IRX-related homeobox gene MKX is a novel oncogene in acute myeloid leukemia

Abstract

Homeobox genes encode transcription factors which organize differentiation processes in all tissue types including the hematopoietic compartment. Recently, we have reported physiological expression of TALE-class homeobox gene IRX1 in early myelopoiesis restricted to the megakaryocyte-erythroid-progenitor stage and in early B-cell development to the pro-B-cell stage. In contrast, sister homeobox genes IRX2, IRX3 and IRX5 are aberrantly activated in the corresponding malignancies acute myeloid leukemia (AML) and B-cell progenitor acute lymphoid leukemia. Here, we examined the role of IRX-related homeobox gene MKX (also termed IRXL1 or mohawk) in normal and malignant hematopoiesis. Screening of public datasets revealed silent MKX in normal myelopoiesis and B-cell differentiation, and aberrant expression in subsets of AML and multiple myeloma (MM) cell lines and patients. To investigate its dysregulation and oncogenic function we used AML cell line OCI-AML3 as model which strongly expressed MKX at both RNA and protein levels. We found that IRX5, JUNB and NFkB activated MKX in this cell line, while downregulated GATA2 and STAT5 inhibited its expression. MKX downstream analysis was conducted by siRNA-mediated knockdown and RNA-sequencing in OCI-AML3, and by comparative expression profiling analysis of a public dataset from MM patients. Analysis of these data revealed activation of CCL2 which in turn promoted proliferation. Furthermore, MKX upregulated SESN3 and downregulated BCL2L11, which may together underlie decreased etoposide-induced apoptosis. Finally, myeloid differentiation genes CEBPD and GATA2 were respectively up- and downregulated by MKX. Taken together, our study identified MKX as novel aberrantly expressed homeobox gene in AML and MM, highlighting the function of IRX1 in normal myelopoiesis and B-cell development, and of IRX-related genes in corresponding malignancies. Our data merit further investigation of MKX and its deregulated target genes to serve as novel markers and/or potential therapeutic targets in AML patient subsets.

Introduction

The hematopoietic system is divided into the myeloid and lymphoid lineages, generating all blood and immune cells lifelong, forming a highly proliferative tissue. On the other hand, reflecting this high proliferative capacity, hematopoietic cells are prone to malignant transformation. Accordingly, specific genetic alterations, like chromosomal and genomic rearrangements or gene mutations in hematopoietic stem and progenitor cells, often underlie aberrant gene activities and the development of leukemia and lymphoma. Subtype-specific aberrations serve as diagnostic markers and may comprise therapeutic targets [1–4]. Therefore, identification of novel mutated or deregulated genes extends the potential for the clinical treatment.

Regulation of developmental processes in hematopoiesis is primarily orchestrated at the transcriptional level [5–10]. Accordingly, many hematopoietic malignancies display mutations and/or deregulations of transcription factors (TFs). Therefore, analysis of TF-encoding genes may reveal novel pathological mechanisms and clinical tracks. The human genome contains about 1.600 TF-encoding genes [11]. These genes and factors are classified according to sequence and structure of their DNA-binding domains. Homeobox genes encode developmental TFs, sharing a conserved homeodomain which contains three alpha-helices connected by two loops [12–14]. TALE-class homeodomain factors share a three amino acid insertion within the first loop of their homeodomain, representing a unique feature [13]. Due to their basic impact in cell and tissue differentiation, homeobox genes are frequently deregulated in cancer including leukemia/lymphoma [15–17]. PBX1 is a well-known member of the TALE-class and physiologically expressed in early stem and progenitor cells but silent in developing B-cells [18]. However, in B-cell progenitor acute lymphoid leukemia (BCP-ALL) PBX1 is part of the fusion gene TCF3::PBX1, and in Hodgkin lymphoma PBX1 is aberrantly activated, demonstrating that deregulated PBX1 acts as an oncogene in B-cell malignancies [18, 19].

In the past, we have reported about selected groups of TFs and described their gene activities in the hematopoietic compartment. The obtained TF patterns were called codes. Accordingly, we have established the so-called “TALE-code” which describes physiological activities of TALE-class homeobox genes in all progenitors and mature cell types of myeloid and lymphoid hematopoiesis [20, 21]. This code informs our understanding of developmental gene regulation in the hematopoietic compartment, and helps identification of deregulated TALE-class homeobox genes in corresponding leukemias and lymphomas [10, 18]. This schema helped disclose that TALE-class homeobox gene IRX1 is physiologically expressed in specific lymphoid and myeloid progenitors, namely pro-B-cells and megakaryocyte-erythroid-progenitors (MEPs), respectively [20, 21]. On the other hand, the closely related homeobox genes IRX2, IRX3 and IRX5 are aberrantly expressed in subsets of BCP-ALL and acute myeloid leukemia (AML) patients and cell lines [20–23]. These findings indicate that in the normal course of lymphoid and myeloid development IRX1 performs specific gene regulations in these progenitors and that continued differentiation and maturation are contingent upon its silencing, while aberrant activities of related IRX-genes disturb these processes [10, 20–22].

