Cancer research has had a bit of a Cassandra problem. For decades, cancer research has had a fundamental understanding of disease biology, including key driver mutations in genes that can lead to many types of cancer. This understanding has been both a blessing and a curse.
Cancer research has been alerting drug developers to drug targets that could lead to exciting new therapies. That’s the blessing. But cancer research has found that drug developers are unable to hit many of the most important targets. That’s the curse.
Fortunately, the curse—which we might call the curse of undruggable targets—may be lifting. Recent advances in chemistry, proteomics, and machine learning, alongside the development of new modalities, have resulted in the approval of a first-in-class drug that hits a formerly undruggable oncotarget. This approval may be just the first of a series of approvals for drugs that hit drug targets that were once deemed
undruggable.
Discovery of cryptic pockets
In May, Amgen announced a U.S. Food and Drug Administration approval for a small-molecule drug called Lumakras (sotorasib) for patients who have non-small cell lung cancer (NSCLC) and who carry the G12C mutant form of the KRAS protein. KRAS is a prominent member of the RAS family of cancer-driving genes, and mutant forms of KRAS are commonly found in patients with lung, colorectal, and pancreatic cancers. Earlier efforts to block KRAS proteins with small molecules had failed because binding sites on the proteins were difficult to find.
The KRAS protein plays a role in the epidermal growth factor receptor (EGFR) signaling pathway, against which Amgen and other companies have approved therapies. But the KRAS protein itself remained one of cancer’s holy grails.
One of the keys for Amgen was the discovery of “cryptic” pockets, potential binding sites on a protein that are evident only when it bends or flexes a certain way. The KRAS G12C mutant protein has a glycine switched to a cysteine at amino acid position 12, close enough to a newly discovered cryptic pocket to provide a hook for a small molecule.
Lumakras is one of a handful of approved irreversible covalent inhibitors, and it takes an approach that is a key part of Amgen’s “secret sauce,” says Margaret Chu-Moyer, PhD, vice president of research at Amgen. “KRAS was used as the poster child for ‘undruggable’ for a long time,” she explains. “What was known about it was that there were no pockets for a small molecule to bind, to stop it from being active. Somehow, you have to get in there and grab onto the protein.”
According to Amgen, about 13% of Americans with NSCLC carry the KRAS G12C mutation, where Lumakras was approved. Additionally, 1–3% of patients with colorectal and pancreatic cancer carry the mutation, and Lumakras is in Phase II testing for colorectal cancer and other solid tumors carrying KRAS G12C.
As the G12C mutation is present in 44% of KRAS-mutant cancers—more than any other KRAS mutation—Lumakras is unlikely to be the last anti-KRAS G12C therapy to market. The next most advanced therapy is Mirati Therapeutics’s Adagrasib (MRTX849), currently in Phase III testing for patients with NSCLC.
The targeting of mutants
G12C, however, is not the end of the KRAS story. Other mutations, like G12D, G12R, and G12V, are more common than G12C in certain cancers, and the small-molecule, cryptic-pocket strategy that is working for the KRAS G12C mutant protein may not play out in other variants.
One approach in development involves hijacking the natural protein degradation system designed to clear faulty proteins from the body. In the system, proteins are marked with ubiquitin to be cleared via proteolysis. Companies like Arvinas are developing heterobifunctional small molecules that bind a target of interest on one end, and an E3 ubiquitin ligase on the other, relying the natural protein degradation process to do the hard work.
But unlike a traditional small-molecule inhibitor, Arvinas’s proteolysis-targeting chimera (PROTAC) degraders “don’t need any other activity besides binding,” says Ian Taylor, PhD, the company’s chief scientific officer. “They can bind to a nook or cranny—that’s what opens up the undruggable space. KRAS is smooth, like a bowling ball, and bound by lots of other proteins, meaning available surfaces are limited for small molecules.”
Taylor thinks protein degradation will prove more reliable for challenging targets, even for other KRAS mutant proteins. Unlike the KRAS G12C protein variant, the G12D and G12V KRAS protein variants have different amino acid chains, and they can’t be attacked with an irreversible covalent inhibitor approach. “You’re back to the same challenges as with KRAS in general,” he remarks.
Arvinas’s two lead PROTAC degraders are in Phase II against well-validated cancer targets. One degrader is targeting the androgen receptor (AR), and one is targeting the estrogen receptor (ER). But Taylor predicts PROTACS will ultimately differentiate themselves on more challenging targets. Currently, the company’s KRAS program includes degraders of G12D and G12V.
The current approach to targeting individual KRAS mutant proteins makes sense given the challenges in targeting KRAS more broadly to this point. But in aggregate, KRAS mutant proteins are present in up to a quarter of all cancers—so a pan-KRAS approach is too attractive to abandon. Arvinas, Taylor notes, is developing pan-KRAS degraders alongside individual mutants.
A vaccine-based approach
Cancer vaccines may be another route to pan-KRAS targeting. Cancer vaccines are designed to trigger immune cells to attack cancer, and so they represent an alternative to therapeutics that attack cancer directly.
