Transcriptomic Profiles of MV4-11 and Kasumi 1 Acute Myeloid Leukemia Cell Lines Modulated by Epigenetic Modifiers Trichostatin A and 5-Azacytidine.

Background: Acute myeloid leukemia (AML) is the most common form of acute leukemias in adults which is clinically and molecularly heterogeneous. Several risk and genetic factors have been widely investigated to characterize AML. However, the concomitant epigenetic factors in controlling the gene expression lead to AML transformation was not fully understood. This study was aimed to identify epigenetically regulated genes in AML cell lines induced by epigenetic modulating agents, Trichostatin A (TSA) and 5-Azacytidine (5-Aza). Materials and Methods: MV4-11 and Kasumi 1 were treated with TSA and/or 5-Aza at IC50 concentration. Gene expression profiling by microarray was utilized using SurePrint G3 Human Gene Expression v3. Gene ontology and KEGG pathway annotations were analyzed by DAVID bioinformatics software using EASE enrichment score. mRNA expression of the differentially expressed genes were verified by quantitative real time PCR. Results: Gene expression analysis revealed a significant changes in the expression of 24,822, 15,720, 15,654 genes in MV4-11 and 12,598, 8828, 18,026 genes in Kasumi 1, in response to TSA, 5-Aza and combination treatments, respectively, compared to non-treated (p<0.05). 7 genes (SOCS3, TUBA1C, CCNA1, MAP3K6, PTPRC, STAT6 and RUNX1) and 4 genes (ANGPTL4, TUBB2A, ADAM12 and PTPN6) shown to be predominantly expressed in MV4-11 and Kasumi 1, respectively (EASE<0.1). The analysis also revealed phagosome pathway commonly activated in both cell lines. Conclusion: Our data showed a distinct optimal biological characteristic and pathway in different types of leukemic cell lines. These finding may help in the identification of cell-specific epigenetic biomarker in the pathogenesis of AML.


INTRODUCTION
Acute myeloid leukemia (AML) is characterized by a block in early progenitor differentiation leading to accumulation of immature and highly proliferative leukemic stem cells (LSCs) in the bone marrow and peripheral blood 1 . The 2017 World Health Organization (WHO) has provided guidelines on the cut-off value of blast percentage of AML by; 200 and 500 cells-leukocytes differential counts in the peripheral blood and in the bone marrow, respectively 2 . For a diagnosis of AML, a marrow or blood blast count of 20% or more is required, except for AML with t(15;17), t (8;21), inv (16) or t (16;16), and some cases of erythroleukemia. AML is the most common form of acute leukemias in adults which affected 32% adults. Although the overall mortality rate has decreased by 1.0% each year from 2001 to 2010, the overall incidence rate was increased by 0.2% each year. In 2018, the American Cancer Society estimated that 19,520 of new cases and 10,670 deaths from AML. The 5years overall survival rate was also poor with only 24% 3 . For many years, gene expression profiling by microarray was used as a traditional method to search abnormalities in cancers, including in AML 4 . These presented data was invaluable and accessible to the identification of disease's class discovery, class prediction, and class comparison. Class discovery refers to the identification of a new subgroup, that later was class predicted by gene expression data. The first and second class already had a diagnostic implication. While the third class, which is class comparison refer to the identification of genes that were deregulated in certain subgroups, that may address biological function 5 . It has long established that AML is clinically heterogeneous disease characterized by an accumulation of continuous genetic abnormalities 6 and prior epigenetic lesions 7 resulting in clonal evolution and expansion. The considerable complexities disrupt the genetic and epigenetic landscapes by changes in gene expression 8 which profoundly affecting treatment response and patients' survival. Earlier epigenetic alteration established cellular identities initiating tumorigenesis by inappropriate activation or inhibition of cellular signaling pathways 9 . For example, promoter hypermethylation of a tumor suppressor genes is commonly implicated in cancer 10 , involving genes controlling the cell cycle and DNA repair 11 . On the other hand, modification to histone protein in nucleosome modulates the transcriptional burst frequency specifically through histone acetylation 12 . Both epigenetic mechanisms endow the regulation in gene expression. Hence, targeting the epigenetically-regulated genes in the control of AML licensed a promising outcome. In this study, high-throughput microarray technique was used to analyze epigenetic-derived molecular mechanism by modulating gene expression using a classical DNA methyltransferase (DNMT) inhibitor; 5-Azacytidine (5-Aza) and a histone deacetylase (HDAC) inhibitor, Trichostatin A (TSA). The aim of this study was to induce the epigenetic response via gene re-expression or down-expression in two types of AML cell lines; MV4-11 and Kasumi 1. It was hypothesized that the silencing of a tumor suppressor gene and the activation of oncogenes in AML were due to epigenetic mechanisms of DNA hypermethylation and histone deacetylation.

