Hypomethylation of MIR‐378 5’‐flanking region predicts poor survival in young patients with myelodysplastic syndrome

Abstract Background Previous studies have disclosed up‐regulation of MIR‐378 in acute myeloid leukemia (AML), and might consequently affect the outcome of the patients. Correspondingly, hypomethylation of MIR‐378 was also identified in AML, particularly for FAB‐M2 subtype with t(8;21) chromosomal translocation. Nevertheless, the methylation status of MIR‐378 has not been illustrated in myelodysplastic syndrome (MDS). Herein we designed to understand the methylation pattern of MIR‐378 involved in MDS and clinical interrelation thereof. Methods Real‐time quantitative methylation‐specific PCR (RQ‐MSP) was performed to evaluate the methylation degree of MIR‐378 5’‐flanking region on bone marrow mononuclear cells collected from 95 de novo MDS patients. Five gene mutations (IDH1, IDH2, DNMT3A, U2AF1, and SF3B1) were detected by high‐resolution melting analysis to further evaluate the clinical relevance of hypomethylation of MIR‐378. Results Unmethylated level of MIR‐378 5’‐flanking region was significantly higher in MDS patients than that in controls (p = .034). Hypomethylated MIR‐378 was identified in 20 of 95 (21%) cases with MDS. Male patients appeared to be more frequent to harbor MIR‐378 hypomethylation compared to female patients (15/55, 27.3% vs. 4/40, 10.0%, p = .04). There was no significant difference in age, white blood cell counts, platelet counts, hemoglobin concentration, and karyotypes between the patients with and without MIR‐378‐hypomethylation. Distinct distribution of five gene mutations was not observed in the two groups as well. However, MIR‐378‐hypomethylated patients had significantly shorter overall survival than those without MIR‐378 hypomethylation (p = .036). Moreover, among patients <60 years, hypomethylation of MIR‐378 was confirmed to be an independent adverse prognostic factor by both Kaplan–Meier and Multivariate Cox analyses. Conclusion Hypomethylation of MIR‐378 5’‐flanking region is an adverse prognosticator in MDS, particularly in patients <60 years.


| INTRODUCTION
Myelodysplastic syndrome (MDS) is a hematopoietic malignance characterized by dyshematopoiesis with a high tendency to transformation of acute myeloid leukemia (AML) (Valent et al., 2007). Cytogenetic as well as genetic alterations have been widely illustrated and considered to play a vital role in the processes of MDS (Adès, Itzykson, & Fenaux, 2014;Lee, Podoltsev, Gore, & Zeidan, 2015). Meanwhile, epigenetic variations, particularly for the aberrant DNA methylation, were also deemed to participate in the development of MDS (Issa, 2013;. In spite of promising clinical application of hypomethylating agents (HMAs) presented recently, however, primary drug resistance and recurrence or disease progress during the treatment are the main obstacles (Carella, 2015), the outcome of patients with MDS generally remains dismal. The knowledge about pathogenesis of MDS is still insufficient. Until recently, a group of small noncoding RNA molecules (19-24 nts), microRNAs, have emerged as crucial roles in nearly all kinds of biologic functions such as development, differentiation, apoptosis, and proliferation (Schmittgen, 2008;Winter, Jung, Keller, Gregory, & Diederichs, 2009). Ectopic expression of miRNAs was implicated in a wide scale of diseases including cancers (Barbarotto, Schmittgen, & Calin, 2008;Calin & Croce, 2006). Increasing number of studies have even supposed that deregulation of various microRNAs contributed to pathogenesis of MDS and transformation into AML (Pons et al., 2009;Votavova et al., 2011), furthermore, large number of microRNAs could be served as useful biomarkers to predict the outcome (Bhise et al., 2016;Cattaneo et al., 2015;Xu, Guo, Zhang, Chen, & Hu, 2016). Strikingly, researchers have reported distinctive influence of MIR-378 in many cancers, such as lung cancer , colorectal carcinoma (Faltejskova et al., 2012), gastric cancer . Our previous study preliminarily observed overexpressed MIR-378 was often related to AML and exerted adverse effects on prognosis . Correspondingly, hypomethylated MIR-378 was revealed to be associated with AML, especially in FAB-M2 subtype (Xiao-Wen et al., 2015). However, there is limited information about MIR-378 methylation status in MDS. This work was to determine the methylation status of the 5'-flanking region of MIR-378, which contains 212 CpG dinucleotides, and further to analyse the clinical role in MDS.  (Bennett et al., 2010;Vardiman et al., 2009). The risk classification for each patient was on the basis of International Prognosis Scoring System (IPSS) for MDS (Greenberg et al., 1997). The clinical features as well as laboratory tests of whole MDS patients were displayed in Table 1.

