For four decades, KRAS was the most famous undruggable target in oncology. It is mutated in roughly a quarter of all human cancers — including the majority of pancreatic cancers, about a third of colorectal cancers, and a significant slice of lung cancers. Killing KRAS would be one of the highest-impact achievements in cancer drug discovery. And for forty years, nobody could do it.
Then in 2013 a single paper from Kevan Shokat's lab changed the trajectory of the entire field, and by 2021 the first KRAS inhibitor had FDA approval. This guide unpacks the biology of why KRAS was so hard, the chemistry of how it was finally drugged, and what the next chapter looks like. You can explore Sotorasib and other oncology drugs in the Drug Discovery Lab.
What KRAS does, in three minutes
KRAS is a small (~21 kDa) GTPase that lives on the inner surface of the plasma membrane. Like all GTPases, it cycles between two states: an inactive GDP-bound state and an active GTP-bound state. When growth factor receptors (such as EGFR) get activated by their ligand, they recruit guanine exchange factors that swap GDP for GTP on KRAS. The active GTP-bound KRAS then binds and activates downstream effectors — principally the RAF-MEK-ERK MAP kinase cascade and the PI3K-AKT pathway — driving cell proliferation and survival.
Normally, KRAS turns itself off relatively quickly by hydrolyzing GTP to GDP, helped by GTPase-activating proteins (GAPs). The whole cycle is tightly regulated. The most common cancer-causing mutations in KRAS — at codons 12, 13, and 61 — cripple GTP hydrolysis. The mutant protein gets stuck in the active GTP-bound state and signals constantly. The cell receives a relentless “divide” signal and tumors grow.
Why nothing worked for 40 years
KRAS was discovered as a Harvey rat sarcoma virus oncogene in the early 1980s. By 1985, it was clear that KRAS mutations were among the most common in human cancer. Drug discovery immediately tried to target it — and immediately ran into serious obstacles.
The GTP problem
The most obvious approach is to block the active site, the same way kinase inhibitors block the ATP pocket. But KRAS binds GTP with picomolar affinity. The cellular concentration of GTP is in the high micromolar range. A small molecule would have to compete with a hundred-million-fold excess of natural ligand at picomolar potency just to start displacing it. That is essentially impossible for any normal drug.
The smooth surface problem
Outside of the GTP pocket, KRAS's surface is famously flat. Druggable proteins usually have well-defined pockets and clefts where small molecules can settle in and form many interactions. KRAS doesn't have those — historically, the protein looked like an unbroken expanse with no obvious place to grab on. Multiple high-throughput screens against KRAS produced no leads.
The membrane problem
KRAS is anchored to the inner leaflet of the plasma membrane by a farnesyl group attached to its C-terminal CAAX motif. Some early efforts tried to block farnesylation with farnesyltransferase inhibitors. They worked beautifully on H-RAS, which is exclusively farnesylated. But KRAS has a backup: when you block farnesylation, KRAS gets geranylgeranylated instead and remains membrane-tethered. Farnesyltransferase inhibitors were a major disappointment in KRAS-driven cancers.
The Shokat breakthrough: covalent G12C
In 2013, Jonathan Ostrem and Kevan Shokat at UCSF published a landmark paper in Nature. They started from a different question: what if you specifically targeted the cysteine that the G12C mutation introduces? G12C is a particularly common KRAS mutation in lung cancer (about 13 percent of NSCLC). In wild-type KRAS, position 12 is a glycine. In G12C mutants, it's a cysteine — and cysteines have a reactive thiol side chain that can be covalently captured by an electrophilic small molecule.
That alone would not be enough. The bigger insight was that when KRAS is in the GDP-bound (inactive) state, a cryptic allosteric pocket opens up just below the switch II loop. This pocket is invisible in most KRAS structures because it only forms transiently when switch II is in a particular conformation. Shokat's lab found that they could design small molecules that bound this switch II pocket and reached over to form a covalent bond with the G12C cysteine.
