KRAS amplification creates a targetable pMHC antigen for T cell engager therapy to overcome KRAS inhibitor resistance

KRAS amplification has emerged as a shared mechanism of resistance to KRAS-targeted therapies, including KRAS G12C and KRAS-multi small molecule inhibitors. Wild-type KRAS allele amplification also defines a subset of aggressive gastroesophageal cancers with poor outcomes. To exploit this alteration, we sought a mutation-agnostic, HLA-A*02:01-restricted KRAS pMHC antigen selectively presented in KRAS-amplified tumors. Using peptide prediction & targeted immunopeptidomics, we identified the amplified antigen “kAMP.A2,” enabling a strategy to selectively target KRAS-amplified tumors with a bispecific pMHC×CD3 T cell engager (TCE). Quantitative immunopeptidomics across multiple models estimated kAMP.A2 at ~10 2 -10 3 copies-per-cell (CPC) in KRAS-amplified cell lines (MKN1*, 119 CPC; COR-L23*, 555; HSKTC*, 1432) and undetectable or <10 CPC in non-amplified cell lines (A375, NCI-H661, ND; NCI-H520*, 9). A*02:01-negative lines (*) were A*02:01-engineered. KRAS pathway inhibition further increased antigen levels: treatment of 10 nM RMC-7977 in COR-L23 cells (G12V; KRAS CN=16), produced a 2.5× rise in surface HLA and an 8-fold increase in kAMP.A2, reaching ~4,000 CPC at 48 h. KRAS amplification thereby establishes an exploitable pMHC therapeutic window-analogous to gp100 or PRAME-that is further broadened by KRAS inhibition. Importantly, NRAS & HRAS paralogs of kAMP.A2 were not detected in any model. Consistent with this, stability profiling showed kAMP.A2 to be markedly more stable at 37 °C (t ½ ≈ 45 min) than NRAS/HRAS peptides (t ½ < 4 min), supporting kAMP.A2 as a selective TCE target with minimal paralog cross-reactivity. Screening our ultra-large human naïve antibody library with phage/yeast display and stringent counter-selection yielded the TCR-mimicking antibody PK313, exhibiting high affinity (K D = 5.3 nM) and strict specificity confirmed by SPR as well as by X-scan and healthy-tissue pMHC libraries (>10 2 & ~10 4 peptides) with no high-affinity off-targets. A 2.8-Å cryo-EM structure revealed extensive kAMP.A2-specific contacts. Reformatted as a TCE, PK313 induced potent cytotoxicity in KRAS-amplified models (MKN1*, EC 50 = 1.1 nM; COR-L23*, 150 pM; HSKTC*, 160 pM) and showed enhanced activity with KRAS inhibitors. Structure-guided maturation is underway to advance PK313 toward development candidate nomination. Our data demonstrate that KRAS amplification-whether intrinsic or emerging under therapeutic pressure-creates a robust, tumor-amplified pMHC antigen with exceptional potential for therapeutic exploitation. PK313 enables selective targeting of KRAS-amplified tumors as a monotherapy or in combination with KRAS small molecule inhibitors to enhance treatment durability and provides a therapeutic strategy for A*02:01 patients (~42% in US population), who are currently excluded from A*03/A*11 KRAS-mutant neoantigen TCR-T/CAR-T/TCE programs.