Development of Humanized CD19-Chimeric Antigen Receptor for B Cell Malignancy

Abstract

VV and LS: Co-first authors CB and XW: Co-senior authors The adoptive transfer of T cells expressing a CD19-targeted chimeric antigen receptor (CD19-CAR) to treat patients with B cell malignancies, including acute lymphoblastic leukemia (ALL) and non-Hodgkin's lymphoma, has resulted in impressive clinical response. Despite initial success of various CD19-CAR T cell products, disease relapse due to lack of persistence is still the leading cause of the treatment failure. One mechanism that contributes to decreased CAR T cell persistence following infusion is the development of humoral and/or cellular immune responses against the CAR. Nearly 90% of clinical CD19-CAR T cell products used worldwide incorporate a CD19scFv derived from the mouse monoclonal antibody FMC63; following infusion, a patient's immune system may mount an anti-CAR response against the mouse-derived scFv and remove the CAR T cells from the patient's body. In this study, murine FMC63scFv was humanized using in silico complementary determining regions (CDR) grafting method. Following humanization, we used protein engineering techniques to optimize the affinity of the humanized CD19scFvs (huCD19scFvs), yielding two lead candidate scFvs, VH4vκ1 and 4D5. The affinity and thermal stability of different scFvs were assessed via yeast surface display. Humanized scFvs had lower affinity, VH4vκ1 (KD = 8.3 ± 1µM), 4D5 (KD = 45 ± 30µM) than FMC63scFv (KD = 4.8 ± 0.7nM). Comparing with FMC63scFv, the humanized variants showed enhanced stability, VH4vκ1 (Tm = 62 ± 0.7°C), 4D5 (Tm = 62 ± 1°C) versus FMC63scFv (Tm = 54 ± 1°C). The two humanized CD19-targeted scFvs (huCD19scFvs) were cloned into CAR constructs (huCD19CAR) containing the ch2Δ spacer, CD4 transmembrane domain, 4-1BB costimulatory and CD3ζ domains followed by truncated EGFR. As a control, murine FMC63scFv was incorporated into the same CAR backbone. HuCD19CAR lentiviral transduction of healthy donor T cells showed efficient CAR expression by EGFR staining with flow cytometry. Both huCD19CARs show similar characteristics in terms of growth and phenotype to the murine CAR. Moreover, the two huCD19 CARs showed specific activity against CD19+ tumor cells in vitro as indicated by degranulation and IFN-γ secretion. Interestingly, among the two huCD19CAR constructs, higher affinity VH4vκ1 CAR showed superior proliferative capacity to 4D5 CAR upon antigen stimulation in vitro. In vivo potency of the huCD19 CARs was evaluated by using 0.5 × 106 Raji tumor-bearing NOD-scid IL2Rgammanull (NSG) mice, treated with 1 × 106 CAR T cells. VH4vκ1 huCD19CAR exhibited comparable antitumor activity to FMC63CAR while outperforming 4D5 huCD19CAR. To investigate the targeting ability to low- and high-CD19 expressing B cell malignancies, we established a CD19-low expressing in vivo Raji tumor model. VH4vκ1 huCD19CAR exhibited more efficient antitumor activity against CD19-low Raji tumor when compared to the 4D5 huCD19CAR (p < 0.01). In summary, our study shows that the VH4vκ1 humanized CAR is similar in efficacy to the FMC63 murine CAR confirmed by both in vitro and in vivo studies. Moreover, comparison of two low affinity humanized CARs indicates that there is a threshold to affinity that can render a CAR ineffective. It is important to note that the VH4vκ1 humanized CAR rescued low-CD19 tumor which could overcome heterogeneity of antigen expression for effective CAR T cell targeting. These characteristics make humanized CD19-CAR T cells generated with VH4Vk1scFv an ideal platform for clinical translation, with the potential for improved potency.