Rational design of TGFbRII dominant negative armored GPC3 CAR-T cells for the treatment of hepatocellular carcinoma

Abstract Chimeric antigen receptor (CAR)-T therapy has yielded impressive clinical results in hematological malignancies and it is a promising approach for solid tumor treatment. However, toxicity is a concern hampering its broader use. In selecting a lead CAR-T candidate against the oncofetal antigen glypican 3 (GPC3), we employed a strategy to manage the toxicity profile by lead selecting a single-chain variable fragment with reduced affinity characterized by a high dissociation rate. Compared to a high-affinity construct, the low-affinity CAR maintained cytotoxic function but showed reduced toxicity in a mouse model of hepatocellular carcinoma (HCC). The HCC microenvironment is rich of the immunosuppressive cytokine TGFβ, prompting us to armor the low affinity GPC3 CAR-T cells with a dominant-negative TGFβRII (TGFβRIIDN). In vitro, expression of TGFβRIIDN nearly abolished TGFβ-induced SMAD2/3 phosphorylation and reduced suppression of IL-2 and IFNg production. In a xenograft model of a human TGFβhigh HCC, armored CAR-T cells had 10/10 complete responders (CR) while the unarmored CAR-T had 4/10 CRs. In four TGFβhigh patient-derived xenograft models armored CAR-T cells achieved 60–90% tumor growth inhibition (TGI) while the unarmored CAR-T achieved no significant TGI. The armored CAR-T cells infiltrated TGFβhigh HCC tumors more abundantly than their unarmored counterparts and expressed lower levels of immune checkpoint markers PD1 and LAG3. In line with these observations, we detected significantly more IFNg at peak response, confirming in vivo functional superiority of TGFβRIIDN armored CAR-T therapy. These data support the clinical development of AZD5851, GPC3 CAR-T armored with TGFβRIIDN.