Credit: Will Ludwig/C&EN/Amgen | Amgen's AMG 510, shown here bound to its protein target, was the first KRAS G12C inhibitor to be tested in humans.
Chemists spent decades trying—and failing—to overcome KRAS, the most commonly mutated oncogene.
Finally, they have succeeded. Designing the first inhibitors of the cancer-causing mutant KRAS G12C was a painstaking process of trial and error, culminating in some of the most closely watched candidates in the drug industry’s pipeline. Along the way, researchers learned a lot about this tough-to-target protein, knowledge they hope will lead to drugs that take down other members of the KRAS family. Read on to learn more about the notorious KRAS.
It wasn’t that long ago that chemists who mentioned in polite scientific society that they were trying to develop a drug to block KRAS would have been met with a sympathetic look—maybe even a condescending chuckle. The subtext? Good luck with that. After all, chasing KRAS, the most notorious cancer-causing protein, has always been something of a fool’s errand. Scientists spent 30 years trying to tame the beast. For 30 years, they failed.
Until now, that is.
In the span of a year, three small molecules that directly block one of KRAS’s several mutant forms started clinical trials. Many other KRAS-targeted therapies, small molecules as well as other modalities, are in the works. With investors convinced that KRAS inhibitors could be multibillion-dollar drugs, they are among the most closely watched compounds in the drug industry’s pipeline.
In an industry enamored with cutting-edge technologies meant to take the guesswork out of drug design, getting to this point has been an exercise in empiricism. It took years of trial and error and thousands of compounds made inside industry labs to get to drug candidates.
Along the way, researchers learned a lot about the wily protein. They always knew KRAS lacks a nice, deep pocket in which to anchor a compound—its compact shape and slick surface have garnered comparisons to Star Wars’ infamous Death Star—but drug hunters now have far more detail about its movements and behavior. All this, they hope, will make it easier to tackle KRAS in all its mutant forms.
The big chemistry question—can a small molecule shut down KRAS—has been answered. Now, everyone involved awaits the answer to the biological one: Does shutting down this cancer villain make a difference in the lives of people with cancer?
KRAS has always been a natural enemy for drug hunters to pursue. Mutated forms of the protein and its family members NRAS and HRAS are found in some 30% of human cancers. About 35% of lung cancers, 45% of colon cancers, and nearly all pancreatic cancers harbor a RAS mutation.
Moreover, people with mutations in KRAS, a key node in cell signaling, often aren’t helped by existing cancer drugs. “The tumors that are driven by KRAS don’t respond to anything else,” notes Frank McCormick, a professor at the University of California, San Francisco’s Helen Diller Family Comprehensive Cancer Center who was instrumental in the current resurgence in RAS-related research. “They’re really the worst kinds of tumors in that respect.”
In short, the target couldn’t be more obvious.
But ever since the early 1980s, when researchers first linked mutated forms of the protein to cancer, KRAS has been a drug discovery dead end. As recently as 2012, chemists leading KRAS programs had a hard time convincing their teams to work on the protein.
“It’s really important to know your enemy, and before early 2012–13, I don’t feel like we knew enough about mutant KRAS,” says Victor Cee, a medicinal chemist at the biotech firm Amgen. There, Cee led a group of researchers that in early 2012 embarked on a quiet, high-risk project to develop an inhibitor of KRAS G12C.
The Amgen researchers’ glimmer of confidence that the KRAS mutant could be conquered came from their success against another cancer-causing protein. About a decade ago, the biotech firm developed a series of molecules against a mutant form of EGFR. Those compounds worked by binding selectively to the mutant protein, and then latching onto a nearby cysteine residue.
Although the EGFR inhibitors never made it into human studies, Amgen scientists started mulling ways to take advantage of cysteines on other proteins. As it happens, the KRAS G12C mutation causes a specific glycine (at the 12-position, hence “G12C”) to morph into a cysteine. The sulfur on that cysteine is like a dab of glue for a drug to covalently stick to and, in theory, irreversibly shut down its activity. And because that cysteine exists only on the mutant form of the protein, healthy KRAS would be spared.
“In those days, it was little more than just an idea that maybe one could target it with a covalent inhibitor someday,” Cee says. And his chemistry team was keenly aware of what it was up against. No one had ever shown that a small molecule could dock to KRAS, much less a crystal structure of the mutant protein. Without those chemical and biological clues, researchers were working in the dark. As Cee bluntly puts it: “We didn’t know how to do it.”
