The Impact of the Chromatin Binding DEK Protein in Hematopoiesis and Acute Myeloid Leukemia

Abstract

Hematopoiesis is an exquisitely regulated process of cellular differentiation to create diverse cell types of the blood. Genetic mutations, or aberrant regulation of gene transcription, can interrupt normal hematopoiesis. This can have dire pathological consequences, including acute myeloid leukemia (AML), in which generation of the myeloid lineage of differentiated cells is interrupted. In this literature review, we discuss how the chromatin remodeling DEK protein can control hematopoietic stem cell (HSC) quiescence, hematopoietic progenitor cell (HPC) proliferation, and myelopoiesis. We further discuss the oncogenic consequences of the t(6;9) chromosomal translocation, which creates the DEK-NUP214 (aka: DEK-CAN) fusion gene, during the pathogenesis of AML. Combined, the literature indicates that DEK is crucial for maintaining homeostasis of hematopoietic stem and progenitor cells, including myeloid progenitors.

Introduction: Leukemia

Leukemia is cancer of the blood or bone marrow which commonly results from rapid over-production of abnormal hematopoietic stem or progenitor cells. Leukemia comprises 4% of all cancer-related deaths and is the 6^th and 8^th leading cause of cancer-related deaths in males and females, respectively.^[1] Although the average age of diagnosis is 68 years old, leukemia is the most common form of cancer in children, constituting 28% of childhood cancers in ages 1-14 years old with an average 5-year survival of 87%.^[1] Leukemia is classified into four subtypes based on the lineage and rate of growth: acute lymphocytic leukemia (ALL), acute myeloid leukemia (AML), chronic myeloid leukemia (CML), and chronic lymphocytic leukemia (CLL). The most common types of leukemia cases are CLL (38%) and AML (31%) in adults and ALL (75%) and AML (17%) in children.^[2] Death rates vary between subtypes with CLL accounting for 18% and AML for 48% of leukemia-related deaths.^[1]

AML is the accumulation of abnormal myeloblasts, or unipotent stem cells, that are precursors to granulocytes in the myeloid lineage. Abnormal myeloblasts may result from genetic or epigenetic mutations that enable proliferation, but not differentiation, leading to expansion in the bone marrow, peripheral blood, and tissues.^[3] This abnormal hematopoiesis can lead to bone marrow failure, abnormally low mature granulocytes, thrombocytes, and erythrocytes, and the presence of circulating blasts.^[3] Thus, according to the WHO’s standards, AML is diagnosed by at least 20% of the myeloid lineage consisting of myeloblasts in the bone marrow or peripheral blood.^[4] The prognosis and treatment of AML is often determined by distinct cytogenetic abnormalities and/or characteristic gene mutations to identify predicted risk stratification, rate of remission, relapse, overall survival, or benefit from targeted therapies.^[3] Common genetic mutations in AML are observed in transcription factors (RUNX1, CEBPA), signal transduction proteins (FLT3), and other various patterns of mutations that may have resulted from secondary AML, de novo, or germline mutations.^[4]

The most prevalent genetic aberrations in adult AML patients include cohorts of NPM1 mutants (30%), chromatin and RNA-splicing mutations (13%), TP53 mutants and chromosomal aneuploidy (10%), bi-allelic CEBPA mutants (4%), isocitrate dehydrogenase 2 (IDH2) mutants (1%), and 14 various chromosomal translocations and inversions (34%).^[5] Gene fusions such as PML-RARA resulting from t(15;17)(q22;q21) and RUNX1-RUNX1T1 from t(8;21)(q22;q22.1) are most common consisting of 13% and 7% of adult AML cases, respectively.^[5] Cytogenetic and genetic mutational status serve as the principal factors in determining resistance to treatment, categorized as favorable-, intermediate-, and adverse-risk groups correlating with prognostic outcome.^[5] One fusion gene resulting from t(6;9)(p23;q34.1) is DEK-NUP214, which represents 1% of genetic abnormalities in AML and often co-occurs with FLT3-ITD (70%) and KRAS (20%) mutations.^[5-7] DEK-NUP214 is identified as an adverse-risk genetic abnormality, thus presenting poor patient prognosis.^{[5, 8]} This review will focus on the role of DEK-NUP214 and full-length DEK in AML and normal hematopoiesis.

