DNA methylation analysis of benign and atypical meningiomas: correlation between RUNX3 methylation and WHO grade

Purpose Although meningiomas are common central nervous system tumors, the biomarkers allowing early diagnosis and progression are still needed. The aim of this study was to evaluate the methylation status of 12 cancer-related genes, namely ERCC1, hMLH1, ATM, CDKN2B (p15INK4B), p14ARF, CDKN2A (p16INK4A), RASSF1A, RUNX3, GATA6, NDRG2, PTEN, and RARβ, in 44 meningioma samples of WHO grade I and II. Methods All genes were analyzed using methylation-specific polymerase chain reaction, while pyrosequencing (PSQ) was used to study NDRG2 promoter methylation. Results The most frequently methylated genes in both types of meningiomas were p14ARF, RASSF1A, and p15INK4B. RUNX3, GATA6, and p16INK4A were methylated to a lesser extent, whereas ATM and RARβ were found to be methylated in a marginal number of patients. The ERCC1, hMLH1, NDRG2, and PTEN genes were unmethylated in all cases. Although tumors of the same grade according to WHO criteria had different genes methylated, the number of methylated genes for each individual patient was low. RUNX3 methylation significantly correlated with meningioma WHO grade, therefore, can be considered as a potential indicator of tumor aggressiveness. The sequence of NDRG2 chosen for PSQ analysis was found methylated in the majority of meningiomas; however, the methylation level was only slightly elevated as compared to non-cancerous brain. Conclusions Overall, the results of this study confirm that DNA methylation plays an important role in the pathogenesis of meningiomas. Further investigations, particularly concerning RUNX3 methylation, are necessary in order to assess the clinical usefulness of the methylation analysis of the studied genes.


Introduction
The most common tumor site in the central nervous system is the meninges (35 %) and meningiomas, with the incidence rate of 7.22 per 100,000 persons, are the most common primary brain tumors (Dolecek et al. 2012). Incidence rate for tumors of the meninges is 2.2 times greater in females as compared to males, and the peak incidence occurs in the sixth and seventh decade of life (Dolecek et al. 2012;Lamszus 2004). They are classified according to the World Health Organization (WHO) grading system 1 3 treatment is therefore needed. In addition to alterations in the DNA sequence, epigenetic modifications are closely linked to cancer initiation and progression. Aberrant DNA methylation is one of the major types of epigenetic modifications in cancer (Paluszczak and Baer-Dubowska 2006). Changes in DNA methylation appearing during early stages of malignancy are potentially reversible and in some cases correlate well with tumor characteristics. For instance, TIMP3 inactivation as a result of gene promoter methylation appears to be involved in meningioma progression as it is associated with more aggressive, high-grade meningioma phenotype (Barski et al. 2010). Recently, three genomewide methylation analyses of benign, atypical, and malignant meningiomas were performed (Kishida et al. 2012;Vengoechea et al. 2013;Gao et al. 2013). All concluded that aberrant hypermethylation of subset of genes occurs in the early stage of meningiomas and highlighted the possibility of malignancy evaluation by assessing methylation status. However, a crucial methylation marker to predict an aggressive subtype in the early stage of meningiomas remains to be established, as existing data are still limited and often inconsistent. Our previous study has shown that a panel of 12 genes including ERCC1, hMLH1, ATM, CDKN2B (p15INK4B), p14ARF, CDKN2A (p16INK4A), RASSF1A, RUNX3, GATA6, NDRG2, PTEN, and RARβ might be useful in evaluation of glioma aggressiveness. In this regard, methylation index (MI) of these genes in individual patient, as well as RUNX3 methylation correlated with glioma WHO grade and higher MI, or aberrant methylation of RUNX3 indicated more aggressive tumors (Majchrzak-Celińska et al. 2015).
The aim of the present study was to evaluate the possible involvement of the hypermethylation of these genes in the malignant profile of benign and atypical meningiomas.

Patients
The study included 44 patients who were referred to the Department and Clinic of Neurosurgery and Neurotraumatology, Poznan University of Medical Sciences, Poznań, Poland, between January 2010 and September 2013. Patients with a histologically confirmed diagnosis of meningioma at different grades of malignancy (I, I/II or II) and a treatment protocol that included surgical resection were enrolled in the study. Histological types of the tumors as well as tumor grades (according to the 2007 WHO classification criteria) were analyzed in the Laboratory of Neuropathology. Thirty-three patients were diagnosed with benign WHO grade I meningioma, two with grade I/II, and nine with atypical meningiomas (grade II). Women comprised 72.73 % (32/44) of all patients, and the average patients' age was 56 years, ranging from 31-87. After resection, tumor samples were directly frozen and stored at −80 °C.