Then, in late 2013, a breakthrough: UCSF chemist Kevan Shokat’s group published a paper showing a compound—a covalent cysteine-targeted inhibitor—tucked into a pocket in KRAS G12C.
Researchers in academia and industry alike lauded the work as launching the first new avenue for targeting KRAS in 30 years, but it came with caveats. Not everyone was convinced by the supporting data showing the compound could knock down the protein. And others doubted Shokat’s strategy could work: his compound bound to a pocket that appears when KRAS is turned off. Some scientists questioned whether a drug would do anything to stop cancer cells from proliferating if it didn’t hit KRAS in its active state.
“At that time, people still didn’t believe this approach was viable,” says Yi Liu, cofounder and chief scientific officer of Wellspring Biosciences, a company based on Shokat’s work. Wellspring scientists went to work to better understand the underlying biology.
In 2016, Liu’s team published a paper proving everyone wrong. “Even oncogenic KRAS cycles between inactive and active states,” Liu says, meaning a compound like Shokat’s could bind to the protein in the inactive state and keep it from turning on.
Although much work remained to turn that original tool compound into a drug, Shokat’s 2013 paper made many companies decide KRAS was worth another shot. “It’s a very interesting approach that certainly sparked a resurgence of a field, so I think a lot of credit is due there to the Shokat laboratory,” says James Christensen, chief scientific officer of Mirati Therapeutics, another firm that is taking on KRAS.
About 6 months before Shokat’s lab showed the world a tool compound tethered to KRAS G12C, the Amgen team captured its first crystal structure of the G12C mutant. Having a snapshot of the protein didn’t tell the chemists how to design a KRAS inhibitor, Cee acknowledges, but the image provided critical insights into the protein’s shape when turned on and off. “We could see there was a clear difference between the active state of KRAS and the inactive state, just in terms of flexibility and micropockets,” Cee says.
The active form of the protein had the telltale Death Star look to it, but the inactive state turned out to have a few crevices to work with. After making roughly 650 acrylamide-containing molecules to screen against KRAS G12C, Amgen researchers found a few hits. “They were terrible, terrible molecules,” Cee recalls.
Still, the company worked on them for about a year and finally started to see what chemists call a structure-activity relationship—that is, they could tweak a feature of a molecule and see, for example, an increase in how rapidly it shut down the mutant protein. But none rose to the level of a drug candidate.
In late 2013, Amgen brought in Carmot Therapeutics, a biotech firm that specializes in making large, fragment-based libraries, to help with the project. Carmot screened on the order of 100,000 unpurified molecules over the course of 2½ years, Cee says. “We learned so much about KRAS over that time period. A lot of what we learned is that many of those—tens of thousands of those—didn’t work out.”
Mirati, which had teamed up with chemists at Array BioPharma to work on KRAS, was having a similarly frustrating experience. “I describe our journey as an iterative cycle of drug discovery–induced manic depression over the span of about 3½ years,” Christensen says, laughing. “We probably ended up making 2,500 molecules.”
What made it so hard? Researchers now understand the shape-shifting nature of KRAS. While all proteins flex, twist, and jiggle, KRAS’s sudden moves turned drug design into a years-long game of whack-a-mole.
For example, Amgen chemists would slightly tweak a compound to try to increase its potency, only to discover the minor change caused a region on the protein called switch 2 to move dramatically, killing potency instead. “I can’t count how many different positions we’ve seen it in—probably more than 20,” Cee says. “It’s so maddening for structure-based drug design.”
Christensen agrees. Some of the changes that KRAS undergoes are so subtle that Mirati’s structural biologists would argue over their meaning, he says. Yet, “it turns out that some of these subtle movements made all the difference in the world.”
The tricky nature of the protein meant many drug discovery programs had to rely on empiricism and chemists’ intuition. The field’s other big breakthrough came in 2015 when a Wellspring patent revealed a new series of compounds with a quinoline scaffold that was much more potent and acted more like a drug than the acrylamide scaffolds everyone had been pursuing on the basis of Shokat’s tool compound.
From there, it was a foot race to get to a viable drug candidate. Amgen, which over the years had continued to keep its program quiet, shocked competitors by being the first to put a KRAS G12C inhibitor into the clinic. The company registered its Phase I/II trial for AMG 510 in July 2018.
