Trib1 promotes acute myeloid leukemia progression by modulating the transcriptional programs of Hoxa9

Abstract

The pseudokinase Trib1 functions as a myeloid oncogene that recruits the E3 ubiquitin ligase COP1 to C/EBPα and interacts with MEK1 to enhance extracellular signal-regulated kinase (ERK) phosphorylation. A close genetic effect of Trib1 on Hoxa9 has been observed in myeloid leukemogenesis, where Trib1 overexpression significantly accelerates Hoxa9-induced leukemia onset. However, the mechanism underlying how Trib1 functionally modulates Hoxa9 transcription activity is unclear. Herein, we provide evidence that Trib1 modulates Hoxa9-associated super-enhancers. Chromatin immunoprecipitation sequencing analysis identified increased histone H3K27Ac signals at super-enhancers of the Erg, Spns2, Rgl1, and Pik3cd loci, as well as increased messenger RNA expression of these genes. Modification of super-enhancer activity was mostly achieved via the degradation of C/EBPα p42 by Trib1, with a slight contribution from the MEK/ERK pathway. Silencing of Erg abrogated the growth advantage acquired by Trib1 overexpression, indicating that Erg is a critical downstream target of the Trib1/Hoxa9 axis. Moreover, treatment of acute myeloid leukemia (AML) cells with the BRD4 inhibitor JQ1 showed growth inhibition in a Trib1/Erg-dependent manner both in vitro and in vivo. Upregulation of ERG by TRIB1 was also observed in human AML cell lines, suggesting that Trib1 is a potential therapeutic target of Hoxa9-associated AML. Taken together, our study demonstrates a novel mechanism by which Trib1 modulates chromatin and Hoxa9-driven transcription in myeloid leukemogenesis.

Graphical abstract

Figure 1.

Trib1 expression induces an aggressive AML phenotype. (A) Morphologies of Hoxa9-expressing AML cells with (hi) or without (null) Trib1 expression (right). Reverse transcription (RT) PCR of Trib1 expression in hi cells (left). Scale bars, 20 μm. (B) Fluorescence-activated cell sorting analysis shows decreased expression of Mac1 and Gr-1 (left) and an increased CD34-positive fraction (right) in Trib1 hi cells. The numbers indicate frequencies (%) of Mac1/Gr-1-double-positive, Mac1-single-positive, and CD34-positive fractions. (C) Decreased expression of the p42 isoform of C/EBPα in Trib1 hi cells. (D) Enhanced and prolonged phosphorylation of ERK1/2 in Trib1 hi cells. (E) Increased proliferation of Trib1 hi cells. ***P < .001. (F) Increased 5-ethynyl-2′-deoxyuridine (EdU) incorporation in Trib1 hi cells. (G) Gene set enrichment analysis shows correlation of the cell cycle pathway (left) and inverse correlation of the C/EBPα network gene sets (right). FDR, false discovery rate; NES, normalized enrichment score; p-val, P value. (H) AML developed by transplantation of Trib1 hi cells with a median survival time of 97.1 days, whereas transplantation of Trib1 null cells failed to show AML development in vivo. Significance between 2 cohorts was examined by a log-rank test.

Figure 2.

Hoxa9-binding sites in Trib1 hi and null cells show close association with enhancers and C/EBPα-binding sites. (A) Global distribution of Hoxa9-binding peaks. (B) MEME suite motif analysis shows enrichment of the C/EBPα motif in Hoxa9-binding peaks and vice versa in Trib1 null cells. The Meis1-binding motif was also enriched in both Hoxa9 and C/EBPα peaks. (C) Venn diagram showing frequent overlaps between Hoxa9- and C/EBPα-binding peaks in Trib1 null cells (left). Density plot of Hoxa9 and C/EBPα ChIP-seq data sets centered on TSSs (right). Each row represents a single peak. (D) Meta-profile of average ChIP-seq signals for Hoxa9 and H3K27ac in Trib1 null (left) and hi (right) cells in the region ± 5 kb around TSSs. (E) ChIP-seq occupancy profiles for C/EBPα, Hoxa9, and H3K27ac at the Chek1, Selp, and Ly6e loci.

Figure 3.

Different super-enhancer distribution between Trib1 hi and null cells identifies Hoxa9/Trib1 target genes. (A) Enhancers were ranked by increasing H3K27ac ChIP-seq signals in Trib1 hi (left, top) and null (left, bottom) cells. Using the ROSE algorithm, 437 and 765 enhancers were defined as super-enhancers in Trib1 hi and null cells, respectively. Overlap between super-enhancers and Hoxa9 DNA-binding peaks are shown in Venn diagrams (right). (B) Enrichment of Gene Ontology biological process for Trib1 hi super-enhancer loci. (C) Schematic diagram for target gene identification. SE, super-enhancer. (D) Quantitative RT-PCR shows increased expression of Hoxa9/Trib1 target genes in Trib1 hi cells. (E) Quantitative ChIP-PCR of H3K27ac signals for super-enhancers of Erg, Spns2, Rgl1, and Pik3cd. The Chek1 locus is shown as a negative control. Three distinct loci for each super-enhancer were examined, and the average accumulation in these 3 loci is indicated. (F) Density plots for ChIP-seq reads of C/EBPα, Hoxa9, and H3K27ac in Trib1 hi and null cells at the +85 enhancer of Erg. The yellow arrowhead indicates the position of the sgRNA for the +85 enhancer. (G) Immunoblotting shows significant decrease of Erg protein expression (arrow) by knockout of the Erg enhancer using a sgRNA for the +85 enhancer. The asterisk in immunoblotting indicates nonspecific bands. *P < .05, **P < .01, ***P < .001; n.s, not significant.

