We have performed our MSP
We have performed our MSP analyses as real-time PCR using SybrGreen (Fig. 2). The results of the beta-actin PCR vary in each sample group because the amount of DNA for bisulfite treatment was adjusted to a constant volume of serum (500μl) rather than to equal amounts of DNA. Because we expected lower amounts of methylated DAPK fragments than beta-actin DNA in serum we used different amounts of bisulfite-treated DNA—in particular 5-fold more DNA for our DAPK MSP. However we still measured lower amounts of DAPK than beta-actin in sera from ovarian cancer and leiomyoma patients (Fig. 2). IVF patients showed slightly lower CT-values for DAPK than for beta-actin corresponding to equal amounts if considering the differences in input DNA (Fig. 2).
To investigate the origin of methylated DNA in leiomyoma patients, we analyzed biopsies from 17 patients including 10 of the patients already analyzed for DAPK methylation in serum. However, only sera from 5 of these patients were evaluable as shown by beta-actin amplification (3 and 2 patients with DAPK positive and negative sera respectively). The frequency of DAPK methylation in primary tissue (35.3%) was higher than in serum (23.8%) but lower than the methylation frequency of ovarian cancer samples (50%, Table 2). These differences did not reach statistical significance as evaluated by the χ2-test (p>0.05). Both patients with DAPK-negative serum were also negative for DAPK methylation in the primary tissue. Moreover 2 of 3 patients with DAPK methylation in serum showed methylation in the analyzed tissue.
To control for MSP amplification bias, we analyzed PCR products from all DAPK positive sera from OvCa, leiomyoma and IVF patients. We used BstUI 8-CPT-Cyclic AMP sodium salt of cloned and uncloned MSP products as well as Sanger sequencing of plasmid clones to investigate the methylation state of internal CpG dinucleotides within the DAPK amplicon (Fig. 3). Importantly all MSP products from DAPK positive sera showed internally methylated CpGs in analyzed DNA fragments. The presence of the BstUI site (CGCG) in bisulfite treated DNA depends on the methylation state in the original DNA. Only fully methylated sites will be cut after bisulfite treatment. Restriction analyses showed completely digested clones from 10 OvCa and 5 leiomyoma samples (3 clones each, Fig. 3A). In contrast, the majority of IVF MSP products (2/3) were only partially or not digested by BstUI (Fig. 3B). Our sequencing experiments from cloned MSP products confirmed the BstUI restriction results. All sequenced OvCa and leiomyoma MSP products were completely methylated whereas DNA fragments of IVF patients showed partially methylated DAPK only (Fig. 3C).
Discussion Methylated DNA is widely studied to define biomarkers for gynecologic malignancies. However, conflicting results are reported for methylation frequencies (MF) of different genes in ovarian cancer . We have analyzed the methylation of DAPK in primary tissue and serum of ovarian cancer and leiomyoma patients. The estimated fraction of 50% methylated OvCa (Table 2) is within the range of published studies. Collins et al. detected DAPK methylation in 20/30 ovarian cancers (67%) . The clinico-pathological data of this cohort are similar to the characteristics of the herein presented patients. Most ovarian cancers were stage III serous carcinoma. However other studies found a lower or even absent methylation of DAPK in ovarian carcinomas. Caceres et al. and Theodoridis et al. [, ] also used methylation specific PCR but detected DAPK methylation in 3/16 OvCa (19% MF) and 0/106 OvCa respectively. Terasawa and colleagues  analyzed methylation by combined bisulfite restriction analyzes (COBRA) and reported an absence of DAPK methylation in 80 ovarian cancer samples. Therefore, both methodological discrepancies but also inherent differences of methylation frequencies in the study populations could contribute to observed discrepancies. If this reflects a common phenomenon studies using methylation markers for early detection, prognosis or treatment response should be validated in several populations.