That moment was gratifying for Cee’s medicinal chemistry team. “There were many periods where it wasn’t clear we were going to be able to get there at all, and it was also hard to see what was going on externally,” he says. “We were super afraid that somebody else was going to figure this out faster than we were.”
Mirati opened enrollment in its trial for its KRAS G12C inhibitor, MRTX849, a few months after Amgen’s trial began. And this summer, Wellspring’s partner Johnson & Johnson began a study of JNJ-74699157.
Now everyone—researchers, oncologists, cancer patients, and investors—is waiting to see if it was all worth the effort.
Drugging KRAS is about more than researchers flexing their med chem muscles. The hope is that a KRAS inhibitor will offer a lifeline to cancer patients with few options.
“I get a lot of emails and calls from patients who say, ‘When can I have your compound?’ ” Wellspring’s Liu says. One person even offered to waive liability just for access to the experimental treatment. “You can tell that patients are desperate.”
But while plenty of evidence in the lab suggests that blocking the signaling node could kill cancer cells, the drugs’ ability to meaningfully shrink tumors in humans has long been an unknown. “You never know until you try,” UCSF’s Shokat says.
In June, the field held its breath as Amgen offered up its first data on AMG 510. The early results, released at a premier oncology meeting, included information on 35 patients, primarily with lung and colon cancer, all of whom harbored the G12C mutation and already had been treated with several other drugs. Despite the small data set, the excitement was palpable: not only was the drug safe, but it also prompted tumor shrinkage in half the 10 evaluable patients with lung cancer; 4 more saw their cancer stabilize.
The responses for colon cancer were less impressive, but KRAS researchers say they had always expected it to be a tougher disease to conquer. “Oftentimes people think a mutation is a mutation is a mutation,” Wellspring executive chairman Troy Wilson says. But, he explains, enough of the other players in KRAS’s signaling web have been drugged for researchers to know that some tumors are more recalcitrant than others.
Earlier this month, Amgen provided additional data supporting the early signs of efficacy in lung cancer, and later this month the firm will offer yet another trial update.
Investors, meanwhile, are obsessively parsing each tidbit of information that Amgen releases. Depending on whom you ask, each new data set either dampens or affirms the AMG 510 buzz. Investors’ primary concern is that the benefit offered by KRAS G12C inhibitors could be fleeting. Cancer is a pernicious foe, known to escape via side streets when a drug shuts down a main artery. Everyone is trying to understand if patients in the Amgen trial are already showing signs of resistance to the drug.
For investors, a shorter-term effect would mean lower sales of the drug. Although stock analysts say they have yet to put AMG 510 into their forecasting models, that hasn’t stopped them from estimating—using the early data from those few dozen patients—that the drug could bring in anywhere from $1.5 billion to $3.5 billion in annual sales.
The worries over resistance are not lost on researchers either. “Simply shutting off RAS is not always enough,” says Mark Goldsmith, CEO of Revolution Medicines, which is working on KRAS inhibitors with a different mode of action from the ones in the clinic. “There’s enough complexity and richness in the RAS cascade that there are workarounds available to the cell.”
Companies are using several strategies to address resistance. Some are simply trying to design better inhibitors. Mirati, for example, claims that it had a drug candidate that could have gone into the clinic a year before MRTX849 actually began trials, but it wanted to improve the drug’s properties.
“We’ve done a lot of preclinical work looking at this whole concept of rebound signaling,” Christensen says, referring to the idea that knocking out one signaling node could prompt cancer cells to shift their reliance to another. Because some of KRAS’s feedback pathways become “supercharged,” Mirati wanted to be sure its compound could bind to and block the protein over the course of a day. The firm ultimately landed on a compound with a half-life in the range of 20–30 hours, he says.
Researchers concede that compound design will get them only so far. They expect to see a proliferation of clinical trials combining KRAS inhibitors with other cancer drugs, including immunotherapies and other targeted small molecules.
One partner of interest is compounds that inhibit SHP2, a phosphatase that itself had been a tough drug target until it was recently conquered with allosteric inhibitors. As Christensen explains, SHP2 can affect the speed at which KRAS cycles between its on and off states. Because the current KRAS G12C inhibitors bind only to the inactive form of the protein, researchers reason that adding a SHP2 inhibitor would improve access to their target. Plus, he adds, “the two molecules are also good at comprehensively blocking downstream signaling.”
