Methylation of miR-34a, miR-34b/c, miR-124-1 and miR-203 in Ph-negative myeloproliferative neoplasms

Background MicroRNA (miR) miR-34a, -34b/c, -124-1 and -203 are tumor suppressor miRs implicated in carcinogenesis. Methods We studied DNA methylation of these miRs in Philadelphia-negative (Ph-ve) myeloproliferative neoplasms (MPNs). Methylation-specific PCR (MSP), verified by direct sequencing of the methylated MSP products, was performed in cell lines, normal controls and diagnostic marrow samples of patients with MPNs. Results Methylation of these miRs was absent in the normal controls. miR-34b/c were homozygously methylated in HEL cells but heterozygously in MEG-01. In HEL cells, homozygous miR-34b/c methylation was associated with miR silencing, and 5-aza-2'-deoxycytidine treatment led to re-expression of both miR-34b and miR-34c, consistent with that both miRs are under the regulation of the same promoter CpG island. miR-34a was heterozygously methylated in MEG-01 and K-562. miR-203 was completely unmethylated in K-562 and SET-2 but no MSP amplification was found in both HEL and MEG-01, suggestive of miR deletion. In primary samples, four each had miR-34b/c and -203 methylation, in which two had concomitant methylation of miR-34b/c and -203. miR-34a was methylated in one patient and none had methylation of miR-124-1. Seven patients (15.6%) had methylation of at least one of the four miRs. miR methylation did not correlate with clinical parameters, disease complications or JAK2 V617F mutation. Conclusion This is the first report of miR hypermethylation in MPNs. miR-203 hypermethylation is not specific to Ph+ve leukemias but also present in Ph-ve MPNs. miR-34b/c methylation was associated with reversible miR silencing. There was no correlation of miR methylation with clinical demographic data or outcome.


Background
Philadelphia-negative (Ph-ve) myeloproliferative neoplasm (MPN) is a stem cell disease with proliferation of myeloid lineage, leading to the development of distinct clinical entities including polycythemia vera (PV), essential thrombocythemia (ET) and primary myelofibrosis (PMF) [1][2][3]. JAK2 V617F mutation, resulting in constitutive activation of JAK-STAT signaling, occurs in about half of the patients with ET and PMF but in more than 90% of patients with PV [1].
Gene methylation is an alternative mechanism of gene inactivation, and various tumor suppressor genes regulating the cell cycle, apoptosis and cell signaling have been shown to be hypermethylated in hematological malignancies [4].
Hypermethylation of miR-34a, a transcriptional target of p53, has been demonstrated in solid and hematopoietic cancers [8,9], whereas restoration of which will inhibit CDK6 translation by complementary binding to the 3' untranslated region (3' UTR) of the CDK6 mRNA and induce apoptosis, thereby showing the tumor suppressor role of miR-34a [8]. Epigenetic inactivation of miR-34b, another p53 downstream target of the miR-34 family, has also been implicated in acute myeloid leukemia (AML), and the re-expression of miR-34b led to suppression of CREB expression and inhibition of cell proliferation [10]. Promoter methylation of miR-124-1, the first tumor suppressor miR found to be regulated by DNA methylation, has been shown to confer poor prognosis of acute lymphoblastic leukemia (ALL) [11]. Moreover, hypermethylation of miR-203 has been reported in chronic myeloid leukemia (CML), conferring a proliferative advantage to the tumor cells by inhibiting the oncogenic BCR-ABL fusion protein [12]. In Ph-ve MPN, little is known about the epigenetic alteration of miR methylation. In this report, we studied the methylation status of miR-34a, miR-34b/c, miR-124-1 and miR-203 in PV, ET and PMF.

