2.1. BISULFITE MODIFICATION Any methylation information that is present in the original DNA is lost as soon as the DNA is amplified in vitro. Thus, in order to study methylation using PCR, the DNA must be modified in some way that can be replicated. Sodium bisulfite treatment, which deaminates cytosine to uracil, has become the method of choice for modification. The rate of deamination of 5-methylcytosine to thymine is very much slower than the conversion of cytosine to uracil (49).
Bisulfite modification must be carried out on single-stranded DNA. After this, the cytosines are sulfonated by sodium bisulfite at a low pH; the sulfonated cytosines then readily undergo deamination to sulfonated uracil, which is then desulfonated using high pH.
The use of bisulfite modification prior to PCR treatment enables methylated cytosines in genomic DNA to be directly identified (50,51). The only cytosines remaining derive from methylated CpGs (the sequence 5'ACATmCGG3' becomes 5'ATATCGG3'), thus enabling a direct readout of methylated cytosines by sequencing. An important consequence of modification is that the two DNA strands are no longer complementary, with the other strand being converted from 5'CmCGATGT3' to 5'TCGATGT3'.
Bisulfite modification is robust and a wide variety of conditions have been used. Several solutions (e.g. sodium hydroxide, sodium bisulfite, and hydroquinone) need to be freshly prepared. Many of the important experimental parameters have been described (52,53).
If bisulfite modification is incomplete, accurate analysis of methylation cannot take place. This can be readily observed by sequencing as some non-CpG cytosines will remain as cytosines. A recent review deals with experimental artifacts that inhibit complete bisulfite modification (54). In particular, it is critical to fully denature the DNA before bisulfite modification.
2.2. METHODOLOGIES USED TO STUDY METHYLATION OF BISULFITE- TREATED DNA The gold standard methodology for analysis of methylation is genomic sequencing.
However, just as a variety of simpler, more cost-effective methods are often used to replace sequencing in mutation detection, a variety of simpler methods are often used to replace genomic sequencing in methylation detection. Methylation-specific PCR (MSP) is widely used, as it is rapidly performed and can detect relatively low levels of methylation (Section 2.4).
The methodologies can be divided into those that interrogate specific sites for DNA methylation and those that scan a region for DNA methylation. The latter have, in general, been adapted from mutation screening techniques that rely on physical separation according to sequence, such as single-strand conformation analysis, denaturing gradient gel electrophoresis, and denaturing high-performance liquid chromatography. All of the PCR-based techniques can potentially use DNA made from paraffin-embedded formalin-fixed tissues. Several bisulfite modification procedures have been described that deal specifically with DNA from paraffin sections or very low numbers of cells (55-58).
2.3. GENOMIC SEQUENCING The bisulfite-modification-based genomic sequencing methodology (50,51) revolutionized the analysis of methylation. Bisulfite-modified DNA is amplified with strand-specific primers framing the region of interest, and the PCR product is then sequenced. The PCR products can be cloned and sequenced to determine the methy-lation of each cytosine in individual DNA molecules. For tumor samples, this involves sequencing of many clones, as a considerable amount of sequence is often derived from normal tissue. This approach remains an extremely powerful research tool because it allows the detailed study of methylation heterogeneity in any given sample.
The PCR products can be directly sequenced, which gives an average estimation of methylation at each CpG site. Direct sequencing is both faster and considerably less expensive than cloning. However, for heterogeneous samples, it might be difficult to detect low-level methylation. A modification in which only the cytosine and thymine residues are sequenced using fluorescent sequencing and GENESCAN software has been described (59). The degree of methylation is obtained by comparison of the cytosine and thymine peaks.
2.4. METHYLATION-SPECIFIC PCR Methylation-specific PCR (MSP) is the most widely used method for the detection of methylation. It uses primers designed to be specific for methylated, bisulfite-modified DNA (60). Unmethylated sequences are not amplified, although a second pair of primers specific for unmethylated sequences is often designed. MSP is based on the principle that oligonu-cleotides with a mismatched 3'-residue will not function as primers in the PCR under appropriately stringent conditions. This principle had previously formed the basis of earlier developed single-nucleotide-polymorphism typing methodologies variously known as allele-specific PCR, ARMS (amplification refractory mutation system), and PASA (PCR amplification of specific alleles), which allowed genotyping on the basis of presence or absence of a PCR product (61-63). If a band is seen on a gel after PCR, it is concluded that the sample is methylated or unmethylated according to which primers were used. It is the most sensitive nonquantitative technique available and can detect as low as 0.01% methylation.
