BET inhibitors rescue anti-PD1 resistance by enhancing TCF7 accessibility in leukemia-derived terminally exhausted CD8+ T cells

Abstract

Many acute myeloid leukemia (AML) patients exhibit hallmarks of immune exhaustion, such as increased myeloid-derived suppressor cells, suppressive regulatory T cells and dysfunctional T cells. Similarly, we have identified the same immune-related features, including exhausted CD8⁺ T cells (TEx) in a mouse model of AML. Here we show that inhibitors that target bromodomain and extra-terminal domain (BET) proteins affect tumor-intrinsic factors but also rescue T cell exhaustion and ICB resistance. Ex vivo treatment of cells from AML mice and AML patients with BET inhibitors (BETi) reversed CD8⁺ T cell exhaustion by restoring proliferative capacity and expansion of the more functional precursor-exhausted T cells. This reversal was enhanced by combined BETi and anti-PD1 treatment. BETi synergized with anti-PD1 in vivo, resulting in the reduction of circulating leukemia cells, enrichment of CD8⁺ T cells in the bone marrow, and increase in expression of Tcf7, Slamf6, and Cxcr5 in CD8⁺ T cells. Finally, we profiled the epigenomes of in vivo JQ1-treated AML-derived CD8⁺ T cells by single-cell ATAC-seq and found that JQ1 increases Tcf7 accessibility specifically in Tex cells, suggesting that BETi likely acts mechanistically by relieving repression of progenitor programs in Tex CD8⁺ T cells and maintaining a pool of anti-PD1 responsive CD8⁺ T cells.

Fig. 1. GEMM mouse model characterization.

A Left: splenocytes isolated from 6 C57BL/6 (WT, black) or 7 Flt3-ITD+/–, Tet2+/–, Lys-Cre+/– (AML, red) mice were stained for live CD11b+ cells and evaluated by flow cytometry. Significance determined by Mann–Whitney t-test. Right: whole spleens isolated from 15 WT (black) and 24 AML (red) mice were weighed. Median weights for each group are displayed on the right side of plot. Significance determined by Mann–Whitney t-test. B Blood isolated from 24 WT (black) and 20 AML (red) mice were stained for live CD11b+ cells and evaluated by flow cytometry. Significance determined by Mann–Whitney t-test. C Representative H&E staining of WT and AML mice-derived spleen (top row, 500 µm scale), liver (middle row, 200 µm scale), and bone marrow (bottom row, 100 µm scale). D Splenocytes from untreated 7 WT (black) and 5 AML (red) mice were assessed for expression (median fluorescence intensity) of markers of immune exhaustion on CD4⁺ T cells (Live, CD11b-, CD3⁺, CD4⁺). Top row, left to right: CTLA-4, TBET, CD44, PD1). Bottom Row, left to right: EOMES, LAG3, TIGIT, TCF1). Significance determined by multiple Mann–Whitney t-tests. CD44, PD1, LAG3 are significantly increased in AML mice. CTLA-4, EOMES, and TCF1 are significantly decreased in AML mice. E Splenocytes from untreated WT (black) and AML (red) mice were assessed for expression (median fluorescence intensity) of markers of immune exhaustion on CD8⁺ T cells, as in (D). Significance determined by multiple Mann–Whitney t-tests. CD44, EOMES, and TIGIT are significantly increased in AML mice.

Fig. 2. AML mouse-derived CD8⁺ T cells are intrinsically dysfunctional and unresponsive to TCR stimulation.

A Genomic DNA was extracted from 5 WT and 5 AML mouse splenocytes and sequenced for TCRb. Simpsons Clonality was calculated and plotted (WT mice in black, AML mice in red). Significance determined by Mann–Whitney t-test. B, C 6 WT (black) and 5 AML (red) splenocytes were isolated, stained with proliferation dye (CFSE), and cultured for 72 h with anti-CD3. Proliferation of A CD8⁺ T cells and B CD4⁺ T cells was then assessed by flow cytometry, staining with viability and markers to identify T cells. Proliferation displayed is percent CFSE diluted relative to unstimulated (HIgG) control for each cell type. Significance determined by Mann–Whitney t-tests. D 4 WT (black) and 4 AML (red) derived T cells were isolated from splenocytes via CD3 negative isolation magnetic beads. The T cells were then stained with CFSE and plated for 72 h with anti-CD3 and anti-CD28 stimulation. The cells were then harvested and assessed by flow cytometry. Significance determined by Mann–Whitney t-test. E, F Splenocytes derived from 12 WT (black) and 12 AML (red) mice were stained for surface and intracellular markers of T cell exhaustion and evaluated by flow cytometry. Terminally Exhausted CD8 T cells (TEx) are represented as a E fraction of all CD8 T cells (%PD1+, TIM3+, TCF1-) and F calculated total number of cells per spleen. Significance was determined by Mann–Whitney t-tests.

Fig. 3. AML mouse T cells are refractory to ICB therapy but partially rescued with BET inhibition.