Methylation-specific polymerase chain reaction (MSP)
DNA was extracted from bone marrow samples at diagnosis and from cell lines by standard method. MSP for aberrant gene promoter methylation was performed as previously described [13,14]. Treatment of DNA with bisulfite for conversion of unmethylated cytosine to uracil (but unaffecting methylated cytosine) was performed with a commercially available kit (EpiTect Bisulfite Kit, Qiagen, Germany). Primers used for the methylated MSP (M-MSP) and unmethylated MSP (U-MSP) were shown in Table 2. DNA from normal bone marrow donors was used as negative control, while enzymatically methylated control DNA (CpGenome Universal Methylated DNA, Chemicon) was used as positive control in all the experiments. MSP was performed in a thermal cycler (9700, Applied Biosystems, Foster City, CA) with the following cycling conditions: 95°C for 5 minutes, specific cycles of 95°C for 30 seconds, specific annealing temperature for 30 seconds (Table 2), 72°C for 30 seconds, and a final extension of 10 minutes at 72°C. The MSP mixture contained 50 ng of bisulfite-treated DNA, 0.2 mM dNTPs, MgCl 2 (Table 2), 10 pmol of each primer, 1 × PCR buffer, and 2.5 units of AmpliTaq Gold DNA Polymerase (Applied Biosciences, Foster City, CA) in a final volume of 25 μl. Ten microliters of PCR products were loaded onto 6% nondenaturing polyacrylamide gels, electrophoresed, and visualized under ultraviolet light after staining with ethidium bromide.   [15]. Conventional RT-PCR for primary miR-34a was performed as previously described [9].

Statistical analysis
Correlation between combined miR methylation status (i.e. those with methylation of any of these four miRs) with categorical variables (gender, MPN subtype, occurrence of myeloid transformation, development of thrombosis [either at diagnosis or after diagnosis of MPN] and presence of JAK2 V617F mutation) and continuous variables (age, presenting Hb, leukocyte and platelet counts) was computed by the Chi-square test (or Fisher Exact test) and Student's T-test. All p-values were two-sided.

MSP Controls
Direct sequencing of the M-MSP products from the methylated positive control confirmed the MSP specificity and complete bisulfite conversion, which methylated cytosine remained as cytosine upon sequencing (underlined) while unmethylated cytosine appeared as thymidine ( Figure 1A).

Cell lines
MSP analysis of the four cell lines showed that miR-34a was hemizygously methylated in MEG-01 and K-562 and completely unmethylated in HEL and SET-2. miR-34b/c was completely methylated in HEL, hemizygously methylated in MEG-01 and completely unmethylated in K-562 and SET-2. miR-124-1 was completely unmethylated in all the four cell lines. miR-203 was completely unmethylated in K-562 and SET-2. However, there was neither U-or M-MSP signals of miR-203 in both HEL and MEG-01, suggesting a possibility of homozygous deletion (Table 3; Figure 1C).

5-AzadC treatment of HEL cells
Untreated HEL cells were homozygously methylated for miR-34b/c (Chr. 11q23). After 5-AzadC hypomethylation treatment, miR-34b/c U-MSP signal emerged, together with about 3-fold increase in expression of both mature miR-34b and miR-34c on day 7 as analyzed by stem-loop RT-PCR ( Figure 3). On the other hand, miR-34a was unmethylated in HEL. By conventional RT-PCR of the primary transcript of miR-34a, pri-miR-34a was not constitutively expressed, and treatment with 5-AzadC did not lead to expression of primary miR-34a (Additional file 1).

