RUNX2 regulates leukemic cell metabolism and chemotaxis in high-risk T cell acute lymphoblastic leukemia

Abstract

T cell acute lymphoblastic leukemia (T-ALL) is an aggressive hematologic malignancy with inferior outcome compared with that of B cell ALL. Here, we show that Runt-related transcription factor 2 (RUNX2) was upregulated in high-risk T-ALL with KMT2A rearrangements (KMT2A-R) or an immature immunophenotype. In KMT2A-R cells, we identified RUNX2 as a direct target of the KMT2A chimeras, where it reciprocally bound the KMT2A promoter, establishing a regulatory feed-forward mechanism. Notably, RUNX2 was required for survival of immature and KMT2A-R T-ALL cells in vitro and in vivo. We report direct transcriptional regulation of CXCR4 signaling by RUNX2, thereby promoting chemotaxis, adhesion, and homing to medullary and extramedullary sites. RUNX2 enabled these energy-demanding processes by increasing metabolic activity in T-ALL cells through positive regulation of both glycolysis and oxidative phosphorylation. Concurrently, RUNX2 upregulation increased mitochondrial dynamics and biogenesis in T-ALL cells. Finally, as a proof of concept, we demonstrate that immature and KMT2A-R T-ALL cells were vulnerable to pharmacological targeting of the interaction between RUNX2 and its cofactor CBFβ. In conclusion, we show that RUNX2 acts as a dependency factor in high-risk subtypes of human T-ALL through concomitant regulation of tumor metabolism and leukemic cell migration.

Trial registration: ClinicalTrials.gov NCT01185886.

Keywords: Cell migration/adhesion; Leukemias; Molecular biology; Oncology.

Conflict of interest statement

Conflict of interest: The authors have declared that no conflict of interest exists.

Figures

Figure 1. RUNX2 is upregulated in T-ALL harboring KMT2A-R and/or immature/ETP phenotype.

(A) RUNX2 expression across distinct molecular/genetic subtypes of T-ALL identified in a cohort of 242 pediatric and young adult T-ALL patients (8). (B) RUNX2 mRNA levels in HOXA-deregulated T-ALL with chromosomal rearrangements involving the KMT2A gene (KMT2A-R, n = 11; other HOXA, n = 21) (8). (C) RUNX2 expression in 189 T-ALL samples classified based on ETP (n = 19), near-ETP (n = 24), and non-ETP (n = 146) phenotype by flow cytometry (8). (D and E) RUNX2 mRNA and protein levels in T-ALL cell lines (n = 9). (F) RUNX2 mRNA expression in primary T-ALL samples depending on maturation arrest (n = 11) according to European Group for the Immunological Characterization of Acute Leukemias (EGIL) classification (103). (G) qRT-PCR analyses of RUNX2 mRNA in primary T-ALL samples with or without KMT2A-R (KMT2A-R, n = 12; others, n = 13). Data are shown as mean ± SD, using 3 independent experiments. (H) Immunoblotting for RUNX2 protein levels in primary T-ALL samples (n = 9). Red font indicates KMT2A-R samples. (I) RUNX2 protein expression in PDX samples (n = 14) by flow cytometry. (A and C) Kruskal-Wallis with Dunn’s multiple comparison test; (B) unpaired Mann-Whitney U test. *P < 0.05; **P < 0.005; ***P < 0.0005; ****P < 0.0001.

Figure 2. RUNX2 is highly expressed in ETPs, but has no major impact on phenotypic markers of normal hematopoiesis or T cell development.

RUNX2 expression in human (A) and mouse (B) T cell subsets. (C) qRT-PCR for Runx2 expression in cKit-enriched and cKit-negative mouse thymocytes. Rel., relative. (D and E) Peripheral blood cell counts measured in Runx2fl/flVaviCre+/+ (Ctrl), Runx2fl/+VaviCretg/+ (Hez), and Runx2fl/flVaviCretg/+ (KO) mice (7–10 weeks). (F) Total cell number extracted from thymuses. (G–J) Percentages of T cell subsets measured by flow cytometry. (G) DP, double-negative T cells, (H) DN, double-positive T cells, (I) CD4 single-positive T cells, (J) CD8 single-positive T cells. (C) Paired 2-tailed t test; (D–J) 1-way ANOVA with Tukey’s multiple comparison test. ***P < 0.0005.