Introduction: DEK

The DEK gene, located on chromosome 6p23, has been found to play an important role in several molecular pathways and systems relating to cellular functionality. DEK is a 375 amino acid protein that consists of the DNA-binding SAP domain located close to the N-terminus and a nearby, structurally similar pseudo-SAP (ΨSAP) domain. It has numerous acidic regions, a central putative nuclear localization signal, an RNA-binding region, and a unique C-terminal DNA binding domain that has been termed the DEK domain, which functions in DNA-binding and DEK multimerization (Fig 1A, top).^[9-13] DEK is most widely known as a DNA binding and chromatin remodeling protein that impacts histone composition as a histone H3.3 chaperone and has been linked to post-translational histone modifications.^[14-16] Specifically, using human mesenchymal stem cells and murine embryonic stem cells, it was determined that DEK restricts HIRA-dependent H3.3 histone loading onto chromatin telomeric regions by recruiting H3.3 to promyelocytic leukemia (PML) nuclear bodies.^[17] This activity of DEK was found to be dependent on DAXX/ATRX function and was required to maintain telomere integrity.^[17] The histone H3.3 chaperone activity of DEK has also been confirmed in Drosophila, indicating an evolutionary function for DEK in regulating chromatin architecture and accessibility.^[15] The ability of DEK to bind to, and regulate, histone H3 has resulted in an association between DEK levels and histone H3 post-translational modification. Drosophila and human models have demonstrated that the changes in DEK expression impact H3K9me3 (trimethylation of Histone H3 at lysine 9) and global changes to histone acetylation at multiple H3 lysine residues.^{[16, 18, 19]} This global change in histone H3, and to a lesser extent histone H4, acetylation can be attributed to the interaction of DEK with proteins such as histone deacetylase II (HDAC2),^[20] histone acetyltransferases p300 and PCAF,^[21] and the NcoR1/HDAC3 complex.^[19] Naturally, these chromatin remodeling functions of DEK have an impact on a cell’s transcriptional profile, particularly affecting genes related to DNA damage responses, proliferation, immune responses, and cell adhesion/migration in multiple cell types.^[22-24] Early studies suggested that DEK had both transcriptional repression and activation functions for specifically identified transcription factors, including the ability to regulate NFκB transcriptional activity,^[25] but there was never a consensus. Its broad roles in histone H3 dynamics and modifications may explain these conflicting reports.

Figure 1: Diagram of the DEK, NUP214, and DEK-NUP214 proteins.

(A) (top) A schematic of the DEK protein demonstrates that this 375 amino acid protein consists of several acidic (A) regions, neighboring SAP and pseudo-SAP domains, a putative nuclear localization signal (NLS), an RNA-binding region, and a unique C-terminal DEK domain that both binds DNA and facilitates DEK multimerization on chromatin, (middle) A schematic of the NUP214 protein, which consists of an N-terminal beta-propeller domain, a coiled-coil domain with a leucine zipper region, and numerous FG repeats at the C-terminus. There is a “C-terminal extension” (CTE) of the beta-propeller immediately adjacent to this domain, (bottom) A schematic of the DEK-NUP214 fusion protein, in which the majority of the DEK gene is fused to the C-terminal region of NUP214. The fusion results in the loss of the beta-propeller and leucine zipper regions of NUP214 and replaces them with the highly acidic regions and the DNA and RNA binding regions of DEK. (B) A diagram depicting human chromosomes 6 (blue) and 9 (orange), and the resulting derivative chromosomes resulting from the t(6;9) translocation. This figure was created with BioRender.com.

DEK is particularly critical for its involvement with mechanisms of DNA repair such as non-homologous end joining (NHEJ) and homologous recombination (HR).^{[26, 27]} Specifically, using fluorescent reporters of homologous recombination, Smith et al. identified a critical role for DEK in efficient HR-mediated DNA repair by binding to Rad51 and facilitating Rad51 filament formation on DNA double-strand breaks.^[27] With regards to NHEJ, Kavanaugh et al. demonstrated that DEK loss in HeLa cells resulted in impaired Ku70/80 dimerization and localization to DNA breaks, while NHEJ reporter assays demonstrated reduced repair efficiency in DEK knockout mouse embryonic fibroblasts compared to DEK wild-type cells.^[26] This DNA repair function for DEK was also validated by Kappes et al.^[28]

There are also non-DNA binding functions of DEK. Recently, an RNA-binding domain was identified in DEK and prior studies suggested it may be involved in RNA splicing.^{[13, 29-31]} In addition, extracellular DEK has been detected to function as a chemokine (discussed below) or as a component of neutrophil extracellular traps.^{[32, 33]} Extracellular, hyper-acetylated and poly(ADP-ribosyl)ated DEK protein is also an auto-antigen in autoimmune diseases such as juvenile idiopathic arthritis in which antibodies target the C-terminus (amino acids 188-350) of extracellular DEK.^{[28, 34]}

Post-translational modifications of the DEK protein are under-studied, but a few seminal reports have indicated their impact on DEK function. Phosphorylation of DEK by protein kinase CK2, primarily in the C-terminal portion of DEK, weakens the affinity of DEK binding to DNA in vitro, while still maintaining chromatin localization and dimerization ability in vivo.^[35] However, pronounced hypophosphorylation of DEK, and enhanced binding to chromatin, occurs during apoptosis although a portion of soluble, hyperphosphorylated DEK protein is also detected in the cytosol and nucleosol.^{[28, 36]} Furthermore, acetylation of DEK also decreases its DNA binding affinity at gene promoters and drives its re-localization to interchromatin granule clusters, which are sites enriched with pre-mRNA splicing factors.^[37] Finally, Kappes et al. also determined that DEK can undergo poly(ADP-ribosyl)ation, which results in loss of DNA-binding activity and the ability to alter DNA topology by EMSA assays.^[28] However, significant research is needed to determine the impact of these modifications for DEK chromatin modifying activity, subsequent cell signaling effects, and oncogenic function.