Figure 4.

Trib1 modulates super-enhancer activity and gene expression via C/EBPα degradation. (A) Trib1 hi cells were treated with the MEK1 inhibitor U0126 (10 μM) for 24 hours. Inhibition of ERK1/2 phosphorylation is evidenced by immunoblotting (left). Quantitative ChIP-PCR showing relative signals of H3K27ac to input DNA (right). The Chek1 locus was used as a negative control. (B) Quantitative RT-PCR showing mild downregulation of Erg and Spns2 expression by U0126 treatment (right). Expression of the Erg protein (arrow) was also diminished (top). (C) Cebpa was silenced by shRNA treatment using 2 different sequences in Trib1 null cells. The effect of shRNA was confirmed by immunoblotting (left). Quantitative ChIP-PCR showing relative signals of H3K27ac to input DNA (right). The Chek1 locus was used as a negative control. Three distinct loci for each super-enhancer were examined and the average accumulation in these 3 loci is indicated. (D) Quantitative RT-PCR showing upregulation of Erg, Spns2, Rgl1, and Pik3cd expression by Cebpa silencing (right). Expression of the Erg protein (arrow) is significantly upregulated (left). (E) Quantitative RT-PCR showing partial reduction of the Cebpa silencing effect on Erg, Spns2, Rgl1, and Pik3cd expression by human CEBPA p42, but not p30, expression (left). Western blot showing moderate decrease of the Erg protein by p42, but not p30, expression (center, top). Myc-tagged p42 or p30 expression (center, middle). Quantitative ChIP-PCR showing relative signals of H3K27ac to input DNA (right). The Chek1 locus was used as a negative control. *P < .05, **P < .01, ***P < .001. DMSO, dimethyl sulfoxide.

Figure 5.

Erg is an important downstream target involved in the effect Trib1 has on Hoxa9. (A) Immunoblotting showing an increase in 55-kDa Erg protein expression (arrow) in Trib1 hi cells. (B) Growth promotion of Trib1 null cells by Myc-tagged Erg overexpression. Immunoblotting of Myc-tagged Erg protein expression (left). Cell numbers of Trib1 null, Trib1 null with Erg, and Trib1 hi cells are shown on the right. (C) shRNA-mediated silencing of Erg. Relative Erg mRNA expression is shown (left). Cell numbers of Trib1 null, Trib1 hi, and Trib1 hi introduced with 2 different shRNA lentivirus are shown on the right. (D) Erg knockout by CRISPR/Cas9 in Trib1 hi cells. Cell numbers of Trib1 null, Trib1 hi, and Trib1 hi cells with sgRNA-mediated Erg knockout (right). (E) A Kaplan-Meier survival curve for Trib1 hi, Trib1 hi cells with sgRNA-mediated Erg knockout, and Trib1 null cells with Erg overexpression. Significance between Trib1 hi and Trib1/Erg knockout, and between Trib1 hi and Trib1 null/Erg overexpression was examined by a log-rank test. ***P < .001.

Figure 6.

JQ1 treatment inhibits the growth of Trib1 hi AML and leads to Erg downregulation. (A) Growth of Trib1 hi (left) and null (right) cells treated with JQ1 or vehicle. (B) Quantitative RT-PCR showing downregulation of Erg and Spns2 but not Rgl1 and Pik3cd expression after JQ1 treatment. (C) Monocytic differentiation of Trib1 hi cells by JQ1 treatment of 24 hours. Left, Giemsa staining; center, flow cytometric analysis for F4/80 expression. The isotype control-stained cells are indicated as yellow and green histograms. F4/80-positive fractions are quantitated as bar graphs; right, qRT-PCR for Il6ra expression. Scale bar: 20 μm. (D) Detection of early apoptosis induced by JQ1 treatment of 48 hours. Annexin V staining shows a significant increase in early apoptotic cells, as evidenced by flow cytometry (left). Bar graphs show an increase in both early and late apoptotic cells (right). (E) Growth of Trib1 hi (left) and null (right) cells treated with THZ1 or vehicle. (F) Quantitative RT-PCR shows no significant changes in target gene expression by THZ1 treatment. (G) In vivo treatment of mice bearing AML with JQ1 and with or without cytarabine (AraC) and daunorubicin (DnR). Schematic illustration of the experiment (left). Kaplan-Meier survival curves showing improvement of survival with JQ1 treatment and with or without AraC and DnR. ** P < .01, *** P < .001. BMT, bone marrow transplant; i.p., intraperitoneal; PI, propidium iodide.

Figure 7.

An important role for the TRIB1/ERG axis in human AML. (A) Quantitative RT-PCR for TRIB1, HOXA9, and ERG in human AML cell lines. (B) Downregulation of TRIB1/HOXA9 downstream genes in KU812 cells by TRIB1 silencing. Relative gene expression to shControl in KU812 cells. (C) Growth inhibition by TRIB1 knockdown. Cell numbers of KU812 cells treated with 2 different shRNA lentivirus. (D) KU812 and P39 cells were treated with JQ1 at the indicated dosages for 3 days. Cell numbers at the indicated time points are shown. (E) Effects of JQ1 on gene expression in KU812 and P39 cells. Relative mRNA expression of the indicated genes is shown. (F) Kaplan-Meier survival curves for patients for patients with normal-karyotype AML between high and low expression of ERG (left) or SPNS2 (right). Significances were measured by a log-rank test. Hazard ratios (HRs) and 95% confidence intervals (CIs) are indicated. *P < .05, **P < .01, ***P < .001.