In this article we will discuss about the analysis of promoter methylation using Denaturing High-Performance Liquid Chromatography (DHPLC).
Introduction to Analysis of Promoter Methylation using DHPLC:
Sequencing and analysis of the human genome revealed the presence of only 30,000 to 40,000 protein-coding genes, a stark contrast to previous estimates of about 100,000 genes. While a large number of putative tumor suppressor genes (TSG), which are down regulated in cancer development, could be identified by array technology, it remains largely problematic to demonstrate possible inactivation mechanisms for putative TSG identified.
A widely known and accepted mechanism of tumor suppressor gene inactivation in mammalian cell cancer is the methylation of CpG islands, particularly in promoter regions, leading to the formation of 5-methyl cytosine. Two techniques are commonly employed for the analysis of promoter methylation pattern.
Methylated and unmethylated genomic DNA can be distinguished by either the use of methylation-sensitive restriction enzymes or by conversion of genomic DNA with sodium bisulfite. Sodium bisulfite converts unmethylated cytosine (C) to uracil (U). Methylated C is not affected by the chemical treatment with bisulfite.
The effects of bisulfite conversion of genomic DNA can be evaluated by several more or less time consuming or sensitive assays. Such assays include subcloning and subsequent sequencing, restriction analysis, as well as other techniques.
Here we describe the use of Denaturing High Performance Liquid Chromatography (DHPLC) as an alternative screening technology for promoter methylation. The technology was first described by Baumer et al. in the context of investigating the methylation status of imprinted genes. DHPLC combines several advantages. It is a highly sensitive high-throughput technique.
The technology, which is applicable to fragments ranging in size from approximately 200 to 400 bp, not only facilitates the simultaneous analysis of multiple CpG sites, it also permits differentiation between different methylation patterns.
Thus, it becomes possible to distinguish between different methylation pattern such as single-site CpG methylation and complete methylation of all CpG sites within a given fragment. Consequently, verification and characterization of the methylation status on the base level has to be performed only on DHPLC-positive samples by PCR and/or subcloning followed by sequencing.
The fundamental principle of the DHPLC technique for mutation detection and methylation analysis is sequence-dependent denaturation of DNA fragments in the presence and absence of mismatches.
In the case of methylation analysis genomic DNA is initially treated with sodium bisulfite, which is followed by PCR amplification, and final analysis of PCR products under partially denaturing conditions. In the case where a single CpG-site is methylated, causing protection of that site from the effects of bisulfite treatment, the resulting PCR product will contain only one cytosine.
All unmethylated cytosines will have been converted to thymidine in the final PCR product. After denaturation/ renaturation differences in the methylation status of a given site will result in heteroduplex formation between respective PCR products. In the case where differences in methylation pattern are limited to a single site, the detection of these differences is equivalent to the detection of point mutations.
Complete methylation of all CpG-sites within an amplified region results in an increased GC content of the amplified DNA fragment after sodium bisulfite treatment compared to the PCR product derived from unmethylated genomic DNA.
Increased GC content results in increased stability and melting temperature of the PCR fragment analyzed, which in turn translates into increased retention times of respective PCR products during DHPLC analysis under partially denaturing conditions.
Analysis of genomic DNA representing different methylation pattern will result in PCR fragments of differing sequences after bisulfite treatment, each of which may or may not have a distinctive melting temperature and retention time. Furthermore, a more or less complex mixture of heteroduplexes is expected to form. This will result in complex DHPLC elution profiles reflecting the complexity of methylation pattern in the PCR-amplified region of interest.
Methylation Analysis by DHPLC:
Implementation and testing of DHPLC as a prescreening method for methylation analysis was performed on a 392-bp promoter fragment containing 35 CpG sites. In our opinion and based on our experience the size of a fragment analyzed should be between 200-400 bp, because fragments of less than 200 bp may not give satisfactory results, as the fragments melt over a very narrow temperature range.
Analysis of fragments larger than 400 bp may reduce the sensitivity of the method due to the multitude of heteroduplexes that may be formed. Moreover, it is difficult to amplify sequences greater than 600 bp in length from bisulfite-treated DNA due to bisulfite-induced strand breakages.
Primers and PCR conditions were optimized to amplify a PCR product from the promoter region of sodium bisulfite converted, methylated and unmethylated DNA. No PCR product needs to be amplified from untreated genomic DNA. Therefore, CpG dinucleotides in the primer sequence have to be avoided. However, several bases those are originally present as cytosines at the 3′-end of the primers will help ensure that unconverted sequences are not amplified.
Amplification of the promoter region by PCR benefits from the use of a low- fidelity polymerase. High-fidelity polymerases, usually preferred for DHPLC analysis, generally do not tolerate the presence of uracil in template DNA. Thus, high-fidelity polymerases are not suitable for the direct amplification from bisulfite-treated DNA, at least during the initial stages of the PCR.
