DNA methylation is an important epigenetic mechanism involved in fundamental biological processes such as development, imprinting, and carcino-genesis. For these reasons, DNA methylation represents a valuable source for cancer biomarkers. Methods for the sensitive and specific detection of methylated DNA are a prerequisite for the implementation of DNA biomarkers into clinical routine when early detection based on the analysis of body fluids is desired. Here, a novel technique is presented for the detection of DNA methylation biomarkers, based on real-time PCR of bisulfite-treated template with enzymatic digestion of background DNA during amplification using the heat-stable enzyme Tsp509I. An assay for the lung cancer methylation biomarker BARHL2 was used to show clinical and analytical performance of the method in comparison with methylation-specific PCR technology. Both technologies showed comparable performance when analyzing technical DNA mixtures and bronchial lavage samples from 75 patients suspected of having lung cancer. The results demonstrate that the approach is useful for sensitive and specific detection of a few copies of methylated DNA in samples with a high background of unmethylated DNA, such as in clinical samples from body fluids.

Introduction

DNA methylation plays an important role in regulating cellular differentiation and development (1). As aberrant DNA methylation of specific loci is also linked to pathologic processes like carcinogenesis (for review, see Reference 2), the analysis of DNA methylation–based biomarkers represents a promising method for tumor diagnosis. The use of such biomarkers in body fluids (e.g., blood or urine) of cancer patients might allow for early detection of the disease (3,4,5,6). However, these samples usually contain high concentrations of background DNA from normal cells and only a small portion of DNA from the tumor. Therefore, one major challenge for the application of methylation biomarkers in diagnostic routine is the sensitive detection of single copies of methylated DNA in a high amount of unmethylated background DNA. Several methods exist that are applicable for DNA methylation analysis. However, most of these techniques do not allow for sensitive detection.

As of now, PCR analysis after bisulfite conversion is the method of choice when sensitive detection is desired (7). Bisulfite treatment of the template DNA leads to deamination of unmethylated cytosines to uracil, leaving only methylated cytosines unaltered and thereby transforming epigenetic information into sequence information. Methylation-specific PCR (MSP), HeavyMethyl, and Headloop PCR are methods that allow for the detection of single copies of methylated DNA. In MSP, primers bind specifically to bisulfite-converted methylated or unmethylated DNA, leading to specific amplification (8). Advantages of this technique are its high relative sensitivity, ease of design, and low complexity of the reaction. However, for certain loci, primer design can be challenging for reaching acceptable specificity. HeavyMethyl technology is based on amplification of bisulfite-treated DNA using methylation-unspecific primers in combination with methylation-specific blocking oligonucleotides (9). In HeavyMethyl PCR, the high number of different oligonucleotides per assay can be challenging with respect to assay design. However, the possibility to adapt blocking conditions leads to increased assay flexibility and allows for the selective amplification of even such loci where specific MSP is hardly possible. In Headloop PCR, the amplification of specific sequences after bisulfite treatment is suppressed through an extension at the 5' end of the primers (10). After synthesis of the strand including the extension, the 5' end can internally bind if the amplicon contains the respective recognition sequence, which prevents further amplification. This technique is highly sensitive and specific, but primer design and the optimization of PCR conditions for specific loop formation are challenging. Methods that allow for sensitive detection—especially at loci where MSP and HeavyMethyl assay designs are difficult—would offer a significant improvement for DNA methylation analysis.

In this paper, a new technique for the sensitive and specific amplification of methylated DNA is presented, which is based on the digestion of bisulfite-converted, unmethylated DNA during each PCR cycle via the restriction endonuclease Tsp509I. Tsp509I, which specifically cuts at AATT sites, was previously used to analyze the efficiency of DNA bisulfite conversion (11). The specific cleavage of unmethylated template is based on the presence of CpG dinucleotides in the sequence context AATCG or AACCG within the region of interest. After bisulfite conversion, these sequences will either be converted to AATCG (CpG methylated) or AATTG (CpG unmethylated). Therefore, only unmethylated CpG sites in the described sequence contexts result in a recognition site after bisulfite treatment, enabling the digestion of the template. Due to its optimal incubation temperature of 65°C, heat stability, and high activity under different buffering conditions, it is possible to use Tsp509I directly in the PCR reaction. During each PCR cycle, the enzymatic activity leads to cleavage of PCR products arising from unmethylated bisulfite-converted template, whereas methylated sequences remain unaffected.

In general, DNA restriction is a common technique for methylation analysis. Methylation-sensitive endonucleases have previously been used to digest non-bisulfite treated unmethylated template prior to PCR (12,13). The major disadvantage of this pre-PCR restriction is that even minor remnants of non-digested DNA will be amplified, leading to false-positive results. However, such by-products are inevitable, as restriction enzymes do not reach complete digestion, especially when the DNA is fragmented or partially single-stranded, as is expected in body fluids. Accordingly, the sensitive and specific detection of methylated DNA is limited. In combined bisulfite restriction analysis (COBRA), bisulfite-treated DNA is subjected to a single restriction step after PCR (14). In COBRA, all DNA is amplified, without a preferred (biased) amplification of methylated DNA. Thus, samples that originally contained only traces of methylated DNA will also only contain minor portions of originally methylated DNA in the PCR product. These small amounts cannot easily be detected via enzymatic restriction. The new Tsp509I assay overcomes these limitations via digestion of unmethylated background during PCR cycling. This leads to highly sensitive and specific detection of methylated DNA as shown in comparative analysis with MSP on predefined DNA mixtures. The high performance of the new technique was confirmed on patient sample material in a case-control study with bronchial lavage samples from 75 patients examined for suspected lung cancer using the lung cancer DNA methylation biomarker BARHL2 (Bar-like homeobox) (15,16).