Figure 3. RUNX2 regulates T-ALL survival.

(A) Cell growth of LOUCY and KARPAS-45 cells transduced with RUNX2 shRNAs (shRUNX2 B, shRUNX2 C) and scrambled control (shNC). Data are shown as mean ± SD for 1 of 3 independent experiments performed in triplicate. (B) Flow cytometry cell-cycle analysis using propidium iodide staining (48 hours). (C) Apoptotic cell death in T-ALL cells stained with annexin V/7-AAD (96 hours). (B and C) Data are representative of 1 of 3 independent experiments. (D) Human LOUCY and KARPAS-45 cells, murine MOHITO cells expressing control plasmid (CON), KMT2A-R (KMT2A-MLLT1, KMT2A-MLLT4), and primary T-ALL cells (PASSPP, 18-169) were transduced with RUNX2 shRNAs and scrambled control. Western blot analysis of the indicated proteins. (E) RUNX2 and H3K27ac ChIP-Seq at the CTNNB1 (β-catenin) and BIRC5 (survivin) loci in KARPAS-45. Forced expression of (F) active β-catenin and (G) survivin in LOUCY with or without shRNA-mediated RUNX2 depletion. (H and I) Annexin V/7-ADD staining in LOUCY (96 hours). Data are shown as mean ± SD for 3 independent experiments. AnV, annexin V. (J) Kaplan-Meier survival curve analyses of NSG mice (n = 10/group) transplanted with 106 LOUCY cells expressing shRUNX2 B, shRUNX2 C, or scrambled control (median log-rank Mantel-Cox test). (K) Flow cytometric quantification of human CD45+ cells in mice (n = 4/group) euthanized 41 days after inoculation with 106 transduced LOUCY cells. (L) The levels of GFP+ cells were determined in blood (day 30), BM, and spleen (day 100) in animals injected with PDX samples harboring KMT2A-MLLT-1. (A) Repeated measure ANOVA with Tukey’s multiple comparisons test; (H, I, K, and L) 1-way ANOVA with Tukey’s multiple comparison test. ***P < 0.0005; ****P < 0.0001. (D, F, and G) Representative blots from at least 2 separate experiments.

Figure 4. RUNX2 genome-wide binding profiles reveal a regulatory feedback loop with KMT2A fusion proteins.

(A and B) Most highly enriched binding motifs according to HOMER upon RUNX2 ChIP-Seq in KARPAS-45 (KMT2A-R) and PER-117 (ETP). (C and D) Distribution of RUNX2 binding over different genomic regions in KARPAS-45 and PER-117. (E) MA plot showing differentially regulated genes (PAdj < 0.05) in KARPAS-45, 72 hours after shRNA-mediated RUNX2 silencing vs. control in 5 biological replicates. (F–H) Top 10 enrichments of RUNX2-binding peaks based on functional annotation provided by GREAT (3.0.0). (I) RUNX2 and H3K27ac binding profile on the KMT2A gene region in KARPAS-45. (J) ChiP-qPCR on KARPAS-45 cells and 2 KMT2A-R primary T-ALL patient samples. Enrichment of RUNX2 on KMT2A promoter (left); enrichment of N-terminus of KMT2A (KMT2AN) on RUNX2 promoter (right). (K) Luciferase reporter assay for KMT2A on HEK293 cells transduced with RUNX2 shRNA. The RUNX2 binding site is indicated as X (–359 to –368 bp upstream of KMT2A coding starting site). (L) Silencing of RUNX2 leads to decreased levels of KMT2AN in primary T-ALL samples. (M) Runx2 expression in KMT2A-MLLT1 transformed cells compared with lineage-depleted mouse progenitor cells. (N) Gradually increased expression of Runx2 measured by qRT-PCR during KMT2A-MLLT1 driven transformation over different replatings. (J) Unpaired 2-tailed t test with Holm-Šidák correction for multiple comparisons; (M) unpaired 2-tailed t test. *P < 0.05; ***P < 0.0005; ****P < 0.0001.