Control non-cancerous brain sample originated from a patient with brain hematoma, and blood sample for the analysis of white blood cells methylation was taken from healthy donor. All patients gave informed consent for the analyses to be undertaken, and the study protocol was approved by the Poznan University of Medical Sciences Clinical Research Ethics Committee (approval no. 505/12).
Isolation of DNA from tumor tissue DNA was isolated using GeneMATRIX Tissue DNA Purification Kit [EurX, Gdańsk, Poland]. Briefly, 25 mg of tumor tissue was incubated with Lyse T Buffer, then RNase A and proteinase K, which was followed by incubation in 56 °C until the tissue was completely digested. Buffer Sol T and ethanol were then added to provide selective conditions for DNA binding during brief centrifugation, and then, DNA bound to minicolumn was washed twice to remove the traces of contaminants. DNA was eluted with Tris buffer, pH 8.5; its concentration and purity were measured using NanoDrop spectrophotometer. DNA was stored at −20 °C for further analysis.

Methylation-specific polymerase chain reaction (MSP)
Using MSP assay, the methylation status of ERCC1, hMLH1, ATM, CDKN2B (p15INK4B), p14ARF, CDKN2A (p16INK4A), RASSF1A, RUNX3, GATA6, NDRG2, PTEN, and RARβ was assessed. Bisulfite modification of 500 ng DNA, preceding MSP, was performed using the EZ DNA methylation kit [Zymo Research Corp., USA, D5002], following manufacturer's protocol. Commercially available completely methylated control DNA [CpG Methylated HeLa Genomic DNA, New England Biolabs], completely unmethylated control DNA [EpiTect Control DNA, unmethylated, Qiagen], and water served as positive, negative, and blank controls of MSP assay, respectively. DNA samples derived from the white blood cells from healthy volunteers as well as a brain fragment from a patient with brain hematoma served as normal control samples for DNA methylation analysis.
MSP assay was carried out in MyCycler Thermal Cycler with Gradient [Bio-Rad]   The reaction conditions differed for each type of polymerase used, but generally after an initial step of heat denaturation at 95 °C for 1-4 min. (depending on the type of polymerase used), 35-38 cycles were carried out (95 °C for 15-30 s, appropriate annealing temperature for 30 s and 72 °C for 10-45 s), which were followed by final elongation at 72 °C for 5-7 min. The PCR products were separated by electrophoresis in 2 % agarose gel stained with ethidium bromide and visualized under UV illumination.

Pyrosequencing (PSQ)
Pyrosequencing (PSQ) of NDRG2 was performed on a PyroMark Vacuum Prep Workstation and Pyrosequencer PSQ™ 96 ID system, using PyroQ CpG™ software 1.0.9 (Biotage, Uppsala, Sweden), for quantification of the methylation level. Bisulfite-treated DNA samples were first amplified using FastStart Taq DNA Polymerase [Roche Diagnostics], with the forward primer: GGGCGGTGT-GATTGATTC, and reverse, biotinylated primer: biotin-CCCACAATCTTCTCCCGTTCT (primers were designed using PyroQ CpG™ software 1.0.9). The 25 μl of reaction mixture contained 5 mM MgCl 2 , 10 × buffer, 0.2 mM dNTPs, 5 U/μl polymerase, 0.4 μM forward primer, 0.2 μl reverse primer, and water. The reaction conditions were as follows: 95 °C 4 min., 45 cycles of 95 °C for 30 s, 58 °C for 30 s, 72 °C for 45 s, followed by 72 °C for 7 min. The PCR product had 327 bp, and 2 μl of the product was visualized after ethidium bromide staining on 2 % agarose gel; the rest of the PCR product underwent PSQ. The analyzed region is presented in Fig. 1. The sequencing primer was as follows: GGGGGAAAGGGTTAGTA.
As controls, completely methylated control DNA (CpG Methylated HeLa Genomic DNA, New England Biolabs), and completely unmethylated control DNA (EpiTect Control DNA, unmethylated, Qiagen), as well as DNA from normal, non-cancerous brain tissue were included in the assay. The reaction without any template DNA (blank) was also performed. Internal control for completion of the bisulfite treatment was included in the sequence. The cutoff value to distinguish between methylated and unmethylated samples was established at 11 %, based on the results obtained on control non-cancerous brain tissue, according to the previously described formula (Majchrzak-Celińska et al. 2015;Håvik et al. 2012).

