Remodeling of the immune microenvironment is linked to adverse outcome in pediatric T cell acute lymphoblastic leukemia

Abstract

Changes in the immune microenvironment are frequent in cancers occurring in adult patients, yet our understanding of the pediatric cancer immune microenvironment and its clinical relevance is limited. We investigate the immune microenvironment in pediatric T cell acute lymphoblastic leukemia (T-ALL), using single-cell CITE-seq and immune repertoire analyses. We identify a T-ALL subgroup characterized by a remodeled immune microenvironment, which is associated with adverse clinical outcome in minimal residual disease low patients. This adverse immune landscape is dominated by the presence of a population of non-malignant CD4^-CD8^-TCRαβ T cells that interact with CXCL16 expressing non-classical monocytes. Leukemia cell intrinsic transcriptional rewiring in these patients is associated with activation of Rap1 signaling. Inhibiting Rap1 signaling results in increased sensitivity to the BCL2/BCL-XL inhibitor navitoclax. Our study provides insights into the immune microenvironment of pediatric hematologic malignancies, forming the basis for identifying potential (immuno) therapeutic targets and risk stratification for treatment.

Fig. 1. Defining the CD45⁺ immune microenvironment in pediatric T-ALL.

a Schematic overview of the single-cell CITE-seq (transcriptome and surface protein expression) and immune repertoire sequencing (TCRαβ/BCR-seq) workflow. The clinical cohort included fifteen pediatric T-ALL patients for whom samples were collected at diagnosis (Dx, PB (n = 10) and BM (n = 5)), short-term treatment (ST, PB (n = 8)) and remission (RM, PB (n = 6). Four healthy donor samples (PB (n = 2) and BM (n = 2)) were taken along as control samples. Samples were sorted for CD45^high immune cells and CD7⁺CD45^dim malignant cells. The 5’ 10x Genomics Immune Profiling sequencing protocol was used to analyze the transcriptome (RNA-seq), surface protein expression (ADT-seq, 204 surface proteins) and immune repertoire (TCRαβ/BCR-seq). b Weighted-nearest-neighbor (WNN) UMAP plot based on transcriptome and surface protein expression (CITE-seq) of 136,671 cells from fifteen T-ALL patients and four healthy donors, colored by cell type. mDCs: myeloid dendritic cells; pDCs: plasmacytoid DCs; NK: natural killer; HSPC: hematopoietic stem and progenitor cells. c Bubble heatmap showing normalized gene expression (left) and surface protein expression (right) of selected signature genes or proteins for cell types shown in (b). d Surface protein expression of CD34, CD1a, CD3, CD4 and CD8 in malignant cells and normal T cells from T-ALL patients at diagnosis. e Heatmap of average RNA expression of known T-ALL oncogenes in malignant cells of T-ALL patients at diagnosis. f Heatmap of percentage of cells with TCRβ motifs detected in the 10x Genomics scTCRαβ-seq data. g Distribution of immune cell types in healthy donors (BM: n = 2, PB: n = 2) and T-ALL patient samples (diagnosis: BM: n = 5, PB: n = 10; prophase: PB: n = 6; treatment: PB: n = 2, remission: PB: n = 6).

Fig. 2. Altered composition of B cell types in T-ALL.

a WNN UMAP plot based on transcriptome and surface protein expression of 14,587 B cells colored by B cell types. b Bubble heatmap showing normalized gene expression (left) and surface protein expression (right) of selected signature genes or proteins for B cell types shown in (a). c Distribution of B cell types in healthy donors (BM: n = 2, PB: n = 2) and T-ALL patients (diagnosis: BM: n = 5, PB: n = 10; prophase: PB: n = 6; treatment: PB: n = 2, remission: PB: n = 6). At.: atypical. d, Log2 fold changes of differential neighborhood abundance testing by MiloR. MiloR neighborhoods (n = 430, where each neighborhood contains on average 15 cells) are grouped by B cell type. Colors indicate an enrichment (red) or depletion (green) of B cell type at T-ALL diagnosis if more than half of the neighborhoods were significantly enriched/depleted (spatial FDR < 0.1). e Boxplot showing percentage of atypical memory B cells of B cells for healthy donors (n = 4) and T-ALL patient samples across treatment (diagnosis n = 15; prophase n = 6; treatment n = 2, remission n = 6). PB/BM samples are indicated by a square (BM) or circle (PB). An unpaired t-test was used for statistical analysis. f Proportion of expressed BCR constant chain (scBCR-seq) in B cells in healthy donors and T-ALL patients at diagnosis. g Heatmap of scaled average gene (left) and surface protein (right) expression of selected differentially expressed genes across B cells. See Supplementary Data 4 for full list of differentially expressed genes.

