Defining the transcriptional control of pediatric AML highlights RARA as a superenhancer-regulated druggable dependency

Abstract

Somatic mutations are rare in pediatric acute myeloid leukemia (pAML), indicating that alternate strategies are needed to identify targetable dependencies. We performed the first enhancer mapping of pAML in 22 patient samples. Generally, pAML samples were distinct from adult AML samples, and MLL (KMT2A)-rearranged samples were also distinct from non-KMT2A-rearranged samples. Focusing specifically on superenhancers (SEs), we identified SEs associated with many known leukemia regulators. The retinoic acid receptor alpha (RARA) gene was differentially regulated in our cohort, and a RARA-associated SE was detected in 64% of the study cohort across all cytogenetic and molecular subtypes tested. RARA SE+ pAML cell lines and samples exhibited high RARA messenger RNA levels. These samples were specifically sensitive to the synthetic RARA agonist tamibarotene in vitro, with slowed proliferation, apoptosis induction, differentiation, and upregulated retinoid target gene expression, compared with RARA SE- samples. Tamibarotene prolonged survival and suppressed the leukemia burden of an RARA SE+ pAML patient-derived xenograft mouse model compared with a RARA SE- patient-derived xenograft. Our work shows that examining chromatin regulation can identify new, druggable dependencies in pAML and provides a rationale for a pediatric tamibarotene trial in children with RARA-high AML.

Graphical abstract

Figure 1.

The SE landscape of pAML. (A) Similarity of primary pAML and adult AML patient samples by unsupervised hierarchical clustering of H3K27ac signal at consensus SE regions. (B) Similarity of pAML cell lines and samples by unsupervised hierarchical clustering of H3K27ac signal at consensus SE regions. Cytogenetic and molecular features depicted for each cell line and sample. Notes: replicates of the cell line Kasumi1 and the sample BM102004 were included (denoted “_1,” “_2,” and “_3”). “CBF” = core-binding factor, “Const tri 21” = a patient with trisomy 21 presenting with transient myeloproliferative disorder, and “NK” = normal karyotype. (C) Core consensus SE regions ranked according to the product of frequency and amplitude of H3K27ac signal (FASE). Consensus SE regions were defined as SE regions present in >25% of samples. (D-E) Core pAML SE-associated genes compared with known Gene Ontology (GO) signatures (D) and MSigDB C2 curated gene sets (E). Gene sets are represented by strength of significance as depicted by –Log10(P-value). (F) Enhancer ranking of three core SE-associated genes across all pAML samples. RARA SE status defined by red (present) or blue (absent) in the right panel.

Figure 2.

The majority of pAMLs contain a RARA SE, which is correlated to unique SE-associated genes. (A) Fraction of pAML samples with or without the RARA SE within cytogenetic and molecular subgroups. (B) Meta tracks of H3K27ac ChIP-seq signal (rpm/bp) at the RARA SE locus of RARA SE+ and SE− samples. (C) Rank ordering of SEs by Pearson correlation (p-corr) to the RARA SE. Select genes associated with SEs with high and low p-corr are highlighted. (D) H3K27ac ChIP-seq meta tracks at loci of SEs with high and low correlation to the RARA SE as seen in panel C. (E-F) Perturbation gene sets (MSigDB C2) enriched by gene set overlap in the SE genes strongly correlated (E, n = 693 genes, 220 SEs; cutoff, top 5% of SEs by RARA p-corr) and poorly correlated (F, n = 264 genes, 220 SEs; cutoff, bottom 5% of SEs by RARA p-corr) to the RARA SE.

Figure 3.

RARA SE+ pAML cells are sensitive to tamibarotene. RARA SE+ pAML samples per cell lines are in red, RARA SE− pAML samples per cell lines in blue, and APL cell line in gray. Experiments were conducted in biologic triplicates (except where noted), and all experiments were repeated at least once as independent, asynchronous experiments to confirm results, with representative data from one independent experiment shown. (A) RARA messenger (messenger RNA [mRNA]) levels in pAML cell lines (Kasumi1, MV4;11, THP1), an APL cell line (NB4), and patient samples (technical replicates n = 3 per cell line or sample). (B) Effect of tamibarotene (100 nM) on cell viability over time. Viability was normalized to the viability of dimethyl sulfoxide (DMSO)-treated cells measured at the same time points. (C) CD38 positivity in response to tamibarotene (100 nM) after 72 hours’ treatment. Expression of other differentiation markers in response to tamibarotene (100 nM) after 72 hours’ treatment: CD66 (D) and CD11c (E). (F) Induction of apoptosis by 100 nM tamibarotene in pAML cell lines and samples after 72 hours’ treatment. (G) DHRS3 mRNA expression fold change after 24 hours’ tamibarotene treatment (100 nM). * P < .05; **.001 < P < .01; ***.0001 < P < .001; ****P < 0.0001.

Figure 4.

Tamibarotene-induced gene expression changes in a RARA SE+ pAML sample consistent with retinoid-differentiating effect. (A) Tamibarotene-induced gene expression changes from RNA-seq in p401 after 24 hours of tamibarotene treatment (100 nM, n = 4 biologic replicates), with significance determined by a Log2FC cutoff of 1 to −1 and a P value <.05. (B) Row-normalized expression of the top 50 most differentially regulated genes in panel A defined by absolute Log2FC. (C-D) Gene set enrichment analysis (GSEA) of significant genes from panel B, with (C) being a quantitative comparison of all gene sets available from the molecular signature database (MSigDB) by gene set enrichment analysis for tamibarotene-treated p401 cells. Data are presented as a scatterplot of the false discovery rate (FDR) vs normalized enrichment score (NES) for each evaluated gene set. Red indicates significant sets based on an FDR cutoff <0.05. (D) Plot of the running sum for the most enriched retinoid gene set within the tamibarotene-treated p401 data set, including the normalized enrichment score and the leading-edge subset of enriched genes. DMSO, dimethyl sulfoxide.

Figure 5.

RARA SE presence predicts pAML in vivo sensitivity to tamibarotene. (A) Effect of tamibarotene (6 mg/kg by oral gavage daily) on peripheral blood leukemic burden in p401, a RARA SE+ pAML PDX mice (n = 5 per arm). (B) Effect of tamibarotene treatment on the liver and spleen weights of p401 PDX mice at time of euthanasia. (C) Effect of tamibarotene treatment on survival of p401 PDX mice. (D) CD38⁺ cells percentage in response to tamibarotene treatment in p401 PDX mice. (E) Effect of tamibarotene on the peripheral blood leukemia burden of p198, an RARA SE− pAML PDX (n = 6 per arm). (F) Effect of tamibarotene treatment on the liver and spleen weights of p198 PDX mice at time of euthanasia. (G) Effect of tamibarotene treatment on survival of p198 PDX mice. (H) Body weights of both p401 and p198 mice treated with tamibarotene over time. **.001 < P < .01; ***.0001 < P< .001. n.s., not significant.