Here, we analyzed TALE-class homeobox gene MKX (also termed IRXL1 or mohawk) which plays a physiological role in tendon development and is closely related to the hematopoietically expressed IRX1 gene [13, 24]. MKX was found to be silent in normal hematopoiesis and aberrantly expressed in specific hematopoietic malignancies. The identified MKX-expressing AML cell line OCI-AML3 was used as a model to investigate the role of this TALE-class homeobox gene in leukemogenesis.

Materials and methods

Bioinformatic analyses of gene expression profiling and RNA-seq data

Gene expression data for normal cell types and patients were obtained from Gene Expression Omnibus (GEO, www.ncbi.nlm.nih.gov). We analyzed myelopoietic cells using datasets GSE42519 and GSE19599, AML patients using datasets GSE15434, GSE21261, GSE14468 and GSE35784, and MM patients using GSE19554 and GSE57317 [25–31]. The associated online tool GEO2R provided comparison of patient subsets, revealing the top 250 significant differentially expressed genes [32]. Furthermore, we analyzed RNA-seq data of normal cell types from The Human Protein Atlas (www.proteinatlas.org), and of 100 leukemia/lymphoma cell lines termed LL-100 using dataset E-MTAB-7721 obtained via ArrayExpress (www.ebi.ac.uk/arrayexpress) [33, 34]. Analysis of LL-100 RNA-seq data was performed via the online tool DSMZCellDive [35]. RNA-sequencing data from siRNA-treated OCI-AML3 cells (performed in triplicate) were generated at Eurofins MWG (Ebersberg, Germany). Sample libraries for control and treated cells were prepared with the strand-specific cDNA library and sequenced 2x150bp on the Illumina NovaSeq 6000 platform by Eurofins Genomics (INVIEW Transcriptome, Ebersberg, Germany), aiming a minimum of 30M reads per sample with an insert size of >150bp. Trimming and quality control of the sequencing reads were performed via fastp and quantification of reads via salmon on the reference human gene code GRCh38, version 37 [36, 37]. Finally, data were analyzed using DESeq2 and R/Bioconductor to retrieve normalized count data [38]. Sequencing data are available from BioStudies (www.ebi.ac.uk/biostudies) via BSST1315.

Cell lines and treatments

Cell lines are held at the DSMZ (Braunschweig, Germany) and were cultivated as described (www.DSMZ.de). All cell lines had been authenticated and tested negative for mycoplasma infection. Inhibition of gene expression levels within cell lines was performed using gene specific siRNA oligonucleotides with reference to AllStars negative Control siRNA (siCTR), all obtained from Qiagen (Hilden, Germany). Overexpression studies were performed using commercial cDNA-constructs cloned into expression vector pCMV6 (Thermo Fisher Scientific, Darmstadt, Germany). SiRNAs (100 pmol) and plasmid-DNA (2 μg) were transfected into 1x10⁶ cells by electroporation using the EPI-2500 impulse generator (Fischer, Heidelberg, Germany) at 350 V for 10 ms. Electroporated cells were harvested after 20 h cultivation. RNA-sequencing analysis was performed in triplicate of OCI-AML3 cells treated by siRNA-mediated knockdown at Eurofins MWG (Konstanz, Germany). We used RNAeasy for RNA extraction (Qiagen), and Bioanalyzer for quality control (Agilent Technologies, Santa Clara, USA). Additional cell treatments were performed using 20 ng/ml recombinant BMP2 or TNFSF11 (R & D Systems, Wiesbaden, Germany), 100 μM azacytidine (AZA), 10 μM dorsomorphin, 14 μM NFkB-inhibitor, and 100 μM etoposide (Sigma-Aldrich, Taufkirchen, Germany) for 20 h. For functional testing, treated cells were analyzed using the IncuCyte S3 Live-Cell Imaging Analysis System and the Cell-by-Cell analysis software (Sartorius, Göttingen, Germany). For detection of apoptotic cells, we additionally used the IncuCyte Caspase-3/7 Green Apoptosis Assay diluted at 1:2000 (Sartorius). Live-cell imaging experiments were performed twice with fourfold parallel tests.

Polymerase chain-reaction (PCR) analyses

Total RNA was extracted from cultivated cell lines using TRIzol reagent (Thermo Fisher Scientific). Primary human total RNA derived from B-cells, T-cells, monocytes, peripheral blood mononuclear cells (PBMC), brain and prostate were purchased from Biochain/BioCat (Heidelberg, Germany). cDNA was synthesized using 1 μg RNA, random priming and Superscript II (Thermo Fisher Scientific). Real time quantitative (RQ)-PCR analysis was performed using the 7500 Real-time System and commercial buffer and primer sets (Thermo Fisher Scientific). For normalization of expression levels, we quantified the transcripts of TATA box binding protein (TBP). Quantitative analyses were performed as biological replicates and measured in triplicate. Standard deviations are presented in the figures as error bars. Statistical significance was assessed by Student´s T-Test (two-tailed) and the resultant p-values indicated by asterisks (* p<0.05, ** p<0.01, *** p<0.001, n.s. not significant).