Discussion
Little information is available about the role of miRs in MPNs. Based on a literature search on the PubMed using keywords of "microRNA, methylation and myeloproliferative", no publication could be found. Therefore, this is likely the first report on methylation of miRs in MPN. In this study, we examined the methylation of miR-34a, -34b/c, -124-1 and -203 for a range of MPNs. In an attempt to identify miRs specifically involved in erythropoiesis, miR expression from in vitro expansion of erythroid cells derived from peripheral blood mononuclear cells were compared with controls, which showed miR-451 upregulation was specific to erythropoiesis [16]. On the other hand, during in vitro differentiation of megakaryocytes derived from CD34+ hematopoietic progenitors, downregulation of miR-10a, -10b, -17, -20, -106 and -126 was observed [17]. Moreover, when comparing the miR expression profiling of megakaryoblastic cell line with in vitro differentiated megakaryocytes, upregulation of miR-99a, -101, -126, and -135 was also found [17]. Therefore, the upregulation and downregulation of these specific miRs were associated with megakaryocytic and erythropoietic differentiation. Despite that multiple TSGs were shown to be silenced by hypermethylation in AML [4,14,[18][19][20][21][22], methylation of tumor suppressor genes was less frequently reported in MPN. Only recently, there were a few reports of methylation of the negative regulators of the JAK-STAT signaling pathway, SOCS1, SOCS2 and SOCS3, in Ph-ve MPN regardless of the JAK2 V617F mutation status [23][24][25][26].
Furthermore, based on previous work by us and others, CDK6 has been shown to be the target of multiple miRs including miR-34a, -34b/c and -124-1 [27,28]. Moreover, CREB is the target of miR-34b and -203 [10,29]. Finally, ABL is an additional target of miR-203 [12]. Consequently, epigenetic silencing of tumor suppressor miR-34a, miR-34b/c, miR-124-1 and miR-203 will confer proliferative advantage to the tumor cells [27][28][29]. In contrast to a previous report which showed miR-203 was methylated in Ph +ve but not Ph-ve MPN or leukemia, using MSP primers in the similar region [12], we demonstrated that miR-203 was hypermethylated in primary MPN samples, which was further verified by direct sequencing of the methylated MSP products. Therefore, it would appear that miR-203 methylation is involved in a wider spectrum of MPNs or leukemias, regardless of their Ph chromosome status. Finally, while two patients had concomitant methylation of miR-203 and -34b/c, none had concomitant methylation of miR-34a and -34b/c, both transcriptional targets of p53, and hence avoiding duplication of tumor suppressor gene inactivation of the same pathway.
In HEL and MEG-01 cells, both U-and M-MSP signals of miR-203 were absent, which might be due to the following possibilities: (1) sample DNA degradation, (2) inappropriate PCR condition, or (3) homozygous deletion of the region. Since simultaneous U-MSP analysis of the same DNA sample for miR-34a, miR-34b/c, and miR-124-1 promoter successfully generated the U-MSP signals, hence the absence of MSP signals for miR-203 in HEL and MEG-01 cells could not be explained by a poor DNA quality. Moreover, as miR-203 U-MSP was successful in all the other samples including cell lines, normal controls, patient samples and methylated positive control, inappropriate MSP conditions appears unlikely. Therefore, the absence of both M-and U-MSP signals in HEL and MEG-01 cells might be caused by deletions of the region. However, karyotypic data of HEL and MEG-01 cells did not reveal homozygous deletion of 14q32, and hence whether absence of MSP amplification of miR-203 might be due to microdeletion of this region requires further study [30]. Furthermore, hypomethylation treatment of the HEL cells, which was homozygously methylated for the miR-34b/c, a microRNA cluster localized to 11q23, showed significant re-expression of mature miR-34b and miR-34c. This  finding is consistent with that both miR-34b and miR-34c are under the promoter regulation of the same CpG island. By contrast, miR-34a, another member of the miR-34 family localized to 1p36, was not constitutively expressed. Moreover, hypomethylating treatment did not lead to expression of primary miR-34a, suggesting additional mechanism, possibly histone modification, in the regulation of miR-34a expression [9]. Furthermore, in addition to showing miR silencing in cell line, it is important to show the correlation of miR methylation and miR expression in the primary sample. In this connection, we shall collect both DNA and RNA from diagnostic bone marrow samples in the future. In order to identify miRs that are methylated specifically at the time of transformation to AML or MDS, and hence implicated for pathogenesis of myeloid transformation, ideally one should analyze the paired marrow samples at both diagnosis and leukemic/myelodysplastic transformation. This is exemplified by our recent publication in the study of epigenetic inactivation of miR-34b/c methylation in myeloma, in which we showed that while miR-34b/c is not methylated at diagnosis, it is frequently methylated at the time of relapse or disease progression. This is as evidenced by the significantly more frequent methylation of miR-34b/c of myeloma samples at relapse in patients with both diagnostic and relapse marrow samples [31].
Finally, unlike the association of TSG with clinical parameters, such as the association of CDKN2B and WIF1 methylation with high presenting leukocyte count in acute promyelocytic leukemia [32,33], methylation of these miRs did not correlate with demographic, presenting blood counts, JAK2 V617F mutation or complications including thrombosis and myeloid transformations.

Conclusion
This is the first report of miR hypermethylation in MPNs. miR-203 hypermethylation is not specific to Ph +ve leukemias but also present in Ph-ve MPNs. miR-34b/c methylation was associated with reversible miR silencing. There was no correlation of miR methylation with clinical demographic data or outcome.