Histone Methylation Mark Analyses Distinguish Carts Manufactured from Distinct Sources and Uncover Novel Transcription Factors Associated with In Vivo Function of Carts after Infusion into DLBCL Patients That Are Not Identified By RNA-Seq

Abstract

Chimeric antigen receptor-modified T cell (CART) therapy has heralded a revolution in cancer immunotherapy. Yet, understanding the qualities that govern effects on in vivo function of CART infusion products (IP) has been challenging. RNA sequencing (RNA-seq), albeit high dimensional in scope, is limited by the dynamic nature of transcription, especially in T cells and CARTs where activation may hinder identification of transiently or minimally expressed genes. We investigated whether epigenomic analyses of histone 3 lysine methylation marks (H3Kme) might uncover genes associated with the potential of starting T cell subsets and CART IP that could not be identified by RNA-seq. We used Cleavage Under Targets and Restriction Using Nuclease (CUT&RUN) to assess whole genome transcriptionally permissive H3K4me2 and repressive H3K27me3 marks in naïve (N), central memory (CM) and effector memory (EM) CD8+ T cells and in CARTs of different potencies and manufactured from different sources. We identify, for the first time, epigenetic predictors of CART cell expansion in a clinical trial (NCT01865617). In preclinical models, CARTs derived from less differentiated N and CM T cells control tumor more efficiently than their EM-derived counterparts. We conducted parallel RNA-seq and histone mark analyses on N, CM and EM CD8+ T cells. We found superior distinction between T cell subsets by individual histone mark analyses compared to RNA-seq. H3K4me2 marks identified 2.7- to 4.5-fold more differentially enriched transcription factors (TFs) than RNA-seq when comparing subsets. Statistical noise in high transcript abundance genes and transcript sparsity in low transcript abundance genes complicates the discriminative capacity of RNA-seq. To further elucidate factors governing T cell qualities we assessed combinatorial patterns of change in histone marks between CD8+ T cell subsets in both high and low transcript abundance genes that were not identified as differentially expressed by RNA-seq. Patterns of histone mark changes between subsets identified novel genes associated with T cell differentiation not seen by RNA-seq or differential enrichment of individual histone marks. We then determined whether histone mark analyses could distinguish CARTs manufactured from similar, yet distinct CD8 + T cell sources. When comparing CM- and EM-derived CART cells from healthy donors (HD) we found that analyses of histone marks identified 33- and 44-fold more genes, respectively, than RNA-seq, including TFs associated with T cell differentiation (KLF7, ZEB2, LEF1) that were not identified by RNA-seq (Fig. 1). By assessing transitional histone mark patterns (H3K4me2+H3K27me3+ and H3K4me2-H3K27me3-), we identified genesets associated with T cell differentiation, cellular stress response and metabolism, as well as master regulators associated with T cell differentiation, exhaustion, and lipid metabolism (POU2F1, RELA, NFKB, MYC, STAT1, SREBF1) that were not identified by RNA-seq (Fig. 2). Similarly, histone mark analyses distinguished and identified increased differentiation in CARTs manufactured from phenotypically CM cells isolated from DLBCL patients on a phase 1/2 clinical trial (NCT01865617) from CARTs manufactured from CM cells from healthy donors. Identification of genetic programming in CARTs associated with in vivo function in clinical trials of CART immunotherapy has been exceptionally challenging. We analysed CART IP from NCT01865617 and found that histone mark analyses, but not RNA-seq, identified differences between IP that associated with the magnitude of CART counts in patient blood after infusion. We demonstrated that the proliferation-augmenting TF, KLF7, and its epigenomic regulome, were associated with in vivo CART cell accumulation. Histone mark analyses enable identification of novel parameters of T cell and CART quality that could be modified to improve outcomes of CART immunotherapy.