In numerous animal and plant studies, DEK expression has been implicated in responses to environmental conditions and adaptation. For instance, in Arabidopsis, DEK (particularly DEK3) prevents precocious flowering and limits germination in high salt environments.^[38-40] In both cotton (Gossypium hirsutum) and Arabidopsis, DEK expression protected the plants against fungus that causes Verticillium wilt.^[41] In Artemia shrimp embryos, which maintain quiescence in harsh environments, DEK expression promotes exit from quiescence, proliferation, and subsequent embryo development. Finally, a recent manuscript indicated there is positive selection for the DEK gene in mountain voles that live at high elevation.^[42] The ability to adjust to environmental cues and survive harsh environments is crucial for cancer cells. Thus, DEK functions in plants and animals may be informative for DEK functions in normal and cancerous human cells. In human models, atypical DEK gene expression, whether over or under-expressed, has been shown to have potentially pathogenic consequences, including tumor promotion and possible neural degeneration, respectively. DEK depletion or deletion has typically resulted in an increase in cell death through apoptosis via the stabilization of the p53 tumor suppressor gene.^[43] In neural cells, DEK depletion correlates with apoptosis and markers of neuropathology, including Tau hyper-phosphorylation.^[44] On the other hand, higher levels of DEK expression have been correlated with increased tumor growth and metastasis in solid malignancies.^{[9, 45-52]} In addition to these oncogenic functions, we recently reported that DEK expression in tumor cells creates an immune-suppressive microenvironment through the M2-like polarization of tumor-associated macrophages.^[23] Interestingly, unlike the consequences of deregulated DEK expression in solid tissues and malignancies, changes in the expression of full-length DEK in leukemia are controversial, with some reports showing down-regulation while others demonstrate over-expression (discussed below). However, the consequences of the t(6;9) translocation interrupting the DEK gene for leukemogenesis are unequivocal.

DEK-NUP214 Fusion: Clinical Presentation

NUP214 is a nucleoporin that binds to the cytoplasmic surface of the nuclear pore complex, which functions to export nuclear proteins and mRNA into the cytoplasm. It consists of an N-terminal β-propeller domain, a central coiled coil domain with a leucine zipper, and numerous FG repeats at the C-terminus (Fig 1A, center).^[53] Five NUP214 (aka: CAN) fusion genes have been described in various types of leukemia and reports indicate that these fusion proteins interfere with the nuclear export functions of NUP214, resulting in the aberrant nuclear accumulation of target proteins and poly-A+ RNAs.^{[54, 55]} One fusion partner of CAN/NUP214 is the DEK gene, which creates the DEK-NUP214 fusion as a result of a t(6;9)(p23;q34) chromosomal translocation (Fig 1B).^[56] Specifically, this translocation fuses nearly the entire DEK gene (amino acids 1-349) with the C-terminus of NUP214 (amino acids 813-2090; Fig 1A, bottom).^[56] NUP214 fusions are commonly seen in ALL and AML via chromosomal translocations, SET-NUP214 and DEK-NUP214, respectively, both of which result in a fragment of the NUP214 nucleoporin protein fusing to a chromatin remodeling protein.^[53] SET-NUP214 has been correlated with increased expression of MEF2C and LYL1, a hematopoiesis regulator associated with lymphoblastic leukemia, while DEK-NUP214 is linked to myelopoiesis (discussed below), potentially explaining the distinct hematopoietic profiles between SET-NUP214 and DEK-NUP214.^{[53, 57]} Interestingly, multiple case reports have documented patients positive for the DEK-NUP214 fusion protein, but with translocations variant to t(6;9)(p23;q34.1). Variant translocations t(1;9)(p22;q34) and t(9;12)(q34;q15) respectively resulted in FAB M4 AML positive for DEK-NUP214 expression, and acute monoblastic leukemia positive for DEK-NUP214, PRRC2B-DEK, and FLT3-TKD.^{[58, 59]} A third case report described a patient with t(6;9)(p23;q34) DEK-NUP214 FAB M2 AML also positive for NUP214-RAC1 and RAC1-COL12A1 fusion proteins, FLT3-ITD, and overexpression of NUP214 and RAC1.^[60] In all cases of variation in t(6;9) DEK-NUP214 AML fusion gene translocations and expression, patients positive for the DEK-NUP214 fusion protein presented with poor prognosis, despite aggressive treatment.^58-60

Beyond NUP214, AFF2 (aka: FMR2) was also identified as a partner gene for DEK fusions in nonkeratinizing, basaloid squamous cell carcinomas of the sinonasal tract. The DEK-AFF2 fusion protein is a result of a t(6;X) translocation that splices the N-terminus of DEK through exon 7, to the C-terminus of AFF2, starting between exons 4 to 9.^[61] The DEK-AFF2 tumors included aggressive, high-grade carcinomas due to local recurrence and distant lymph node metastases, along with low-grade carcinomas and benign papillomas.^[61-64] Little is known about the function of the AFF2/FMR2 protein, although it is a putative transcriptional activator or elongation factor, can bind to RNA quadruplexes, and is the target of GCC repeat expansion in Fragile X E (FRAXE) syndrome.^[65-67] AFF2 is one of the four protein family members that also includes AF4, LAF4, and AF5q31, all three of which have been found fused to MLL in acute leukemias.^[68-70] It is interesting to note that, regardless of the type of cancer, DEK fusion proteins always consist of the N-terminal acidic regions and central pseudo-SAP/SAP domains.