Protein analysis

Western blots were generated by the semi-dry method. Protein lysates from cell lines were prepared using SIGMAFast protease inhibitor cocktail (Sigma-Aldrich). Proteins were transferred onto nitrocellulose membranes (Bio-Rad, München, Germany) and blocked with 5% dry milk powder dissolved in phosphate-buffered-saline buffer (PBS). The following antibodies were used: alpha-Tubulin (Sigma-Aldrich, #T6199), and MKX (Santa Cruz Biotechnology, Heidelberg, Germany # sc-515878). For loading control, blots were reversibly stained with Poinceau (Sigma-Aldrich) and detection of alpha-Tubulin (TUBA) performed thereafter. Secondary antibodies were linked to peroxidase for detection by Western-Lightning-ECL (Perkin Elmer, Waltham, MA, USA). Documentation was performed using the digital system ChemoStar Imager (INTAS, Göttingen, Germany). CCL2 protein was quantified in cell line supernatants by Enzyme-Linked Immunosorbant Assay (ELISA) using Human CCL2 Quantikine ELISA Kit (R & D Systems, #DCP00). Supernatants were harvested from 1x10⁶ cells growing in 1 ml medium for 24h and stored at -20°C.

Results

MKX expression in normal tissues and hematopoietic malignancies

TALE-class homeobox gene IRX1 is physiologically expressed at the MEP stage of early myelopoiesis and at the pro-B-cell stage of early B-cell differentiation, while the closely related homeobox gene MKX was found to be absent in these cells following analysis of public gene expression profiling datasets GSE42519 and GSE19599 (S1 Fig) [20, 21]. Furthermore, RNA-seq data from the Human Protein Atlas discounted MKX expression in mature myeloid and lymphoid hematopoietic cells, but prominent activity in brain, prostate and ovary (S1 Fig). Thus, in contrast to IRX1, MKX is silent in normal hematopoietic cells including MEPs and pro-B-cells.

Recently, we sequenced RNA from 100 leukemia/lymphoma cell lines, covering diverse hematopoietic malignancies, derived from myeloid and lymphoid lineages [34]. Screening of these RNA-seq data demonstrated prominent MKX expression in AML cell line OCI-AML3, low levels in multiple myeloma (MM) cell lines KMM-1, L-363, LP-1 and OPM-2, while cell lines derived from other entities including BCP-ALL tested negative (Fig 1A). Analysis of patients with corresponding malignancies using public gene expression profiling datasets showed MKX activity in about 1% of AML, and 10% of MM patients (S2 Fig). Thus, we identified aberrantly expressed MKX in subsets of AML and MM cell lines and patients, albeit at different frequencies.

Fig 1. MKX expression analysis.

(A) MKX expression in 100 leukemia/lymphoma cell lines using RNA-seq data. The following entities are indicated: megakaryocytic AML, myelomonocytic AML, myeloma, and pre-B-cell ALL. MKX-positive cell lines are OCI-AML3 (AML), and KMM-1, L-363, LP-1 and OPM-2 (myeloma). (B) RQ-PCR analysis of MKX in AML and MM cell lines, and selected primary cells. The expression level of OCI-AML3 was set as 1 while the remaining samples are related to that value. (C) Western blot analysis of MKX in AML cell lines. TUBA served as loading control.

RQ-PCR analysis confirmed enhanced MKX RNA expression in OCI-AML3, lower transcript levels in MM cell lines, and absent gene activity in AML control cell lines and primary immune and blood cells, including B-cells, T-cells, monocytes and peripheral blood mononuclear cells (Fig 1B). OCI-AML3 expressed even higher MKX transcript levels as compared to primary brain and prostate samples, supporting aberrant activation in this AML cell line (Fig 1B). Finally, Western blot analysis showed MKX expression in OCI-AML3 at the protein level, endorsing this cell line as a suitable model for MKX in AML (Fig 1C). Of note, screening of 32 AML cell lines using public gene expression profiling dataset GSE59808 confirmed MKX activity in OCI-AML3 while the remaining cell lines tested negative (S3 Fig).

Taken together, MKX is aberrantly expressed in subsets of AML and MM patients and cell lines thereby highlighting its likely oncogenic role in these malignancies. Due to prominent MKX expression in OCI-AML3, we used this cell line as a model to investigate deregulation and function of this TALE-class homeobox gene in AML.

Regulation of MKX expression in OCI-AML3

Chromosomal rearrangements and/or genomic aberrations frequently underlie gene deregulation in hematopoietic malignancies including AML [39, 40]. However, the reported karyotype from OCI-AML3 discounted conspicuous chromosomal abnormalities at the MKX gene located 10p12.1 (www.DSMZ.de). Furthermore, we performed genomic profiling of OCI-AML3, similarly discounting copy number alterations at chromosome 10, including the cryptic del(10)(p12.1-p12.3) which can generate fusion gene MLLT10::MKX, as described in pediatric BCP-ALL (Fig 2A) [41]. Of note, chromosomal aberration t(10;11)(p12;q23) is a frequent rearrangement in AML which also targets MLLT10 at position 10p12.3, but generates the fusion gene KMT2A::MLLT10 [42]. Thus, neither chromosomal nor genomic aberrations at position 10p12 were detected in OCI-AML3, such as might contribute to MKX activation.

Fig 2. Aberrant regulation of MKX in OCI-AML3.

