Hpa II-McrBC whole-genome-amplification PCR (HM-WGA-PCR) uses two types of restriction enzyme. One is either Hpa II (CCGG) or Hha I (GCGC), which digests unmethylated DNA, but not methylated DNA (i.e., methylation-sensitive enzymes). The other type is McrBC (RmCN40-80RmC), which cuts methylated DNA, but not unmethylated DNA (i.e., methylation-dependent enzyme) [13]. In HM-WGA-PCR, genomic DNA is first digested with either Hpa II/Hha I or McrBC and then subjected to whole-genome-amplification by phi29 DNA polymerase [14], which can uniformly amplify genomic DNA (Figure 1). Finally, PCR amplification of each amplicon is conducted by using the two whole-genome-amplified DNA samples as the templates (i.e., one from Hpa II/Hha I digestion and the other from McrBC digestion) (Figure 1).
Note that each amplicon is selected to contain both Hpa II/Hha I and McrBC sites. If an amplicon is biallelically methylated (i.e., complete methylation), it is amplified only from Hpa II-digested genomic DNA, but not from McrBC-digested DNA (Figure 1). Conversely, if an amplicon escapes methylation biallelically (i.e., null methylation), it is amplified only from McrBC-digested DNA, but not from Hpa II-digested DNA (Figure 1). If an amplicon is composed of both methylated and unmethylated alleles (i.e., composite methylation), it is amplified from both templates, because Hpa II/Hha I and McrBC fail to digest methylated and unmethylated alleles, respectively (Figure 1). If an amplicon is partially methylated on both alleles (i.e., incomplete methylation), it is amplified from neither template, because both enzymes digest both alleles (Figure 1). Accordingly, four different allelic methylation patterns (i.e., complete, null, composite, and incomplete methylation) can be clearly discriminated as four distinct amplification patterns (Figure 1).
HM-WGA-PCR is based on HM-PCR, which includes only two steps in Figure 1, namely "Digestion" and "PCR" [8]. A drawback of HM-PCR is that it requires 200 ng of genomic DNA to analyze methylation in one amplicon; 50 ng each for Hpa II-, McrBC-, Msp I- and mock-digestion, the latter two of which serve as controls. Therefore, it is not suitable for a comprehensive analysis of methylation when the quantity of available genomic DNA is limited. In HM-WGA-PCR, we input 500 ng of genomic DNA for each digestion and used 1/10th of the digested DNA (i.e., 50 ng) for WGA, which likely yielded ~5 μg of amplified DNA. We confirmed that 1/100th or 50 ng of the WGA product, which is equivalent to 0.5 ng of original genomic DNA, is enough for each PCR.
Since HM-WGA-PCR is basically identical to HM-PCR except for the WGA process, both methods should provide the same results on the same amplicon. We thus examined whether or not HM-WGA-PCR can distinguish between the four allelic methylation patterns as clearly as HM-PCR. For this purpose, we used primer pairs for eight CGIs on human chromosome 21q (#85, #95, #100, #102, #106, #112, #120, and #142) as representatives of the four allelic methylation patterns (Additional file 1: Supplemental Table S1). As shown in Figure 2A~2D, HM-PCR and HM-WGA-PCR provided the same patterns. In other words, both HM-PCR and HM-WGA-PCR determined that CGIs #85 and #120, #100 and #102, #112 and #142, #95 and #106, are modified by complete-, null-, composite-, and incomplete-methylation, respectively. These results suggested that HM-WGA-PCR can distinguish the four methylation status patterns on CGIs as distinctly as HM-PCR.
To test the robustness of HM-WGA-PCR more rigorously, we applied it to the 147 CGIs on human chromosome 21q that had been analyzed in our previous study (Additional file 1: Supplemental Table S1). While we previously analyzed 149 CGIs in total, we omitted CGIs #103 and #75 from this study for the following reasons. The former CGI lacks recognition sites for methylation-sensitive restriction enzymes and hence cannot be analyzed by HM-PCR; it had been analyzed by bisulfite sequencing in our previous study [8]. The latter CGI was omitted, because updated reference sequence data disproved the validity of the primers used for this CGI in the previous study (data not shown) [8].
We applied HM-WGA-PCR to the 147 CGIs and successfully amplified all of them, suggesting largely unbiased amplification by WGA. Furthermore, in every case of the 147 CGIs, allelic methylation status revealed by HM-WGA-PCR was exactly the same as that revealed by the HM-PCR; we thus achieved 100% consistency between the two methods (data not shown). Note that these HM-WGA-PCR assays consumed only 300 ng of genomic DNA, whereas our previous HM-PCR study of the same 147 CGIs had required 30 μg.
We applied HM-WGA-PCR to several CGIs on other chromosomes, including CGIs #160, #592, and #610 on chromosome 11q (Figure 2E), and confirmed that the results of HM-PCR and HM-WGA-PCR are consistent in all the cases examined (data not shown). We also confirmed that HM-WGA-PCR works well on placental DNA to give consistent results with our previous HM-PCR (data not shown).
Taken together, HM-WGA-PCR is much more efficient than HM-PCR and retains comparable accuracy to HM-PCR. It would thus provide a reliable alternative to HM-PCR, when examining allelic methylation status using a limited quantity of DNA.
Allelic methylation status can be also examined by methods based on bisulfite conversion of unmethylated cytosine (e.g., bisulfite sequencing, COBRA [15], and methylation-specific PCR [16]). While these methods as well as HM-WGA-PCR require a few ng of DNA for each assay and are thus comparable in terms of the amount of DNA required, each of them has its own advantages and drawbacks. For instance, bisulfite sequencing has single nucleotide resolution, but requires tedious steps for sequencing and, often, cloning. By contrast, PCR-based methods are much easier to perform, although their resolution is limited. Among the PCR methods, amplicons for bisulfite-based ones are forced to be short due to the effect of bisulfite-induced DNA fragmentation, whereas those for HM-WGA-PCR are free from such concern and can be much longer. On the other hand, HM-WGA-PCR can interrogate only the regions bearing both Hpa II/Hha I and McrBC sites, whereas bisulfite-based PCR methods are free from such restriction. Incomplete bisulfite-conversion leads to false methylation signals in bisulfite-based methods, and incomplete digestion leads to false-positive PCR signals in HM-WGA-PCR. In anyway, HM-WGA-PCR complements bisulfite-based methods and would be a method of choice in various applications for its simplicity.