Although it is relatively uncommon, representing only 1% of AML and myelodysplastic syndrome (MDS) cases in adults and 2% in pediatric AML, the t(6;9)(p23;q34) DEK-NUP214 fusion AML is a particularly aggressive subtype.^{[4, 6]} The median age of diagnosis is 23 years (range: 2-66 years), which is strikingly younger than the median age of diagnosis with AML in general (67 years).^{[6, 71]} There is a male predominance in patients with t(6;9) AML, but no significant difference between males and females.^[71] Upon diagnosis with t(6;9) DEK-NUP214 AML, the median survival is 13.5 months (12.5 months in children and 14.4 months in adults), with a 5-year survival of 28% in children and 9% in adults.^{[6, 71]} Numerous clinical studies report that a majority of pediatric and adult patients with t(6;9) DEK-NUP214 AML clinically present with highly concurrent FLT3-ITD and other somatic mutations, yet do not significantly affect patient outcomes.^{[7, 8, 71-74]} Patients with t(6;9) AML predominantly present with FAB M2 AML classification^{[72, 73]} granulocytic or multilineage dysplasia,^{[7, 8, 71, 74]} anemia and thrombocytopenia^{[7, 74]} presence of Auer rods,^{[71, 74]} and some report basophilia^{[7, 71]} while others do not.^{[8, 74]} Patients with t(6;9) MDS compared to t(6;9) AML presented with significantly lower blast count, WBC count, and incidence of FLT3-ITD co-occurrence, but significantly higher frequency of multilineage dysplasia and platelet count.^[75] Clinical studies reported that adult and pediatric patients with t(6;9) AML presented with poor prognosis despite aggressive chemotherapy,^{[7, 8, 71-74]} however, allogeneic hematopoietic stem cell transplantation (HSCT) in addition to chemotherapy improved patient outcomes.^{[71, 73, 74]}

DEK-NUP214 Fusion: Treatment

Recent studies have determined that in addition to chemotherapy, HSCT improves prognosis and overall survival (OS) for patients with t(6;9) MDS and AML. For example, a matched-pair analysis between t(6;9) and normal karyotype AML patients who received allogeneic HSCT demonstrated that OS, disease-free survival, non-relapse mortality, and cumulative incidence of relapse were not significantly different, suggesting that HSCT may provide more favorable prognosis for t(6;9) AML patients.^[76] Furthermore, treatment with HSCT during complete remission (CR) and in FAB M2 AML significantly improved outcomes, presenting with a 3-year OS of 76% compared to a 3-year OS of 0% in non-CR and non-M2 AML patients. Incidence of relapse significantly decreased in patients who received HSCT during CR compared to patients not in CR, specifically when HSCT was performed during CR1 instead of CR2/3 or active disease.^{[77, 78]} Some studies concluded that FLT3-ITD status and other chromosomal abnormalities do not impact patient outcomes,^[78] whereas others stated that additional mutations such as FLT3-ITD can accelerate the kinetics of relapse for DEK-NUP214 positive patients.^[79]

While multiple studies have been reported for patients with t(6;9) AML, fewer studies have investigated t(6;9) MDS. However, Fang et al. compared clinical characteristics and outcomes of patients with t(6;9) AML versus MDS in a significantly large cohort for this rare disease. Prognosis of t(6;9) MDS is generally similar to t(6;9) AML, with 61% of t(6;9) MDS patients progressing to AML at 13 months.^[75] In both t(6;9) MDS and AML cases, patients who received HSCT treatment had significantly longer OS than patients without HSCT (62 months versus 17 months, respectively).^[75] Furthermore, t(6;9) MDS and AML cases maintained a similar prognosis and OS in patients who received HSCT, but in groups not treated with HSCT, t(6;9) MDS maintained a significantly longer OS than t(6;9) AML (26 months versus 13 months, respectively).^[75] Therefore, the initial diagnosis of MDS versus AML and treatment with transplantation versus not were the only significant prognostic differences between t(6;9) AML and MDS cases.

DEK-NUP214 Fusion: Oncogenic Functions

Numerous questions about the functional role of the truncated and fused DEK gene in the DEK-NUP214 fusion remain unanswered. The acidic regions and SAP/Ψ-SAP domains of the DEK gene, which are necessary for the histone and DNA binding functions of DEK (discussed above), are preserved in the DEK-NUP214 fusion protein.^{[53, 56]} However, it remains unclear if, or how, DEK-NUP214 affects these functions of DEK, and alternatively, how DEK individually contributes to the pathogenesis of the DEK-NUP214 fusion protein. The mechanistic function of the DEK-NUP214 fusion gene and protein is also not well understood, but a few studies have elucidated some answers. In vitro assays showed that compared to t(6;9)-negative AML bone marrow (BM), t(6;9)-positive AML BM cells hold increased sensitivity to cytokines such as G-CSF, GM-CSF, IL-3, and SCF, which differentially regulated growth and differentiation of t(6;9) leukemic cells into blasts, neutrophils, and basophils.^[80] t(6;9)-positive AML cells increased in the number of blasts and neutrophils in response to G-CSF and IL-3 stimulation more than t(6;9)-negative AML cells.^[80] Further in vitro studies demonstrated that in DEK-NUP214 cells, protein synthesis was up-regulated at the translational level through significantly increased phosphorylation of EIF4E, specifically in the myeloid lineage.^[81] These findings were supported by the work of Sanden et al., who discovered that DEK-NUP214 induced proliferation of myeloid cells via post-transcriptional upregulation of mTOR and increased mTORC1 activity, contributing to an increase in global protein synthesis and a metabolic shift toward oxidative phosphorylation.^[82] These studies were suggestive of oncogenic functions, however additional studies were conducted to determine the transforming ability of DEK-NUP214.