Cell Viability Assay
Percentage viability of non-treated and treated MV4-11 and Kasumi 1 after the 24 hours exposure to TSA and 5-Aza treatments were measured by Trypan Blue Exclusion Assay (Life Technologies, CA, USA). The half maximal inhibitory concentration (IC50) was determined by GraphPad Prism 6.0 (GraphPad, CA, USA).

Total RNA extraction and quality control
Total RNA was extracted from treated and untreated MV4-11 and Kasumi 1 using Total RNA Isolation Kit (Promega, SA, USA) according to the manufacturer's protocol. The final elution step was performed using 30 µl of elution buffer for a highly concentrated RNAs. The isolated RNA concentration and purity were determined by Nanodrop ND-1000 spectrophotometer (Thermo-Fisher Scientific, WA, USA). Prior to the gene expression profiling, the RNA integrity was assessed by 1.5% agarose gel electrophoresis and their RIN (RNA integrity number) values were determine by Agilent 2100 Bioanalyzer (Agilent, CA, USA). The qualified RNAs (absorbance 280/260 1.8-2.1 ratio; highly intact 28S and 18S ribosomal RNA and RIN above 7) were stored at -80 ºC until further analysis.

Microarray analysis
Whole genome expression profiling was performed using One-Color SurePrint G3 Human Gene Expression v3, 8 x 60K slides contained array probe (Agilent Technologies, CA, USA). Prior to Cyanine 3 (Cy3) labeling, RNA spiked-In dilution was prepared using RNA spiked-In Kit (Agilent Technologies, CA, USA) to each sample using T7 RNA polymerase (RNA reference target) for normalization. Cy3-labeled cRNA was generated from 25 ng input total RNA using Low Input Quick Amp Labeling Kit (Agilent Technologies, CA, USA). The fluorescent-labeled cRNA was purified by RNAeasy Mini Kit and RNAasefree DNAase Set (Qiagen, CA, USA) and quantified by Nanodrop ND-1000 spectrophotometer. 25 ng of fluorescein-labeled and amplified cRNA was hybridized into array slides containing 60,000 probes (Agilent Technologies, CA, USA) at 65 degree Celsius for 17 hours. After hybridization and washing steps, the array slides were scanned using SureCan Microarray Scanner (Agilent Technologies, CA, USA) to measure the fluorescence intensity of Cy3 labeled RNA bound to the microarray slide. The resulted images were processed using the Feature Extraction (FE) software v.12 (Agilent Technologies, CA, USA) for data filtering. Raw data obtained was analyzed by Genespring GX v12.6 software (Agilent Technologies, CA, USA).

Database screening
Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis annotations were utilized by the Database for Annotation, Visualization and Integrated Discovery (DAVID) Bioinformatics Resources v6.8 (https://david.ncifcrf.gov/) to characterize and predict epigenetically regulated genes in treated AML cell lines. The Enhanced AL Scoring Engine (EASE) scoring system (a modified Fisher Exact pvalue, p<0.1) was implemented for statistical analysis to provide enriched GO terms and pathways annotation within gene lists. EASE analysis produces a consistent and similar functional annotation with numerous analytical methods 13 , and Venn diagram was constructed to analyze genes with differential expression pattern after TSA and 5-Aza treatment in MV4-11 and Kasumi 1. The analysis was conducted by the Venny 2.1 software (http://bioinfogp.cnb.csic.es/tools/venny/).

Quantitative Real-time PCR (qRT-PCR)
To validate microarray data, qRT-PCR analysis on selected up-regulated and down-regulated genes was performed by Taqman gene expression assays and analyzed using Applied Biosystem (ABI)® 7500

RESULTS
A significant decrease in cell viability was observed after the TSA and 5-Aza treatments (One-way ANOVA, p<0.05). The half maximal inhibitory concentration (IC50) was acquired at 2.2 µM and 2.3 µM for MV4-11 and; 6.25 µM and 6.95 µM for Kasumi 1 in TSA and 5-Aza, respectively. TSA and 5-Aza treatments have higher potency in MV4-11 due to their lower IC50 value compared to Kasumi 1 ( Figure 1).

