Have drug hunters finally cracked KRas?

Chemist Kevan Shokat remembers vividly the moment he had his first clue that scientists in his University of California, San Francisco, lab were on their way to cracking one of the most notorious—and elusive—cancer-causing proteins.

It was December 2011, and he was snowbound on the road to Lake Tahoe when he got an e-mail from one of his postdocs, Ulf Peters. The message contained little more than columns of numbers. But Shokat knew that, when fed into special software, those numbers would translate into a three-dimensional image of a small molecule bound to KRas, a member of a protein family mutated in 30% of cancers that for decades has stymied drug developers. Those cryptic numbers were the culmination of several years of hard, often discouraging, work.

Shokat really needed to get to a computer.

But first, he had to make it out of the snowstorm. “I couldn’t wait to get there, and the car just wouldn’t make it up the hill,” he recalls. Finally, he got to a computer and downloaded the image. Shokat describes what he saw as “just, like, unbelievable.”

Shokat’s lab had managed to find a new, albeit shallow, pocket on a mutant form of KRas that is known to be a driver of lung cancer. The 3-D image allowed chemists in his lab to tweak small molecules to better fit into the pocket—and ultimately to design the first covalent inhibitor of the protein.

That Shokat was stuck on a mountainside during that breakthrough moment was fitting: KRas is considered by many drug developers to be the Mount Everest of targets.

The ubiquity of KRas mutations in cancer has made it one of the most desirable drug targets, and yet it is also one of the toughest to tackle. Despite decades of work, researchers still lack detailed information about what the surface of the protein looks like as it moves and interacts with its ligands. That makes finding cracks where small molecules can slip in like trying to climb the sheer face of a mountain while blindfolded.

Shokat’s research is part of a resurgence in activity around KRas. Armed with fresh ideas and aided by sophisticated technology, scientists think they will finally be able to tackle this target. The newfound confidence has spawned multiple biotech firms devoted to KRas inhibition. Patent filings have surged since 2012, when Shokat and other scientists started reporting small molecules that could, if only weakly, inhibit the protein.

But along with the swell in drug discovery efforts around KRas comes realism. Drug developers are uncharacteristically humbled by the task ahead. “This will take a lot of perseverance and resilience,” says Rosana Kapeller, chief scientific officer of Nimbus Therapeutics. “This target has been a graveyard, so we want to be very smart about how we approach it so we don’t suffer the same fate.”

A sheer peak

For the past two decades, most of the small molecules developed to treat cancer targeted protein kinases, enzymes that tack phosphate groups onto other proteins. Kinases make good targets not just because of their link to cancer—mutations in specific kinases are known to drive the growth of certain tumors—but because they are “druggable.” Most kinases get their phosphate groups from the nucleotide adenosine triphosphate (ATP) when it’s docked in a binding pocket. When ATP is not home, a small molecule can conveniently slip in, turning off the kinase’s activity.

In contrast, the activity of the Ras family—which includes KRas, NRas, and HRas—is dictated by the nucleotides guanosine triphosphate (GTP), which switches it on, and guanosine diphosphate (GDP), which turns it off. Whereas ATP wafts in and out of the binding pocket on kinases, GTP and GDP stick to KRas like glue.

“With Ras, the whole protein folds up around GTP or GDP,” explains Frank McCormick, a UCSF cancer researcher. Their attraction is so strong that only a tiny amount of the ligand needs to be around to latch onto Ras. Drug hunters haven’t been able to find molecules that can kick the ligands out of their home.

With that avenue blocked, the next obvious strategy for drug developers is to look for other pockets on KRas where small molecules can bind and alter its activity. But there’s a problem: The enzyme is so smooth that researchers liken it to a billiard ball. The few alternate pockets that have been discovered are shallow, and it has proven nearly impossible to design a small molecule that can forge a strong bond with the protein.

In the 1990s, scientists focused on compounds that blocked farnesyl transferase, an enzyme that adds a lipid tail to Ras that allows it to hook into the cell membrane. The strategy worked like gangbusters in mice, and a few farnesyl transferase inhibitors developed by big pharma firms made it to late-stage studies in humans. All of them failed. It turned out KRas had a backup method of gaining that lipid tail.