In 2010, Oancea et al. investigated if DEK-NUP214 had transforming ability to induce leukemia from hematopoietic stem cells (HSCs). They transduced Sca1⁺Lin⁻ murine HSCs with a DEK-NUP214 over-expression construct, which resulted in a consistent but low efficiency of leukemia induction accompanied by splenomegaly. Transformation efficiency reached 100% when Lin−/Sca1+/c-Kit⁺/Flk2⁻ long-term HSCs (LT-HSCs) were transduced with DEK-NUP214, while transduction of short-term HSCs (Lin−/Sca1⁺/c-Kit⁺/Flk2⁺) did not result in leukemia.^[83]

Qin et al. later tested the transforming ability of DEK-NUP214 for human HSCs. In the first established humanized mouse model of t(6;9) AML, DEK-NUP214 induced transformation of human CD34+ hematopoietic progenitor cells with a CD45+CD13+CD34+CD38+ immunophenotype, potentially via aberrant upregulation of HOXA and HOXB genes.^[84] Additional studies utilized the patient-derived FKH1 cell line, the only t(6;9) AML cell line available, and cells transduced with the DEK-NUP214 (DEK-CAN) gene to investigate the molecular pathways deregulated in the presence of the DEK-NUP214 fusion. Chiriches et al. identified that the FKH1 cell line also harbors a ETV6-ABL1 fusion, but treatment with imatinib could be used to narrow down DEK-NUP214 associated signaling. They identified significantly activated STAT5, mTOR, and Src family kinases (SFKs), which were required for cell growth and survival.^[85] Additionally, DEK-NUP214 expressing cells have been found to induce AML through maintained stem cell capacity in LT-HSCs and activated JAK2, thus significantly upregulating STAT3 and STAT5 levels.^[86] Finally, analysis of previous clinical studies demonstrated that in t(6;9) AML patient samples, there was significantly increased expression of EYA3, SESN1, PRDM2/RIZ, and HIST2H4, overexpression of HOXA and HOXB genes,^[86] and RAS was a common gene mutated in both t(6;9) AML and t(6;9) MDS patients.^[75] Therefore, the DEK-NUP214 fusion protein is involved in promoting myeloid proliferation, capable of leukemic transformation, and up-regulates numerous signaling pathways.

Recently, Chiriches et al published a follow-up study to identify how the DEK-NUP214 fusion protein may interfere with normal DEK and NUP214 localization and function. They reported that the DEK-NUP214 fusion causes the mis-localization of NUP214 away from the perinuclear region associated with nuclear pores to an intra-nuclear micro-speckled pattern by immunofluorescence.^[87] The DEK-NUP214 fusion protein maintained the chromatin localization of full-length DEK, but it was no longer able to interact with modified histone H3, H3K9me3, as determined by co-immunoprecipitation.^[87] Chiriches et al also utilized tandem affinity proteomics to identify interacting proteins for the DEK-NUP214 fusion protein. They determined that DEK-NUP214 can bind to nuclear pore protein NUP88 and nuclear export protein XPO1, similar to NUP214. However, they also identified interacting proteins associated with RNA processing and splicing, ribosome biogenesis, epigenetic mechanisms of chromatin remodeling, and DNA repair, all of which have been previously affiliated with DEK interacting proteins and function.^[87-89] Specifically, they identified that the DEK-NUP214 protein could interact with full-length DEK, DEAD-box helicases DDX-5, −17, and −21, DHX9, YBX1, HNRNPM, XRCC5 (encodes Ku80 protein), SF3B1, and RP3.^[87] However, they did not identify known DEK interacting proteins such as CK2 and histone H3 through this method, which suggests that the DEK-NUP214 fusion protein may cause loss of some DEK chromatin remodeling functions and its post-translational modification (i.e.: phosphorylation by CK2).^{[87, 88]} Additional work is needed to determine if the loss of DEK-mediated histone H3 binding and regulation, the mis-localization of NUP214, changes to the role of DEK in RNA processing and ribosome biogenesis, or a gain of function is responsible for the oncogenic activity of DEK-NUP214.

DEK in AML

While t(6;9) translocation AML has been a point of clinical focus, the role of full-length (non-translocation) DEK in AML is far less studied. DEK is extensively described to be an overexpressed, tumor-promoting protein in solid tumors compared to normal tissues, most notably in breast cancer, melanoma, human papilloma virus (HPV) positive cervical cancer, and head and neck squamous cell carcinoma.^{[9, 45, 46, 48, 90-94]} However, for AML, there is conflicting evidence regarding changes in DEK expression levels during leukemogenesis. Logan et al. performed an in silico analysis using gene expression data from the MILE study database and noted that DEK mRNA was under-expressed in AML across molecular subgroups compared to normal bone marrow (NBM). All individual subtypes, including normal karyotype, translocations, and other cytogenetic abnormalities, showed significantly reduced DEK expression compared to NBM.^[95] However, these results could not be validated by quantitative RT-PCR for DEK mRNA in primary AML samples compared to unfractionated NBM, while immunohistochemical staining for DEK protein on bone marrow biopsies suggested either decreased expression or no significant changes compared to NBM. Finally, they found that levels of DEK expression did not correlate with overall survival of AML patients. However, when stratifying AML subtypes into risk groups (favorable, intermediate, adverse), they reported that the lowest DEK protein expression correlated with favorable subtypes, and the highest DEK expression correlated with intermediate subtypes.^[95]

