Betacellulin drives therapy resistance in glioblastoma

Background. The transcription factor signal transducer and activator of transcription 3 (STAT3) drives progression in glioblastoma (GBM), suggesting STAT3 as a therapeutic target. Surprisingly however, GBM cells generally show primary resistance to STAT3 blockade. Methods. Human glioblastoma cell lines LN229, U87, SF767, and U373, and patient-derived xenografts (PDXs) GBM8 and GBM43 were used to evaluate epidermal growth factor receptor (EGFR) activation during STAT3 inhibition. Protein and gene expression experiments, protein stability assays, cytokine arrays, phospho-tyrosine arrays and EGFR-ligand protein arrays were performed on STAT3 inhibitor-treated cells. To evaluate antitumor activity, we administered a betacellulin (BTC)-neutralizing antibody alone and in combination with STAT3 inhibition. BTC is an EGFR ligand. We therefore treated mice with orthotopic xenografts using the third-generation EGFR inhibitor osimertinib, with or without STAT3 knockdown. Results. We demonstrate that both small-molecule inhibitors and knockdown of STAT3 led to expression and secretion of the EGFR ligand BTC, resulting in activation of EGFR and subsequent downstream phosphorylation of nuclear factor-kappaB (NF-kappa B). Neutralizing antibody against BTC abrogated activation of both EGFR and NF-kappa B in response to inhibition of STAT3; with combinatorial blockade of STAT3 and BTC inducing apoptosis in GBM cells. Blocking EGFR and STAT3 together inhibited tumor growth, improving survival in mice bearing orthotopic GBM PDXs in vivo. Conclusion. These data reveal a feedback loop among STAT3, EGFR, and NF-kappa B that mediates primary resistance to STAT3 blockade and suggest strategies for therapeutic intervention.