Statistical analysis
Css Statistica, version 10, was used for statistical analysis. Fisher's exact test was used to determine the correlations between methylation frequency and WHO grade of the tumors (Stat Exact program). Spearman's rank correlation coefficient was used to measure the dependence between MI and WHO grade of the tumor, as well as patient's age. Student's t test was applied to determine the relevance of Pearson's R coefficient. Chi-squared test and Pearson's R coefficient were used to determine the correlation between methylation analysis results and patient's age. Kruskal-Wallis test was used to determine the dependence between histological type of the tumor and MI, differences in methylation  GGTTTTTTGGGTTTTAGTTTAGGATATTGCGTTTTTTTTAAGTTTTTATTTTATTTTCGT  AGGGTTAATTTCGTTTTTGCGTTTGTTTATTTAGGAGTTAGAGTTTTTGGGGGTTTTCGG  TTTTTTTTTCGCGTTTTTTGAGGTATTGATTTTAGAGTTTTTGTTGGTTATTTTTTTTTT  ATTTTTCGTTTGTTCGCGATCGTTTTTTATTTATAGCGGTTTTTCGTATTTTTCGTTTTT  ATTTTTTCGTTTTGTTGTTTAATTTTTTTCGGAGTAGTCGGAGAGTAGGCGTCGGGACGT  AGTAAAGAGAGGAGAGGTATTAGGATTTTGGGTACGAGGGAGTCGGAATTTTTGCGTTTT  CGACGTTTTTTCGTGTTTTTTTTCGGCGTTAGTATATTTGTTTTTGTTAGTCGGCGAGAT  TTTCGTTTTTTTAGCGGCGGTTTCGTTTATTTTTTTTTTTTTTAATTTAGTTTTTTGTGT  TCGGGCGGGGGGAGTCGAATTCGTGGGGTGTTAGTATTTCGGGTTCGTTTTTTTATTATT  GGTTTTTATTATTGAGTTTGGGGGCGAAGGAAGGAGTTAGAGTGTAATTGTAGATTTAGG  TCGTGGATGGGGCGGTGTTGAGGGTAGTCGAGGGATTTTTTTTAGGGTTGAGGGATTTTT  GTATTTTTTATATCGTTTTAATGCGGGGAGGGGGTGGTGGAATGTTTTGGTGAGTCGAGT  AGAGTTTGGAAGCGTTTAGTCGTTCGTTTTTCGTTTTTCGGTTTTAGTATAATTTTTTTT  TGAGCGGTAGTTTGGGCGGGGTGAAAGGGGGGGCGTGTTTATTGGGGGTTTGGGGCGGGT  TTGCGGGGAGTTTAGCGTTCGTGGGCGGGTTGGGGCGGGTCGGGGGCGGGTTTGCGGGAA  GTTCGAGTCGGGGCGGGTTCGAGTTAAAGGTAAGTGAAGGTGGAAGCGGTCGCGGCGGTA  GTAGGTAGGGGGAGGGGTAGGCGAGGGGTCGTAGGGTTTGGAAGGCGTTAGTCGGGTCGG  CGGGCGGTGTGATTGATTCGCGTTTTTTGGAGTTGGAGGTTCGGGGGAAAGGGTTAGTAC  GGAGCGGGCGTTCGGTTGTTGCGTATAAAGGTTGAGGTTTTAAGAGTTGTAGGGCGTGTT  TGGGGTGCGCGCGAGGTTGTGTGTAAGGTTCGGGGCGTTTGGTATCGGTTACGTAAGGGT  GTTAGTTTTTACGCGGACGGGTTGCGTAGGGATTCGTAGGGACGGGTTTCGGGTGTTGGG levels determined by PSQ between tumors of different WHO grade and the dependences between methylation levels, when analyzed with PSQ and patient's age. Mann-Whitney U test was used to determine the statistical significance of methylation levels in each CpG site in terms of patient's sex.

NDRG2 pyrosequencing (PSQ)
Twenty-five meningioma samples were analyzed using PSQ. Sixty-eight percent of the samples (17/25) represented benign tumors (WHO grade I); however, one sample was classified as I/II grade, and seven were classified as grade II. Representative pyrogram is shown in Fig. 3. Totally, 84.00 % (21/25) of tumors showed a mean methylation signal above the level represented by normal, non-cancerous brain tissue. The highest methylation level was found for CpG4, whereas the lowest for CpG1. The average methylation of the five CpG sites analyzed in the methylated group were as follows: CpG1-8.17 %, CpG2-14.05 %, CpG3-17.14 %, CpG4-27.60 %, and CpG5-14.09 %. The mean methylation levels for tumor WHO grade I and I/II were 15.90 % (±5.53 %) and for grade II 13.71 % (±8.41 %) and thus could not differentiate between tumor WHO grades. The relationship between average NDRG2 methylation level in each CpG dinucleotide and WHO grades of analyzed meningiomas is presented in Fig. 4. Kruskal-Wallis test revealed that the average methylation level of all five CpG sites correlates with patient's age-higher methylation levels were observed in older patients, as compared with younger ones (p = 0.03). According to Mann-Whitney's U test, women had statistically higher methylation levels in CpG2 as compared to men (p = 0.01).