Those failures came at a time when genomics was revealing cancer-causing mutations in kinases, and researchers largely moved on. But today, drug developers are running out of kinases to chase, and they are also recognizing the limitations of kinase-targeted therapies. “The era of new kinase targets is pretty much over,” UCSF’s McCormick says. “Now, I think most companies are interested in immunotherapy and Ras.”

Thankfully, kinase fatigue is coinciding with innovative strategies for how to tackle KRas. “Two or three decades of working on KRas exhausted most people’s best ideas,” says Don Nicholson, Nimbus’s chief executive officer. “But most people’s best ideas were from 2000 and earlier. We’re in a whole new generation of technical capabilities.”

A foothold

Shokat had been mulling over how to target KRas for nearly a decade before he finally put resources to the project. From the day he joined UCSF in the late 1990s, his colleague McCormick, a giant in the field of Ras biology, kept pestering him to work on KRas inhibitors.

Finally, in 2008, Shokat devoted a few students to developing compounds that inhibit KRas. Shokat’s team focused in on G12C, a mutant form of KRas that features a cysteine in place of a glycine at the 12th amino acid and is a common driver of lung cancer. After several strategies fell flat, the researchers decided to screen the mutant protein against a colleague’s library of “tethering” molecules—small chemical fragments that would react with the cysteine residue.

They were delighted when the screen yielded a few hits and even more encouraged to find that the two best fragments didn’t affect normal KRas. That selectivity for the mutant protein could help minimize side effects of any drug eventually developed using the tethering approach.

The researchers spent more than six months trying to optimize compounds, but without a crystal structure, they were working in the dark. Shokat says the data he received on the mountainside in late 2011 were just the boost his lab needed: They revealed an allosteric site that previously had not been known and allowed the chemists to improve upon the compound. They spent more months in a cycle of optimizing molecules, solving crystal structures, and gauging the activity of the compounds.

In 2012, those early inhibitors became the foundation for Wellspring Biosciences, a biotech firm started by the same team that had formed Shokat’s previous company, Intellikine. Johnson & Johnson licensed the program early the next year.

Shokat’s covalent inhibitor was published in late 2013, and although everyone was impressed with his clever feat of chemistry, some people had a concern. The compounds bind to the mutant protein when it’s in the inactive state, and not all were convinced that this would translate into a clinical benefit.

While in talks with potential partners, “People always said, ‘Yeah, you showed us a big and beautiful crystal structure, but what does it do in biology?’ ” recalls Yi Liu, Wellspring’s chief scientific officer. Adding to the skepticism, Shokat’s best initial molecule wasn’t binding to KRas with the potency needed for a drug.

Wellspring researchers plugged away at the problem and managed to make compounds that were up to 1,000 times as potent. The researchers have since shown that their KRas inhibitors block signaling from the protein and prevent growth of cells with the G12C mutation.

Computational climbers

A year before Shokat published his work and Wellspring licensed his compounds to J&J, two other labs reported molecules that bind to different shallow pockets on KRas.

Genentech scientists and Vanderbilt University’s Stephen Fesik both used nuclear magnetic resonance to identify chemical fragments that interact with KRas. They then either optimized those fragments or stitched them together into molecules that land in adjacent spots and increase overall binding strength. Fesik pioneered the NMR-based approach to drug design in the late 1990s while working at Abbott Laboratories, and it has proven a good way to find inhibitors of tough targets.

Unfortunately, neither the molecules from Genentech nor Fesik’s lab have proven capable of shutting down KRas, UCSF’s McCormick notes. Fesik continues to look for ways to turn off the protein.

But that research, along with Shokat’s, did reveal critical structural insights and tool compounds for others interested in pursuing KRas. Nimbus, for one, decided that both the science and the technology had evolved enough to devote resources to KRas. With its partner Schrödinger, a leader in computational chemistry, Nimbus began working on KRas about 18 months ago.

Computer-based methods have been tried with tough targets like KRas before, but the difference today is that computational chemists now recognize that their focus needs to expand beyond 3-D structural insights, quantum physics, and molecular interactions to include the time-based movement of the protein.