86
International Journal of Hematology Oncology and Stem Cell Research ijhoscr.tums.ac.ir

DISCUSSIONS
It was recognized that epigenetic changes serve as a mediator in cancer progression by the changes of gene expression. Epigenetic alterations are reported to concurrently disrupt the essential signaling pathway predisposed cell to uncontrolled growth, longer survival, and metastasis 14 . Histone modifications and DNA hypermethylation are two known epigenetic mechanisms that largely impact the regulation of gene transcription. Histone modification by acetylation has been found to be significantly deficient in acute leukemia patients, compared with the normal individual 15 . In this study, TSA acts by increasing the acetylation level by inhibiting HDAC activity in human leukemic cell lines. Histone acetylation is known to enhance the expression of specific genes that elicit extensive cellular morphology and metabolic changes, such as growth arrest, differentiation, and apoptosis 16 . Aberrant DNA methylation was the most common epigenetic alteration in leukemia in which an increased level of DNA methylation was observed in AML at remission 17 . 5-Aza reverts DNA methylation to induce antineoplastic activity either by global hypomethylation and direct cytotoxicity on abnormal hematopoietic cells in the bone marrow 18 . 5-Aza inhibits DNMT thus to induce reexpression of the silenced genes to halt tumor growth, and to cause modest differentiation in transformed leukemic cell lines and primary AML 19 . The current study found that both TSA and 5-Aza inhibit the growth of MV4-11 and Kasumi 1 cell lines in a dose-dependent manner. The IC50 of both treatments at 24 hours were lower in MV4-11, compared to Kasumi 1 which could suggest the inhibitory effect of the drugs were less sensitive in Kasumi 1 harboring t(8;21) than in MV4-11 with FLT3-ITD mutation. The variation in the IC50 values would also represent different expression signature in response to TSA and 5-Aza treatments. It is proposed that the genes which were commonly expressed within TSA, 5-Aza and TSA+5-Aza treatments were epigenetically regulated and involved in the pathogenesis of AML and may serve as candidates for potential biomarkers although they did not share similar GO profile and targeted different signaling pathways. DEF8, NDUFC2, GUSBP1, ARIH2, STX12 and HIST1H2BN were highly re-expressed (more than 100 folds) in either treatment of MV4-11, have not been previously discussed on their role in cancer except for HIST1H2BN. DEP8 is located at chromosome 16 encodes for an activator of intracellular signal transduction reported to carry single nucleotide polymorphism (SNP) rs4268748 at 16q24 with significantly associated with cell cycle regulator, CDK10 expression 20 . GUSBP1 which was located at chromosome 5 were involved in transcriptional regulation by putative alternative promoters (PAPs) 21 . ARIH2 primarily functions in neuronal differentiation was found to be tumor-specific in Glioblastoma multiforme (GBM) correlated with growth suppression in GBM cell lines 22 . Treatment with 5-aza-2′-deoxycytidine resulted in gene reexpression of HIST1H2BN in malignant ovarian cancer 23 .
Differential down-regulation of HIST1H2BN was observed in meningiomas was associated with malignant progression 24 . RAB12 is a member of RAS oncogene family, function as small GTPase for intracellular protein transport, activated in stimulus-dependent pattern and promote microtubules-dependent of the cell secretarygranule in mast cell 25 and its up-regulation has been linked with colorectal cancer 26 . The most optimal GO in MV4-11 were Cytoskeleton organization involving TUBA1C, JAK-STAT cascade involving SOCS3 and STAT6 and the cell cycle involving CCNA1, associated with Phagosome, JAK-STAT pathway and Viral carcinogenesis, respectively, CCNA1 was expressed after TSA treatment with high fold-change (298.44) in MV4-11, but was slightly re-expressed at a low level in 5-Aza and combination treatment (fold-change: 5.67 and 2.81, respectively) (results not shown). CCNA1, located at chromosome 13, encodes for activating regulatory subunit which binds to cyclin-dependent kinases 2 (CDK2) and cell division cycle 2 (CDC2) for the cell cycle machinery to progress into S phase 27 . In normal cells, CCNA1 was prominently expressed in testes, hematopoietic cells, and brain 28 . CCNA1 acts as tumor suppressor gene (TSG) which is epigenetically silenced by hypermethylation in cervical cancer 29 , ovarian, renal and lung carcinoma 30 . In AML, CCNA1 was found to be overexpressed especially in M3 and M2 AML with significant worse overall survival 31 39 and breast cancer 40 . Other candidate genes convoluted in the JAK-STAT pathway associated with hematological malignancies are STAT6 and RUNX1. TUBA1C, located at chromosome 12 is a member of tubulin family of microtubules ubiquitously expressed in the esophagus, bone marrow, appendix, brain, colon, bladder and placenta 41 . TUBA1C expression was significantly increased in hepatocellular carcinoma (HCC) on both mRNA and protein level, which predict a poor prognosis 42 , reduced expression