Critically, the inhibitors only work when KRAS is in the GDP-bound state. That means the drug effectively traps KRAS in its inactive form — exactly what you want. And because it's covalent, the bond forms permanently. Once KRAS G12C is hit, it stays hit until the protein is degraded by normal turnover.
Sotorasib: the first KRAS inhibitor
Amgen took the Shokat insight and built Sotorasib (originally AMG 510). It is a covalent G12C inhibitor that binds the switch II pocket and forms a Michael adduct with the cysteine 12 thiol via an acrylamide warhead. Sotorasib was approved by the FDA in May 2021 for adult patients with locally advanced or metastatic NSCLC carrying a KRAS G12C mutation, after at least one prior line of therapy.
It was the first KRAS inhibitor of any kind to reach approval — more than 40 years after KRAS was identified as a cancer driver. Response rates in pivotal trials were in the 30 to 40 percent range, with median progression-free survival around 6 to 7 months. These numbers are not curative, but they are meaningful in a heavily pretreated patient population that had no targeted options whatsoever before Sotorasib.
Adagrasib and the second wave
Mirati Therapeutics followed with Adagrasib (Krazati), a second covalent G12C inhibitor with somewhat different properties. Adagrasib has a longer half-life (around 24 hours) and shows central nervous system penetration, which is important for brain metastases that are common in NSCLC. Adagrasib was approved by the FDA in December 2022.
Both drugs work by the same general mechanism but have different chemistry, pharmacokinetics, and side effect profiles. The two-drug landscape has given clinicians more options and given researchers a chance to study how resistance to G12C inhibitors develops in patients.
Resistance and combinations
Like every targeted therapy, KRAS G12C inhibitors eventually face resistance. Several mechanisms have been characterized:
- Secondary KRAS mutations that prevent drug binding (mutations in switch II or that lock KRAS in the GTP-bound state).
- Activation of bypass pathways like MET amplification or EGFR upregulation that drive downstream signaling without going through KRAS.
- Restoration of upstream signaling. Cells can compensate by reactivating the MAPK pathway through alternative mechanisms.
- Histological transformation from adenocarcinoma to squamous cell carcinoma in some cases.
These observations are driving the next phase of KRAS drug development: combinations with EGFR inhibitors, SHP2 inhibitors, MEK inhibitors, and immunotherapy. Several of these combinations are being tested in trials and some have shown promising activity.
The pan-KRAS frontier
KRAS G12C is only about 13 percent of NSCLC and a smaller fraction of pancreatic and colorectal cancers. The much larger pie is the rest of the KRAS mutation spectrum: G12D, G12V, G13D, Q61H, and others. None of these mutations introduce a reactive cysteine in the same place, so the Shokat covalent trick doesn't apply directly.
Several companies are now pursuing different strategies:
- Mutation-specific noncovalent inhibitorsdesigned to fit the slightly different switch II pocket of each mutant
- Pan-KRAS inhibitors that bind a conserved region of switch II and work against multiple mutations (and possibly even wild-type)
- SOS1 inhibitors that block the upstream guanine exchange factor that activates KRAS
- SHP2 inhibitors that disrupt the membrane-recruitment scaffold KRAS depends on
Several of these molecules are in early clinical trials in 2026. The pan-KRAS problem is still unsolved, but for the first time it looks tractable.
Explore Sotorasib in the Drug Discovery Lab
The Sotorasib workspace in the Drug Discovery Lab includes the full structure, SMILES, drug-like property calculations, and notes on the covalent binding mechanism. You can also load other oncology workspaces and compare KRAS G12C inhibitors against other targeted kinase inhibitors and traditional cytotoxics.
Bottom line
KRAS went from undruggable to drugged in roughly a decade thanks to one critical insight — that the G12C mutation introduces a reactive cysteine that you can covalently capture from a previously unrecognized switch II allosteric pocket. Sotorasib and Adagrasib turned that insight into approved medicines with real clinical benefit. The scientific lesson is broader: “undruggable” usually means “not yet drugged with the chemistry we've tried so far.” The next frontier — pan-KRAS inhibitors that work against the other mutation types — is now an active area of clinical development.