A motif-vocabulary model of CAR T-cell intracellular domains identifies determinants of immunophenotype differentiation

Chimeric antigen receptor (CAR) T-cell efficacy depends critically on the costimulatory domain, which shapes downstream signaling and the immunophenotype of manufactured products. Despite mechanistic evidence that immune receptors function as motif-based signaling scaffolds, CAR engineering has focused on a narrow set of costimulatory domains--principally CD28 and 4-1BB--leaving much of the available signaling design space unexplored. Here, we screened 1,243 naturally occurring intracellular domains as costimulatory modules in an anti-CD20 CAR backbone in primary human CD8+ T cells and quantified construct enrichment across memory-differentiation and PD-1-defined immunophenotypic compartments. Using Eukaryotic Linear Motif (ELM) annotations, we analyzed motif-phenotype associations via complementary statistical approaches: Mann-Whitney screening and negative binomial regression identified ELM features associated with differential construct representation, while Dirichlet-Multinomial modeling--which properly accounts for the compositional structure of FACS-partitioned data--revealed that individual ELMs do not significantly alter phenotype distributions. This discrepancy indicates that single motifs primarily affect proliferation or survival rather than differentiation fate. In contrast, construct-level analysis using a leave-one-out compositional test identified specific costimulatory domains with significant phenotype-shifting effects, demonstrating that particular combinations of ELMs--rather than individual motifs--determine immunophenotype. These results suggest that CAR T-cell differentiation state is governed by the integrated output of multiple signaling motifs and provide a combinatorial framework for rational costimulatory domain engineering.