Low-affinity CAR T cells exhibit reduced trogocytosis, preventing rapid antigen loss, and increasing CAR T cell expansion

Chimeric antigen receptor (CAR) T cells, specifically CAR T cells targeting CD19, have greatly improved outcomes for patients with B cell lymphoma. However, disease relapse remains a common occurrence in many patients [ 1 , 2 ]. Different mechanisms of relapse have been demonstrated, such as CD19 open-reading frame mutations, upregulation of coinhibitory ligands, and CAR T cell exhaustion [ 3 ]. However, in many relapsed patients no clear immune escape mechanisms can be identified and additional processes, such as trogocytosis, may play an important role. Trogocytosis is a process originally described in the TCR context whereby TCR internalization upon binding of peptide/MHC results in stripping of MHC from antigen presenting cells and subsequent display of MHC on the surface of the recipient T cell [ 4 ]. Trogocytosis has since been observed in many facets of natural immunity and plays an important role in immune cell development and function [ 5 ]. Recently, CAR T cells were also shown to exhibit trogocytosis when co-cultured with different types of cancer cells, including CAR T cells based on the high-affinity CD19 binding domain FMC63, which were shown to strip CD19 from lymphoma cells and incorporate it into their own plasma membrane [ 6 ]. In addition, it was shown that T cells engineered to express CD19 can be killed by CD19 CAR T cells suggesting that such fratricide might also occur spontaneously after CAR T cells acquire CD19 by trogocytosis [ 6 ]. To date, no strategies exist to limit trogocytosis in the CAR T cell context to prevent fratricide and the emergence of antigen-negative tumor cells. We, therefore, aimed to develop an approach to limit CAR T cell-mediated trogocytosis. Trogocytosis has previously been shown to correlate with TCR-peptide/MHC avidity [ 7 ], and we hypothesized that CAR affinity may have a similar relationship to CAR T cell-mediated trogocytosis. Furthermore, as current iterations of antigen-binding domains used for the construction of CARs likely exceed the affinity required for efficient target cell killing [ 8 ], we hypothesized that it may be possible to separate the pro-tumorigenic effects mediated by trogocytosis from the respective CAR’s antitumor activity by reducing CAR affinity (Fig. 1A ). This assumption is supported by the recent finding that T cells expressing a CAR targeting CD19 (“CAT”) with ~ 40-fold lower affinity ( K on = 2.2 × 10 5 M −1 s −1 , K off = 3.1 × 10 −3 s −1 , K D = 14 nM) than the clinically approved FMC63-based CAR ( K on = 2.1 × 10 5 M −1 s − 1, K off = 6.8 × 10 −5 s −1 , K D = 0.328 nM) exhibit higher efficacy and persistence than FMC63-based CAR T cells in a mouse model [ 9 ], as well as robust antitumor efficacy and persistence in two clinical trials [ 9 , 10 ]. We hypothesize that reduced trogocytosis may have contributed to this enhanced in vivo persistence. We therefore generated non-binding (∆scFv), low affinity (CAT) and high affinity (FMC63) CD19 CAR T cells (Fig. 1B ) in order to determine the impact of CAR affinity on trogocytosis. In addition, we explored the generalizability of this approach by using the CD229-specific CAR binding domain 2D3 and a single amino acid substitution variant with reduced affinity [ 11 ] (Fig. 1C ) and determined the impact on trogocytosis (Fig. 1D ). We found that low affinity CD19 and CD229 CAR T cells showed killing of target cells indistinguishable from high affinity receptors (Fig. 1E, F ) as well as unchanged interferon-γ production and proliferation when stimulated by beads covalently coated with CD19 (Supplementary Fig. 1 ). Importantly, treatment with low affinity constructs was associated with significantly higher levels of the targeted antigen on the tumor cells at the end of in vitro co-cultures (Fig. 1G, H ), suggesting reduced levels of antigen loss. We next explored whether this finding might be the result of reduced trogocytosis. Indeed, we found that high-affinity CD19 and CD229 CAR T cells had acquired membrane from their respective target cells during the co-culture, a hallmark of trogocytosis, but that use of low affinity CARs had significantly reduced this membrane transfer (Fig. 1I, J ). Additionally, we found that use of low affinity CD19 (Supplementary Fig. 2A ) and CD229 (Supplementary Fig. 2B ) CARs had also resulted in substantially reduced transfer of the respective target antigen from the tumor cells to the CAR T cells. We further observed that, while target antigen levels were higher, CAR expression had decreased significantly on FMC63 but not CAT CAR T cells upon culture with lymphoma cells, possibly indicating increased CAR internalization of high affinity CD19 CAR T cells (Supplementary Fig. 3 ). Finally, in a short-term B cell acute lymphoblastic leukemia (B-ALL) mouse model (Fig. 1K ) showing incomplete tumor control (Supplementary Fig. 4 ), we also observed significantly reduced CD19 expression on tumor cells from mice treated with FMC63 CAR T cells but not CAT CAR T cells. Taken together, these data show that it is possible to significantly limit trogocytosis by reducing CAR affinity while maintaining antitumor activity as well as clinical efficacy [ 9 , 10 ]. Fig. 1: Low affinity CAR T cells exhibit reduced trogocytosis while preserving antitumor activity. A Schema of relationship between CAR affinity, cytotoxicity, and trogocytosis. B Schema of high and low affinity CD19 scFvs. C Structure of CDR3 regions and key residues of CD229 scFv 2D3 as determined by AlphaFold [ 11 ]. D Schema of high and low affinity CD229 scFvs. Low and high affinity CAR T cells were cultured with either E CD19+ primary CLL cells or F CD229+ multiple myeloma cell line U266 at the indicated effector:target ratios for 4 h. CLL cell viability was assessed by flow cytometry using counting beads. Multiple myeloma cell viability was determined by luminescence. G CD19 MFI on CLL cells, H CD229 MFI on DB cells, I % PKH26+ CAR T cells, and J % BioTracker 555 (BT555)+ CAR T cells following a 4 h coculture with CAR T cells and PKH26-labeled primary CLL cells or BT555-labeled DB cells at an effector:target ratio of 4:1. Data represent mean ± SD from three replicates. Statistical significance was determined by two-sided Student’s t test. K Schema of trogocytosis mouse model. L Percentage of CD19+ NALM6 tumor cells in the spleens of NRG mice treated with high and low affinity CD19 CAR T cells. Data represent mean ± SD from ∆scFv ( N = 4) treated and FMC63 ( N = 4) or CAT ( N = 5) treated mice. Statistical significance was determined by two-sided Student’s t test. Full size image