Cellular immunotherapies, particularly CAR T cell therapies, have shown remarkable success against hematological malignancies. However, their application to solid tumors remains a significant challenge due to factors such as resistance to CAR T cell-mediated killing, antigen escape, limited persistence, and treatment-related adverse events like cytokine release syndrome (CRS) and neurotoxicity. Current strategies to improve CAR T cell efficacy often involve additional genetic modifications or drug combinations, increasing treatment complexity. This study aims to develop a high-throughput method for rapidly engineering CAR T cells with diverse functional profiles to overcome these limitations. The inherent modularity of CAR signaling domains presents an opportunity to systematically explore novel combinations, which has largely been unexplored to date due to the low-throughput nature of traditional CAR design and functional assays. Recent advances in single-cell sequencing provide the technological platform needed for high-throughput screening of a large library of CAR variants. The authors thus aim to build on previous work using pooled CAR libraries to improve CAR design and expand the diversity of CAR synthetic proteins. The approach will make use of single-cell sequencing to provide multidimensional and translatable CAR functional profiles, thus moving away from the use of engineered cell lines and reporter genes.

Existing CAR T cell therapies primarily utilize a limited set of signaling domains, mainly CD3ζ, CD28, and 4-1BB. While these combinations have shown efficacy, their limitations in solid tumors highlight the need for greater diversity in CAR design. Previous studies have explored the use of pooled CAR signaling domain libraries, employing fluorescence-activated cell sorting (FACS) and amplicon sequencing to identify novel functional variants. However, these methods often rely on limited readouts and do not fully capture the complexity of the T cell response. Single-cell RNA sequencing (scRNA-seq) has emerged as a powerful tool for characterizing CAR T cell responses, offering a high-resolution view of gene expression patterns and functional states. This study utilizes scRNA-seq combined with a novel single-cell CAR sequencing (scCAR-seq) to identify combinations of signaling domains that may elicit diverse, and improved responses compared to what has been used to date. The aim is to achieve a high-throughput screening and functional profiling methodology that could help rapidly improve CAR design and accelerate their development for various therapeutic applications.

The speedingCARs method involves a modular cloning and assembly strategy for shuffling intracellular signaling elements. A library of 180 unique CAR variants is generated by combinatorially shuffling intracellular domains from various immune co-receptors and viral proteins. This approach ensures that each CAR variant retains a total of three immunoreceptor tyrosine activation motifs (ITAMs) to trigger downstream signaling. The design also incorporates a trastuzumab-based scFv for HER2 antigen specificity, allowing for the use of HER2-expressing tumor cell lines to test functionality. CRISPR-Cas9-mediated genome editing is used to genomically integrate the CAR library into the TCR alpha chain (TRAC) locus of primary human T cells from healthy donors. This approach ensures monoclonality and consistent transgene expression in each cell. Following genome editing, CAR T cells are isolated by FACS. The resulting pooled CAR T cell library is then co-cultured with HER2-expressing breast cancer cells (SKBR3). A second cell line, MCF-7, is also used, though at lower expression levels of HER2. After co-culture, CAR T cells are isolated by FACS and subjected to scRNA-seq and scCAR-seq. scRNA-seq profiles the transcriptional phenotypes of the CAR T cells, while scCAR-seq de-multiplexes the library, linking each cell's transcriptome to its specific CAR variant. The resulting data are then analyzed using bioinformatic methods such as unsupervised clustering, differential gene expression analysis, and gene set enrichment analysis to identify CAR variants with unique functional profiles. Selected CAR variants are further characterized using functional assays, including cytokine secretion analysis, and cytotoxicity assays using both 2D and 3D tumor models. Finally, the scRNA-seq data of CD8+ CAR T cells is mapped onto a published dataset of tumor-infiltrating lymphocytes (TILs) from lung cancer patients treated with immune checkpoint blockade (ICB) therapy to identify promising CAR variants that might be associated with improved anti-tumor response in solid tumors.

The scRNA-seq analysis identified thirteen distinct T cell clusters based on marker gene expression and functional characteristics (e.g., memory, effector, cytotoxic, exhausted). The study identified five clusters strongly enriched in CAR-stimulated cells and depleted in control groups (negative controls and unstimulated 28z CAR T cells). This suggests these clusters are related to CAR-mediated activation. Analysis of CAR variant enrichment within these CAR-induced clusters (CICs) revealed several variants with unique transcriptional profiles and functional characteristics, differing from the standard 28z and BBz CARs. Principal component analysis (PCA) of the scRNA-seq data demonstrated that different signaling domain combinations induce distinct transcriptional signatures. Gene set enrichment analysis identified several enriched pathways, including those related to cytokine and chemokine signaling in CICs, as opposed to pathways associated with CD3ζ/CD28 stimulation in non-CIC clusters. Ten CAR variants were selected for further functional characterization based on their scRNA-seq profiles and mapping to TIL data. These variants displayed varying levels of CAR expression, cytotoxic capacity, and cytokine secretion. Cytotoxicity assays, using both 2D and 3D tumor models (SKBR3 and MCF7 cell lines), revealed that several variants exhibited similar or enhanced tumor killing compared to standard 28z and BBz CARs. Interestingly, some variants showed reduced proinflammatory cytokine secretion while maintaining robust cytotoxic function. Finally, mapping of the CD8+ CAR T cell scRNA-seq data onto a published TIL dataset from lung cancer patients undergoing ICB therapy revealed a high degree of overlap between the CAR-induced phenotypes and clinically relevant T cell states. This indicates a potential for these novel CAR variants in solid tumor therapies.

The speedingCARs platform successfully identified CAR variants with unique transcriptional profiles and functional characteristics, offering a significant advance in CAR T cell engineering. The high-throughput nature of the approach, combined with the comprehensive analysis of single-cell data, allows for the rapid exploration of a vast combinatorial space of signaling domains. The identification of CAR variants with reduced proinflammatory cytokine secretion while maintaining potent antitumor activity is particularly significant, as it may contribute to safer and more effective CAR T cell therapies. The mapping of CAR-induced phenotypes to TILs from responding patients provides translational relevance, potentially highlighting CAR variants suitable for treating solid tumors. The approach also allows for the investigation of signaling domains that have not been previously explored, potentially leading to the discovery of novel immunological profiles and functions. While further in vivo studies are necessary to validate the clinical potential of these new CAR variants, the results strongly suggest that speedingCARs is a powerful tool for developing next-generation CAR T cell therapies.

The speedingCARs method provides a robust and efficient platform for engineering CAR T cells with diverse functional profiles. The integration of signaling domain shuffling, single-cell sequencing, and mapping to clinically relevant TIL data enables the rapid identification of promising CAR variants with potentially enhanced therapeutic properties. This study showcases the power of this approach in expanding the repertoire of CAR T cells, paving the way for more personalized and effective cancer immunotherapies. Future work should focus on in vivo validation of the identified CAR variants and exploring their applications in different cancer types and therapeutic settings.

The study primarily focused on in vitro assays. While the in vitro results are promising, further in vivo studies are needed to confirm the efficacy and safety of the identified CAR variants in preclinical models. The co-culture system used might not fully recapitulate the complexities of the tumor microenvironment. Mapping to TIL data from lung cancer patients might not be fully generalizable to all types of solid tumors. The relatively short co-culture time of 36 hours could have limited the identification of variants that might exhibit unique phenotypes at later time points.