Generation of T-cell-receptor-negative CD8αβ-positive CAR T cells from T-cell-derived induced pluripotent stem cells

Abstract

The production of autologous T cells expressing a chimaeric antigen receptor (CAR) is time-consuming, costly and occasionally unsuccessful. T-cell-derived induced pluripotent stem cells (TiPS) are a promising source for the generation of 'off-the-shelf' CAR T cells, but the in vitro differentiation of TiPS often yields T cells with suboptimal features. Here we show that the premature expression of the T-cell receptor (TCR) or a constitutively expressed CAR in TiPS promotes the acquisition of an innate phenotype, which can be averted by disabling the TCR and relying on the CAR to drive differentiation. Delaying CAR expression and calibrating its signalling strength in TiPS enabled the generation of human TCR^- CD8αβ⁺ CAR T cells that perform similarly to CD8αβ⁺ CAR T cells from peripheral blood, achieving effective tumour control on systemic administration in a mouse model of leukaemia and without causing graft-versus-host disease. Driving T-cell maturation in TiPS in the absence of a TCR by taking advantage of a CAR may facilitate the large-scale development of potent allogeneic CD8αβ⁺ T cells for a broad range of immunotherapies.

Extended Data Fig. 1.. T lymphoid commitment of hES, FiPS and TiPS on OP9-mDLL1

a, Flow cytometric analysis of pluripotency marker expression on H1, FiPS and WT-TiPS b, Flow cytometric analysis of T lymphoid markers of H1, FiPS and WT-TiPS during differentiation on OP9-mDLL1 at indicated timepoints. Plots depicting CD7/CD5 are gated on live CD45⁺ cells, plots depicting CD3/TCRαβ, CD4/CD8α and CD8α/CD8β are gated on live CD45⁺CD7⁺ cells. CD3/TCRαβ and CD4/CD8α at D40 are as presented in Fig. 1b.

Extended Data Fig. 2.. Generation, validation, and differentiation of TRAC^−/−-TiPS

a, CRISPR/Cas9-targeted integration of EF1a-GFP-P2A-Puromycing-bGHpA (G2AP) expression unit into the TRAC locus. Top, TRAC locus; middle, plasmid containing the G2AP expression unit flanked by homology arms; bottom, edited TRAC locus. ‘FWD’ and ‘REV’ indicate the location of the forward and reverse primers used in b. b, PCR validation of G2AP integration into the TRAC locus of TiPS clones. c, Flow cytometric analysis of pluripotency marker expression on TRAC^−/−-TiPS. Gated on live cells. d, T lymphoid makers of WT-TiPS and TRAC^−/−-TiPS during differentiation on OP9-mDLL1 at the indicated timepoints. Gated on live CD45⁺ cells. D40 is as presented in Fig. 1c.

Extended Data Fig. 3.. Early T lymphoid commitment of WT-TiPS and CAR-TiPS on human Notch ligands

a, SFG γRV plasmid design to transduce human Notch ligands (DLL1, DLL4, JAG1 or JAG2) into parental OP9 cells. b, Notch ligand expression on engineered OP9 lines. Filled grey histogram are stained parental OP9 cells, open black histogram are transduced OP9 cells. c, DTX1 induction in WT-TiPS by OP9 expressing indicated Notch ligand. D20 differentiating WT-TiPS cells were co-cultured with indicated OP9. DTX1 induction was measured by ddPCR, relative to endogenous RPL13A. The fold change was calculated relative to 0 h. Data shown is average of n = 2 technical replicates. d, g, Flow cytometric analysis of T lymphoid commitment marker expression (CD7, CD5, TCRαβ and CD56) of WT-TiPS (d) and CAR-TiPS (g) differentiated on OP9 expressing indicated human Notch ligands. Gated on live CD45⁺ cells. e, Flow cytometric analysis of pluripotency marker expression on CAR-TiPS. Gated on live cells. f, Phosphorylated-ERK1/2 levels in WT-TiPS (blue) and CAR-TiPS (red) on D35 (n = 3 technical replicates). h, Phenotype (left panels) and apoptosis levels (right panels) of WT-TiPS (top) and CAR-TiPS (bottom) from D27 – D35 of differentiation on OP9-DLL4. Percentage of apoptotic cells in each T lineage developmental stage was based on percentage live Annexin-V⁺ cells. * P<0.05, ** P<0.001, *** P<0.001, Welch’s 2-sample t test, data are means ± s.d (f).