(A) Genomic copy number analysis of OCI-AML3 showing the complete chromosome 10 (above), and the chromosomal region 10p12.1-p12.3 highlighting MLLT10 and MKX (below). (B) Data obtained from the UCSC genome browser indicating a CpG-island at MKX (left). RQ-PCR analysis of MKX and NANOG in AML cell line HL-60 treated with DNA-methyltransferase inhibitor AZA (right). The expression level of HL-60 control was set as 1 while the remaining samples are related to that value. (C) RQ-PCR analysis of IRX3 and MKX in OCI-AML3 treated with BMP2 (left) and dorsomorphin (middle). RQ-PCR analysis of SMAD4, IRX3 and MKX in OCI-AML3 treated for siRNA-mediated knockdown (right). (D) RQ-PCR analysis of IRX3 and MKX (left), and IRX5 and MKX (right) in OCI-AML3 treated for siRNA-mediated knockdown. (E) RQ-PCR analysis of JUNB and MKX in OCI-AML3 treated for siRNA-mediated knockdown. The expression level of OCI-AML3 siCTR was set as 1 while the remaining samples are respectively related to that value. Statistical significance was assessed by Student´s T-Test (two-tailed) and the resultant p-values indicated by asterisks (* p<0.05, ** p<0.01, *** p<0.001, n.s. not significant).

Inspection of the MKX locus at the UCSC genome browser (www.genome.cseUCSC.edu) indicated the presence of a CpG island in the promoter region which may underlie its (de)regulated expression (Fig 2B). This hypothesis was supported by recent publications, reporting that DNA methylation at MKX and other homeobox genes was detected in solid tumors via a pan-cancer screen, and that OCI-AML3 carries a pediatric AML-associated mutation in DNMT3A [43, 44]. However, treatment of AML control cell line HL-60 with DNA methyltransferase inhibitor AZA promoted activation of the accordingly regulated homeobox gene NANOG but left MKX unchanged (Fig 2B) [45], discounting aberrant DNA demethylation in MKX deregulation in AML.

Recently, we reported aberrant expression of TALE-class homeobox genes IRX3 and IRX5 in OCI-AML3, showing that elevated BMP2-SMAD4-signalling activates IRX3 [20]. However, neither treatment with BMP2, inhibition of BMP-signalling by dorsomorphin, nor siRNA-mediated knockdown of SMAD4 altered the expression of MKX, excluding its activation via this pathway in OCI-AML3 (Fig 2C). In contrast, siRNA-mediated knockdown of IRX5 (but not of IRX3) resulted in slightly reduced MKX expression levels (Fig 2D). Furthermore, MKX expression was also reduced by knockdown of JUNB (Fig 2E), which has been shown to play activating roles for IRX3 and IRX5 [20]. Together, these findings showed that IRX5 and JUNB support MKX activation in OCI-AML3.

To examine the potential role of other TFs in MKX deregulation we screened TF binding sites within its gene sequence using public UCSC genome browser data (www.genome.cseUCSC.edu) as indicated in Fig 3A. We focused on HOXA, IRF, MEF2, STAT5, GATA and NFkB–a panel representing basic regulators of normal and malignant myelopoiesis. We performed siRNA-mediated knockdown of HOXA10 which, though reportedly regulating IRX3 and IRX5 [20], showed no regulatory impact on MKX (Fig 3B). MEF2C and MEF2D regulate tendon and muscle development while IRF8 and MEF2D co-operate oncogenically in AML [46–48]. Furthermore, RNA-seq data from AML cell lines demonstrated highly expressed IRF8 exclusively in OCI-AML3, indicating aberrant overexpression in this cell line (S4 Fig). However, knockdown of neither IRF8 nor MEF2D altered the expression of MKX (Fig 3B), excluding these factors as MKX regulators in OCI-AML3. In contrast, siRNA-mediated knockdown of STAT5A and STAT5B enhanced MKX expression (Fig 3C), demonstrating that these factors are transcriptional repressors of MKX. Interestingly, RNA-seq data from AML cell lines showed reduced expression levels of both STAT5A and STAT5B in OCI-AML3 (S4 Fig), thereby supporting MKX activity in this cell line. Next, knockdown of myeloid differentiation factor GATA2 resulted in elevated MKX expression, also demonstrating a repressive role (Fig 3C). Similar to STAT5, RNA-seq data showed reduced GATA2 expression in OCI-AML3 (S4 Fig). Finally, we investigated the role of NFkB in MKX regulation. ChIP-seq data from the ENCODE-project demonstrated direct binding of NFkB to the indicated binding site at MKX (Fig 3D). RNA-seq data from 23 AML cell lines for selected TNF-signalling components revealed for OCI-AML3 elevated expression of TNFSF11 and its receptor TNFRSF11A which reportedly activate NFkB TFs (Fig 3E) [49]. RQ-PCR analysis confirmed their high expression levels in OCI-AML3 (Fig 3F). While stimulation of OCI-AML3 with additional TNFSF11 spared MKX expression in OCI-AML3, treatment with a pharmacological NFkB-inhibitor suppressed MKX transcription, showing a significant activating impact of NFkB in MKX expression (Fig 3F).

Fig 3. (De)regulation of MKX in OCI-AML3.