KRAS has been a challenging target for immunotherapies because it’s small enough to evade immune detection, observes Christopher Haqq, MD, PhD, executive vice president, head of research and development, and chief medical officer, Elicio Therapeutics. Nonetheless, KRAS is being targeted by ELI-002, a therapeutic vaccine that Elicio is developing with its amphiphile (AMP) platform. The company says that the AMP platform can deliver investigational immunotherapeutics directly to the “brain center” of the immune system—the lymph nodes.
ELI-002 is designed to boost the signal for immune detection by tagging KRAS peptide antigens with amphiphilic albumin binders. It consists of seven different mutant KRAS peptide antigens (that is, seven AMP-modified mutant KRAS peptide antigens) and an AMP-modified immune-stimulatory oligonucleotide adjuvant.
“By turning on costimulatory signals in the dendritic cells that are doing the job of presenting these antigens, we now get these formerly undruggable targets to raise these really robust immune responses,” Haqq explains. Elicio is now enrolling patients with KRAS mutant solid tumors, including colorectal cancer and pancreatic ductal adenocarcinoma, in Phase I/II trials for ELI-002.
Lasting gains in transient forms
While the proof remains in the drug approval pudding, the set of truly undruggable targets seems to be shrinking. Historically, only 10% of the human proteome has been effectively druggable. Druggable targets are rigid, with bent portions that stay bent, the inner surfaces of which present small-molecule drugs places to dock, says Chris Varma, PhD, co-founder, chairman, and CEO of Frontier Medicines. Supposedly undruggable targets are usually looser and more stringlike. They lack obvious docking surfaces—except if they adopt certain conformations.
In the context of living cells, most proteins are flexible and dynamic. Yet if these stringlike entities are “shaken,” Varma points out, “transient pockets or curves appear.”
Varma suggests that one way to “drug the undruggable” is to target the transient pockets that occur when proteins adopt their active conformations. To illustrate this approach, he discusses how it could be used to improve the targeting of the KRAS G12C mutant protein. To date, most development programs have targeted inactive forms of the mutant protein. Hence, these programs may yield drug candidates of limited efficacy.
Frontier identifies transient but druggable docking sites by using a development platform that melds machine learning with chemoproteomics, an approach to characterizing interactions between proteins and small molecules. The platform is helping Frontier advance its lead program, which targets the active form of KRAS G12C. “With our technology,” Varma asserts, “we can do that in a really broad way across the proteome.”
Targeted delivery with exosomes
Even in instances where a protein itself remains a challenging target, drugmakers have had increasing success targeting the transcription process upstream from the protein with tools like RNA interference, says Douglas E. Williams, PhD, the president and CEO of Codiak BioSciences, an engineered exosomes company. He suggests that targets are called undruggable only because they haven’t been drugged yet—and that the difference between undruggable and demonstrably drugged could be a matter of drug delivery.
Exosomes—a type of natural nanoparticle released by all mammalian cells as part of an intracellular communication system—have been explored as a novel drug delivery system for proteins, antisense oligonucleotides (ASOs), gene therapies, CRISPR system components, vaccines, and more. Williams says that in some cases, making a target druggable is really a question of “getting a therapeutic index that allows you to give enough drug to see the desired therapeutic effect without triggering unacceptable toxicity.”
Because exosomes and other extracellular vesicles can carry different payloads, target specific cells, and deliver drugs inside a cell, Codiak is exploring them in a variety of settings. Its two lead cancer programs entered the clinic last year. One delivers the natural immunotherapeutic cytokine interleukin-12. The other delivers an agonist of the STimulator of InterferoN Genes (STING) pathway. Both therapeutic payloads have showed promise for decades in petri dishes but dose-limiting toxicities in vivo.
“Our platform expands the therapeutic index and gives you that window to operate in,” Williams asserts. “It allows you to get the target in the cell type of interest and minimize the off-target toxicity. That’s how we think about the value of our platform.”
Codiak has disclosed an early-stage ASO-loaded exosome program in partnership with Jazz Pharmaceuticals. The program is targeting NRAS—another traditionally undruggable RAS family oncotarget. Though less common than KRAS, NRAS is found in up to 20% of melanomas.
Exosomes may also be able to reach another group of proteins that small molecules have struggled to target: transcription factors. The transcription factor STAT6 is being targeted by Codiak’s exoASO-STAT6, an ASO-loaded exosome. It is expected to enter human trials next year.
Small-molecule drugs targeting RNA
The last white whale among cancer targets may be the MYC family of oncogenes, transcription factors found in more than half of all cancers. Michael Gilman, PhD, CEO of Arrakis Therapeutics, says the biology of MYC has been well understood for some time. “I worked on MYC as a postdoc in the 80’s,” he recalls. “The question remains, how do you mobilize this biology to put it to work for patients?”
Arrakis’s approach to targeting transcription factors involves modifying small-molecule chemistry to target RNA, to block translation, interfere with splicing, or both. The company undertook large deep-sequencing experiments to determine what a druggable site on an RNA looks like, Gilman says. The company hasn’t yet disclosed which targets it is pursuing, but it says the challenges for undruggables extend beyond just the biology or the chemistry.
“We’re talking to MYC key opinion leaders,” Gilman says. “But by definition, you don’t know how to develop undruggables clinically. There’s no path established for those.”