On the other hand, a large number of PCR cycles or two consecutive rounds of PCR amplification are often required in order to generate sufficient PCR product for analysis due to low PCR efficiency. In order to nevertheless achieve a PCR product with a low overall rate of misincorporation, addition of a proofreading polymerase may be beneficial after the initial cycles of a PCR.
Several DNA samples with different, well-characterized methylation pattern for the chosen promoter were investigated. Two tumor cell lines that had previously been characterized as fully methylated or unmethylated in the promoter region analyzed and DNA derived from a tumor showing methylation of twelve of the 3′-terminal CpG sites were available for testing. Additionally, positive controls with only a few, well defined methylated CpG sites were generated using specific methylases.
While Hhal methylase modifies the internal cytosine residue in the sequence 5′-GCGC-3′, Hpall methylase modifies the internal cytosine residue in the sequence 5′-CCGG-3’. Unmethylated genomic DNA was treated with HpaII and/or HhaI. Use of the individual methylases results in the methylation of three sites each in the promoter region of interest. When used in combination, a total of six sites are methylated in the promoter region of interest.
Complete methylation of the promoter fragment of interest can be achieved by treatment of unmethylated genomic DNA with SssI methylase, which modifies cytosine residues in the sequence 5′-CG-3′. Mixing of methylase-treated DNA with wild type DNA is not necessary to achieve heteroduplex formation, because a complete conversion of all putative methylation sites is not probable.
DNA samples derived from the blood of young human subjects can be used as controls representing completely unmethylated DNA. Genomic DNA from young human subjects should be used in order to avoid the effects of age-related methylation. In this way it is possible to generate positive controls for every promoter fragment of any putative TSG known so far, which enables one to optimize specifically conditions for DHPLC analysis.
Choosing DHPLC Analysis Conditions:
Several issues need to be considered when determining optimal running conditions for DHPLC analysis on the WAVE Nucleic Acid Fragment Analysis System. Optimal gradient conditions and analysis temperatures are determined using WAVEMAKER software.
The software is used to predict melting curves for PCR products resulting from fully methylated, i.e.: all cytosines are conserved in the sequence of interest, and completely unmethylated DNA, i.e.: all cytosines in the sequence of interest are converted to thymidine, after bisulfite treatment. It is apparent that the melting temperature of the PCR product resulting from fully methylated genomic DNA has a higher stability due to its increased GC content relative to unmethylated DNA.
While the PCR product resulting from unmethylated genomic DNA after bisulfite treatment is partly denatured at 61°C, the corresponding fragment from methylated genomic DNA still remains largely double stranded. At 63°C the latter fragment is only partly denatured while the PCR amplicon resulting from unmethylated genomic DNA after bisulfite treatment is predominantly in its single-stranded form (Figure 10-1).
Thus, the optimal temperature for detecting a single methylated CpG site in the fragment analyzed here is predicted to be 61°C. Considering that methylation of more than one CpG site is generally expected, the resulting number of anticipated base changes will result in destabilization of the DNA helix and a concomitant lowering of the melting temperature of the fragment analyzed.
Thus, the optimal temperature for DHPLC analysis should be in a temperature range below the WAVEMAKER software predicted temperature for the fully unmethylated region of interest. At this temperature a shift in retention time between the PCR product derived from methylated and unmethylated fragments can be observed.
However, the differences in retention times are more obvious at higher temperatures, i.e., the melting temperature predicted for the PCR product resulting from the fully methylated region of interest, which is at approximately 63°C in the fragment analyzed here.
The differences in retention times can be correlated directly to the different degrees of denaturation between fragments. This is a direct consequence of the higher affinity of the DNASep Cartridge for double-stranded DNA as compared to single-stranded DNA (Figure 10-2).
When determining gradient conditions for methylation analysis by DHPLC with the aid of the WAVEMAKER software, one needs to consider that multiple methylation patterns may be present. This is particularly the case when analyzing genomic DNA from tumors, which may contain different methylation patterns in and by themselves, but may also contain genomic DNA from surrounding healthy tissues.
Heterogeneity in the methylation pattern of the genomic DNA analyzed will lead to extensive heteroduplex formation. Contrary to the hypothesis proposed by Deng et al., which assumes that the presence of multiple sequence differences inhibit heteroduplex formation, it is our opinion that this is not the case? We conclude from our observations and experience that heteroduplex formation is efficient even if multiple sequence variants are present.
The ratio of homo- and heteroduplex formation is, however, dependent on the relative amounts of sequence variants present. If a particular sequence or sequences is/are dominant in a given mixture of sequence variants, this will result in prominent homoduplex peaks. Sequence variants present at low abundance wilt in this case form heteroduplexes efficiently, albeit at low abundance and with a wide spread of retention times.
Increasing the elution gradient to 3 – 4% per minute of buffer B, which contains 25% acetonitrile, instead of the 2% per minute slope of the standard gradient recommended by WAVEMAKER software, has the effect of causing the elution of heteroduplexes formed in a shortened time interval.