Conversely, Casas et al. analyzed mRNA expression of DEK, in addition to HOXA9, CBL, and CSF1R, in bone marrow samples from adult patients with AML via qRT-PCR, and associated cases by age, FAB, immunophenotype, and karyotype aberrations.^[96] It was determined that DEK mRNA was overexpressed in 97.6% (40/41) of AML cases compared to two NBM controls, independent of the t(6;9)(p23;q34) translocation as none of the patients had t(6;9)-positive AML. Additionally, CD34− bone marrow samples were associated with overexpression of DEK compared to CD34+ samples, which potentially shows a contribution of DEK to leukemic cell differentiation.^[96] Interestingly, this work replicated the team’s prior findings, in which they reported that DEK mRNA was overexpressed in 9/15 AML patient samples and under-expressed in none compared to healthy control samples, as determined by microarray (4.92-fold) and qRT-PCR (1.8-fold) analyses.^[97]

The expression of full-length DEK in AML samples with other cytogenetic abnormalities has also been investigated. Sanden, Nilsson, and Gullberg found that NUP98-HOXA9 and BCR-ABL1 fusions upregulated DEK protein expression. Meanwhile, there was a trend towards AML1-ETO slightly upregulating DEK and PML-RARα downregulating DEK protein expression, although these differences were not statistically significant. Furthermore, DEK protein expression somewhat correlated with the proliferation of transduced cells. For example, BCR-ABL1 resulted in the highest DEK upregulation and was also the most proliferative, while NUP98-HOXA9 only modestly upregulated DEK and had proliferation rates similar to control primary CD34+ cells. The only cells with mildly down-regulated DEK (PML-RARα) was the least proliferative cell line. It was suggested that DEK upregulation was not an initiating event for leukemogenesis, but rather a secondary effect to the expression of BCR-ABL1 and NUP98-HOXA9.^[98] Finally, Koleva et al. performed western blotting for DEK protein on myeloblasts from AML patients with wild-type FLT3 versus FLT3-ITD. While the blot was not quantified by densitometry, there is a potential trend towards decreased DEK protein levels in FLT3-ITD primary AML blasts.^[99] Overall, the role and expression of full-length DEK in AML and leukemogenesis is not fully understood, as contradictory findings demonstrate both under and overexpression of DEK mRNA and protein in AML samples and a lack of correlation with patient prognosis. Further studies are needed to determine if intracellular up- or down-regulation of DEK expression impacts leukemogenesis as a modifier of other genetic abnormalities. Additionally, the effects of DEK post-translational modifications and extracellular DEK functions are completely unstudied during leukemogenesis and normal cell signaling. Thus, the field is ripe for mechanistic molecular and biochemical studies.

Unfortunately, functional studies of full-length DEK in leukemia are limited, likely because there is no clear consensus for whether there is an impact on changes in DEK expression for leukemogenesis. One example of a study of full-length DEK function in leukemia was conducted by Sanden et al. who performed chromatin immunoprecipitation sequencing (ChIP-Seq) for DEK in U937 human myeloid leukemia cells. Their work indicated that DEK protein binds to regions of open chromatin rich in euchromatic marks and at transcriptional start sites.^[100] However, this should be interpreted with caution, as ChIP-Seq studies, especially those performed in the early stages of this technology, are biased towards open chromatin and highly expressed gene regions.^[101]

DEK in Hematopoiesis

The role of DEK in normal hematopoiesis has been more clearly elucidated, serving as an important player in differentiation, particularly myelopoiesis, hematopoietic stem and progenitor cell (HSPC) proliferation, and both extracellular and intracellular regulation. Logan et al. concluded that DEK levels were associated with hematopoietic progenitor cells (HPC) and specific cell stages in normal hematopoietic differentiation. Specifically, in silico analysis with the Hemaexplorer database showed that DEK mRNA expression was highest in human hematopoietic stem cells (HSCs) and progressively decreased in a stepwise manner during differentiation from myeloid progenitors to mature myeloid cells in the peripheral blood and bone marrow (7-fold difference).^[95] Chen et al. noticed a similar pattern in mouse HSCs (Lin⁻cKit⁺Sca-1⁺CD48⁻CD150⁺).^[19] In contrast, Logan et al. also determined with the Hemaexplorer database that in murine bone marrow, DEK expression was lower in long-term HSCs (LT-HSCs), peaked to the highest expression level in the granulocyte-monocyte progenitor (GMP) population, then decreased to the lowest levels in mature, fully differentiated monocytes and granulocytes.^[95] However, it is not clear which cell markers were used to distinguish HSCs from progenitors with this method. Therefore, these specific levels of DEK expression in normal hematopoiesis demonstrate that it may have a distinct function in differentiation.^[95]

To understand how DEK expression in HSCs and HPCs (collectively, HSPCs) influence their function, Broxmeyer et al. compared the number and proliferation of HSPCs in constitutive DEK knockout (KO) mice to DEK wild-type (WT) mice. Analysis of the bone marrow and spleen determined that DEK KO mice showed higher numbers of colony forming unit (CFU) CFU-GM, CFU-GEMM, and BFU-E populations, which were more likely to be proliferating based on cell cycle analysis, compared to control animals. Furthermore, the addition of recombinant DEK protein suppressed proliferation of HPCs isolated from both murine bone marrow and human cord blood cells.^[102] The increased number of CFU progenitors in DEK KO mice was also confirmed by our group in 2017, although we did not notice any significant difference in the number of long-term (LT) or short-term (ST) HSCs or multipotent progenitors (MPPs) between KO and WT animals in our constitutive DEK KO model.^[24] However, we did report decreased numbers of LK progenitor cells (Lin⁻c-Kit⁺Sca-1⁻). The mice in our studies were 8-10 weeks old, which prompted Chen et al. to study the effects of DEK deficiency in hematopoietic stem cells over time and in a conditional DEK knockout model. While their work validated our prior studies that young DEK knockout animals do not demonstrate differences in HSPC numbers, mice >3 months old began to demonstrate loss of LSK, ST-HSCs, LT-HSCs, HPCs, and MPPs, indicating that DEK is important for maintaining the HSC pool.^[19]