Extended Data Fig. 4.. CD8αβ single positive CAR⁺ iT cell development

WT-TiPS were differentiated on OP9-DLL4 and transduced to express the 1928z CAR at D35 utilizing γRV SFG-1928z-P2A-LNGFR. Cells were expanded for 7 days in expansion media supplemented with IL-2. a, CD4/CD8αβ expression prior to transduction (D35) and on D42 in LNGFR⁺ cells, LNGFR⁻ cells and untransduced control cells which remained in differentiation on OP9-DLL4. Gated on live CD45⁺ cells. b, Cytotoxic activity of CAR⁺ iT cells in a 18 h bioluminescence assay, using FFLuc- NALM6 as target cells (n = 3 technical replicates, data are mean ± s.d). c, CRISPR/Cas9-targeted integration of CAR transgene into the TRAC locus. Top, TRAC locus; middle, plasmid containing the CAR transgene cassette flanked by homology arms; bottom, edited TRAC locus. d, f, PCR validation of CAR integration into the TRAC locus of TRAC-1928z-TiPS (d) and TRAC-1XX-TiPS (f) clones. e, g, Pluripotency marker expression on TRAC-1928z-TiPS (e) and TRAC-1XX-TiPS (g), gated on live cells.

Extended Data Fig. 5.. T lineage commitment of TRAC-CAR-TiPS

a, T lineage commitment marker expression (CD7/CD5, CD4/CD8α, CD8α/CD8β) of WT-TiPS (left), TRAC-1928z-TiPS (middle) and TRAC-1XX-TiPS on OP9-DLL4 at the indicated timepoints. CD7/CD5 is gated on live CD45⁺ cells, others are gated on live CD45⁺CD7⁺ cells. b, Flow cytometric analysis of T cell phenotype markers of D35 DP TRAC-1XX-iT cells. Gated on live CD45⁺CD7⁺CD4⁺CD8αβ⁺ cells. c, Intracellular and cell-surface expression of CD3 and TCRαβ on D35 TRAC-1XX-iT cells.

Extended Data Fig. 6.. Tonic ITAM phosphorylation in CAR⁺ T cells

a, Representative flow cytometry plot of CAR expression and pITAM1 (top panel) or pITAM3 (bottom panel) in PBMC-derived T cells expressing γRV-1928z, TRAC-1928z or TRAC-1XX (gated on live CAR⁺), or in control TRAC^−/− cells (gated on live CAR⁻). b, c, Percentage of pITAM1⁺ (b) and pITAM3⁺ (c) in the populations shown in a (n = 4–5 biological replicates, data are means ± s.d.).

Extended Data Fig. 7.. DP TRAC-1XX-iT cell mature to CD8αβ SP iT cells on 3T3-CD19–41BBL

a, c, Flow cytometric analysis of D42 cells matured on 3T3-CD19 (a) or 3T3-CD19–41BBL (c). Gated on live CD45⁺CD7⁺ cells. b, Flow cytometric analysis of D35 and D42 phenotypes of stimulated DP TRAC-1XX-iT cells. D35 TRAC-1XX-iT cells were sorted for a CD4⁺CD8αβ⁺ DP phenotype, stimulated on 3T3-CD19–41BBL and expanded for seven days. Gated on live CD45⁺CD7⁺ cells. d, Fold Expansion and T cell phenotype marker expression of TRAC-1XX-iT cells matured on 3T3-CD19–41BBL (3T3) or recombinant CD19-Fc. e, 4 h cytotoxicity assay of 3T3-CD19–41BBL stimulated TRAC-1XX-iT cells in response to NALM6-CD19⁺ an NALM6-CD19^−/− target cells (n = 3 technical replicates, data are means ± s.d.) f CD19 expression on primary CLL cells.

Extended Data Fig. 8.. Comparison of CD8αβ TRAC-1XX-iT cells and peripheral blood lymphocytes

a, Representative examples of lymphoid phenotype marker expression in CD8αβ TRAC-1XX-iT (red), CD8αβ αβTCR-T (blue), CD4 αβTCR-T (orange), γδTCR-T (green) and NK cells (purple). CD8αβ TRAC-1XX-iT cells are the same as represented in Fig. 5a. b, Variability of lymphoid phenotype marker expression in CD8αβ TRAC-1XX-iT cells (n= 3–4 biological replicates, data are means ± s.d). Biological replicates shown are samples utilized in RNA analysis (Fig. 5b, c). c, Principal Component Analysis comparing TRAC-1XX CD8αβ αβTCR-T cells (CD8, n = 4), TRAC-1XX CD4 αβTCR-T cells (CD4, n = 3), γRV-1XX γδTCR-T cells (γδ, n = 4), γRV-1XX NK cells (NK, n = 4) and CD8αβ⁺ TRAC-1XX-iT cells (iT CD8αβ, n = 4).