| Therapeutic regimen
Treatment options including supportive care, low-or highintensity therapies, were given generally according to the risk classification and the wishes of patients. Briefly, symptomatic Low/Int-1 patients were managed with hematopoietic growth factors, thalidomide or immunosuppressive therapy, and supportive treatment. The treatment for Int-2/High patients included decitabine or chemotherapy with CAG (aclacinomycin, cytarabine, and G-CSF) or HAG (homoharringtonine, cytarabine, and G-CSF) protocol.

| DNA isolation, bisulfite modification
The BM mononuclear cells (BMNCs) were obtained by using of Ficoll-Hypaque gradient. Genomic DNA extraction from the BMNCs was conducted via genomic DNA purification kit (Gentra), and then CpGenome ™ DNA Modification Kit (Chemicon) was used to treat the production of genomic DNA according to the manufacturer's protocol. Conclusion: Hypomethylation of MIR-378 5'-flanking region is an adverse prognosticator in MDS, particularly in patients <60 years.

K E Y W O R D S
hypomethylation, MIR-378, myelodysplastic syndrome, prognosis | 3 of 9 WU et al.

| Methylation-specific PCR
The methylation-specific PCR (MSP) reactions for methylated MIR-378 and unmethylated MIR-378 were performed on Step One Plus (Applied Biosystems) using two sets of methylated (M) and unmethylated (U) specific primers (Table 2). For the control, we chose ALU repetitive sequence as was described before . The amplification reaction system consisted of 10 μM SYBR Premix Ex Taq II, 0.4 μM of primers, 0.4 μL 50× ROX (Takara), and 20 ng of bisulfite-modified DNA. Amplification for methylated specific primers was conducted in the conditions of 95°C for 30 s, cycled at 95°C for 5 s, 62°C for 30 s, and 72°C for 30, and a fluorescence collection step at 75°C for 30 s (40 cycles),   ) was used to evaluate unmethylated level of MIR-378. N unmethylation-MIR-378 was determined using the 2 −ΔΔCT method comparing the ALU gene level. This means that a higher value of N unmethylation-MIR-378 represents a lower methylation status or hypomethylation of MIR-378. The products of MSP were resolved in the concentration of 2% agarose gels to visualize under UV illumination.

| Bisulfite-sequencing PCR
In our experiment, Bisulfite-sequencing PCR (BSP) was used to further determine DNA methylation density of MIR-378 in samples from three MIR-378-hypermethylated and three MIR-378 hypomethylated MDS patients according to the results of RQ-MSP. The PCR was conducted with the reaction system containing 6.25 μM of dNTP mixture, 10× PCR buffer (0.25 mM KCl), 0.75 U of Hot start DNA polymerase (Takara), 0.5 μM of primers, and 20 ng of modified DNA on iCycler Thermal Cycler (Eppendorf). The bisulfitetreated DNA was amplified with sequencing primers in Table  2. The conditions of amplification were 98°C for 10 s, then cycled at 98°C for 10 s, 56°C for 30 s, 72°C for 30 s, and entered to extension at 72°C for 7 min. After the program of purification, the PCR products were cloned into pMD ® 19-T Vector (Takara), and 10 clones of each BSP product were sequenced (BGI Tech Solut ions Co., Ltd). The percentage of methylation was calculated from the number of methylated CpG divided by total CpG loci of all 10 clones.

