3D chromatin hubs as regulatory units of identity and survival in human acute leukemia

Abstract

Cancer progression involves genetic and epigenetic changes that disrupt chromatin 3D organization, affecting enhancer-promoter interactions and promoting growth. Here, we provide an integrative approach, combining chromatin conformation, accessibility, and transcription analysis, validated by in silico and CRISPR-interference screens, to identify relevant 3D topologies in pediatric T cell leukemia (T-ALL and ETP-ALL). We characterize 3D hubs as regulatory centers for oncogenes and disease markers, linking them to biological processes like cell division, inflammation, and stress response. Single-cell mapping reveals heterogeneous gene activation in discrete epigenetic clones, aiding in patient stratification for relapse risk after chemotherapy. Finally, we identify MYB as a 3D hub regulator in leukemia cells and show that the targeting of key regulators leads to hub dissolution, thereby providing a novel and effective anti-leukemic strategy. Overall, our work demonstrates the relevance of studying oncogenic 3D hubs to better understand cancer biology and tumor heterogeneity and to propose novel therapeutic strategies.

Keywords: HiChIP; MYB; chromatin structure; heterogeneity; hubs; leukemia; modules; scATAC.

Figure 1.. Mapping of leukemic 3D chromosomal landscapes by H3K27ac HiChIP

(A) Hierarchical clustering of primary T-ALL, ETP-ALL, and hematopoietic cell samples, based on RNA expression of leukemia-subtype marker genes (Table S2). (B) Scheme of integrated analysis of primary samples by H3K27ac HiChIP and RNA-seq. (C) Principal-component analysis of samples analyzed by H3K27ac HiChIP, based on interactivity (IA) scores. (D) Log₂ fold changes of RNA expression and interactivity levels of genes comparing primary T-ALL and ETP-ALL samples, highlighting disease-specific markers.

Figure 2.. Establishment of 3D neo-hubs after chromosomal translocations

(A–H) Hi-C contact matrices (A and B–F) and virtual 4C analyses (C and D–H) at the t(10;14)(q24;q11) translocation in pt-06 and pt-11 (A–D, TLX1 promoter as viewpoint) and at the t(1;14)(p32;q11) in pt-03 (F–H, TAL1 promoter as viewpoint), analyzed by H3K27ac HiChIP. (E–G) H3K27ac HiChIP and RNA-seq signals in T-ALL patients at the chr14 and chr10 (E) and at the chr14 and chr1 (G) regions involved in translocations.

Figure 3.. In silico and CRISPR-interference screening of leukemia-specific 3D hubs

(A) Experimental design of CRISPRi screenings. (B) In silico perturbation analysis of ATXN7L3B promoter by C.Origami. Predicted Hi-C matrices, insulation scores and peak counts, and changes in DEL (ATXN7L3B promoter deletion) and REF (wild-type CUTTL1 prediction) are indicated. (C) H3K27ac HiChIP, RNA-seq, CTCF ChIP-seq, and CRISPRi and CRISPRko analyses at chr12:74,862,748–77,172,303 in the indicated samples. (D and E) CRISPRi of ATXN7L3B, KRR1, and BBS10 promoters (D) and CRISPRko of ATXN7L3B, KRR1, and BBS10 coding regions (E) and analysis of percentage of mCherry+ cells over time expressed as ratio over day 4 post-transduction. (F) qPCR analysis of ATXN7L3B, KRR1, and BBS10 after CRISPRi (mean ± SEM), 7 days post-infection with sgRNAs. Samples compared with sgCTRL by one-way ANOVA.

Figure 4.. CRISPRi of acute leukemia surface markers

(A) Percentage of H3K27ac HiChIP anchors falling in the same category across the four cell types analyzed. (B and C) Cell-type specificity scores for promoter hubs (B) or genes interacting with hubs (C, considering the maximally expressed one or a mean of all of them) shared by all (all 4) or specific for only one cell type (only 1). Specificity scores were compared by Mann-Whitney test. (D) Percentages of T-ALL or ETP-ALL disease marker genes (Table S2) centered on a hub or interacting with a hub across primary samples. (E) CD33 expression by RNA-seq in CD34⁺ HSPCs, double-positive thymocytes (DP THYs), single-positive thymocytes (SP THYs), T-ALL, and ETP-ALL samples. (F and G) CD33 expression by RNA-seq in ETP-ALL, or near ETP-ALL or T-ALL, in two patient cohorts (TARGET dataset “phs000464” in F, EGAS00001004810 in G) was compared by Kruskal-Wallis test (F) and Mann-Whitney test (G). (H) H3K27ac HiChIP, RNA-seq and location of sgRNAs used in (I) and (J) at chr19: 51,663,246–51,765,862 in the indicated samples. (I and J) CD33 qPCR (I, mean ± SEM) and flow cytometry (J) analyses of CUTLL3 cells 7 days after CRISPRi at the indicated loci. Samples compared with sgCTRL by one-way ANOVA. (K) H3K27ac HiChIP, RNA-seq and the location of sgRNAs at chr2: 24,181,676–24,381,913 in the indicated cell types. (L) RNA-seq expression of MFSD2B in CD34⁺ HSPCs, thymocytes, and T-ALL and ETP-ALL cases. (M and N) MFSD2B expression by RNA-seq in ETP-ALL, or near ETP-ALL or T-ALL, in two patient cohorts (TARGET dataset phs000464 in M, EGAS00001004810 in N) was compared by Kruskal-Wallis test (M) and Mann-Whitney test (N). (O) MFSD2B qPCR analysis (mean ± SEM) of CUTLL3 cells 7 days after CRISPRi at the indicated loci. Samples compared with sgCTRL by one-way ANOVA.