Interestingly, the increased proliferation of DEK deficient HSPCs, as noted by Broxmeyer et al., did not translate into improved engraftment after transplantation. On the contrary, in primary and secondary mouse bone marrow transplantations, DEK played a positive role in engraftment of longer-term repopulating HSCs, potentially even in the self-renewal capacity of bone marrow HSCs. Thus, DEK has a negative regulatory effect on proliferation of HPCs but can increase LT-HSC engraftment.^[102] This was validated by Chen et al. who recently demonstrated in Tie2-Cre and Mx1-Cre/DEK^fl/fl conditional DEK knockout mice that DEK deficiency caused decreased HSC self-renewal capacity and poor engraftment.^[19]

Follow-up studies by Capitano et al. aimed to elucidate the mechanism by which DEK suppresses HPC numbers. It was determined that DEK has a Glu-Leu-Arg (ELR) motif, similar to myelosuppressive chemokines IL-8 and CXCL2/MIP2, and that extracellular DEK protein can inhibit the proliferation of target progenitor cells through the CXCR2 receptor, heparan sulfate proteoglycans, and downstream Gαi protein signaling.^[103] Interestingly, it was also determined that extracellular DEK did not need to be internalized or translocated to the nucleus of the target cell in order to function. Extracellular DEK was found to mediate its suppressive effects on HSPCs, specifically LSK and LK cells, via activation of signaling pathways downstream of CXCR2, as determined by phosphorylation of ERK1/2 (pT202 and pY204), p38 MAPK (pT180 and pY182), and AKT (pS473). Transcriptomic analyses of Lin− bone marrow cells from DEK KO mice treated with recombinant mouse DEK (rmDEK) protein revealed increased expression of several inflammatory cytokine genes, such as Tnf and Cxcl10.^[103] Further analyses of this transcriptomic data found that genes associated with chemotaxis (cell chemotaxis, leukocyte migration, and cell-cell adhesion regulators) were significantly up-regulated in BM HSCs and HPCs treated with recombinant DEK compared to untreated cells.^[104] It was then determined that all subtypes of LSK cells (i.e. LT-HSCs, ST-HSCs, and MPPs) migrated toward rmDEK in a dose-dependent, chemotactic manner. Additionally, DEK was a more potent chemoattractant for LSK cells than SDF1a, which is one of the few known potent chemoattractants for HSPCs. This chemotactic function of DEK was found to be dependent upon CXCR2, since administration of anti-CXCR2 antibodies and a Gαi inhibitor prevented migration of LSK cells toward rmDEK.^[104] Combined, this work indicated that extracellular DEK can function as a pro-inflammatory cytokine and chemokine to control HSPC function in vitro and in vivo via CXCR2 receptor signaling.^{[103] [104]}

Aside from functioning as an extracellular regulator of hematopoiesis, there are also cell intrinsic functions of DEK that are important for HSPC function and maintenance. Notably, we observed that DEK loss did not impact the bone marrow microenvironment, as donor cells isolated from CD45.1+ C57Bl/6-Ptprc^a mice (aka: BoyJ), which are congenic to standard CD45.2+ C57Bl/6 mice, were able to engraft equally well into DEK WT and KO recipient mice.^[24] In determining the role of DEK in response to radiation-induced DNA damage, we found that constitutive DEK KO mice had increased survival after four doses of sublethal (7Gy/dose) irradiation. This correlated with increased short-term repopulation from irradiated KO cells in a competitive bone marrow transplant assay compared to DEK WT mice. Thus, indicating that DEK loss leads to increased resistance to radiation potentially due to the survival of transplantable progenitor cells. Increased radio-resistance in DEK KO HPCs (specifically Lin⁻c-Kit⁻Sca1⁺CD48⁺CD150⁺ cells) was attributed to decreased phosphorylation of p38, an increase in the number of quiescent cells, and a transcriptomic signature that indicated irradiated DEK KO progenitor cells maintained quiescence and suppressed apoptosis compared to irradiated DEK WT bone marrow cells.^[24] Therefore, it seems likely that DEK functions to regulate the cell cycle in response to radiation stress in HSPCs, potentially through its DNA repair activities and/or impact on p53 activity.^{[24, 26, 27, 43]}

