is a high-risk endeavor,” acknowledges Crews, “and there’s still a strong bias that this can’t be done in an academic lab. But I would like to hope that our example encourages others.”