Fig. 3. Remodeling of the T cell landscape in T-ALL.

a WNN UMAP plot based on transcriptome and surface protein expression of 46,754 TNK cells colored by T cell types. b Bubble heatmap showing normalized gene expression (left) and surface protein expression (right) of selected signature genes or proteins for T cell types shown in (a). c Distribution of T cell types in healthy donors (BM: n = 2, PB: n = 2) and T-ALL patients (diagnosis: BM: n = 5, PB: n = 10; prophase: PB: n = 6; treatment: PB: n = 2, remission: PB: n = 6). d Log2 fold changes of differential neighborhood abundance testing by MiloR. MiloR neighborhoods (n = 2038, where each neighborhood contains on average 26 cells) are grouped by T cell type. Colors indicate an enrichment (red) or depletion (green) of T cell type at T-ALL diagnosis if more than half of the neighborhoods were significantly enriched/depleted (spatial FDR < 0.1). e Surface protein expression of TCRαβ, CD4 and CD8 in T cell types (n = 39,006 T cells across 4 healthy donors and 15 T-ALL patients). f Chao1 TCRαβ diversity index scores (Immunarch) of normal T cells in healthy donors (n = 4) and T-ALL patients at diagnosis with at least 75 T cells (n = 10). An unpaired t-test was used for statistical analysis. g Circle packing plot showing T cell type (color) in relation to TCRαβ clonotype for healthy donors (left), T-ALL diagnosis (right). Grouped T cells in bold circles indicate clonal cells (i.e. same TCRαβ cdr3 motif). h Heatmap of percentage of cells with TCRβ motifs detected in the 10x Genomics scTCRαβ-seq results. ‘Other TCRβ’ indicates any other TCRβ motif than the clonal TCRβ motifs found in malignant cells (see Fig. 1f). i Heatmap of t-test statistics of TCRαβ feature enrichment (Mann-Whitney U test FDR < 0.01, ConGA) in T cells.

Fig. 4. Unique transcriptional programs in DNαβ T cells.

a Heatmap of scaled RNA expression of top 200 differentially expressed genes across all T cells. Genes indicate top marker genes of DNαβ T cells (purple) or selected marker genes of other T cells that have low expression in DNαβ T cells (green). b Bar graph of significant gene ontologies based on top 200 marker genes in DNαβ T cells analyzed by DAVID Gene ontology (P-value is derived from a Modified Fisher’s Exact Test). See Supplementary Data 5 for full list of differentially expressed genes. Immunoreg.: immunoregulatory. c Scatterplot of CD69 and CD137 surface protein expression in T cells from healthy donors and T-ALL patient samples at diagnosis. d Heatmap of z-normalized TIL signature scores^,, for DNαβ T cells from T-ALL patients, T cells from normal donors and pan-cancer enriched TILs (ProjecTILs). e T cell stemness scores for T cell types (CD8⁺naïve: n = 2593 cells; CD4⁺naïve: n = 5839 cells; CD4⁺memory: n = 4169 cells; CD8⁺GZMK⁺: n = 1356 cells; CD8⁺GZMB⁺: n = 1547 cells; DNαβ: n = 1481 cells). A one-way Anova test followed by Tukey’s HSD post-hoc test was used for statistical analysis (p < 0.001 for all comparisons with DNαβ T cells). f Concentration of IL10 in serum of healthy donors (n = 2) and T-ALL patients at diagnosis and short-term treatment for group 1 (n = 4) and group 2 (n = 4) T-ALL patients. A Kruskal-Wallis test followed by Dunn’s post-hoc test with Holm correction was used for statistical testing and only significant comparisons are shown. g Regulon activity (pySCENIC) in DNαβ T cells. h Boxplots showing DNαβ T cell transcriptomic signature scores for DNαβ T cells from T-ALL patients, T cells from healthy donors and pan-cancer enriched TILs (Project TIL). i Boxplot showing percentage of T-ALL associated DNαβ T cells of T cells across treatment (diagnosis n = 4; prophase n = 3; treatment n = 1, remission n = 3). j Boxplots showing DNαβ T cell transcriptomic signature scores for DNαβ T single cells from healthy donors (PB, n = 14 cells) and group 2 T-ALL patients at diagnosis (PB, n = 901 cells) and prophase treatment (PB, n = 2040 cells). A one-way Anova test followed by Tukey’s HSD post-hoc test was used for statistical analysis.

