Acquired resistance to sotorasib in KRAS ^G12C mutant NSCLC is vulnerable to PI3K-mTOR pathway inhibition mediated by 4E-BP1 regulator of cap-dependent translation

Sotorasib (AMG510) and adagrasib (MRTX849) have shown significant efficacy in KRAS ^G12C mutant NSCLC, but acquired resistance occurs within 6–12 months. While some resistance arises from new mutations, over half of the resistant cases lack identifiable genomic alterations. We hypothesize that resistance is driven by signaling network rewiring, creating new therapeutic vulnerabilities. To investigate acquired resistance (AR) mechanisms, multiple AR models, including cell lines (H23AR & H358AR), PDXs (TC303AR & TC314AR), CDXs (H358AR CDX), and PDXOs (PDXO303AR & PDXO314AR) were developed. H23AR and H358AR cells displayed >600-fold and 200-fold and PDXO303AR and PDXO314AR, exhibited >300-fold and >100-fold resistance to sotorasib, respectively compared to their parental counterparts, however, no additional mutations in KRAS or other potential genetic alterations were identified. The AR cells and PDXOs also showed comparable resistance to adagrasib. Proteomic and phosphoproteomic analyses in TC303AR & TC314AR PDXs identified distinct protein signatures associated with KRAS reactivation, mTORC1 signaling upregulation, and PI3K/AKT/mTOR pathway activation. PI3K protein levels were significantly elevated in AR PDXs, H23AR, and H358AR cells. Pharmacological inhibition of PI3K with copanlisib or genetic knockout via CRISPR-Cas9 restored sotorasib sensitivity, suppressed colony formation, and inhibited downstream effectors, including p-AKT, p-mTOR, p-S6, p70S6K, p-GSK3β, and p-PRAS40 in AR cells. copanlisib also sensitized both acquired and primary resistant PDXOs and synergized with sotorasib in restoring drug sensitivity. We found that p4E-BP1 was significantly upregulated in H23AR and H358AR cells, and copanlisib suppressed its expression. The level of p4E-BP1 expression was correlated with Sotorasib sensitivity in PI3K knockout clones, where the most sensitive clone displayed reduced or no p4E-BP1 expression. CRISPR-Cas9-mediated knockout of 4E-BP1, either alone or in combination with PI3K knockout, dramatically restored sotorasib sensitivity to levels comparable to parental cells. Suppression of 4E-BP1 hyperphosphorylation required dual inhibition of mTORC1 and mTORC2, and treatment with AZD8055 or sapanisertib (mTORC1/2 dual inhibitors) significantly dephosphorylated 4E-BP1 and restored sotorasib sensitivity in resistant cells and PDXOs. In contrast, everolimus (a mTORC1-selective inhibitor) did not restore sotorasib sensitivity. In PDX, CDX, and xenograft models in vivo, the combination of sotorasib with either copanlisib or sapanisertib resulted in robust, synergistic, and durable tumor regression at well-tolerated doses. These findings showed the critical role of PI3K/mTOR signaling as a bypass mechanism of resistance to KRAS ^G12C inhibitors. We conclude that mTORC1/2 mediated inhibition of p4E-BP1 and combination strategies targeting this pathway effectively overcomes acquired resistance to KRAS ^G12C inhibitors in NSCLC.