The comparison of NDRG2 promoter methylation profile obtained by MSP and PSQ revealed that 4 out of 25 samples gave concordant results when analyzed using MSP and PSQ, with the cutoff value determined to be 11 %. The high discordant rate was expected since two different fragments of the promoter were analyzed (the distance between them was 643 nt).

Discussion
Recent genome-scale DNA methylation analyses of different subtypes of meningiomas although provided novel 1 3 insights into the molecular pathophysiology of malignant transformation did not indicate the definite markers of the development of these tumors (Kishida et al. 2012;Gao et al. 2013). Therefore, a crucial methylation marker to predict a greater malignant potential of meningiomas still remains to be established.
In our study, DNA methylation of a panel of genes was assessed in a group of patients with benign or atypical meningiomas. The panel included genes encoding DNA repair proteins (ERCC1, hMLH1, and ATM), cell cycle regulators (genes encoded in INK4B-ARF-INK4A locus, RASSF1A, and PTEN), and genes involved in cell differentiation and proliferation (RARβ, GATA6, NDRG2, and RUNX3). Our earlier study revealed that MI of these genes, i.e., the number of methylated genes divided by the number of all genes analyzed in the panel (Chen et al. 2011)   . The results of this study showed that such correlation does not occur in meningiomas. Although tumors of the same grade according to WHO criteria had differently methylated genes, the number of methylated genes for each individual patient was low, with the median of one and the maximum of four methylated genes out of 12. However, the analysis of individual genes methylation provided some interesting observations. The most important finding of our current study was the methylation of RUNX3, which was found in more than half of atypical meningiomas [55.56 %] and also in a subset of benign meningiomas [16.67 %]. More importantly, according to Fisher's exact test, RUNX3 methylation correlated with the presence of more aggressive, atypical meningiomas (p = 0.03). The Runt-related transcription factors (RUNX) are involved in developmental processes and, through multiple-interacting proteins, have been implicated in diverse signaling pathways and cellular processes (Chuang et al. 2013). The frequent inactivation of RUNX genes in cancer indicates crucial roles for RUNX in tumor suppression; however, they have also been characterized as oncogenes, depending on the context (Blyth et al. 2005). The role of RUNX3 methylation has been intensively studied in gastric cancers (Fan et al. 2011); however, data relating CNS tumors are very limited and apply only to small groups of patients (Avci et al. 2014). Thus, although the results of our present study require confirmation on larger patient cohorts, they suggest that RUNX3 methylation might be considered as the promising candidate biomarker for the assessment of meningioma aggressivenes.
Recently, significant differences in the expression of NDRG2 gene were reported between primary and recurrent meningioma groups, and benign and atypical meningiomas (Skiriute et al. 2011). Earlier, the downregulation of this gene was described in a subset of lower-grade meningiomas, including atypical meningiomas with clinically aggressive behavior, which was associated with hypermethylation of the NDRG2 promoter (Lusis et al. 2005). In this study, we analyzed NDRG2 promoter methylation using PSQ technique, because the sequence, reported to be a target of aberrant methylation in colon cancer (Piepoli et al. 2009) and used in our MSP analysis, was found unmethylated in all meningioma samples analyzed. The analysis of five CpG sites in NDRG2 promoter revealed methylation in a large subset of meningiomas. Eighty-four percent of tumor samples showed a mean methylation signal above the level found in normal, non-cancerous brain tissue. However, the mean methylation levels of five CpGs were below 20 % for all meningioma grades and whether this mild methylation has any significant impact on mRNA levels remains to be determined. However, higher methylation levels in CpG2 were observed in women than men, and the average methylation level of all five CpG sites correlated with patient's age. Thus, the results of our study confirm that mild NDRG2 methylation is a common event in meningioma formation, but clinical application of this observation seems to be limited. Moreover, PSQ analysis of a sequence within 5′UTR was proven to be a reliable method for its analysis.
