December 30, 2014;
A role for BMP-induced homeobox gene MIXL1 in acute myelogenous leukemia and identification of type I BMP receptor as a potential target for therapy.
Inducer in Xenopus Like1 (MIXL1), a paired-type homeobox transcription factor induced by TGF-β family of ligands is required for early embryonic specification of mesoderm
. Retrovirally transduced Mixl1 is reported to induce acute myelogenous leukemia (AML
) with a high penetrance. But the mechanistic underpinnings of MIXL1 mediated leukemogenesis are unknown. Here, we establish the protooncogene c-REL
to be a transcriptional target of MIXL1 by genome wide chromatin immune precipitation. Accordingly, expression of c-REL
and its downstream targets BCL2L1
and BCL2A2 are elevated in MIXL1 expressing cells. Notably, MIXL1 regulates c-REL
through a zinc finger binding motif, potentially by a MIXL1-Zinc finger protein transcriptional complex. Furthermore, MIXL1 expression is detected in the cancer genome atlas (TCGA) AML
samples in a pattern mutually exclusive from that of HOXA9
suggesting the existence of a core, yet distinct HOX transcriptional program. Finally, we demonstrate MIXL1 to be induced by BMP4
and not TGF-β in primary human hematopoietic stem and progenitor cells. Consequently, MIXL1 expressing AML
cells are preferentially sensitive to the BMPR1 kinase inhibitor LDN-193189. These findings support the existence of a novel MIXL1-c REL
mediated survival axis in AML
that can be targeted by BMPR1 inhibitors. (MIXL1- human gene, Mixl1- mouse ortholog, MIXL1- protein).
Disease Ontology terms:
acute myeloid leukemia
LEUKEMIA, ACUTE MYELOID; AML
[+] show captions
Figure 1. MIXL1 expression confers decreased sensitivity to doxorubicin in AML cells(A) Stable transfectants of U937 cells express MIXL1 at levels similar to those of endogenous MIXL1 in AML cell lines. MIXL1 was detected by probing 30 μg of whole cell lysates resolved on SDS-PAGE and transferred to PVDF membrane, with rabbit antibodies against N-terminal epitope of MIXL1 and with β-actin for a loading control . (B) MIXL1 expression reduces sensitivity of U937 cells to doxorubicin. The cell lines were treated with 0–1.75 μM doxorubicin on day 0. Cell survival was measured at 24 hours by MTS assay as detailed in Materials and Methods. Absorbance of untreated cells was normalized to 1. Relative viability at varying concentrations of doxorubicin is denoted.
Figure 2. Identification of direct MIXL1 transcriptional targets by ChIP-Sequencing(A) Venn diagram 1MIXL-Flag normalized to control-Flag and control-IgG (Set 1), 2MIXL-Flag normalized to control-Flag and control-IgG (Set 2), and 1MIXL-Flag and 2MIXL-Flag combined and normalized to control-Flag and control-IgG (Set 3). A total of 179 peaks shared by the three groups is denoted. (B) Pie chart depicting localization of MIXL1 in the human genome. Peaks were classified according to distance from the nearest transcribed gene using the following criteria: upstream was 5–25 kbp 5′ of the transcription start site, promoter was 0–5 kbp upstream of the transcription start site, body was between the transcription start site and end, TSE was 0–5 kbp downstream of the transcriptional end, downstream was 5–25 kbp downstream of the transcriptional end, and distant peaks were those not allocated to a gene. Note that the majority of peaks (64%) localized to gene promoters. (C) ChIP of five candidate peaks identified by ChIP-Se1. FLAG antibodies were used and ChIP-Seq (EIF1, c-REL, SLC39A13, SMYD5, and ZP3) showed specific MIXL1 binding to both 1MIXL and 2MIXL clones by ChIP normal mouse IgG served as control. Error bars represent standard deviation between triplicates. (D) The most common motif in the ChIP-seq peaks are Zinc-Finger binding sites. Motif1 and Motif2 were the two most statistically significant generated using the Multiple EM for Motif Elicitation [MEME] against the peak regions identified in the ChIP-seq analysis. Both motifs are C/G-heavy regions with similarity to known zinc finger motifs.
Figure 3. MIXL1 up regulates c-REL expression to enhance anti apoptotic gene transcription(A)
MIXL1-expressing clones show enhanced transcript levels for c-REL, BCL2A1, and BCL2L1. Quantitative RT-PCR results show the differences in c-REL, BCL2A1, and BCL2L1 expression levels between the U937 control, 1MIXL, and 2MIXL cells. Expression was normalized to 18S rRNA transcript levels. Error bars represent standard deviation between triplicates. *p < 0.05. (B) ChIP localizes endogenous MIXL1 to c-REL promoter in KG1 cells. Quantitative genomic PCR analysis shows specific enrichment of endogenous MIXL1 immunoprecipitated with either N-terminal or C-terminal MIXL1 antibodies on the c-REL promoter whereas an internal locus within the c-REL gene showed no MIXL1 occupancy. Error bars represent standard deviation between triplicates. (C) Knockdown of MIXL1 decreased while enforced expression of c-REL increased c-REL, BCL2A1, and BCL2L1 transcript levels. MIXL1 shRNA lentivirus and c-REL retrovirus were transduced into KG1 cells. RT-qPCR was performed in triplicate on RNAs isolated 48 hours after transduction. Expression was normalized to 18S rRNA levels, and error bars represent standard deviation between triplicates. (D)
c-REL over-expression rescues MIXL1 knockdown–mediated growth arrest in KG1 cells. Growth was measured by MTS assay every 24 hours over a 4-day period in KG1 cells transduced with MIXL1 shRNA lentivirus and c-REL retrovirus. Absorbance was normalized to that of a non-transfected control sample. *p < 0.05.