“It’s a bit like Jell-O,” Nicholson says. “There’s constant movement, and other small molecules modify the structure. To make computational chemistry usable, you have to predict those things mathematically.”

Nimbus is trying to build such movement into its KRas program. Already, the effort has yielded four so-called “cryptic” binding sites on the enzyme—pockets that are revealed only when KRas flexes and bends. Nimbus is now combining computational chemistry with NMR-based technologies and crystallography to look for compounds that bind to those sites. “Is that easy?” Kapeller asks. “No, it isn’t easy.”

Nimbus is also trying to capitalize on natural products’ ability to bind to tough proteins like Ras. The goal isn’t to use natural products as drugs but to study how they bind to the protein and incorporate key elements of that interaction into molecules built from scratch. The method proved effective in other programs at Nimbus, notably in the design of allosteric inhibitors of acetyl-CoA carboxylase, compounds recently licensed to Gilead Sciences for fatty liver disease.

Because KRas represents a ginormous challenge, Nimbus is looking for collaborators that can bring complementary expertise to the table. Nicholson says the company hopes to secure a partner for its KRas program by early next year.

Columbia University chemist Brent Stockwell also relied on computational chemistry to design small molecules that block Ras. Knowing that drug developers have long been stymied by the protein’s shallow pockets, Stockwell decided to increase the binding strength by finding two footholds rather than one. Using computational drug design, Stockwell docked molecules into adjacent pockets and then linked them to create noncovalent inhibitors of KRas.

Those compounds were spun off of Columbia as a company called Kyras Therapeutics, which recently opened labs in New York City. “One of the things that was truly unique about Brent’s program is that he not only developed the molecules but he took them in vivo,” showing they have reasonable pharmacokinetic properties and activity in mice, says Carlo Rizzuto, a partner with the investment firm Versant Ventures and CEO of Kyras. “I don’t know too many Ras inhibitors that have achieved that.”

Kyras is now building up the new labs while its scientists work to improve the potency and other properties of its initial compounds.

Another foothold

Around the time Shokat started to work on Ras, another prominent chemist, Greg Verdine, was mulling over new ways to use small molecules to inhibit “flat” proteins such as KRas. Human biology offers several examples in which a natural product can block its target only with help from an intracellular protein. The most prominent example is the natural product rapamycin, which binds to mTor by forming a ternary complex with FKBP, a protein-folding chaperone found throughout the body.

In 2012, Verdine launched Warp Drive Bio with the goal of finding what he calls small-molecule-assisted receptor-targeting, or SMART, drugs. The idea is to design compounds that act as the Velcro between two naturally occurring proteins—FKBP and a flat drug target.

KRas poses the ultimate flat target challenge, and Warp Drive scientists have spent the past two years trying to assemble molecules that, with the help of FKBP, block it. In October 2014, Warp Drive researchers captured their first crystal structure of KRas, a small molecule, and FKBP in a happy embrace. The team has come up with a new crystal structure of other ternary complexes every two to three months thereafter, including structures of Ras in both the active (GTP bound) and inactive (GDP bound) forms.

The project has yielded surprises, Verdine says. For example, when ternary complexes of KRas are in the active form, the small molecule is merely an anchor, and the interaction is primarily between the two proteins. Meanwhile, the inactive form of KRas uses an entirely different location to interact with FKBP.

In January, Sanofi, an investor in Warp Drive since its inception, agreed to co­develop drugs against three different Ras mutants. Warp Drive chemists are now working on four chemical scaffolds with the goal of getting a KRas-targeted drug into the clinic by the end of 2018.

Verdine is not the only scientist inspired by the therapeutic potential of the ternary complexes found in nature. Ohio State University researcher Roger Briesewitz, who worked on generalizing the concept of ternary complexes when he was a postdoc in Gerald Crabtree’s lab at Stanford University, teamed up with OSU chemist Dehua Pei to test whether peptides featuring a moiety that recognizes FKBP could sandwich between it and KRas.

Although the pair did find some peptides that shut down Ras by forming the ternary complex, they were surprised to find peptides that did not seem to require FKBP to bind to Ras. Pei constructed a library of cyclic peptides that replaced the moiety, and lo and behold, the peptides turned out to block multiple forms of Ras.