Extended Data Fig. 9.. Functional comparison of TRAC-1XX-iT, CAR-iT and CD8 TRAC-1XX

Functional comparison of healthy-donor peripheral blood TRAC-1XX CD8αβ αβTCR-T (CD8 TRAC-1XX), CAR-iT and TRAC-1XX-iT cells. CD8 TRAC-1XX cell doses represent number of CAR⁺ cells utilized in the assay. a, CAR and CD3 expression in CD8 TRAC-1XX, CAR-iT and TRAC-1XX-iT cells (black line) compared to unstained control (grey filled histogram). b, 18 h incucyte cytotoxicity assay with NLR⁺ CD19^−/− NALM6 target cells (n = 3 technical replicates). c, 4 h intracellular cytokine detection in T cells stimulated with NALM6 CD19⁺ target cells (at a 1:1 E:T), PMA/Ionomycin, NALM6 CD19^−/− target cells (at a 1:1 E:T) unstimulated controls (n = 3 technical replicates). d, 24 h cytokine secretion using NALM6-CD19^−/− as target cells at a 1:1 E:T (n = 11–18 biological replicates, left panel) or unstimulated control (n = 11–18 biological replicates, right panel). e, Schematic representation of the NALM6 in vivo tumour model. f, Kaplan-Meier analysis of tumour free survival (2×10⁶ TRAC-1XX-iT vs 2×10⁶ CD8 TRAC-1XX p=0.0062, 2×10⁶ TRAC-1XX-iT vs 1×10⁵ CD8 TRAC-1XX p=0.0034). * P<0.05, ** P<0.01, *** P<0.001, Chi-Square test (b) log-rank Mantel-Cox test (f). All data are means ± s.d.

Extended Data Fig. 10.. TRAC-1XX-iT function compared to healthy donor peripheral blood-derived CD8 TRAC-1XX T cells.

In vivo functional comparison of healthy-donor peripheral blood TRAC-1XX CD8αβ αβTCR-T (CD8 TRAC-1XX), TRAC-1XX-iT cells. CD8 TRAC-1XX cell doses represent number of CAR⁺ cells utilized in the assay. a, CAR and CD3 expression in CD8 TRAC-1XX and TRAC-1XX-iT cells (black line) compared to unstained control (grey filled histogram). b, Enumeration of tumour cells in the bone marrow and T cells in bone marrow, spleen and blood 6 or 12 days post T cell infusion (n = 2–3 mice, T cell in bone marrow day 12, CD8 TRAC-1XX vs TRAC-1XX-iT p=0.0161, T cells in spleen day 12 CD8 TRAC-1XX vs TRAC-1XX-iT p=0.0052). c, Phenotype of CD8 cells prior to infusion (day 0, n = 1) and of cells derived from the bone marrow on day 6 (n = 3 mice) and 12 (n = 3 mice). d, Kaplan-Meier analysis of overall survival. * P<0.05, ** P<0.01, *** P<0.001 Welch’s 2-sample two-sided t test (b) All data are means ± s.d.

Fig. 1.. DLL4 supports in vitro αβTCR-T cell development of WT-TiPS but not CAR-TiPS

a, Schematic representation of in vitro T cell differentiation protocol. Microscope images are at 4x magnification, scale bar represents 750μm. b, Flow cytometric analysis of T lineage commitment of H1, FiPS and WT-TiPS on OP9-mDLL1, gated on live CD45⁺CD7⁺ cells at day 40 (D40) in the differentiation. c, Flow cytometric analysis of T lineage commitment of WT-TiPS and TRAC^−/−-TiPS on OP9-mDLL1, gated on live CD45⁺ cells at D40 in the differentiation. d, f, Representative flow cytometric analysis of T lineage commitment of WT-TiPS (d) and CAR-TiPS (f) on D35 in differentiation on OP9 expressing the indicated human Notch ligand, gated on live CD45⁺CD7⁺ cells. e, g, Phenotype distribution of WT-TiPS (e, n = 6 biological replicates) or CAR-TiPS (g, n = 6 biological replicates) on D35 of differentiation on OP9-DLL4, gated on live CD45⁺CD7⁺ cells. All data are means ± s.d.