in breast cancer associated invasive stage 43 and their expression was susceptible to colorectal cancer risk 44 . Cytochrome P450 (CYP4F2) was the highest re-expressed gene in TSA treatment with more than 1000 fold-change in MV4-11. CYP4F2 is a drug-metabolizing enzyme gene reported to have an epigenetic regulatory role with clinical implication 45 . Inhibition of DNMT and histone deacetylase (HDAC) by 5-Aza and TSA induced the demethylation of CYP1A1 and CYP1A2 leading to their up-regulation 46 . In Kasumi 1, three common differentially expressed genes in either treatments were ANGPTL4, TUBB2A, and ADAM12 associated with angiogenesis, microtubule-based process, and cell-adhesion, respectively. ANGPTL4, located at chromosome 19 encodes a glycosylated, secreted protein containing a fibrinogen-like C-terminal domain, mainly induced by a nuclear receptor protein, peroxisome-proliferator-activated receptor (PPAR) 47 . It is the most studied among ANGPLT family, functions primarily in the regulation of lipid metabolism, glucose homeostasis, and insulin sensitivity 48 . ANGPTL4 has not been previously discussed in the context of AML. However previous studies have reported ANGPTL4 in various cancer types, including breast cancer, colorectal cancer, prostate cancer, hepatocarcinoma, and renal cell carcinoma, suggesting its important roles in cancer cell growth and progression 49 . In the current study, ANGPTL4 was mutually up-regulated in TSA treatment in both MV4-11 and Kasumi 1 cell lines, thus has wide potential for gene-specific therapy in AML. TUBB2A, located at chromosome 6 is another putative gene in AML with cell-specific expression. It forms a class ll beta-tubulin from six families of tubulins, including, alpha, gamma, delta, epsilon and zeta, and their protein may localize in extracellular exosome, cytoplasm and nucleus, involved in small GTPase activity, GTP binding, nucleotide binding acetylation and methylation 50 . Alpha and beta tubulin sub-families were studied for mutational analysis in human brain tumor and malformations was found in TUBB2A affecting the spectrum of "tubulinopathy" phenotypes 51,52 . Mutations in TUBB2A were also explored in epilepsy 51 , gastric carcinoma and lung cancer 53 but not hematological malignancies. ADAM12, located at chromosome 10 was over-expression in non-Hodgkin's lymphoma that lead to accelerate of proliferation and celladhesion 54 and was commonly methylated in chronic lymphocytic leukemia 55 . The roles of ADAM12 in leukemia pathogenesis is still obscure and need further study since the expression of this gene was similarly down-regulated in both treatments.
PTPN6 (or SHP1) located at chromosome 12 was differentially regulated in TSA and 5-Aza treatments (re-expressed only in 5-Aza but not TSA). Our previous study showed a positive correlation of PTPN6 re-activation due to hypomethylation in MV4-11 that carry a FLT3-ITD mutation after the 5-Aza treatment 56 . PTPN6 expression has been studied in lymphoma, leukemia and other cancers such as breast cancer, ovarian cancer, prostate cancer, and pancreatic cancer 57 , and in hepatocellular carcinoma 58 . PTPN6 is a downstream mediator in the JAK-STAT pathway, and together with SOCS3 they potentially serve as molecular indicators for pathway-targeted therapy in AML. Another example of the methylationrelated gene is PRG2. In the Venn diagram, PRG2 was exclusively expressed in 5-Aza treatment, but not in TSA treatment. The differentially expressed PRG2 was reported in three human leukemic cell lines (K562, THP1, and HL-60) 59 . We also previously reported that the expression of PRG2 was restored after 5-Aza treatment in PKC-412 (Midostaurin) resistant leukemic cell line 56 . DHRS2 and LMTK3 were another highly up-regulated genes in TSA treatment in Kasumi 1 with up to 500 fold change.
Their up-regulation was due to histone acetylation. Finally, despite thousands of genes generated by microarray expression profiling, the highly reexpressed and down-expressed genes perceived in this study were thought to be convoluted with epigenetic regulation of gene transcription in AML.
Although only several genes were selected for validation by qRT-PCR, there were many other genes as discussed earlier that may have important roles in cancer pathogenesis.

CONCLUSION
In conclusion, we have identified common differently expressed genes that are importants in epigenetic regulation of AML. Our finding also revealed that Phagosome pathway was the most optimal and common in both MV4-11 and Kasumi 1 AML cell lines. Although MV4-11 and Kasumi 1 transduced different optimal signaling pathways in response to drug treatment, it was shown that MV4-11 mainly targeted the genes in the JAK-STAT signaling, while Kasumi 1 targeted the genes in transcriptional misregulation in cancer, PI3K-Akt and MAPK signaling, which are all critical pathways in oncogenesis. These were due to their different molecular characteristics (FLT3-ITD vs t(8;21) AML1-ETO). The data presented here may serve as a preliminary finding and are useful for further study to explore epigenetic involvement in the pathogenesis of AML.