Briesewitz teamed with the contract research organization Evotec to further develop the program. Because the pharmacokinetic properties of the peptides are poor, the partners used them to understand how they bind to KRas. Briesewitz now hopes to use that structural information to design small molecules against the target.

A challenging route

Still others are convinced there is a way to directly target the active binding site on KRas. In 2013, a team led by Harvard University chemical biologist Nathanael Gray reported the development of selective, direct-acting covalent inhibitors of the G12C mutant. Their two molecules—a GDP analog and its prodrug derivative—marked the first time scientists had been able to irreversibly target the protein’s active site.

The work was a lovely addition to the scientific compendium, but the compounds have a critical liability: They don’t easily slip inside cells. The team is now looking for small molecules that are cell-permeable.

It is a tough road, says Ken Westover, an assistant professor at the University of Texas Southwestern Medical Center who played a key role in finding the GDP analog while a postdoc in Gray’s lab. The first step is finding compounds that can bind in the active site, which for the partners means looking for GDP-like compounds that preserve some of the interactions of the native ligand or looking for completely novel ones. But the method “is technically difficult,” Westover, who continues to collaborate with Gray, acknowledges. Moreover, those GDP-like compounds will need to be made covalent. That, Westover says, will first require laborious experiments to study the structure of Ras in complex with new ligands and then many iterations of medicinal chemistry.

Westover concedes that some scientists dismiss the idea of finding a compound that can compete with the natural Ras ligands. “Many Ras experts will think that’s a crazy approach” he says. But he counters that skepticism by noting that the partners have theoretical work to support the strategy and, importantly, have published proof-of-concept experiments showing that a compound—albeit one that is not a drug—can compete for that spot on Ras.

Next questions

In parallel with the hunt for drugs, researchers are trying to close some of the yawning gaps in their understanding of the fundamental biology and the dynamic movement of KRas—knowledge that could lead to new ideas about how to block its activity.

In 2013, the National Cancer Institute launched the Ras Initiative with $10 million in funding to support scientists working on the protein. Led by UCSF’s McCormick, the initiative has a goal of developing tools, such as reagents and assays, and filling in some of the gaps in the science.

One focus of the Ras Initiative is solving crystal structures of the various mutants of KRas as they interact with their cellular companions. “We all know that proteins move around, but everybody thinks about crystal structures as a snapshot in time and space,” notes Nimbus’s Kapeller, who is not involved in the NCI initiative. “With KRas, we’re finding out the crystal structures are very deceiving.”

McCormick points to two flavors of structures that, if solved, would benefit researchers: Ras in its active state—when GTP is bound to the mutant protein—and Ras when it snuggles up against the cell plasma membrane, a location necessary for its function. “The future of Ras biology and chemistry is understanding the biochemical and biophysical properties in the plasma membrane,” he explains.

Researchers involved in the Ras Initiative have made it their goal to publish several structures of mutant proteins bound to GTP this year. Meanwhile, scientists at Nimbus and elsewhere are considering whether techniques such as capturing crystal structures at room, rather than cryogenic, temperature could help develop more meaningful information about the protein in action.

Even as details emerge about how various forms of Ras work, drug developers know some questions can only be answered by testing their inhibitors in humans. They hope to see one or more drug candidates begin clinical trials in the next two years. That would provide important information about the cancer target and whether liabilities—side effects, compensatory pathways, drug resistance—will emerge.

“Nobody really knows ultimately what will be the most effective and well-tolerated” approach to inhibiting this target, Kyras’s Rizzuto notes. “This is an issue the field is grappling with. We just don’t know, and probably won’t know until we have good compounds in the clinic.”

Although answers to critical questions about KRas are tantalizingly close, researchers are quick to stress that KRas has quashed enthusiasm many times before. Even the most determined drug hunters concede that all of the clever approaches now being tried might, as in the past, fail.

That challenge is also powerful motivation. “Like Everest, KRas deserves a lot of respect,” Nimbus’s Nicholson says. “But we think we have a good chance.”