Methylation of RASSF1A promoter is considered as an early and frequent event in tumorigenesis, and there are attempts to apply RASSF1A methylation as a diagnostic marker in cancer screening (Richter et al. 2009). The results of our study revealed that RASSF1A is methylated in 27.27 % of meningioma samples, and the frequency of gene promoter methylation is increasing with the tumor grade (25.00 % of tumors grade I and I/II, and 55.56 % of tumors grade II). Thus, we confirm the earlier findings of Nakane et al. (2007), which showed RASSF1A promoter methylation in 18.2, 63.6, and 42.9 % of the grade I, II, and III of meninigiomas, respectively. Overall, these results suggest that RASSF1A methylation is likely to play an important role in both meningioma development and progression. However, because this gene is frequently methylated in other malignancies (Richter et al. 2009), its role as meningioma biomarker is limited and should be considered only in a panel of other frequently methylated genes.
The panel of genes whose promoter methylation was analyzed in this study included also other commonly hypermethylated ones, such as those located in the INK4B-ARF-INK4A locus. Inactivation of the G1/S-phase cell cycle checkpoint was reported to be an important aberration in anaplastic meningiomas (Boström et al. 2001), and it was suggested that deregulations of p14-MDM2-p53 pathway may contribute to the malignant progression of those tumors (Amatya et al. 2004 Fig. 4 The relationship between average NDRG2 methylation level in each CpG dinucleotide and WHO grades of analyzed meningiomas. Each CpG dinucleotide is represented by different color (CDKN2A) (9.09 %). Methylation of these genes characterized both benign as well as atypical meningiomas. These findings indicate that INK4B-ARF-INK4A locus is a target of aberrant methylation in meningiomas and that at least in some of those tumors the loss of cell cycle control is due to epigenetic silencing of p14ARF, p15INK4B, and/or p16INK4A genes. According to the two-hit model of Knudson, DNA methylation can be a second hit inactivating those tumor suppressors, following genetic mutations which are also often present in meningiomas in this locus (Boström et al. 2001;Knudson 1996). Similar findings concerning DNA methylation in meningiomas were recently published by Aydemir et al. (2012), who found p16INK4A to be methylated in 8.3 % [3/36] of cases and Liu et al. (2005) reporting methylation of this gene in 10 % of samples. Amatya et al. (2004) found methylation of p14ARF gene in 5 of 58 cases of benign meningiomas (8.6 %), 2 of 10 cases of atypical meningiomas (20 %), and 2 of 4 cases of anaplastic meningiomas (50 %). However, the utility of INK4B-ARF-INK4A locus methylation in terms of clinically useful biomarker seems limited. Linsler et al. (2014) reported that methylation of p16INK4A is not associated with recurrence, higher grade, or chromosomal aberrations observed in meningiomas. What is more neither age nor sex had significant influence on tumor recurrence, with regard to the p16INK4A methylation status (Linsler et al. 2014).
Recently, GATA6 gene was found to be frequently methylated in glioma patients (Skiriute et al. 2012;Cecener et al. 2012;Martinez et al. 2009). Moreover, significant association of this gene methylation with unfavorable patient survival was found (Martinez et al. 2009). In meningiomas, to our best knowledge, the methylation of this gene has not been analyzed. Our results indicate that similarly to gliomas, GATA6 promoter is also methylated in meningiomas, however, with significantly lower frequency (13.64 vs 30-68 % in glioma). Thus, GATA6 promoter methylation in meningioma may play certain role in menigioma development, although cannot be considered as a biomarker candidate.
The other genes analyzed in this study, namely ATM and RARβ, were found to be methylated in a marginal number of patients (ATM-6.82 %, RARβ-2.38 %), while ERCC1, hMLH1, and PTEN were not methylated at all. The low frequency or lack of methylation of those genes suggests that their promoter methylation does not play a major role in the development of meningioma.
In summary, this study provided evidence that p14ARF, RASSF1A, p15INK4B (CDKN2B), RUNX3, GATA6, p16INK4A (CDKN2A), NDRG2 and to lesser extent ATM and RARβ promoter methylation are associated with the development of meningiomas. RUNX3 methylation correlates with meningioma WHO grade and, therefore, can be used as a potential indicator of tumor aggressiveness. Despite the methylation of NDRG2 in the majority of meningiomas, it cannot be considered as a diagnostic biomarker because its methylation level is only slightly elevated in comparison to non-cancerous brain tissue. Overall, the results of this study confirm that DNA methylation plays an important role in the pathogenesis of meningiomas. Further investigations, particularly concerning RUNX3 methylation, are required in order to assess clinically useful potential of the genes analyzed in this study.