Maturation and persistence of CAR T cells derived from human pluripotent stem cells via chemical inhibition of G9a/GLP

Abstract

Elucidating mechanisms of T cell development can guide in vitro T cell differentiation from induced pluripotent stem cells (iPSCs) and facilitate off-the-shelf T cell-based immunotherapies. Using a stroma-free human iPSC-T cell differentiation platform, we screened for epigenetic modulators that influence T cell specification and identified the H3K9-directed histone methyltransferases G9a/GLP as repressors of T cell fate. We show that G9a/GLP inhibition during specific time windows of differentiation of hematopoietic stem and progenitor cells (HSPCs) skews cell fates toward lymphoid lineages. Inhibition of G9a/GLP promotes the production of lymphoid cells during zebrafish embryonic hematopoiesis, demonstrating the evolutionary conservation of G9a/GLP function. Importantly, chemical inhibition of G9a/GLP facilitates the generation of mature iPSC-T cells that bear transcriptional similarity to peripheral blood αβ T cells. When engineered to express chimeric antigen receptors, the epigenetically engineered iPSC-T cells exhibit enhanced effector functions in vitro and durable, persistent antitumor activity in a xenograft tumor-rechallenge model.

Keywords: CAR-T cells; G9a/GLP; T cell differentiation; cancer immunotherapy; chemical screen; epigenetic regulation; hematopoiesis; lymphoid development; pluripotent stem cells.

Fig. 1. Small molecule screens identify epigenetic modulators that promote in vitro T cell specification.

A) Schematic illustration of small molecule screening using iPSC-derived 5F-HSPCs. B) Graph showing the result of screen using 5F-HSPCs. Small molecules with Z scores ≥ 3 were identified as primary hits. C) Relative proT cell numbers of primary hits (n = 3, mean ± SEM, * P ≤ 0.05, *** P≤0.001). C) Schematic illustration of small molecule screening using iPSC-derived CD34⁺ cells. E) Graph showing the result of screen using iPSC-HE cells. Small molecules with Z scores ≥ 3 were identified as primary hits. F) Relative proT cell numbers of primary hits (n = 3, mean ± SEM, *** P≤0.001). G) Result of a dose-response assay to detect number of proT cells generated from iPSC-HE cells treated with UNC0224 (n = 3, mean ± SEM, * P ≤ 0.05, ** Pɤ0.01, *** P≤0.001)

Fig. 2. G9a/GLP repression promotes lymphoid commitment at the expense of myeloid potential.

A) Schematic illustration of experimental approach. B) Expression of myeloid (CD33) and lymphoid (CD7) marker following lymphoid differentiation from DMSO or UNC0224 treated iPSC-HSPCs, gated on CD45⁺ cells. C) Numbers of colonies formed per 30K of iPSC-HSPCs treated with DMSO or UNC0224 during the EHT (n = 3). G, granulocyte; M, monocytes; E, erythrocyte; GM, granulocyte, monocyte; GEMM, granulocyte, erythrocyte, monocyte, megakaryocyte. D) Schematic illustration of experimental approaches using G9a CRISPRi iPSCs. E) Numbers of colonies formed per 10K of control or G9a CRISPRi iPSC-HSPCs (n = 3). G, granulocyte; M, monocytes; E, erythrocyte; GM, granulocyte, monocyte; GEMM, granulocyte, erythrocyte, monocyte, megakaryocyte. F) Frequency of lymphoid progenitor cells derived from control or G9a CRISPRi iPSC-HSPCs after 2 weeks of differentiation (n = 6, mean ± SEM, ** P≤0.01). G) Frequency of CD56⁺ NK cells derived from control or G9a CRISPRi iPSC-HSPCs after 4 weeks of NK cell differentiation (n = 3, mean ± SEM, * P ≤ 0.05). H) Frequency of CD19⁺ B cells derived from control or G9a CRISPRi iPSC-HSPCs after 5 weeks of B cell differentiation (n = 3, mean ± SEM, *** P≤0.001).

Fig. 3. G9a/GLP regulates chromatin accessibility of lymphoid genes.