| Statistical analysis
The SPSS software 22.0 package was used for Statistical analysis. Mann-Whitney/Wilcoxon test was performed to compare two groups with continuous variables. For the comparison of categorical variables between patient teams, the statistical method utilized was Pearson Chi-square analysis or Fisher exact test. Kaplan-Meier method was employed for survival analysis, and the identification of the independent risk factors affecting the survival was carried out by Cox regression model. A two-tailed p < .05 was considered statistically significant.

5'-flanking region in MDS patients
The human MIR-378 gene locates in chromosome 5q32, according to bioinformatics analysis, there are three CpG islands within −2.0 kb consisted in MIR-378 5'-flanking region (Xiao-Wen et al., 2015). As was disclosed by RQ-MSP, the unmethylated level of MIR-378 5'-flanking region was substantially higher in MDS patients (median 0.021, range 0-23.990) than that in controls (median 0.005, range 0-1.000) (p = .034, Figure 1). The products of RQ-MSP for MIR-378 methylation and MIR-378 unmethylation were analyzed by agarose gel electrophoresis and the representative electrophoretogram is displayed in Figure 2. Bisulfite sequencing was further performed to examine three patients with MIR-378 hypomethylation and three patients with MIR-378 hypermethylation on the basis of RQ-MSP, and the results are shown in Figure 3. There was a significantly negative correlation between the results of RQ-MSP and BSP (R = −.886, p = .019). Thus, our BSP analysis confirmed the results of RQ-MSP. M  GGATGAGTTTTGAGTCGTTT  CCATACAAACCGCTCACTCC  129   U  TAGGATGAGTTTTGAGTTGT  ATAATCCCATACAAACCACT  137   BSP  AGGATTTTTTGGTGATTTTTG  TCACCCTCTAACTACATAATCCC  200 Abbreviations: BSP, bisulfite-sequencing PCR; RQ-MSP, real-time quantitative methylation-specific PCR.

5'-flanking region methylation and clinical characteristics
For clinical relevance analysis, the MDS patients were divided into two groups, the one with hypomethylated MIR-378 and the other without hypomethylated MIR-378, based on the value of the mean plus 1 SD (0.252) in normal controls. Accordingly, hypomethylated MIR-378 was identified in 20 of 95 (21.1%) cases, however, there is no statistical significance as compared to healthy donors (6.3% [1/16], p = .188). No significant difference was observed in age, white blood cell counts, platelets, hemoglobin, as well as karyotypic groups between the two groups (p > .05). The incidence of hypomethylation of MIR-378 in male (15/55, 27.3%) was significantly higher than that in female (4/40, 10.0%) (p = .040).
There is no significant difference in distribution of MIR-378 hypomethylaton in the FAB classifications, WHO classifications, or IPSS subgroups. It was also failed to find distinction of the five gene mutations between the two groups (p > .05) ( Table 1).

5'-flanking region methylation and prognosis
A total of 77 cases with available follow-up data were included in our survival analysis. The median survival of all patients was about 24 months. Hypomethylated patients presented significantly worse survival compared to hypermethylated patients (median 8 vs. 30 months, respectively, p = .036, Figure 4a). Subgroup analysis revealed that patients with MIR-378 hypomethylation presented slightly worse OS (median 7 months) than those without MIR-378 hypomethylation (median not obtained) in cytogenetically normal patients (p = .152). Multivariate analysis including sex (male vs. female), IPSS (Low/Int-1/Int-2/High), MIR-378 hypomethylation (yes vs. no), five gene mutations (mt vs. wt) failed to demonstrat independent prognostic factor of MIR-378 hypomethylation in whole MDS patients (Table  3). However, among the age <60 years, patients with MIR-378 hypomethylation had significantly shorter OS (median 9 months) than those without MIR-378 hypomethylation (median not obtained) (p = .027, Figure 4b). The negative prognostic impact was also confirmed by multivariate analysis (Table 4).