Figure 5.. Identification of SH-defined gene modules

(A) SH analysis in primary samples. (B–E) Heatmap representation of combined Z scores in the indicated primary samples using T-ALL (B and C) or ETP-ALL (D and E) SH module genes. (F) Spearman correlations between CD1B and TSHR Z scores over combined T-ALL SH module scores in leukemia (TARGET dataset phs000464). (G) CD1A/B/C/D/E expression by RNA-seq in ETP-ALL and T-ALL patients (EGAS00001004810), compared by Mann-Whitney test. (H) H3K27ac HiChIP, RNA-seq and location of sgRNAs at chr1:158,127,840–158,490,917 in the indicated samples. (I and J) CD1B and CD1E qPCR (I) (mean ± SEM) and flow cytometry (J) analyses of CUTLL1 cells 7 days after CRISPRi at the indicated loci. Samples compared with sgCTRL by one-way ANOVA. (K and L) TSHR expression by RNA-seq in ETP-ALL, or near ETP-ALL or T-ALL, in two patient cohorts (TARGET dataset phs000464 in K, EGAS00001004810 in L) was compared by Kruskal-Wallis test (K) and Mann-Whitney test (L). (M) H3K27ac HiChIP and RNA-seq of the indicated samples and location of CRISPRi sgRNAs used in (N) at chr14:81,360,751–81,749,969. (N) TSHR and GTF2A1 qPCR analysis (mean ± SEM) of CUTLL1 cells 7 days after CRISPRi at the indicated loci. Samples compared with sgCTRL by one-way ANOVA. (O) RNA-seq expression of TSHR and GTF2A1 in CD34⁺ HSPCs, thymocytes, T-ALL, and ETP-ALL were compared by two-way ANOVA.

Figure 6.. SH heterogeneity defined by scATAC-seq and identification of high-risk disease groups

(A–F) Uniform manifold approximation and projections (UMAPs) of pt-03 (A) and IA-06 (E) cells analyzed by scATAC-seq. (B–F) Combined gene scores of chromatin accessibility (ChrAcc) for pt-03 clusters, using T-ALL SH modules (B) and for IA-06 clusters, using ETP-ALL SH modules (F). (C and D) pt-03 cells colored according to combined gene scores of ChrAcc, using T-ALL SH module 7 (C) and 2 (D) genes. (G and H) IA-06 cells colored according to combined gene scores of ChrAcc, using ETP-ALL SH module V (G) and VI (H) genes. (I) H3K27ac HiChIP, RNA-seq, and pseudobulk ATAC-seq analysis for IA-06 at chr6:31,574,498–31,622,596 (left) and chr6:31,684,841–31,714,907 (right). Cluster tracks were ordered by decreasing module VI scores. (J) Hierarchical clustering of leukemic samples (TARGET dataset phs000464), based on the combined Z scores of T-ALL SH-associated gene modules, into two groups (cycling vs. stressed). (K) Minimal residual disease (MRD) status at day 29 post-induction therapy of leukemia patients belonging to the groups in (J).

Figure 7.. Targeting MYB-dependent 3D hubs in T cell acute leukemia

(A) LOLA analysis of transcription factors enriched at ETP-ALL vs. T-ALL hubs. (B) Changes in IA of genes undergoing downregulation (DOWN), upregulation (UP), or that remained stably expressed (STABLE) after MYB knockdown. (C) H3K27ac HiChIP and RNA-seq in CUTLL1 sgCTRL and sgMYB cells, MYB ChIP-seq (GSM1519635) in Jurkat cells at chr22:19,314,234–19,519,795 and chr11:61,553,260–61,637,028. (D) Comparison of log₂ fold changes in H3K27ac activity and IA in CUTLL1 knockdown for MYB (sgMYB) vs. control (sgCTRL). (E) Log₂ fold changes of interactivity at anchors (hubs vs. non-hubs) after MYB knockdown by Cas13. (F) SH ranking in sgCTRL and sgMYB CUTTL1 cells, considering anchors with a decrease in IA (blue dots in D). (G) Annexin V/DAPI staining by flow cytometry and quantification (mean ± SEM) of alive, apoptotic, and necrotic cells of PDX T-ALL cells treated ex vivo with 20 μM TG3, MYBMIM, or vehicle (PBS). Samples were compared by two-way ANOVA.