334 Computational mapping of CAR-T cell exhaustion through single-cell perturbation screens

Le résumé fourni par la source

Background While CAR-T cell therapies in hematologic cancers have shown great promise, in solid tumors additional cell engineering will be required to overcome T cell exhaustion, insufficient persistence and the inhibitory tumor microenvironment. We have developed T cell engineering and automation platforms that enable the discovery of combinatorial genetic perturbations to enhance therapeutic T cell function. Methods To identify potential modulators of T cell states and functions, we used a single-cell perturbation screening strategy in pooled and arrayed format. Pooled screens enable interrogation of large numbers of perturbations in vitro and in vivo, whereas arrayed in vitro screens also allow for complex functional assessment. In all screens, single cell transcriptional and surface protein expression profiles were collected, allowing us to map the effect of genetic perturbation to the cell states and the potential functional impact on CAR-T anti-tumor response. We performed a screen including hundreds of gain-of-function (GOF) perturbations and loss-of-function (LOF) perturbations (including paired combinatorial perturbations) in pooled and arrayed format. Functional readout of the perturbations showed a wide dynamic range of T cell proliferation and target cell killing, spanning 4 and 9 orders of magnitude, respectively. Results Computational analysis of single-cell molecular profiles revealed that perturbations can be partitioned into distinct modules based on similarities in transcription profiles. We identified perturbation modules with significant changes of gene programs such as T cell exhaustion, activation and metabolism. We have derived transcriptional signatures that correlate with desirable T cell phenotypes of increased proliferation and tumor control. Furthermore, we showed that transcriptional profiles from the pooled screen are predictive of the functional readouts from the arrayed screen, demonstrating capability of transfer learning between the two screen modalities. This also highlights the potential to use the pooled screen as the first pass for larger scale screens, with complementary arrayed screens to collect more in-depth functional readouts from nominated hits. Conclusions These results demonstrate a powerful discovery platform that may accelerate the development of efficacious CAR-T cell therapies for solid tumors.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.