ABSTRACT
BACKGROUND: Golgi phosphoprotein 2 (GOLPH2) has been shown to be involved in chronic inflammatory processes and carcinogenesis. GOLPH2 is prominently overexpressed in hepatocellular carcinoma, melanoma, glioblastoma, prostate, lung, and colorectal cancer. With a low and tightly regulated expression in non-malignant tissues, GOLPH2 has been proposed as an attractive target for cancer therapy. However, GOLPH2 is predominantly located intracellularly and when situated outside of the cell it is proteolytically cleaved and shed from the cell surface. Until now, GOLPH2 has been regarded as an "undruggable" target. OBJECTIVE: We sought to create antibodies that specifically bind to GOLPH2 overexpressing tumor cells. PATIENTS AND METHODS: Antibodies binding to membranous GOLPH2 despite shedding of the protein were generated from a scFV library screening. These antibodies target the part of GOLPH2 that remains at the cell surface after proteolytic cleavage. These antibodies were then tested in vitro and in vivo. RESULTS: Two candidates (G2-1 and G2-2) showed target specific binding in vitro. Utilizing a tumor array (n = 128 tumors) with G2-2 and a reference antibody, a GOLPH2 expression scoring system was established. Rapid internalization of the antibodies was noted so this was exploited to deliver a toxic payload of pyrrolobenzodiazepine (PBD). In two patient-derived xenograft (PDX)-models, colorectal and lung cancer, the G2-2 antibody drug conjugate (ADC) displayed high efficacy with significant tumor responses (P = 0.001; P = 0.013) and improved survival (P = 0.0001; P = 0.0011) compared with controls. CONCLUSIONS: Treatment with GOLPH2-directed antibodies induces durable responses in colorectal and lung cancer models. With a robust companion assay for GOLPH2 positivity at hand our findings prepare for the translation into a clinical trial.