This leads to a decrease in resolution between heteroduplexes, but increased peak intensity of the mixture of heteroduplexes present. Using an elution gradient that spans a wider range of acetonitrile concentration than that recommended by WAVEMAKER software is beneficial in that hetero- and homoduplex signals are guaranteed to elute within the analysis portion of the elution gradient (see Table 10-1).
Analysis of DHPLC results can further be improved by increasing the injection volume from usually 5µl, at a concentration of approximately 20ng of PCR product/µl, to 10-15µl.
Quantity and quality of PCR products must be confirmed prior to methylation analysis by DPHLC. Quantification and quality determination of PCR products can be performed on the WAVE Nucleic Acid Fragment Analysis System quantitatively and by gel electrophoresis semi-quantitatively.
Analysis of DNA Samples with Different Methylation Patterns:
PCR products derived from fully methylated or completely unmethylated genomic DNA after bisulfite treatment result in the formation of a single homoduplex for each sample. As shown above, the retention times observed differ due to differences in melting temperatures as a function of GC content of the fragment analyzed.
The PCR fragment with the higher GC content, i.e.: derived from fully methylated genomic DNA, exhibits an increased retention time over the PCR amplicon derived from completely unmethylated genomic DNA. The presence of different methylation patterns in the genomic DNA analyzed leads to the formation of PCR products of differing sequences. Hence, heteroduplexes are formed after denaturation and reannealing of the PCR products.
The latter is demonstrated in a model system using genomic DNA treated with different combinations of the methylases HpaII and HhaI followed by sodium bisulfite conversion. Three or six CpG sites, respectively, are methylated by the individual methylases or the combined action of both.
While peak shapes/heteroduplex pattern observed are very similar for genomic DNA treated with the individual methylases HpaII and HhaI (Figure 10-3b and c), the heteroduplex pattern of the PCR products generated from genomic DNA treated with both enzymes is distinctly different (Figure 10-3d) from the patterns shown in Figure 3, panel b and c.
In the case of the heteroduplex pattern shown in Figure 3 e, which is derived from tumor DNA, the heteroduplex pattern generated appears to be more complex and spans a wider retention-time window than in the examples shown above. From this it is concluded that at least 12 CpG sites are methylated in the tumor’s genomic DNA.
From the examples shown so far, it is apparent that based on the elution profiles and peak patterns it is not only possible to distinguish between fully methylated and completely unmethylated genomic DNA, but it is also possible to qualitatively determine the extent and diversity of methylation patterns present in the genomic DNA sample analyzed. Determination of the various heteroduplex patterns require a clear wild type signal obtained by complete conversion of unmethylated cytosines to uracil by bisulfite.
It is essential that formation of double-stranded DNA during bisulfite conversion be avoided and that the DNA analyzed is of high quality.
Analysis of Heterogeneous Tumor Material:
In DNA derived from tumor material the methylation patterns can be extremely variable due to the heterogeneity of the tumor material. As mentioned above, the mixture of cells with complete CpG-site methylation, wild type cells without CpG-site methylation, and cells with methylation of different degrees result in formation of the two homoduplexes with distinct retention times as well as a mixture of heteroduplex peaks (Figure 10-4). Dependent on the portion of the cells with variable degrees of methylation the different peak signals can be more or less intensive, making it necessary to inject an increased amount of PCR product.
DHPLC is a suitable prescreening technology for the analysis of promoter methylation. Due to its high-throughput capability, DHPLC is a quick and cost effective alternative to conventional methods.
This is particularly the case when a great number of so far unknown putative tumor suppressor genes need to be screened for their methylation status in order to determine promoter methylation as a possible inactivation mechanism.
Unlike restriction analysis, which is limited to the restriction enzyme recognition site, the DHPLC technology enables the determination of the methylation status at multiple CpG sites simultaneously within an amplified PCR fragment.
Compared to subcloning and sequencing, DHPLC is significantly less laboriously and quicker. The DHPLC technology provides information about the general extend of methylation, i.e.: whether a fragment is fully methylated, completely unmethylated, or partially methylated.
Sequencing of PCR products of DHPLC-positive samples, either directly or after cloning, following bisulfite treatment of genomic DNA is still required in order to determine the methylation pattern with resolution on the level of individual bases.
The high sensitivity of DHPLC enables the characterization of single CpG- site methylation as well as complete CpG-site methylation within a fragment analyzed. Even the degree of CpG-site methylation can be estimated up to a certain point by the extent of heteroduplex formation.
Using specific methylases, positive controls for completely methylated (SssI) or partly methylated genomic DNA (HpaII or HhaI) can be generated. Thus, before analyzing unknown promoter fragments optimal experimental conditions for the analysis by DHPLC, like gradient conditions and analysis temperatures, can easily be established.