Fig. 2.. TRAC-controlled 1928z-1XX CAR expression facilitates DP T cell development

a, Induction of αβTCR (upper panel) and CAR (lower panel) expression in WT-TiPS, CAR-TiPS and TRAC-CAR-TiPS throughout T lymphoid development on OP9-DLL4 at the indicated timepoints. Gated on live CD45⁺CD7⁺ cells. b, d, Representative flow cytometric analysis of T lineage commitment markers of TRAC-1928z-TiPS (b) and TRAC-1XX-TiPS (d) gated on live CD45⁺ cells at D35 in differentiation on OP9-DLL4. c, e, D35 phenotype distribution of TRAC-1928z-TiPS (c, n = 3 biological replicates) and TRAC-1XX-TiPS (e, n = 11 biological replicates) at D35 gated on live CD45⁺CD7⁺ cells. All data are means ± s.d.

Fig. 3.. CAR regulation influences Notch and TCR target gene induction

a, Schematic representation of Notch and (pre)TCR signalling interactions as reported in the literature. b, ddPCR analysis of Notch and (pre)TCR target genes at D24, 27, 31 and 35 of T cell differentiation (normalized to RPL13A) (n = 3 technical replicates). c, t-distributed Stochastic Neighbour embedding (tSNE) analysis of cell surface expression of CD4, CD8α, CD8β and pTα on D35 TRAC-1XX-TiPS iT cells. Colour scale represent level of marker expression. d, Distribution of pTα expression on D35 TRAC-1XX-TiPS (n = 3 biological replicates). All data are means ± s.d.

Fig. 4.. 4–1BBL co-stimulation enhances CD8αβ TRAC-1XX-iT proliferation and function

a, Representative phenotype of TRAC-1XX-iT cells matured on 3T3-CD19 for 7 days (D35-D42), gated on live CD45⁺CD7⁺ (left and right) and CD45⁺CD7⁺CD8α⁺ (middle). b, Distribution of CD8αα and CD8αβ phenotype in the CD8α⁺ compartment (n = 3 biological replicates). c, Expansion of TRAC-1XX-iT cells from D35-D42 after maturation on 3T3-CD19 (n = 6 biological replicates). d, 4–1BB cell-surface expression on TRAC-1XX-iT cells on D35 8 h after exposure to parental 3T3 (black), 3T3-CD19 (red) or left unstimulated (grey). e, Representative phenotype of TRAC-1XX-iT cells matured on 3T3-CD19–41BBL, gated on live CD45⁺CD7⁺ (left and right) and CD45⁺CD7⁺CD8α⁺ (middle). f, Distribution of CD8αα and CD8αβ phenotype in the CD8α⁺ compartment (n = 6 biological replicates). g, Expansion of TRAC-1XX-iT cells from D35-D42 after maturation on 3T3-CD19–41BBL (n = 6 biological replicates). h, Total cell expansion from D0-D42 in iT differentiation with maturation on 3T3-CD19±41BBL (n = 6 biological replicates for each group, p=0.0008). i, Cytotoxic activity measured in an 18 h bioluminescence assay, using firefly luciferase (FFLuc)-expressing NALM6 at the indicated effector-to-target (E:T) ratios (n = 3; technical replicates). j, 4 h intracellular cytokine detection of 3T3-CD19±41BBL-matured cells in response to NALM6 (n = 3 technical replicates). k, Representative expansion of D42 cells matured on 3T3-CD19±41BBL upon repeated weekly antigen exposure on 3T3-CD19. l, Schematic representation of NALM6 in vivo tumour model. m, Tumour burden (total flux in photons per second) of NALM6-bearing mice treated with 2×10⁶ D42 TRAC-1XX-iT cells (n = 4–5, line = one mouse). n, Kaplan-Meier analysis of overall survival (p<0.0001). o, Flow cytometric quantification of iT cells (left panel, p=0.0339) and tumour cells (right panel) in bone marrow 6 days after T cell infusion (n = 3). p, Cytotoxic activity measured in a 6 h flow cytometry assay, using primary CD19⁺ CLL cells at the indicated E:Ts with 3T3-CD19–41BBL-matured TRAC-1XX-iT cells (n = 3 technical replicates) * P<0.05, *** P<0.001, Welch’s 2-sample two-sided t test (h, o), log-rank Mantel-Cox test (n). All data are means ± s.d.

