PI3K orchestration of the in vivo persistence of chimeric antigen receptor-modified T cells

Abstract

In vivo persistence of chimeric antigen receptor (CAR)-modified T cells correlates with therapeutic efficacy, yet CAR-specific factors that support persistence are not well resolved. Using a CD33-specific CAR in an acute myeloid leukemia (AML) model, we show how CAR expression alters T cell differentiation in a ligand independent manner. Ex vivo expanded CAR-T cells demonstrated decreased naïve and stem memory populations and increased effector subsets relative to vector-transduced control cells. This was associated with reduced in vivo persistence. Decreased persistence was not due to specificity or tumor presence, but to pre-transfer tonic signaling through the CAR CD3ζ ITAMs. We identified activation of the PI3K pathway in CD33 CAR-T cells as responsible. Treatment with a PI3K inhibitor modulated the differentiation program of CAR-T cells, preserved a less differentiated state without affecting T cell expansion, and improved in vivo persistence and reduced tumor burden. These results resolve mechanisms by which tonic signaling of CAR-T cells modulates their fate, and identifies a novel pharmacologic approach to enhance the durability of CAR-T cells for immunotherapy.

Figure 1. CD33 and CD19 CAR-T cells control AML growth

Mice were injected with MOLM-13-CD19 tumor cells and treated with CD33 CAR, CD19 CAR, or control T cells. Mice were monitored for survival and tumor burden, or organs were harvested 18 days after transfer. (A) Kaplan-Meier survival analysis. (B) Xenogen images and bioluminescent signal intensities (C) MOLM-13-CD19 tumor cell numbers *** p<0.001, **** p<0.0001.

Figure 2. Poor CAR-T cell survival is tumor independent

(A) Mice were injected with MOLM-13-CD19 tumor cells and treated with CD33 CAR, CD19 CAR or control T cells. Organs were harvested 5 days after transfer. Total numbers of CD3⁺CD8⁺ CAR or control T cells are shown. (B–C) CD33 CAR-T cells (GFP) and control T cells (RFP) were co-transferred at ratios of 0.9:1–1.1:1 into mice with or without tumor. Organs were harvested 5 days after transfer. (B) Numbers of CD8⁺ CAR and control T cells with and without tumor. (C) Ratio of CD8⁺ control T cells to CD33 CAR-T cells with and without tumor, normalized to input ratio. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

Figure 3. CAR-T cells exhibit increased effector differentiation

(A) Composition of T_N, T_CM, T_EM and T_EFF CD8⁺ T cell subsets in CD33 CAR and control T cells after ex vivo activation. (B) Percent of T_SCM cells after ex vivo activation. (C) Methylation analysis of genomic DNA CpG sites within the IFNγ promotor. Naïve CD8⁺CD45RA⁺CD45RO⁻CCR7⁺CD95⁻ T cells were sorted from donor samples, transduced with CD33 CAR or control vector, and assessed 9 days after ex vivo activation. Each line represents an individual clone. Bar graphs show % CpG methylation at each site of the locus in CD33 CAR or control T cells. * p<0.05, ** p<0.01, **** p<0.0001.

Figure 4. Altered CAR-T cell differentiation due to CAR CD3ζ ITAM signaling

(A) Composition of CD8⁺ T_N, T_CM, T_EM, and T_EFF subsets in CD33 41BB, CD33 ITAM, and CD33 41BB ITAM T cells relative to CD33 CAR and control T cells 12 days after activation. (B) Activation and exhaustion marker expression on CD8⁺ CD33 CAR and CD33-ITAM T cells 12 days after activation. (C) MOLM-13-CD19 cells were incubated with CD33 CAR, CD33 CAR ITAM mutant, or control T cells as indicated for 24–48h. Tumor cells were quantified by flow cytometry and normalized to cultures without added T cells. * p<0.05, *** p<0.001, **** p<0.0001.

Figure 5. Differentiation status of CAR-T cells influences in vivo persistence

(A–B) CD33 CAR and control T cells were mixed 1:1 and transferred with or without MOLM-13-CD19 tumor cells into mice. Cell numbers were determined 5 days after transfer. (A) Proportions of and (B) Numbers of CD8⁺ T_N, T_CM, T_EM and T_EFF subsets. (C) CD8⁺CCR7⁺CD45RA⁺ T_N and CD8⁺CCR7⁻CD45RA⁻ T_EM cells were sorted from activated CD33 CAR or control T cells and transferred individually into mice. Cell numbers were normalized to the CD33 CAR T_EM group. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

Figure 6. CD33 CAR-T cells exhibit a distinct transcriptional profile

(A) Differentially expressed genes from PI3K/AKT and glycolysis pathways identified by Ingenuity Pathway Analysis (IPA) of CD8⁺ CD33 CAR and control T cells 12 days after ex vivo activation. (B) Phospho-flow staining of pS6 and p4EBP-1 in CD33 CAR and control T cells. CD33 CAR-T cells were stimulated with PMA/Ionomycin as a positive control. ** p<0.01, **** p<0.0001.

Figure 7. PI3K inhibition restrains aberrant CAR-T cell differentiation ex vivo

CD33 CAR or control T cells were stimulated for 5 days, treated with inhibitor for 4 days, and analyzed by flow cytometry. PI3K inhibitors LY294002 (LY) and IC87114 (IC), mTOR inhibitors rapamycin (RAPA) and PP242, AKT inhibitor API-2, and glycolysis inhibitor DCA were used as indicated. (A) Percentages of T_N, T_CM, T_EM and T_EFF subsets of CD8⁺ T cells. (B) Numbers of T_N, T_CM, T_EM and T_EFF subsets of CD8⁺ T cells normalized to the number of untreated CD33 CAR-T cells. (C) Relative numbers of total CD8⁺ T cells after 4 days of inhibitor treatment, normalized to untreated CAR-T cells. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

Figure 8. PI3K inhibition during ex vivo expansion improves CAR-T cell persistence and reduces tumor burden in vivo

Mice were injected with MOLM-13-CD19 tumor cells and either untreated or LY-treated CD33 CAR-T cells. Organs were harvested 14 days after transfer. (A) Total CD8⁺ T cell numbers (top) and total MOLM-13-CD19 tumor cell numbers (bottom). (B) Proportions of CD8⁺ T_N, T_CM, T_EM and T_EFF subsets. (C) Kaplan-Meier survival analysis of mice with MOLM-13-CD19 tumors treated with CD33 CAR, CD33 CAR + LY treatment, or vector-transduced control T cells. ** p<0.01, *** p<0.001, **** p<0.0001.