(A) Data obtained from the UCSC genome browser indicating potential TF-binding sites (blue arrow-heads) at the MKX locus. (B) RQ-PCR analysis of HOXA10 and MKX (left), IRF8 and MKX (middle), and MEF2D and MKX in OCI-AML3 treated for siRNA-mediated knockdown. (C) RQ-PCR analysis of STAT5A and MKX (left), STAT5B and MKX (middle), and GATA2 and MKX in OCI-AML3 treated for siRNA-mediated knockdown. The expression level of OCI-AML3 siCTR was set as 1 while the remaining samples are respectively related to that value. (D) ChIP-seq data from the ENCODE project showing direct binding of NFkB at the MKX locus. (E) Heatmap, based on RNA-seq data, showing gene expression levels from selected NFkB-signalling components, demonstrating elevated activity of TNFRSF11A and TNFSF11 in OCI-AML3. (F) RQ-PCR analysis of TNFRSF11A (left) and TNFSF11 (middle) in selected AML cell lines. RQ-PCR analysis of MKX in OCI-AML3 treated with TNFSF11 and NFkB-inhibitor (right). The expression level of OCI-AML3 control was set as 1 while the remaining samples are related to that value. Statistical significance was assessed by Student´s T-Test (two-tailed) and the resultant p-values indicated by asterisks (* p<0.05, ** p<0.01, *** p<0.001, n.s. not significant).

Taken together, we identified IRX5, JUNB and NFkB as activators, and downregulated STAT5A/B and GATA2 as repressors for MKX in AML cell line OCI-AML3 which collectively control aberrant expression of this homeobox gene.

MKX target genes and function in AML

To identify target genes regulated by MKX in AML we subjected OCI-AML3 to siRNA-mediated knockdown of MKX and subsequent RNA sequencing analysis. After confirming knockdown at transcript and protein levels (Fig 4A), the most significantly expressed genes differing between knockdowns and controls were listed, showing 78 activated and 28 genes repressed by MKX (S1 Table). For more detailed analysis we selected activated CCL2 (also known as MCP-1) which encodes a ligand that binds and activates its cognate receptor CCR2, plays multifaceted roles in myeloid immune cells, and acts as an oncogene in some cancer types [50, 51]. Knockdown of MKX in OCI-AML3 was confirmed by Western blot analysis, and the concomitant reduction of CCL2 protein in the supernatant by ELISA analysis (Fig 4A). RQ-PCR analysis showed enhanced CCL2 expression levels in AML cell lines OCI-AML3 and MUTZ-3, and corresponding expression was detected at the protein level by ELISA (Fig 4B). Thus, CCL2 is activated by MKX in OCI-AML3 and, accordingly, strongly expressed. However, forced expression of MKX in MKX-negative AML cell line NB-4 reduced CCL2 transcription (Fig 4C), indicating the absence of additional factors implicated in the regulation of CCL2 or cooperation with MKX.

Fig 4. MKX activates CCL2 in OCI-AML3.

(A) Western blot analysis of MKX in OCI-AML3 treated for siRNA-mediated knockdown of MKX. TUBA served as loading control (left). ELISA analysis of CCL2 protein in supernatants from OCI-AML3 treated for siRNA-mediated knockdown of MKX (right). (B) RQ-PCR analysis of CCL2 in selected AML cell lines (left). ELISA analysis of CCL2 protein in supernatants from selected AML cell lines (right). (C) RQ-PCR analysis of NB-4 treated for MKX-overexpression resulted in reduced CCL2 transcript levels. (D) RQ-PCR analysis of IRX3 and CCL2 (left), and IRX5 and CCL2 (right) in OCI-AML3 treated for siRNA-mediated knockdown. (E) Gene expression profiling data for CCL2 from hematopoietic stem and developing myeloid cells using public dataset GSE42519. (F) RQ-PCR analysis of CCL2 in OCI-AML3 treated for siRNA-mediated knockdown of CCL2 (left). Live-cell imaging analyses of OCI-AML3 treated for siRNA-mediated knockdown of CCL2, detecting apoptosis (right) and (G) proliferation (left). The data for the last point of imaging showed no significant difference for apoptosis but for proliferation (p = 0.0059). RQ-PCR analysis of CCL2 and CCND1 in OCI-AML3 treated for siRNA-mediated knockdown of CCL2 (right). The expression levels of NB-4 vector and OCI-AML3 siCTR were set as 1 while the remaining samples are respectively related to those values. Statistical significance was assessed by Student´s T-Test (two-tailed) and the resultant p-values indicated by asterisks (* p<0.05, ** p<0.01, *** p<0.001, n.s. not significant).

Analysis of the CCL2 promoter region revealed three binding sites for MKX (at -1418 bp, -663 bp, -510 bp) indicating direct regulation (S5 Fig). This region also contained a binding site for IRX factors (at -574 bp) which slightly differ from that for MKX, suggesting transcriptional regulation of CCL2 by IRX3 and/or IRX5 as well (S5 Fig). Consistently, siRNA-mediated knockdown of IRX3 and IRX5 resulted in reduced CCL2 expression, confirming stimulation by both factors (Fig 4D). Thus, CCL2 is activated by aberrantly expressed TALE homeodomain TFs MKX, IRX3 and IRX5 in AML cell line OCI-AML3. Of note, CCL2 is not significantly expressed during early myelopoiesis, indicating stage-independent effects when ectopically activated (Fig 4E). Functional analysis of CCL2 was performed in OCI-AML3 via its knockdown which was confirmed by RQ-PCR (Fig 4F). Live-cell imaging analysis of the treated cells discounted any role in apoptosis (Fig 4F) but rather that of driving proliferation (Fig 4G). Consistently, siRNA-mediated knockdown of CCL2 resulted in downregulation of CCND1, demonstrating that CCL2 activates this proliferation driver in OCI-AML3 (Fig 4G).

