Gene regulatory network determinants of rapid recall in human memory T cells

Rapid recall is the hallmark of memory T cells. While naïve T cells require days to mount an effector response to a new threat, antigen-experienced memory T cells can produce cytokines within hours of the repeatencounter. The establishment of memory and control of rapid recall across lifespan is poorly understood, yet the mechanisms are fundamental to pathogen defense and immunological diseases. Epigenetic poising is a likely mechanism. Indeed, compared to naïve, memory cells exhibit enhanced chromatin accessibility proximal to rapid recall genes, but the transcription factors (TFs) that establish, maintain and utilize these putative regulatory elements are unknown. Here, we leverage single-nuclei (sn)multiome-seq (simultaneous snRNA-seq and snATAC-seq) to (1) characterize the dynamic activation responses of naïve and memory CD4+ T cell subsets and (2) reconstruct the underlying gene regulatory networks, at genome scale. Our analysis uncovered thousands of genes and putative regulatory elements with rapid-recall dynamics, shared and unique across the several memory populations. A core of memory-associated TFs (MAF, PRDM1, RUNX2, SMAD3 and KLF6) was predicted to orchestrate rapid recall and maintain accessible chromatin at rapid-recall gene enhancers in resting memory cells. KLF6 binding to its predicted target genes was confirmed by ChIP-seq, while the memory-associated activities of all five factors replicated in independent scRNA-seq studies. Using GWAS to link our T cell populations and their dynamic chromatin landscapes to human phenotypes, we nominate CD4+ T cell populations, rapid recall responses and gene regulatory mechanisms that might mediate genetic risk to autoimmune and inflammatory diseases. Overall design: We applied sc-multiome-seq technology (tandem transcriptome and chromatin accessibility) to a TCR activation time course of naïve and memory human CD4+ T cells, to elucidate the molecular drivers of rapid recall at genome scale. WGS was used to deconvolute donors. ChIP-sequencing of a CD4T cells stimulated for 5 hours was performed with two separate KLF6 antibodies to confirm the GRN-predicted role of KLF6. Suspensions of primary hCD4 T cells from healthy donors were partially depleted of CD27+ cells (TCM and naive populations) to improve resolution of TEM populations. TCR stimulation time course includes four time points: resting, 2.5h (early activation), 5h (peak activation for memory cells), and 15h (late activation), for a total of 16 scMultiome-Seq experiments (4 time points x 4 donors; each sample includes fifth spike-in resting control donor cells).