Identification and functional characterization of a novel best-in-class anti-CD79b CAR T cell immunotherapy using a multidimensional screening strategy

Abstract Introduction: Chimeric antigen receptor T cell therapy targeting CD19 (CART19) has revolutionized the treatment of large B cell lymphomas (LBCL). However, most patients either fail to respond or relapse, frequently due to CD19 antigen loss. In this context, CD79, which is highly expressed on B-cell malignancies, emerges as a promising alternative target. The clinical success of polatuzumab vedotin further supports the therapeutic relevance of CD79 targeting, particularly with CAR T cells. However, for clinical utility, this product should be at least equally effective as CART19. Therefore, we sought to develop a best-in-class anti-CD79 CART (CART79) for B-cell malignancies using a multidimensional screening strategy.Methods and Results: To develop an optimal CART79, we implemented a two-phase screening strategy. In the first phase, we engineered and screened a library of 24 CAR constructs using 6 different public domain single-chain variable fragments (scFvs), two scFv orientations (H-L or L-H) and 2 hinge domains (CD8a or truncIgG4). All constructs included a CD8 transmembrane domain and a 4-1BB co-stimulatory domain. Constructs were evaluated in vitro for expansion potential (CTV proliferation), cytotoxicity (luciferase-based), and cytokine production (ELISA) using the CD79+ LBCL cell line OCI-Ly18. This functional screen led to the selection of 6 top candidates (polatuzumab L-H truncIgG4, 4447 L-H CD8a, 4447 L-H truncIgG4, 4447 H-L truncIgG4, 4450 H-L CD8a, and 4450 L-H truncIgG4). In particular, these 6 CART79 demonstrated significantly higher specific killing compared to CART19 (p<0.001) at day D6 of coculture. Further, all of them had a significantly increased interferon-gamma production when cocultured with OCI-Ly18 compared to the other CART79 (p<0.05). These 6 top candidates were evaluated in vivo using NSG mice injected with 5x106 OCI-Ly18 cells subcutaneously on D0, followed by 2.5x106 CAR+ T cells intravenously on D15. At D15 post-CART injection, polatuzumab L-H truncIgG4 emerged as the most effective CART79 candidate in vivo, achieving tumor control comparable to CART19 (p=0.25) and demonstrating the most consistent response among the 6 tested CART79 constructs. In the second phase, we aimed to discover novel anti-CD79b binders. Balb/c mice were immunized with human CD79b antigen, followed by hybridoma generation and immunohistochemical screening using frozen human tonsil sections to identify specific binders. Two new murine anti-CD79b scFvs (clones 128 and 151) were identified and incorporated into 4 new CART constructs (2 orientations per clone, truncIgG4 hinge, 41BB-CD3z construct). These new CART79 were compared head-to-head with polatuzumab L-H truncIgG4 in vitro using OCI-Ly18 for cytotoxicity, cytokine secretion, and proliferation assays. The 151 H-L CART demonstrated the highest specific killing activity with a mean of 85.5%, showing significantly greater cytotoxicity than polatuzumab L-H truncIgG4 (70.8%, p = 0.049) as well as numerically better than CART19 (77.4%, p = 0.16). This was in line with our in silico mechanical properties modeling of novel scFv showing that 151 H-L exhibited the highest mechanical stability among all tested, with a peak unbinding force of 648.92 kJ/mol/nm, well above the range observed for other clones (361.77-531.91 kJ/mol/nm) using steered molecular dynamics. These observations were confirmed in vivo in NSG mice injected subcutaneously with 5×106 OCI-Ly18 cells and treated with 2.5×106 CART+ cells intravenously. At D14 post-injection, the 151 H-L truncIgG4 CART showed the best in vivo expansion compared to both CART19 and polatuzumab L-H truncIgG4 (p=0.040 and 0.045, respectively). At D21 post-injection, 151 H-L truncIgG4 CART treated mice demonstrated effectively controlled tumor growth in the same range as polatuzumab L-H truncIgG4 (p=0.45) compared to UTD treated mice (p=0.048). Conclusions: This study presents a rational, stepwise strategy to develop CART79 by combining engineering of publicly available scFvs with the generation of novel anti-CD79b scFvs. Through comprehensive in silico, in vitro, and in vivo evaluation, we identified several lead candidates with superior functional properties. Among them, the 151 H-L truncIgG4 construct emerged as the most effective, demonstrating robust binding, potent cytotoxicity, and strong in vivo tumor control. This CART79 represents a top candidate option for clinical development, particularly in CART19-refractory LBCL.