Additional MKX target genes identified by our knockdown/RNA-sequencing approach revealed activated SESN3, and repressed GATA2 (S1 Table). SESN3 (SESTRIN 3) encodes a stress-regulator which reduces intracellular reactive oxygen species (ROS), and GATA2 a myelopoietic TF [52, 53]. Regulation of these genes by MKX was confirmed by RQ-PCR analysis after siRNA-mediated knockdown (Fig 5A). In MEPs, expression profiling data from developmental stages of myelopoiesis demonstrated reduced levels of SESN3, and elevated levels of GATA2 (Fig 5B). Therefore, deregulation of these genes at that stage may disturb survival and differentiation processes during early myelopoiesis.

Fig 5. MKX impacts apoptosis and differentiation.

(A) RQ-PCR analysis of MKX and SESN3 (above) and of MKX and GATA2 (below) in OCI-AML3 treated for siRNA-mediated knockdown of MKX. (B) Gene expression profiling data for SESN3 (above) and GATA2 (below) from hematopoietic stem and developing myeloid cells using public dataset GSE42519. (C) RQ-PCR analysis of MKX in OCI-AML3 treated for siRNA-mediated knockdown of MKX (left). Analysis of apoptosis by live-cell imaging of OCI-AML3 treated for siRNA-mediated knockdown of MKX (middle, p = 0.0052) and additionally with etoposide (right, p = 0.0148). The calculated p-values refer to the last point of imaging. (D) RQ-PCR analysis of MKX and BCL2L11 in OCI-AML3 treated for siRNA-mediated knockdown of MKX. (E) Gene expression profiling data for BCL2L11 from hematopoietic stem and developing myeloid cells using public dataset GSE42519. (F) RQ-PCR analysis of HL-60 treated for MKX-overexpression resulted in reduced transcript levels of GATA2 and BCL2L11, and elevated SESN3 levels. The expression levels of OCI-AML3 siCTR and HL-60 vector were set as 1 while the remaining samples are respectively related to those values. Statistical significance was assessed by Student´s T-Test (two-tailed) and the resultant p-values indicated by asterisks (* p<0.05, ** p<0.01, *** p<0.001, n.s. not significant).

Identification of MKX-deregulated survival factor SESN3 prompted analysis of the impact of MKX on apoptosis via knockdown and live-cell imaging. Consistent with our other findings, the data revealed a prominent role for MKX in cell survival (Fig 5C). Moreover, simultaneous treatment with apoptosis-inductor etoposide enhanced the observed MKX knockdown effect (Fig 5C). BCL2L11 encodes a fundamental activator of apoptosis, acts in the context of etoposide, and is regulatory connected with SESN3 [52, 54]. Accordingly, knockdown of MKX in OCI-AML3 resulted in BCL2L11 activation (Fig 5D), indicating that combined activation of SESN3 and suppression of BCL2L11 is responsible for the positive role of MKX in cell survival. Moreover, analysis of the promoter region of BCL2L11 revealed MKX binding sites at -1424 bp and -1291 bp (S5 Fig), suggesting direct transcriptional regulation by MKX. Interestingly, BCL2L11 is elevated during myelopoiesis at the MEP stage and may, therefore, physiologically predispose these progenitors to apoptosis (Fig 5E). Finally, forced expression of MKX in AML cell line HL-60 suppressed GATA2 and BCL2L11, and activated SESN3 (Fig 5F), confirming the regulatory impact observed in OCI-AML3.

Next, we analyzed expression profiling data from MM patients comparing MKX-high with MKX-low samples using public data set GSE19554 and the associated online tool GEOR, which identifies 250 most significantly differing genes. This exercise revealed downregulated GATA2 and SESN3 in MKX-high MM patients (S6 Fig), showing target genes similar to those identified in AML, although inhibition of SESN3 contrasts with its activation by MKX in OCI-AML3. Furthermore, we found downregulated CEBPA, CEBPD and IRF8 in MKX-high MM patients (S6 Fig). These three genes are physiologically downregulated in MEPs (Fig 6A), indicating a specific role in myeloid differentiation at that developmental stage. RQ-PCR analysis of these genes in OCI-AML3 after MKX knockdown showed that MKX activated CEBPD while IRF8 and CEBPA were not regulated by this factor (Fig 6B). Forced expression of MKX in NB-4 and HL-60 resulted in elevated CEBPD (Fig 6C), confirming MKX-mediated activation of CEBPD. Thus, analysis of MKX in MM identified CEBPD as an additional aberrantly activated target gene in AML.

Fig 6. MKX activates CEBPD.