Fig. 5.. CD8αβ TRAC-1XX-iT cells resemble peripheral-blood derived CD8αβ T cells

a, Phenotype analysis of 3T3-CD19–41BBL-matured D42 CD8αβ TRAC-1XX-iT cells for TCR-T cell markers (left and middle panel) and NK-cell markers (right panel). Data is representative of four independent experiments, gated on live CD45⁺CD7⁺CD8αβ⁺ cells. b, Dendrogram of hierarchical clustering analysis based on Euclidian distance matrix comparing the transcriptome of TRAC-1XX CD8αβ αβTCR-T cells (CD8, blue, n = 4 biological replicates), TRAC-1XX CD4 αβTCR-T cells (CD4, orange, n = 3 biological replicates), γRV-1XX γδTCR-T cells (γδ, green, n = 4 biological replicates), γRV-1XX NK cells (NK, purple, n = 4 biological replicates) and CD8αβ⁺ TRAC-1XX-iT cells (iT CD8αβ, red, n = 4 biological replicates). c, Correlation matrix using Pearson’s statistics comparing same groups as in (b).

Fig. 6.. TRAC-1XX-iT have improved persistence and function over CAR-iT cells

Functional comparison of healthy-donor peripheral blood TRAC-1XX CD8αβ αβTCR-T cells (CD8 TRAC-1XX), CAR-iT, and TRAC-1XX-iT cells (matured on 3T3-CD19–41BBL). CD8 TRAC-1XX doses reflect number of CAR⁺ T cells utilized in the assay. a, Cytotoxic activity using a 18 h Incucyte assay, using NLR-expressing NALM6 as target cells (n = 3 technical replicates). b, NALM6 rechallenge assay. NLR⁺ NALM6 and T cells were co-cultured at a 1:1 E:T. Every 72 h T cells were rechallenged with 1x NLR⁺NALM6 and cytokines. NALM6 clearance was measured in NLR⁺ surface area reduction compared to the timepoint of rechallenge (n = 3 technical replicates). c, Twenty-four h cytokine secretion using NALM6 as target cells at a 1:1 E:T ratio (CD8 TRAC-1XX n = 15, TRAC-1XX-iT n = 18, CAR-iT n = 11 biological replicates, IL-2 CD8 TRAC-1XX vs CAR-iT p=0.02, IL-2 CD8 TRAC-1XX vs TRAC-1XX-iT p=0.0222, IFNγ CD8 TRAC-1XX vs CAR-iT p=0.004, IFNγ CD8 TRAC-1XX vs TRAC-1XX-iT p=0.0293, IFNγ TRAC-1XX-iT vs CAR-iT p=0.0363, TNFα CD8 TRAC-1XX vs CAR-iT p=0.001, TNFα CD8 TRAC-1XX vs TRAC-1XX-iT p=0.0027, TNFα TRAC-1XX-iT vs CAR-iT p=0.0262). d, Tumour burden (total flux in photons per second) of NALM6-bearing, untreated mice (n = 4) or mice treated with 4×10⁶ TRAC-1XX-iT (middle) or CAR-iT (right) cells (n = 6, line = one mouse). e, Kaplan-Meier analysis of overall survival (p=0.002). f,g, Enumeration of iT cells (f) and tumour cells (g) in the bone marrow, spleen and blood 12 days post T cell infusion (n = 4 mice, iT in bone marrow TRAC-1XX-iT vs CAR-iT p=0.0329, iT in spleen TRAC-1XX-iT vs CAR-iT p=0.0369). * P<0.05, ** P<0.01, *** P<0.001, Welch’s 2-sample two-sided t test (c, f, g), log-rank Mantel-Cox test (e). All data are means ± s.d.

Fig. 7.. TRAC-1XX-iT cells cure systemic NALM6 tumour model without inducing graft-versus-host disease

a, Schematic representation of systemic NALM6 tumour model. b, Tumour burden (total flux in photons per second) of NALM6-bearing untreated mice (n = 4), or mice treated with 4×10⁶ CD8 TRAC-1XX (n = 5) or TRAC-1XX-iT cells (n = 7, line = one mouse). c, Kaplan-Meier analysis of tumour-free survival (left) and overall survival (right). d, Flow cytometric quantification of tumour cells (left) and T cells (right) in bone marrow 12 days after T cell infusion (n = 2–3 mice). e, Phenotype of persisting TRAC-1XX-iT cells prior to infusion (day 0, n = 1) and of cells derived from the bone marrow on day 6 and 12 days after TRAC-1XX-iT cell infusion (n = 3 mice). * P<0.05, ** P<0.01, *** P<0.001, log-rank Mantel-Cox test (c), Welch’s 2-sample two-sided t test (d). All data are means ± s.d.