Fig. 5. Increased abundance of non-classical monocytes in T-ALL.

a Distribution of myeloid cell types in healthy donors (BM: n = 2, PB: n = 2) and T-ALL patients (diagnosis: BM: n = 5, PB: n = 10; prophase: PB: n = 6; treatment: PB: n = 2, remission: PB: n = 6). b Log2 fold changes of differential neighborhood abundance assessed by MiloR comparing healthy donors (n = 4) and T-ALL patients at diagnosis (n = 15). MiloR neighborhoods (n = 1680, where each neighborhood contains on average 31 cells) are grouped by myeloid cell type. Colors indicate an enrichment (red) or depletion (green) of myeloid cell type in T-ALL if more than half of the neighborhoods were significantly enriched/depleted (spatial FDR < 0.1). c Boxplot showing ratio of NC/C monocytes for healthy donors (BM: n = 2, PB: n = 2) and T-ALL patients (diagnosis: BM: n = 5, PB: n = 10; prophase: PB: n = 6; treatment: PB: n = 2, remission: PB: n = 6). An unpaired t-test (left) or a one-way Anova test followed by Tukey’s HSD post-hoc test (right) was used for statistical analysis and only significant comparisons are shown. d Bubble heatmap showing normalized gene expression (left) and surface protein expression (right) of canonical myeloid marker genes or proteins in myeloid cells. e Boxplot showing ratio of NC/C monocytes (left) and percentage of atypical memory B cells (right) for healthy donors (n = 4) and T-ALL group 1 (n = 10) and T-ALL group 2 (n = 5) T-ALL patient samples at diagnosis. A one-way Anova test followed by Tukey’s HSD post-hoc test was used for statistical analysis and only significant comparisons are shown. f Heatmap showing sum of interaction specificity weights for T cell, B cells, myeloid and leukemia cell-cell interactions in group 2 T-ALL patients (NATMI).

Fig. 6. CXCL16 promotes DNαβ T cell transcriptional programs in chronically activated T cells in vitro.

a Interaction specificity weights of ligand-receptor interactions between NC monocytes and DNαβ T cells in T-ALL group 2 patients (NATMI). b Bubble heatmap showing expression of genes involved in the CXCL16-CXCR6 axis in group 2 T-ALL patients at diagnosis. c mRNA expression of CXCL16, ADAM10 and ADAM17 in peripheral blood NC monocytes for group 2 T-ALL patients at diagnosis (n = 1603 cells), prophase (n = 740 cells) and remission (n = 103 cells) and healthy donors (n = 168 cells). A Kruskal-Wallis test followed by Dunn’s post-hoc test with Holm correction was used for statistical testing and only significant comparisons are shown. d Barplot showing concentration of soluble CXCL16 in serum of primary peripheral blood samples grouped by percentage of NC monocytes of monocytes. Data are presented as mean values +/- SEM (0-30%: n = 3, 30-60%: n = 2, 60–100%: n = 2) including technical duplicates. A Kruskal-Wallis test followed by Dunn’s post-hoc test with Holm correction was used for statistical testing and only significant comparisons are shown. e Experimental approach for in vitro induction of CD8^low cells. CD8⁺ T cells were activated with anti-CD3/CD28 beads and cultured with 150 U/ml IL-2 alone or IL-2 with 100 ng/ml CXCL16 for 7 days. Upon harvest, CD8^high, CD8^mid and CD8^low CD3⁺ T cells were sorted and processed for low-input RNA-seq using the Smart-seq2 protocol. f, Heatmap of scaled normalized expression of low-input RNA-seq using the Smart-seq2 protocol of genes in the DNαβ T cell transcriptome score for in vitro induced CD8^high (n = 3), CD8^mid (n = 4) and CD8^low T cells at day7 in the absence (n = 2) or presence (n = 4) of CXCL16. g Heatmap of regulon activity (RScenic) of top DNαβ T cell regulons shown in Fig. 4g for sorted CD8^high, CD8^mid and CD8^low T cells from cultured CD8⁺ T cells harvested at day 7. h Boxplots showing regulon activity for in vitro induced CD8^low T cells at day 7 in the presence or absence of CXCL16 shown in (g) for each hierarchical cluster (n = 3, n = 3 and n = 5 regulons in each cluster, respectively). A paired t-test was used for statistical analysis; n.s.: not significant.