(A) Gene expression profiling data for IRF8 (above), CEBPA (middle), and CEBPD (below) from hematopoietic stem and developing myeloid cells using public dataset GSE42519. (B) RQ-PCR analysis of MKX and IRF8 (left), MKX and CEBPA (middle), and MKX and CEBPD (right) in OCI-AML3 treated for siRNA-mediated knockdown of MKX. (C) RQ-PCR analysis of NB-4 (left) and HL-60 (right) treated for MKX-overexpression resulted in elevated CEBPD transcript levels. The expression levels of OCI-AML3 siCTR, NB-4 vector and HL-60 vector were set as 1 while the remaining samples are respectively related to those values. Statistical significance was assessed by Student´s T-Test (two-tailed) and the resultant p-values indicated by asterisks (* p<0.05, ** p<0.01, *** p<0.001, n.s. not significant).

Taken together, we identified BCL2L11, CCL2, CEBPD, GATA2 and SESN3 as MKX target genes in AML cell line OCI-AML3. These genes impact apoptosis, proliferation and differentiation, thus demonstrating several oncogenic functions for deregulated MKX in AML. Except CCL2, these MKX target genes show conspicuous activity in myelopoiesis at the IRX1-positive MEP stage, highlighting this developmental progenitor as a candidate cell of origin in AML cases aberrantly expressing IRX1-related MKX.

Discussion

TALE-class homeobox genes encode developmental TFs, regulating basic processes in cell and tissue differentiation [13]. Deregulation of these genes promotes carcinogenesis including leukemia [18, 19, 22]. Accordingly, IRX1 is physiologically expressed in MEPs and pro-B-cells, while aberrant expression of IRX2, IRX3 and IRX5 deregulates differentiation processes in AML and BCP-ALL [20, 21, 23]. Hence, these progenitors represent persuasive candidate cells of origin for malignancies aberrantly expressing IRX-related homeobox genes [22]. Here, we show that MKX is absent in hematopoiesis and aberrantly activated in AML and MM, mimicking the picture for IRX2, IRX3 and IRX5 in AML and BCP-ALL, and staging a functional reenactment of the role played by the closely related homeobox gene IRX1 in early myeloid and B-cell development. Although not in frame, the reported fusion gene MLLT10::MKX in pediatric BCP-ALL may indicate an oncogenic role for MKX in this type of B-cell malignancy as well [41], even though MM is B-cell-derived and rather originates from GC-B-cells [55], contrasting with pro/pre-B-cell originating BCP-ALL. However, this study focused on AML–the role of MKX in MM remains unclear and requires additional investigation.

Aberrant expression of MKX has also been described in colorectal cancer, suggesting a wider role in carcinogenesis [56]. Physiologically, MKX is active in tendon differentiation, inhibiting muscle development via suppression of the master gene MYOD [24, 57–59]. MKX-mediated repression of lineage specific regulators occurs via recruitment of co-repressors including SIN3A/HDAC [60]. In tendon development, BMP2-signalling activates MKX expression [61], which, however, had no effect in OCI-AML3 cells. In contrast, TNFa-signalling regulates MKX expression in tendon as well [62], supporting our discovery of an activating role for TNFa-signalling factor NFkB. OCI-AML3 expresses high levels of TNFSF11 and TNFRSF11A which may control NFkB activation in this cell line.

In addition to NFkB, in AML cell line OCI-AML3 we revealed that MKX is activated by IRX5 and JUNB, but inhibited by GATA2 and STAT5 (Fig 7). IRX5 and JUNB have been recently described to play an oncogenic role in AML and OCI-AML3 [20]. Our data presented here indicate aberrant downregulation of GATA2 in OCI-AML3 which in turn supports MKX expression. GATA2 encodes a basic developmental TF, driving myeloid differentiation. Accordingly, GATA2 deficiency causes several abnormalities in the myeloid lineage [63]. In contrast, STAT5 inhibits myelopoiesis by repression of CEBPA [64]. Consequently, OCI-AML3 expresses elevated CEBPA (S4 Fig) and MKX. However, the mechanism of STAT5 downregulation in OCI-AML3 remains to be elucidated.

Fig 7. Aberrant gene regulatory network of MKX in AML.

This diagram summarizes the results of this study. MKX is located in the center, regulated by JUNB, STAT5, NFkB, IRX5 and GATA2, and regulates the genes GATA2, CEBPD, SESN3, BCL11B and CCL2. These deregulated target genes are involved in proliferation, apoptosis and differentiation.

Our downstream analyses revealed that MKX activates CCL2 in OCI-AML3. However, forced expression of MKX in NB-4 mediated downregulation of CCL2, indicating that its transcriptional activation requires additional factors which are absent in NB-4. CCL2 is a secreted ligand and plays together with MKX a role in tenogenic differentiation, indicating aberrant reactivation of this regulatory pathway [24, 65]. In carcinogenesis including AML, CCL2 acts oncogenic by organizing the tumor microenvironment [51, 66, 67]. Furthermore, CCL2 enhances proliferation of AML cells via activation of CCND1 [68], which was also demonstrated in OCI-AML3. Interestingly, the CCL2-CCR2-pathway activates NFkB in dendritic cell maturation, and both NFkB and AP1 in renal cells [69, 70]. Since we showed that MKX is activated by NFkB and AP1-factor JUNB, these data may suggest a self-reinforcing cycle of MKX and CCL2 in AML. However, this hypothesis is speculative and requires experimental examination. Of note, a non-coding mutation or SNP in the upstream region of CCL2 creates a novel binding site for the TFs PREP1 and PBX2, representing two members of the TALE-class of homeodomain TFs [71, 72]. This SNP is associated with increased CCL2 expression, and particular inflammatory and liver diseases [73, 74]. Thus, pathological overexpression of CCL2 is mediated by certain TALE homeodomain TFs albeit via unrelated mechanisms: (i) a specific sequence alteration enabling binding of constitutive PREP1 and PBX2, or (ii) ectopically expressed MKX using a pre-existing binding site.