Figure 4. MIXL1 binds to the c-REL promoter(A)
c-REL promoter peak region identified by ChIP-Seq, as generated by the University of California, Santa Cruz, genome browser is shown. The location and size of each promoter fragment used for the luciferase reporter assay is displayed underneath. (B) MIXL1 binds to a 550-bp region within the c-REL promoter. Regions of the DNA depicted in 4A were cloned into the reporter vector pBV-Luc luciferase, which were then transiently co-transfected into HEK293T cells with MIXL1, MIXL1 Homeobox-less, or empty expression vector. Equal amount of Renilla luciferase co-transfected with the reporter constructs allowed normalization. Luciferase activity of each combination was tested in triplicate after 48 hours. Error bars represent standard deviation between triplicates. (C) MZF1 binds to the same locus as MIXL1 on c-REL promoter. Quantitative PCR analysis of the identified c-REL promoter region and c-REL intron control region compared the abundance of each genomic locus immunoprecipitated by either IgG or MZF1 antibodies, normalized to a standard curve. Error bars represent standard deviation between triplicates. *p < 0.05. (D) Model depicting a potential MIXL1-TBX-MZF1 multiprotein complex activating c-REL transcription.
Figure 5. High MIXL1 expression denotes a distinct subset of AML(A) Mutually exclusive expression of MIXL1 and HOXA9 in distinct FAB subsets. The RNA-seq data are publicly available from the TCGA website (https://tcga-data.nci.nih.gov/tcga/). 177 samples of Acute Myeloid Leukemia classified by leukemia French American British morphology code (FAB) and a total of 20319 genes with expression values in the RPKM format were included. The data were quantile normalized using the normalize Quantiles function from the limma package. The expression levels of MIXL1 and HOXA9 of the 177 samples from 8 FAB categories were plotted in the heatmap using the heatmap.2 function in the gplots package of R 3.1.1. Within each FAB category, the samples were ordered according to the expression value of the MIXL1 gene. (B)
MIXL1 upregulation identifies a non-overlapping AML subset from those expressing CDX2, HOXA9, or HLX. TCGA AML patient dataset was queried for alterations in expression as determined by RNA-Seq across 166 AML cases through the cBioPortal database. Each column represents a case of AML. MIXL1 is amplified or upregulated in 13% of the total AML cases. Note the predominantly non-overlapping expression patterns of MIXL1, CDX2, HOXA9, and HLX. (C) Seventy-six percent (16/21) of MIXL1-expressing cases in TCGA AML dataset harbored somatic mutations common in AML (NPM1, FLT3, DNMT3A, IDH1, RUNX1 JAK3, and TP53). Each column represents a case. (D) Relapse-free survival of MIXL1-expressing cases is lower than that of non–MIXL1-expressing cases. TCGA AML cases were separated into two groups: MIXL1-expressing (increase in expression or amplified) and non–MIXL1-expressing. Relapse-free survival was then compared between the two groups using the Kaplan-Meier method.
Figure 6. BMP4 induced MIXL1, an important survival axis and therapeutic target in AML(A) MIXL1 expression increased 2-fold in CD34+ HSPCs treated with BMP4. CD34+ HSPCs from three cord blood donors (unique donor number 47, 51, 60) were cultured with either 50 ng/ml BMP4 or 2 ng/ml TGF-β1 for 2 hours. MIXL1 transcript levels were quantified by RT-qPCR using 18S rRNA as normalization control. Error bars represent standard deviation between triplicates. *p < 0.05. (B) LDN-193189 at 3 μM was cytotoxic to OCI-AML2, ML3, KG1, and K562 cells but not U937 and HL60 cells. Each cell line was treated with vehicle or 3 μM LDN-193189 on day 0, and viability was measured every 24 hours by MTS assay in triplicate. (C) OCI-AML2, ML3, KG1, and K562 were sensitive to 200 nM LDN-193189, while the non–MIXL1-expressing lines U937 and HL60 were unaffected. Each cell line was treated in triplicate with 0–700 nM LDN-193189, with the drug or control medium replenished every 24 hours, for 4 days. Viability on day 4 was assayed by MTS assay. Absorbance was normalized to that of control samples treated with vehicle only.
Figure 7. BMP4 induced MIXL1 an important survival axis and therapeutic target in AML(A) Canonical BMP signaling pathway. Upon BMP binding to type II receptor (BMPRII), type I receptor BMPRI is phosphorylated. Activated BMPRI phosphorylates SMAD1 or 5 which heterodimerizes with SMAD4 to up regulate MIXL1 expression. (B) BMPR1 mediated signal is hypothesized to induce endogenous MIXL1 expression in KG1, OCI-ML2, ML3 and K562 cell lines. These cells are sensitive to BMPR1 inhibitor LDN-193189. Enforced expression of MIXL1 in U937 cells 1MIXL1 is independent of BMP signaling and therefore insensitive to LDN-193189. In HL60 and U937 cells which are insensitive to LDN-193189, the pathway may be absent or modified.