Document resource
Background Metastatic castration-resistant prostate cancer (mCRPC) is a lethal, treatment-refractory stage of disease, frequently associated with bone metastases and poor prognosis. While chimeric antigen receptor (CAR) T cell therapy holds promise for mCRPC, clinical translation has been limited by toxicities associated with validated targets such as PSMA. In our first-in-human trials of PSMA-CAR T cells armored with a dominant-negative TGF-β receptor II (dnTGFβRII) ( NCT03089203, NCT04227275), and in the ongoing REGN5678 αPSMA×CD28 bispecific study (NCT03972657), good responses were observed but accompanied by grade ≥3 cytokine release syndrome (CRS), hemophagocytic lymphohistiocytosis (HLH), and immune effector cell-associated neurotoxicity syndrome (ICANS). In addition to toxicity, antigen heterogeneity—specifically, the growth of PSMA-negative tumor clones— decreases the therapeutic efficacy of CAR T cells.Methods To address these challenges, we developed a multi-pronged, all-in-one CAR T cell platform with triple-armoring and dual-targeting functionality. Our design includes: (1) targeting the prostate cancer-associated antigen STEAP2 to reduce on-target/off-tumor toxicity; (2) localized secretion of αPSMA×CD28 bispecific to activate bystander T cells against PSMA-positive variants while limiting systemic toxicity; (3) fine-tuning CD28 affinity to enhance anti-tumor activity and reduce CRS risk; and (4) overcoming immunosuppression by armoring CAR T cells with dnTGFβRII. Bispecific secretion is restricted to NFAT-activated CAR T cells upon STEAP2-engagement, confining CD28 co-stimulation to the tumor microenvironment (TME) and broaden safety profile.We tested our all-in-one CAR T cell platform in vitro using a panel of TGF-β-secreting human prostate cancer cell lines with varying STEAP2 and PSMA expression to discriminate STEAP2-CAR- versus αPSMA×CD28-mediated cytotoxicity. For in vivo efficacy, we established subcutaneous (s.c.), intravenous (i.v.), and intraosseous (i.o.) xenograft models in immunodeficient mice to mimic primary tumors, systemic disease, and bone metastases, respectively. To evaluate safety, we successfully validated a novel huPSMA knock-in (KI) mouse model that recapitulates human PSMA tissue expression.Results We successfully engineered multiple αPSMA×CD28 constructs within a dnTGFβRII-armored STEAP2-CAR T cell backbone. Bispecific secretion was confirmed by ELISA and dual-binding assays to PSMA+ tumor cells and CD28+ T cells. Engineered CAR T cells demonstrated robust in vitro cytotoxicity against heterogeneous prostate cancer cell lines. Furthermore, our preliminary in vivo results show high potency in tumor killing using the LNCaP (STEAP2+, PSMA+) tumor model.Conclusions This all-in-one platform integrates tumor specificity, dual targeting, localized immune activation, and TME resistance, offering a promising strategy for improving the safety and efficacy of CAR T cell therapy in mCRPC and other solid tumors.