A) Heatmap of upregulated and downregulated ATAC peaks in DMSO and UNC0224 treated iPSC-HSPCs (n = 2). UNC low, 500nM of UNC0224 treatment; UNC high, 1μM of UNC0224 treatment. B) GO terms of top enriched biological processes of ATAC peaks in UNC0224-treated iPSC-HSPCs compared to DMSO-treated cells. C) GO terms of top enriched biological processes of upregulated genes in UNC0224-treated iPSC-HSPCs compared to DMSO-treated cells. D) Violin plot showing normalized expression of genes encoding H3K9 methyltransferases in human primary hematopoietic progenitor and differentiated cells .

Fig. 4. G9a/GLP inhibition promotes lymphoid development during zebrafish embryonic hematopoiesis.

A) Schematic illustration of experimental approach in zebrafish. B) Frequency of Rag2:GFP⁺ cells in control and UNC0224 (1μM) treated zebrafish embryos, determined by flow cytometry assay (n = 10 embryos/sample X 6 biological replicates, * P ≤ 0.05). C-E) Phenotypic distribution plots of embryonic in situ hybridizations for Rag1, ikaros, and Lck in control and UNC0224 (1μM) treated embryos at 120 hpf (n = 20 embryos/sample X 3 biological replicates, * P ≤ 0.05, ** P≤0.01,). F) Frequency of Rag2:GFP⁺ cells in zebrafish embryos treated with DMSO, UNC0224 (1μM), ezh1 morpholino, and UNC0224 plus ezh1 morpholino (n = 11, * P ≤ 0.05, ** P≤0.01,).

Fig. 5. G9a/GLP inhibition promotes iPSC differentiation into αβ-like T cells.

A) Number of CD3⁺ T cells on week 4 of differentiation, generated from 1x10⁵ iPSC-HE cells treated with DMSO or UNC0224 during T cell specification (n = 4, * P ≤ 0.05). B) Expression of TCRαβ and TCRγδ in T cells generated at 4 weeks from iPSC-HE cells treated with DMSO or UNC0224 during T cell specification, gated on CD3⁺ cells. C) Frequencies of αβTCR⁺ T cells and γδTCR⁺ T cells in CD3⁺ iPSC-T cells treated with DMSO or UNC0224 during T cell specification (n = 4, mean ± SEM, *** P≤0.001, **** P≤0.0001). D) Heatmap showing Spearman correlation analysis of RNA-seq samples. CON-iPSC-T, iPSC-T cells treated DMSO; UNC-iPSC-T, iPSC-T cells treated with UNC0224; PBMC-T, donor PBMC-derived T cells. E) Scatterplot showing the overlapping differentially expressed genes between PBMC-T vs. DMSO iPSC-T samples and UNC iPSC-T vs. DMSO iPSC-T samples. F) Heatmap showing expression of alpha-beta T cell lineage commitment genes, defined by GO biological process (n = 4).

Fig. 6. Single-cell RNA-seq analysis of iPSC-T cells derived via G9a/GLP inhibition.

A) UMAP visualization of DMSO-treated or UNC0224-treated cells at week 4 of T cell differentiation. B) Proportion of iPSC-derived cells treated with DMSO or UNC0224 at week 4 of T cell differentiation. C) GO biological processes induced by UNC0224 treatment during T cell specification. D) Transcription factors with their regulons significantly enriched in the UNC-induced signature, ranked based on the 𢄒log10 (P-value). E) Heatmap showing gene expression levels of KLF13 regulon. F) UMAP visualization of non-activated or activated week 6 iPSC-T cells derived via G9a/GLP inhibition. G) Proportion of iPSC-T cells before and after activation.

Fig. 7. G9a/GLP inhibition generates iPSC-T cells with enhanced activity.

A) CD107a degranulation following PMA/ionomycin stimulation in iPSC-T cells treated with DMSO or UNC0224 during T cell specification (n = 3, mean ± SEM, * P ≤ 0.05). B) Specific lysis of OCI-Ly1 tumor cells by CD19 CAR iPSC-T cells treated with DMSO or UNC0224 during T cell specification (n = 3, mean ± SEM, * P ≤ 0.05, *** P≤0.001). C) Schematic illustration of tumor rechallenge model. D)Representative bioluminescent images of tumor xenografts. E) Kaplan-Meier curve showing percentage survival of untreated animals and animals treated with CD19 CAR T cells generated from control iPSC-T or UNC iPSC-CAR-T cells. (n=6).