| DISCUSSION
Increasing numbers of microRNAs have been evidenced to participate in tumor initiation, evolution, drug-resistance, and disease recurrence (Eitan et al., 2009;Esquela-Kerscher & Slack, 2006;Fattore, Sacconi, Mancini, & Ciliberto, 2017;Ipn, Ng, Baharuddin, & Zakaria, 2017;Nicoloso, Spizzo, Shimizu, Rossi, & Calin, 2009;Ventura & Jacks, 2009), moreover, several of them have been confirmed to be useful tumor diagnostic and prognostic biomarkers. It is reported that MIR-378 had the abilities to enhance some types of cancer cell proliferation, reduce the cell apoptosis in vitro, as well as promote neoplasm growth, angiogenesis, and metastasis in vivo Lee, Deng, Wang, & Yang, 2007;Ma et al., 2014;Skrzypek et al., 2013;Yu et al., 2014). Clinically, a meta-analysis presented recently has demonstrated the prospective diagnostic value of MIR-378 in human cancers (Li, Shen, Li, Chen, & Chen, 2016). In AML, we firstly reported aberrant overexpression of MIR-378 presented in FAB-M2 subtype, particularly in cytogenetical classification of t(8;21) chromosomal translocation . Our RQ-MSP analysis correspondently shown high incidence of hypomethylation of MIR-378 presented in FAB-M2 subtype, as well as for the cohort with t(8;21) chromosomal aberration (Xiao-Wen et al., 2015).  (Erdogan et al., 2011;Merkerova et al., 2011). Here in, we highlighted MIR-378 methylation status in MDS, and the result demonstrated the generally hypomethylated MIR-378 in MDS, those were compared to normal controls. Hypomethylation of MIR-378 could be detected in 21% cases in our experiment. However, the distribution hypomethylation was not distinguishing either in cytogenetic classifications or in IPSS-risk subgroups. Interestingly, our data indicated that MIR-378 hypomethylation occurred more frequently in male patients. This phenomenon may be attributed to the prevalence of smoking in male people in China. It is crucial that the change of DNA methylation is another contribution by tobacco smoke in cancer (Alghanim, Wu, & Mccord, 2018;Tsaprouni et al., 2014). Whether MIR-378 hypomethylation is correlated with smoking needs epidemiological investigation. In the present study a panel of gene mutations (IDH1, IDH2, DNMT3A, SF3B1,and U2AF1), genetically reported to be the important regulator in DNA methylation and RNA spliceosome and highly associated with MDS by recent studies (Adès et al., 2014;Gill, Leung, & Kwong, 2016;Lee et al., 2015;Lin et al., 2011Lin et al., , 2012, was analyzed by HRMA to further explore the correlation of hypomethylated MIR-378 and genetic changes in MDS. However, no significant difference was found in the frequencies of five gene mutations (IDH1, IDH2, DNMT3A, SF3B1, and U2AF1) between patients with and without MIR-378 hypomethylation. Up-expression rather than hypomethylation of MIR-378 has been disclosed to be an adverse prognosticator in AML by our former studies Xiao-Wen et al., 2015). Notably, in this study we demonstrated that hypomethylation of MIR-378 predicted poor outcome in MDS, particularly in patients <60 years. In our multivariate analysis, IPSS stratification, DNMT3A and SF3B1 mutations could significantly predict the outcome of younger patients (<60 years), in addition to these, hypomethylation of MIR-378 also acts as a risk factor independently, providing helpful information for the prognosis of MDS. Further studies are needed to confirm the results before it can be used routinely as a potential marker for risk stratification in the patients with MDS.
Briefly, our study revealed that hypomethylaiton of MIR-378 was a common epigenetic phenomenon in MDS, and predicted poor outcome of more young patients (<60 years).