In July, Mirati and Novartis teamed up to conduct a study of MRTX849 and Novartis’s SHP2 inhibitor TNO155 in people with solid tumors harboring the G12C mutation. The combination is of interest to Revolution Medicines, which has a SHP2 inhibitor in two clinical studies and last year added a KRAS program through its acquisition of Warp Drive Bio.
Researchers also believe they can combine KRAS inhibitors with checkpoint inhibitors, immunotherapies that help immune cells spot and attack cancer cells. Next year, Amgen says, it will have data showing whether AMG 510 and a checkpoint drug that blocks PD-1 can kill tumors better than either drug alone.
Some believe a durable response will require therapies that can more broadly address KRAS. Moderna, for example, is testing a messenger RNA–based cancer vaccine that addresses the four most common KRAS mutations. In contrast to the years-long effort to find small molecules that block KRAS, the Moderna vaccine program went from “idea on a whiteboard to actually filing an IND”—an investigational new drug application—in roughly a year, says Moderna’s chief medical officer, Tal Zaks. Moderna partnered with Merck & Co. to develop the therapy both alone and in combination with Merck’s checkpoint inhibitor Keytruda.
“It always made sense with me scientifically to combine a cancer vaccine with a checkpoint inhibitor,” Zaks says. “If you’re going to immunize and get a whole bunch of T cells going, you want to make sure they’re unleashed and can go find cancer.”
While they await more human data from KRAS G12C inhibitors, academic and industry researchers are pushing ahead with other ways to block members of the RAS family. The next targets on their list are the other two common mutants, G12D and G12V.
“The general consensus right now would be that targeting other specific mutants with the same kind of attack on the amino acid is going to be much more difficult than the one with cysteine,” UCSF’s McCormick says.
Because the current crop of drug candidates uses a chemical warhead targeting a cysteine only seen on G12C, chemists will have to come up with new strategies for broaching that same pocket on G12D and G12V.
Researchers at Boehringer Ingelheim, which has been working on KRAS inhibitors since 2012, recently revealed the first tool compound that can slip into a pocket that exists on both the active and inactive forms of KRAS. Darryl McConnell, research head for Boehringer’s site in Vienna, notes that significant work remains to make compounds potent and selective enough to bind to that pocket. “That pocket is druggable, no doubt,” he says, but the effort required to drug it will be intense. The pocket is “really small. It’s really shallow, and it’s really polar.”
Nevertheless, the advance is important. “We’ve got two pockets on KRAS that we’ve identified, which is two more than we had 10 years ago,” McConnell says. “How much can we exploit those pockets?”
“I don’t think it’s going to be easy—I mean, KRAS G12C wasn’t easy,” Amgen’s Cee says. “It might be possible.” Researchers now have the benefit of 6 or so years of work on the individual mutants, he says, giving them a sense of the protein’s flexibility, how it cycles between the active and inactive states, and its overall lifestyle.
All those lessons have been enough to get new players to invest in the field. Pfizer, for one, played up Array BioPharma’s vast KRAS knowledge when announcing its $10.6 billion acquisition of the firm. And patent filings suggest many other big firms are putting internal efforts into the protein family.
As Wellspring’s Liu points out, the KRAS G12C story shows that it takes only one good tool compound to open up the field. “If you can get that first step—find a small molecule and demonstrate cellular proof of concept—you can treat it as a normal drug discovery program.”
Company sources are reluctant to give even a whisper of their chemistry strategy for tackling the mutant everyone believes will be the next to be overcome, KRAS G12D. But efforts are definitely underway, and industry executives seem confident that drug candidates are less than 3 years away.
“There are important learnings from G12C in terms of the way the structure moves, the way the nucleotide cycles, the nature of the binding pockets that led us to believe that targeting other mutants is possible,” Mirati’s Christensen says. But, he cautions, KRAS G12D still has some “critical differences” that are presenting the firm’s scientists with a new learning curve.
Those who have been working on KRAS since the early days say they are gratified to see the field reach a tipping point. People no longer wonder whether someone is going to be able to conquer KRAS family members, UCSF’s McCormick says, but rather who will be the first to do it.
“It’s no longer undruggable; it’s a question of which drug is going to be the best,” McCormick says. “That’s a big difference.”