Conformational regulation of the KRAS molecular switch—Structural basis for targeting "undruggable" targets

Keywords: KRAS structure, conformational dynamics, Switch I/Switch II, GTP hydrolysis, effector binding, small-molecule binding pockets

KRAS is the most frequently mutated subtype of the RAS family, and its dysregulation drives approximately one-quarter of human cancers. The KRAS protein belongs to the small GTPase family, cycling between the active GTP-bound state and the inactive GDP-bound state to function as a molecular switch. For a long time, KRAS was considered a classic "undruggable" target due to its nearly spherical smooth surface, lack of deep hydrophobic binding pockets, and picomolar-level GTP affinity. However, the breakthrough success of covalent KRAS G12C inhibitors has completely overturned this perception, sparking a wave of research and development targeting other mutants and pan-RAS inhibitors. Understanding the three-dimensional structure of KRAS, the dynamic equilibrium of its conformations, and the molecular mechanisms by which mutations affect it is a core prerequisite for designing novel inhibitors.

KRAS consists of approximately 189 amino acids with a molecular weight of about 21 kDa. Its core is the G domain (GTPase domain), which includes the phosphate-binding loop (P-loop), Switch I region (residues 30-38), and Switch II region (residues 59-76). The G domain is responsible for nucleotide (GTP or GDP) binding and hydrolysis. The C-terminal hypervariable region (HVR, residues 165-189) anchors KRAS to the inner cell membrane through lipid modifications such as farnesylation, which is essential for its subcellular localization and signal transduction.

KRAS exists in two major conformational states: the GTP-bound state ("ON" state) and the GDP-bound state ("OFF" state). In the ON state, the Switch I and Switch II regions adopt stable closed conformations, exposing the effector-binding interface and enabling interaction with downstream proteins such as RAF and PI3K. In the OFF state, the two switch regions become loose and disordered, losing their effector-binding capacity. The dynamic equilibrium between these two states is governed by GTP hydrolysis (catalyzed by GAP proteins) and GDP/GTP exchange (catalyzed by GEF proteins), forming the biochemical basis of its cycle.

KRAS mutations are primarily concentrated at codons 12, 13, and 61. These residues are located near the phosphate-binding loop and switch regions, playing critical roles in GTP hydrolysis and nucleotide exchange. Glycine 12 mutations (e.g., G12C, G12D, G12V) disrupt GAP-assisted hydrolysis, causing KRAS to preferentially remain in the GTP-bound active state and achieve ligand-independent activation. Glutamine 61 mutations (Q61H/L/R) directly interfere with the positioning of water molecules and the arginine finger during hydrolysis, also leading to loss of GTPase activity.

Importantly, different mutants exhibit variations in protein conformational dynamics and effector affinity. The G12C mutation exposes an accessible pocket in the Switch II region of the GDP-bound state, providing a binding site for covalent inhibitors. The G12D mutation alters the interaction dynamics with GEFs like SOS1 through changes in the hydrogen bond network. The G12V mutation induces stronger hydrophobic perturbations, more thoroughly inhibiting GAP-mediated hydrolysis. These mutation-specific conformational differences form the molecular basis for developing selective inhibitors.

Traditionally, the challenges of directly targeting KRAS included its extremely high GTP-binding affinity (picomolar level), intracellular GTP concentrations far exceeding GDP, and the absence of deep hydrophobic pockets suitable for drug binding. For the G12C mutation, researchers discovered an inducible binding pocket in the Switch II region of the OFF state (Switch II pocket, S-IIP). This led to the design of acrylamide-based compounds that covalently bind to the introduced cysteine, irreversibly locking KRAS in the GDP-bound inactive state. Sotorasib and adagrasib are representative examples of this approach.

For the G12D mutation (lacking a reactive cysteine), research strategies shifted toward non-covalent inhibitors targeting mutation-induced electrostatic and hydrophobic surface changes. Molecules like MRTX1133 stabilize the GDP-bound state by occupying switch regions, blocking effector binding. Another strategy involves developing pan-RAS inhibitors targeting conserved regions shared by all RAS subtypes, such as the Switch II region or the C-terminal membrane-binding interface. Additionally, PROTAC technology, which targets KRAS protein degradation by recruiting E3 ubiquitin ligases, is being actively explored to achieve directed clearance of mutant KRAS.

X-ray crystallography and cryo-electron microscopy have provided critical insights for the rational design of KRAS inhibitors. Numerous structures of KRAS mutants in complex with inhibitors have been resolved, revealing binding modes and induced conformational changes. Molecular dynamics simulations and free energy perturbation calculations are used to predict the affinity and selectivity of new molecules for mutants, significantly accelerating lead compound optimization. Combined with biochemical detection platforms such as surface plasmon resonance (SPR) and TR-FRET, medicinal chemists can perform high-throughput screening of candidate compounds and expand chemical space through structure-guided virtual screening.

The molecular switch mechanism of KRAS has evolved from a simple GTPase cycle to a complex landscape encompassing mutation-specific conformational perturbations, membrane localization dynamics, and the dynamic assembly of effector complexes. In-depth structural biology insights and innovative chemical probe designs have collectively propelled KRAS from "undruggable" to a paradigm of breakthrough. Future inhibitor development targeting G12D, G12V, G13D, and other mutations—as well as pan-inhibitors targeting all RAS mutants—will continue to rely on a refined understanding of KRAS conformational dynamics.