A Cells from AML mice passaged for ~1 month were subjected to an inhibitor library panel, as previously described [5]. Cells were seeded into multiple 384-well plates containing titrations of 188 inhibitors, incubated for 72 h, and viability assessed by MTS assay. Plot represents each areas under the curve (AUC) for every inhibitor on the panel. BET inhibitors JQ1, OTX-015, and CPI-0610 are highlighted in red. B, C Splenocytes were isolated from 7 AML (B) or 6 WT mice (C), stained with CFSE, and cultured for 72 h without TCR stimulation (HIgG), anti-CD3 alone, anti-CD3 with anti-PD1 or titrations of anti-CD3 with JQ1 or anti-CD3 with both JQ1 and anti-PD1. Cells were stained and analyzed by flow cytometry. Plots represent the fold-change in proliferation in CD8⁺ T cells, as measured by percent CFSE diluted relative to anti-CD3 stimulated alone. Significance determined by Kruskal–Wallis multiple comparisons t-tests. D, E Effect of BETi, anti-PD1, or BETi + anti-PD1 treatment on the percent of TPEx CD8⁺ T cells (D) and TEx CD8⁺ T cells (E) after culturing for 72 h as previously in B and C. Significance determined by Kruskal–Wallis multiple comparisons t-tests. F Representative dot plots showing the proliferation of CD8⁺ T cells vs. TCF1 expression from an AML mouse proliferation assays as described in B and C. Each panel shows a different condition. X-axis denotes proliferation dye CFSE. Y-axis denotes TCF1 expression. G Splenocytes from 4 AML mice were stained with CFSE and plated for 72 h without TCR stimulation (HIgG), anti-CD3 alone, anti-CD3 with anti-PD1 and titrations of anti-CD3 with JQ1 or anti-CD3 with both JQ1 and anti-PD1. Proliferation of TPEx (black) and TEx (red) CD8⁺ T cells from AML mice was assessed by flow cytometry. Significance determined by Kruskal–Wallis multiple comparisons t-tests. H Fresh mononuclear cells from bone marrow aspirates or peripheral blood obtained from 7 patients with AML were stained with CTV and cultured for 5 days without TCR stimulation (mIgG), anti-CD3, anti-CD3 with anti-PD1, anti-CD3 with 120 nM JQ1 or anti-CD3 with 120 nM JQ1 and anti-PD1. Cells were stained and analyzed by flow cytometry. Left panel plot represents the fold-change CD8⁺ T cell proliferation for each condition vs. respective anti-CD3 alone control for each patient sample. Right panel plots the actual percent proliferation for each AML patient sample with MIgG or CD3 alone.

Fig. 4. In vivo treatment with BETi and anti-PD1 synergizes in reducing tumor burden and enhances CD8+ T cell activity.

A Schematic detailing in vivo BETi + anti-PD1 treatment strategy and functional readouts of efficacy. B Mice treated with RIgG, JQ1, JQ1 with anti-PD1, or anti-PD1 alone were bled periodically over a 2-week period and assessed for WBC. Data display fold-change WBC (k/µL) normalized per mouse in comparison to pre-treatment bleed WBC. Significance determined by one-way ANOVA. Timepoints are Pre-Bleed, Mid-bleed (day 7), Endpoint (day 14). C, D Bone marrow cells were isolated from treated AML and WT mice and assessed by flow cytometry. Graph denotes %CD3⁺ T cells in the bone marrow (C) and %CD8⁺ T cells in the bone marrow as a percent of all T cells (D). Significance derived from combining two experimental replicates and determined by one-way ANOVA. E CD8⁺ T cells were isolated from JQ1-treated and RIgG-treated AML mice and RNA harvested and analyzed by Nanostring. Volcano plot shows the fold-change in normalized transcript levels of JQ1-treated mice vs. RIgG-treated AML mice vs. –log₁₀ P value as determined by multiple t-tests with Bonferroni correction. Hits of interest are highlighted in red. Dashed red line denotes significance threshold (0.05).

Fig. 5. Single-cell chromatin landscape of in vivo BETi-treated CD8⁺ T cells.

A AML mice were treated in vivo daily for 5 days with vehicle or JQ1 before enriching CD8⁺ T cells via magnetic sort before sequencing via S3-ATAC. Cells were batched and clustered using the ArchR package and cell cluster identities were determined via inferred gene scores of defining gene features. B, C Overlay of cell proportions based on treatment group. Highlights the distribution of B vehicle-treated cells in blue and C JQ1-treated cells in red. D Heatmap generated from column Z scores of significant features for each cluster identified in A. Marker genes of interest are highlighted in black text. Cluster identities are described on the right axis. Cells are clustered via Euclidean distance with the corresponding dendrogram drawn on the left axis. E Proportions of each cell type were determined as a fraction of all cells in vehicle and JQ1 cells. Each point represents either vehicle-treated (left point) or JQ1-treated (right point) and is connected by a line.

Fig. 6. In vivo BETi increases TCF7 accessibility at later stages of TEx differentiation only.

A (Top) cartoon schematic describing the cell types fit to the differentiation trajectory (bottom). Colors represent pseudotime values throughout TEx differentiation (C6–>C8–>C7–>C5). B, C Pseudotime trajectory plot with corresponding TCF7 gene score throughout TEx differentiation for B Vehicle-treated cells and C JQ1-treated cells. Vertical red dashed marks the third quartile throughout pseudotime (late TPex/Early Tex) and the horizontal red dashed line markers the TCF7 accessibility at pseudotime 100. Significance was calculated by Mann–Whitney t-test comparing the aggregate TCF7 gene scores for each cell throughout pseudotime in vehicle and JQ1-treated groups. D A portion of splenocytes derived from the vehicle or JQ1-treated AML mice were analyzed by flow cytometry. Data represent TCF1 median fluorescence intensity of TCF1 on all CD8⁺ T cells. Significance determined by Mann–Whitney t-test. E–G Volcano plot describing the mean difference in gene scores (JQ1 – Volcano treated) from E early TPEx cells only (Cluster 8), F late TPEx cells only (Cluster 7), and G Tex cells only (Cluster 5). Significance was calculated via multiple corrected t-tests. Genes with a mean difference >0.25 are highlighted in red and genes <–0.25 are highlighted in blue. Genes of interest (Tcf7, Tcf3, Pdcd1, and Cxcr3) are highlighted in black. The dashed horizontal line marks the significance cutoff. Genes below the line are not significant.