Furthermore, we showed that MKX activated SESN3 and inhibited BCL2L11 in AML cells. Both target genes are part of the FOXO3-network, regulating ROS-mediated apoptosis as described in neuronal cells [52], explaining our observed survival-effect for MKX. Moreover, BCL2L11 is activated by etoposide treatment, thus driving apoptosis [75]. Both, BCL2L11 and SESN3 show contrasting expression levels in MEPs, increasing the potential of apoptosis-induction at this developmental stage. Therefore, aberrant MKX activity may curb this seeming apoptotic predisposition, enhancing survival of the tumor cells.

In this study we revealed several genes showing conspicuous activity during myelopoiesis in MEPs. In addition to the reported homeobox gene IRX1, BCL2L11 and GATA2 are upregulated while SESN3 and CEBPD are downregulated at this developmental stage. In contrast, MKX inhibits BCL2L11 and GATA2, and activates SESN3 and CEBPD. Furthermore, we showed that GATA2 inhibits expression of MKX, demonstrating that these factors are mutual repressors. This type of regulatory connection stabilizes the activity of one partner, namely MKX in OCI-AML3. On the other hand, repression of GATA2 may have fundamental consequences for myeloid differentiation as reported for the GATA2 deficiency syndrome [63]. Finally, a recent study by Sato and colleagues detected IRX1 mutations in Down syndrome associated myeloid leukemia, and showed that IRX1 boosts megakaryocyte/erythroid differentiation, highlighting its role in early myelopoiesis [76]. Together, these findings may indicate that aberrantly expressed MKX operates oncogenically in MEPs which may, therefore, represent the cell of origin of MKX-positive AML tumors.

In conclusion, we identified ectopic expression of TALE-class homeobox gene MKX in AML and MM patients and cell lines (Fig 7). MKX is closely related to IRX1 which is physiologically expressed in early myeloid and B-cell development. Our data support the idea that MKX as well as IRX2, IRX3 and IRX5 act oncogenically at least in part by interfering the developmental activity of IRX1 in both hematopoietic lineages. However, the role of deregulated MKX in the mature B-cell malignancy MM remains unclear and deserves further investigation. Nevertheless, our study throws light on normal and abnormal hematopoiesis and pinpoints novel potential markers and/or therapeutic targets for respectively subtyping and treatment of AML and MM patient subsets.

Supporting information

S1 Fig. Public gene expression data for MKX and IRX1.

(A) Gene expression profiling data for IRX1 (above) and MKX (below) from hematopoietic stem and developing myeloid cells using public dataset GSE42519. Expression level 3 is indicated by an arrowhead, demonstrating absent expression of MKX in these cells. (B) Gene expression profiling data for IRX1 and MKX from selected hematopoietic cell types using public dataset GSE19599. The cell types MEP and pro-B-cells are indicated by black and blue boxes, respectively. (C) Expression levels of MKX in normal human tissue types (above) and mature hematopoietic cell types (below) using RNA-seq data obtained from the Human Protein Atlas. Selected tissues with high and low MKX expression are indicated with red and green arrowheads, respectively.

(TIF)

S2 Fig. MKX expression in AML and MM patients.

Gene expression profiling data for MKX and IRX1 in AML patients (above) and for MKX in MM patients (below) using public datasets.

(TIF)

S3 Fig. MKX expression in AML cell lines.

Gene expression profiling dataset GSE59808 contains 32 AML cell lines and shows significant MKX activity just in OCI-AML3.

(TIF)

S4 Fig. Gene expression levels in selected AML cell lines.

RNA-seq data for selected genes were used to indicate their expression levels in cell lines via DSMZCellDive. OCI-AML3 is indicated by a blue arrowhead.

(TIF)

S5 Fig. Potential binding sites for MKX and IRX factors.

Potential binding sites for MKX (red) and IRX factors (blue) are indicated in the promoter regions of CCL2 (above), BCL2L11 (middle), and GATA2 (below). The sequences of the first exons are highlighted in bold.

(TIF)

S6 Fig. Comparative gene expression profiling analysis of MM patient data.

Public gene expression profiling dataset GSE19554 covers MM patients. We used the online tool GEO2R to compare MM patients expressing high MKX levels with patients expressing low MKX levels. Selected genes showing statistically significant differences are indicated: MKX, GATA2, SESN3, CEBPA, CEBPD and IRF8.

(TIF)

S1 Table. RNA-seq data from OCI-AML3 treated for siRNA-mediated knockdown of MKX.

The table shows the combined data obtained from three biological replicates, indicating the log2-fold change and the statistical significance. Highlighted are genes selected for detailed analysis in this study, including MKX, CCL2, SESN3, and GATA2.

(PDF)

Data Availability

All relevant data are within the paper and its Supporting information files.

Funding Statement

José Carreras Foundation for SN (DJCLS 02 SP/2023).