Figure 5. RUNX2 upregulation promotes T-ALL cell migration and leukemia progression.

(A) Growth of CCRF-CEM and PF382 cells transduced with RUNX2-expressing (RUNX2 OE) or negative control (NC) plasmids. Data are shown as mean ± SD, 1 of 3 independent experiments performed in triplicate, repeated measure ANOVA with Tukey’s multiple comparisons test. (B) CXCR4 levels; representative histograms (left); median fluorescent intensity (MFI) ± SD, 3 independent experiments (right). Leukemic cell migration (6 hours) through (C) 3 μm porous membrane or (D) HUVEC monolayer ± CXCL12 (50 ng/μl). Data are shown as mean ± SEM, 3 separate experiments performed in duplicate. Two-way ANOVA with Tukey’s multiple comparison correction. (E) RUNX2 and H3K27ac binding at the CXCR4 locus in KARPAS-45. (F) CCRF-CEM and PF382 transduced with RUNX2 OE or negative control; MOHITO cells expressing KMT2A-R (KMT2A-MLLT1, KMT2A-MLLT4) or negative control and primary T-ALL cells (PASSPP, 18-169) transduced with shRUNX2 B, shRUNX2 C, and scrambled control. Immunoblotting with indicated antibodies (n = 2). (G) Adhesion to fibronectin (2 hours). Data are shown as mean ± SD, 3 independent experiments performed in duplicate. (H) VLA-4 levels; representative histograms (left); MFI ± SD, 3 separate experiments (right). (I) Kaplan-Meier plot of NSG mice (n = 10/group) injected i.v. with 106 PF382 cells (log-rank test). (J) Human CD45+ cells (day 34) and (K) bioluminescence imaging (NSG mice, n = 7/group) after interfemoral implantation of PF382 cells (3 × 105). (L) BM homing of leukemic cells (24 hours); NSG mice (n = 5/group) received i.v. PF382 RUNX2 OE and negative control cells (107; noncompetitive) or the mixture of fluorescently labeled PF382 RUNX2 OE (DsRed) and negative control (AmCyan) cells (1:1; 107; competitive). Unpaired and paired 2-tailed t tests, respectively. (M) Human cytokines in blood serum of NSG mice inoculated with 106 PF382 cells (day 28). (B, G, H, J, and M) Unpaired t test with Holm-Šidák multiple comparisons correction. **P < 0.005; ***P < 0.0005; ****P < 0.0001.

Figure 6. RUNX2 potentiates glycolytic and oxidative metabolism in T-ALL cells.

(A) CCRF-CEM and PF382 cells were transduced with RUNX2 expressing (RUNX2 OE) and negative control plasmids. ECAR in glucose stress assay tested on the Seahorse XF24 Bioanalyzer. (B) Representative histograms for GLUT1 levels (top); MFI ± SD, 3 separate experiments (bottom). (C) Migration of T-ALL cells pretreated with glucose inhibitor (2DG, 1 mM; 16 hours) ± CXCL12 (50 ng/μl; 6 hours, 3 μm porous membrane). Data are represented as mean ± SD, 3 independent experiments performed in duplicate. (D) ECAR quantification and (E) GLUT1 levels in MOHITO-negative control cells, MOHITO-expressing KMT2A-R (KMT2A-MLLT4, KMT2A-MLLT1) and transduced with RUNX2 shRNAs (shRUNX2 B, shRUNX2 C) or scrambled control. Representative histograms for GLUT1 expression (left); MFI ± SD, 3 independent experiments (right). OCR upon RUNX2 forced expression in (F) CCRF-CEM and (G) PF382 cells analyzed using Mito Stress assay on the Seahorse XF24 Bioanalyzer. (H) Plot representing OCR/ECAR ratios in RUNX2 OE vs. control cells. (I) OCR and (J) ratio of OCR/ECAR for MOHITO control cells and MOHITO cells with or without KMT2A-R followed by shRNA-mediated RUNX2 silencing. (K) RUNX2 and H3K27ac binding on the LDHA, PGK1, and CHCHD2 gene regions in KARPAS-45. (L) Western blot analysis using indicated antibodies. Representative blots from at least 2 independent experiments. (A, D, and F–J) Data from 1 (mean ± SD) of 2 independent experiments performed in triplicate. (A, B, F, and G) Unpaired 2-tailed t test with Holm-Šidák correction for multiple comparisons; (C, D, and I) 2-way ANOVA with Tukey’s multiple comparison test; (E) 1-way ANOVA with Tukey’s multiple comparison test. *P < 0.05; **P < 0.005; ***P < 0.0005; ****P < 0.0001.