Fig. 7. Inhibition of Rap1 signaling enhances apoptotic priming in T-ALL.

a Bar graph of significant gene ontologies based on the top 100 differentially expressed genes in malignant cells of group 2 T-ALL patients compared to group 1 T-ALL patients (Supplementary Data 7) as analyzed by DAVID Gene ontology. P-value is derived from a Modified Fisher’s Exact Test. Immunoreg: Immunoregulatory interactions between a lymphoid and non-lymphoid cell. b Ligand-receptor interactions (NATMI) between NC monocytes and malignant cells from T-ALL group 2 patients. c, d Violin plots showing surface protein expression of CD54 in NC monocytes from healthy donors (n = 102 cells (BM), n = 168 cells (PB)), T-ALL group 1 (n = 881 cells (PB), n = 1531 cells (BM)) and T-ALL group 2 (n = 1603 cells (PB), n = 1226 cells (BM)) at diagnosis (c), and CD11a and CD18 in malignant cells from T-ALL group 1 (n = 4 (BM), n = 6 (PB)) and T-ALL group 2 (n = 1 (BM), n = 4 (PB)) patients (downsampled to 300 cells/sample) at diagnosis (d). A one-way Anova test followed by Tukey’s HSD post-hoc test (c) or an unpaired t-test (d) was used for statistical analysis. e Viability as assessed by CellTiter-Glo assay upon GGTI-298 treatment in T-ALL cell lines (Jurkat, HPB-ALL and CCRF-CEM) for three days and two primary T-ALL samples from group 2 (P8 and P14) for five days. Viability is normalized to vehicle control DMSO. Each dot represents a data point from an independent replicate. Bars and error bars represent the mean and SEM of triplicates. f Heatmap of normalized cytochrome c^- cells upon 12-hour treatment with GGTI-298 or vehicle DMSO in T-ALL cell lines Jurkat, HPB-ALL and CCRF-CEM. g Most synergistic area score (MSA) (Bliss, SynergyFinder) of combination treatment with Rap1 inhibitor GGTI-298 and indicated drugs in T-ALL cell lines (Jurkat, HPB-ALL and CCRF-CEM) and two primary T-ALL patients samples from group 2. MSA > 10 is considered strongly synergistic (dashed line). Each dot represents the MSA and error bars indicate 95% confidence interval. h Dose-response curves (CellTiter-Glo assay) of T-ALL cell lines (Jurkat and HPB-ALL) and primary T-ALL P14 treated with various concentrations of GGTI-298 and navitoclax (top) or venetoclax (bottom) for three or five days, respectively. Viability of GGTI-298 without addition of BCL2 inhibitors were normalized and set to 100%. Most synergistic area score (MSA) (Bliss, SynergyFinder) is shown (see (g)). Each dot represents the mean and error bars represent SEM of three independent replicates.

Fig. 8. Distinct leukemia transcriptional signatures are linked to patient outcome.

a Heatmap of scaled normalized RNA expression of top 15 genes differentially expressed in leukemia cells of group 1 T-ALL patients and group 2 T-ALL patients. A score for each group was computed based on the sum of expression of these genes. b Violin plots showing surface protein expression of genes present in the T-ALL transcriptomic scores shown in (a) for malignant cells from T-ALL group 1 (n = 4 (BM), n = 6 (PB)) and T-ALL group 2 (n = 1 (BM), n = 4 (PB)) patients (downsampled to 300 cells/sample) at diagnosis. An unpaired t-test was used for statistical analysis; n.s.: not significant. c Group 2 scores for primary T-ALL patient samples of an independent dataset (GSE181157) split by percentage of CD4^-CD8^-CD7^- T cells analyzed by flow cytometry ( < 3% n = 11, >3% n = 3, see Supplementary Fig. 8b). A Wilcoxon rank-sum test was used for statistical analysis. d Visualization of T-ALL oncogenic groups ordered by group 2 scores for patient samples from the TARGET cohort (n = 265). A chi-square test comparing lower tertile and upper tertile was used for statistical analysis. e Kaplan-Meier curve of event-free survival (EFS, n = 580) and overall survival (OS, n = 597) of MRD low T-ALL patients from the Gabriella Miller Kids First Pediatric Research Program dbGaP phs002276.v2.p1 grouped based on score for group 2. A log-rank test was used for statistical analysis to compare high versus low score groups, also see Supplementary Fig. 8c. f Illustration of the remodeled immune system and leukemic Rap1 signaling in group 2 T-ALL patients.