Additional evidence that intracellular DEK expression contributes to regulating the balance between quiescence and proliferation of HSPCs was recently published by Chen et al. As previously mentioned, this group noted that Tie2-Cre/DEK^fl/fl and Mx1-Cre/DEK^fl/fl conditional knockout mice demonstrated a loss of HSPCs during aging.^[19] Whereas we noted that DEK deficient HSPCs aberrantly maintained quiescence upon irradiation, Chen et al. observed that in non-stressed HSPCs, conditional loss of DEK was associated with impaired maintenance of quiescence in HSCs and LSK cells, and increased cycling within these cell populations.^[19] Transcriptomic analyses identified a loss of quiescence-associated proteins p21 and p27, as well as increased Akt1/2 and KRas expression, in DEK-deficient HSCs. Phospho-flow analyses also identified an increase in mTOR activation, which was associated with more glucose consumption in vitro, in DEK conditional knockout HSPCs.^[19] These gene expression changes and pathway activation was associated with chromatin remodeling, as determined by ATAC-Seq, primarily at transcriptional start sites for genes including Akt1 and Akt2. This was due, at least in part, to increased acetylation of histone H3 at lysine 27 (H3K27ac).^[19] They additionally found that DEK could interact with histone deacetylase 3 (HDAC3) and NCoR1, such that the loss of DEK may impair HDAC3 function, leading to increased histone acetylation. In summary, Chen et al. used conditional DEK knockout mice to determine that the chromatin remodeling function of DEK, via histone H3 post-translational modification, is necessary to maintain HSPC quiescence and self-renewal capacity through the downstream regulation of master quiescence proteins p21 and p27, and essential proliferation signaling like the Akt-mTOR pathway.^[19] Together, the work of Chen et al. indicated that DEK loss was associated with a more oncogenic cellular and molecular profile, including KRas and Akt/mTOR activation, increased glucose consumption, and increased proliferation. Therefore, their work mechanistically supports previously discussed findings by Logan et al. that full-length DEK expression may be decreased in a subset of AMLs to promote leukemogenesis.^{[19, 95]} As previously eluded to, the work of Chen et al.^[19] and our work^[24] both studied the role of DEK for quiescence maintenance in HSPCs, but with conflicting results. We noted that DEK loss maintains quiescence and decreases proliferation of HPCs during times of stress (radiation), but Chen et al. observed that DEK loss in unstressed cells does not, but rather, induces cell cycle entry. Both studies used similar sample sizes. Potential explanations for these differences may include the use of constitutive versus conditional knockout models, which may have varying levels of efficiency for DEK loss, different markers to distinguish HSPC subpopulations (CD34 and CD135 in Chen et al. vs. CD48 and CD150 in Serrano-Lopez et al.), and different methods for stressing the DEK-deficient HSPCs.^{[19, 24]}

Given the association between the DEK-NUP214 fusion gene and AML, it is reasonable to hypothesize that DEK plays an essential role in myelopoiesis. Indeed, a quantitative proteomics screen for C/EBPα identified DEK as an interacting protein.^[99] C/EBPα is a leucine zipper transcription factor and a driving factor in myeloid lineage commitment and production of mature granulocytes, especially neutrophils.^[105] Of relevance, C/EBPα is frequently mutated in AML, resulting in loss of function or dominant negative versions of the protein.^[105] DEK interacts with C/EBPα on chromatin as a function of S21 phosphorylation, as C/EBPα and DEK colocalize in the nucleus on chromatin and when the S21 residue of C/EBPα is phosphorylated, the interaction diminishes. The DEK-C/EBPα interaction enhances transcriptional activation of GCSFR3, a known target gene of C/EBPα that promotes myelopoiesis, through direct binding of the DEK-C/EBPα complex to the GCSFR3 promoter. Finally, Koleva et al. found that DEK is necessary for normal granulocytic differentiation of CD34+ human bone marrow cells. DEK deficient cells had double the number of bipotent CFU-GM colonies but significantly decreased the number of granulocytic (CFU-G) colonies.^[99] DEK loss was accompanied by reduced expression of CEBPE and GCSFR3 mRNA levels and reduced numbers of differentiated granulocytes, as determined by fewer numbers of CD11b+ and CD15+ cells after G-CSF stimulation of CD34+ human bone marrow cells. Importantly, these effects were unique to granulocytes, since expression of the CD14 monocyte marker was independent of DEK levels and DEK deficient cells demonstrated no changes in monocyte colony formation.^[99] Overall, this work indicates that DEK interacts with C/EBPα to control expression of myelopoiesis genes to promote granulocytic differentiation in CD34+ hematopoietic progenitors.^[99]

Conclusion

The findings discussed in this review reveal the complex nature of the DEK protein and its multifaceted presence in molecular processes relevant to hematopoiesis and myelopoiesis. Overall, DEK is an important regulator of HSPC quiescence and engraftment and promotes granulopoiesis. The DEK-NUP214 can transform CD34+ human bone marrow cells and contributes to a particularly aggressive form of acute myeloid leukemia. The molecular mechanisms of these functions are poorly described, but likely are mediated through DEK’s chromatin remodeling function and regulation of histone H3 post-translational modifications to direct transcription of proliferation- and differentiation-associated genes. Additional work is needed to better understand the molecular functions of full-length DEK protein in myelopoiesis, the role(s) of full-length DEK in AML, and the mechanism(s) behind the transforming ability of the DEK-NUP214 fusion in t(6;9) AML. Ideally, future studies will identify a therapeutic strategy to target the highly aggressive t(6;9) subtype of AML to improve patient survival. As studies continue to relate DEK to critical interests such as cancer progression, it becomes an enticing target for potential genetic disease markers and chemotherapeutic drugs.

Highlights.

• DEK-NUP214 is an oncogenic fusion protein in t(6;9) acute myeloid leukemia

• Non-fusion DEK protein supports hematopoietic stem cell maintenance and engraftment during bone marrow transplantation

• DEK promotes myelopoiesis and disruption of its function(s) are associated with cellular and molecular correlates of leukemogenesis