Single-cell metabolic profiling of post-infusion CD22 and CD19/CD22 CAR T-cells reveals a shift toward amino acid–driven oxphos informing SLC-armored CAR design

Abstract

Abstract Chimeric antigen receptor (CAR) T-cell therapies have revolutionized treatment for hematologic malignancies, yet many patients fail to achieve durable remission. Insufficient in vivo CAR T-cell expansion consistently correlates with treatment failure across clinical trials. Robust anti-tumor T-cell responses require extensive bioenergetic support, with metabolic fitness emerging as a key determinant of CAR T-cell potency. However, the metabolic pathways that support CAR T cell persistence and function in vivo after infusion into patients remain poorly defined. To address this gap, we developed a comprehensive immuno-metabolic pipeline combining single-cell metabolic assays and plasma metabolomics to analyze CAR T-cells from patients enrolled in CD22 and CD19/CD22 CAR clinical trials for relapsed/refractory B-ALL (NCT02315612, NCT03448393, NCT05098613). Initial high-throughput profiling of healthy donor–derived CD19, CD22, and CD33 CAR T-cells, incorporating either CD28 or 4-1BB costimulatory domains, revealed construct-specific metabolic and functional phenotypes. Across constructs and donors, protein translation—measured by puromycin incorporation—emerged as a robust marker of metabolic activity and cytokine polyfunctionality. Based on these data, we developed protein translation-based single-cell assays to map metabolic dependencies across major metabolic pathways—including glycolysis, oxidative phosphorylation (OXPHOS), glutamine metabolism, and fatty acid oxidation—in patient-derived CAR T-cells pre- and post-infusion. Pre-infusion CD22 and CD19/CD22 CAR T-cells exhibited a highly glycolytic phenotype with minimal reliance on oxidative phosphorylation (n=20 samples). In contrast, post-infusion peripheral blood CAR T cells underwent marked metabolic reprogramming, characterized by a reduced glycolytic dependence and increased reliance on oxidative phosphorylation (OXPHOS) and glutamine uptake (n=29 samples, day 7–14 post infusion). To further delineate metabolic heterogeneity among CAR T-cell subsets, we combined translation-based assays with spectral cytometry in the CD22 CAR cohort (n=12 patients). Notably, in pre-infusion samples, enrichment of CAR T-cell clusters with high OXPHOS dependence correlated with higher expansion, a memory-like phenotype, and complete remission. In post-infusion samples, globally elevated protein translation was associated with higher expansion, and OXPHOS-dependency characterized a CAR T-cell cluster exhibiting a stem-memory phenotype (CCR7High CD62LHigh CD127High TCF1High). Collectively, these data identify OXPHOS-driven translation and amino acid metabolism as key metabolic programs sustaining in vivo CAR T-cell function. Building on these findings, we explored the amino acid environment of post-infusion CAR T-cells. Plasma metabolomics from CD22CAR trial patients (n=20) revealed significant depletion of glutamine and arginine in individuals experiencing cytokine release syndrome (CRS). While such an amino acid-scarce environment may limit CAR T-cell function, this effect could potentially be overcome by augmented expression of metabolite solute carrier (SLC) transporters. Supporting this hypothesis, reanalysis of published scRNA-seq data from post-infusion CD19 CAR T-cells (Haradhvala et al., Nat Med, 2022) showed that CD8+CAR+ T-cells from complete responders expressed higher levels of SLCs, particularly those mediating amino acid uptake. Functional perturbation studies further confirmed the role of amino-acid SLCs: knockdown of either the glutamine (SLC1A5) or arginine (SLC7A1) transporter impaired OXPHOS, reduced stem-memory frequency, and diminished cytotoxicity upon repeated antigen challenge. Guided by these results, we engineered “MetaboArm” CAR T cells co-expressing SLC transporters to enhance amino acid uptake and improve metabolic fitness. Constructs incorporating glutamine or arginine transporters—SLC1A5, SLC7A1, or SLC38A9—significantly increased OXPHOS activity and enhanced anti-leukemic efficacy both in vitro and in vivo. Together, this study establishes the first clinical-trial–based metabolic atlas of post-infusion CAR T cells, identifying amino acid–driven OXPHOS via SLC transporters as a central determinant of therapeutic efficacy and guiding rational metabolic engineering of next-generation CARs.