Figure 7. RUNX2 positively regulates mitochondrial dynamics and biogenesis.

(A) Mitochondria membrane potential (MitoTracker Red CMXRos) and (B) ROS levels (CellROS Deep Red) were tested by flow cytometry in CCRF-CEM and PF382 cells transduced with RUNX2-expressing (RUNX2 OE) and negative control plasmids. Representative histograms (left); MFI ± SD, 3 separate experiments (right). (C) Mitochondrial membrane potential and (D) ROS production in MOHITO negative control cells and MOHITO-expressing KMT2A-R (KMT2A-MLLT4, KMT2A-MLLT1) and transduced with RUNX2 shRNAs (shRUNX2 B and shRUNX2 C) or scrambled control. Representative histograms and MFI ± SD, 3 independent experiments. (E and F) CCRF-CEM and PF382 cells with forced RUNX2 expression; MOHITO cells expressing control plasmid, KMT2A-R (KMT2A-MLLT1, KMT2A-MLLT4), and primary T-ALL cells (PASSPP, 18-169) were transduced with RUNX2 shRNAs and scrambled control. Representative Western blot for the expression of specified proteins (n = 2). (G and H) Migration of CCRF-CEM and PF382 cells upon RUNX2 overexpression. Cells were pretreated with (G) a mitochondrial fission inhibitor, Mdivi-1 (1 μΜ, 16 hours), or (H) AMPK pathway inhibitor, dorsomorphin (1 μM, 16 hours), followed by migration through the 3 μm porous membrane in serum-free medium ± CXCL12 (50 ng/μl; 6 hours). Data are shown as mean ± SD, 3 separate experiments performed in duplicate. (A and B) Unpaired 2-tailed t test with Holm-Šidák correction for multiple comparisons; (C and D) 1-way ANOVA with Tukey’s multiple comparison test; (G and H) 2-way ANOVA with Tukey’s multiple comparison test. **P < 0.005; ***P < 0.0005; ****P < 0.0001.

Figure 8. Immature T-ALL and KMT2A-rearranged leukemias are vulnerable to pharmacological targeting of the RUNX-CBFβ interaction.

(A–D) Gene expression in KMT2A-MLLT1 transformed mouse (Runx2fl/fl; Cre Ert2tg/+) BM progenitor cells. Full Runx2 KO was achieved by treatment of transduced progenitor cells with 400 nM 4-hydroxytamoxifen for 24 hours prior to replating; vehicle-treated (EtOH) cells served as control (Runx2 WT). RNA-Seq normalized gene counts for (A) Runx1, (B) Hoxa10, (C) Hoxa11, and (D) Hoxb13. (E) A panel of T-ALL cell lines showed reduced survival in vitro upon treatment with AI-10-104 (Cell-Titer Glo assay). (F) MA plot showing significantly (PAdj < 0.05) upregulated (red) and downregulated (blue) genes in KARPAS-45 upon 24-hour treatment with 20 μM AI-10-104 compared with DMSO controls. Enriched gene set of mitochondrial inner membrane (MIM) genes represented in dark blue. Data based on 4 independent biological replicates. PDX samples from KMT2A-MLLT1 (G) and KMT2A-MLLT4 (H) patients showed reduced survival upon 24-hour treatment with AI-10-104. (A–D) Unpaired 2-tailed t test. *P < 0.05; **P < 0.005; ***P < 0.0005.