A positive MSP signal should only be obtained when both the cytosines recognized by the 3' end of each of the primers are methylated. This makes MSP an ideal method to screen islands that become heavily methylated. It might be less suited when islands show highly variable methylation, as has been reported for the p15INK4B and CX26 (connexin 26) genes (64,65). MSP could be modified so that only one of the primers recognized a potentially methylatable cytosine, which would allow mapping of one site at a time but MS-SNuPE (Section 2.6) is a better approach to determining single-site methylation.
There are some important limitations of the MSP approach. The first is the susceptibility of MSP to false positives. This can be either a false priming or a sensitivity issue. In the first case, a positive arises even if there is no methylation in the target sequence. If the annealing temperature is too low, amplification can occur across the 3' mismatch. This type of false positive can be detected by the use of an appropriate control such as a known negative cell line. Raising the annealing temperature and the use of a hot-start methodology or the design of new primers can eliminate this problem. Second, the sensitivity of MSP might lead to false positives because of the amplification of a rare subpopulation of methylated sequences. The tumor sample might be extremely heterogeneous, with only a small proportion of methylated cells. In this case, it would not be correct to call the tumor methylated for that particular gene. Considerable overestimates for the methylation frequency of particular genes have been reported when MSP is used (66). There might also be low levels of methylation in the normal somatic tissue (67) or incomplete bisulfite conversion.
These limitations arise because MSP is a nonquantitative methodology. It is difficult to tell whether the signal arises from the predominant proportion of cells or from a small subpopulation thereof. Despite these limitations, MSP remains extremely useful, particularly in a preliminary screen of a large number of tumor specimens because of the rapidity with which it is performed.
2.5. BISULFITE PCR FOLLOWED BY RESTRICTION ANALYSIS Several restriction-enzyme-based approaches have been devised to exploit the differences in sequence between methylated and unmethylated DNA after bisulfite modification. The amplified region could include a number of sites that can be analyzed by restriction analysis using enzymes that have a CpG but no other cytosine in their recognition site (such as BstUI, which cuts at CGCG, or Taq 1, which cuts at TCGA) or by enzymes in which the recognition sequence terminates in a C that is the first base of a CpG residue (such as Hinfl, which cuts at GANTC).
BstU1 digestion was first used to show that MSP products were methylated at other CpG sites than those at the 3' end of the primers (60). More commonly, restriction digestion is used with methylation independent PCR (MIP) primers that amplify both methylated and unmethylated strands. After amplification, the DNA is digested with the appropriate restriction enzymes and examined on an agarose or polyacrylamide gel. This approach is sometimes called COBRA (combined bisulfite restriction analysis) (68,69). A variant methodology looks at sites that are newly created following bisulfite modification (70). Because many CpG sites cannot be analyzed, these methods will miss methylation restricted to untested sites.
Restriction analysis can be useful to estimate the degree of methylation, although precise quantitation might be complicated by heteroduplex formation in which strands containing the restriction site anneal with strands not containing the restriction site. This effect becomes more pronounced as the cycle number is increased. In general, these heteroduplexes do not digest with restriction enzymes. A method that eliminates the effect of heteroduplex formation when restriction digestion is used to quantitate alleles has been described (71). Heteroduplex formation might be less likely, however, when there are multiple differences in sequence, as would be seen for methylated and unmethylated PCR products.
2.6. METHYLATION-SENSITIVE SINGLE-NUCLEOTIDE PRIMER EXTENSION ASSAY Methylation-sensitive single-nucleotide primer extension (MS-SNuPE) assay is an adaptation of the single-nucleotide primer extension assay originally introduced to type single-nucleotide polymorphisms (72). It is used to quantitate the relative levels of cytosine and thymine at a single CpG site. Bisulfite-modified DNA is amplified, made single-stranded and then hybridized with an internal primer whose sequence abuts the cytosine of a CpG residue (73,74). Then, DNA polymerase and 32P-dCTP or 32P-dTTP are added in parallel tubes. The internal primer can only be extended if the appropriate deoxynucleotide triphosphate has been added. If the cytosine is methylated, only a deoxcytidine triphosphate can be added, and if it is unmethylated, only a thymidine triphosphate can be added. The reaction is then denatured, elec-trophoresed on polyacrylamide gels to separate the extended primer from the unincorporated nucleotides, and viewed following autoradiography.
The MS-SNuPE assay is quantitative but seems to have a limited dynamic range. The original methodology only examined the range of 1-100% methylation (73,74). This is sufficient for the analysis of most tumor material and could be used to distinguish homozygous from heterozygous methylation in microdissected material. However, it is not suitable where assays that are sensitive over several orders of magnitude are required. Although several CpG sites can be analyzed in a single reaction by using different length primers, this methodology is too labor-intensive in general for screening applications and is mainly applicable to the detailed study of methylation of known genes.
The original method suffered the limitation that radioactivity was used to detect primer extension. In a more recent method, MS-SNuPE products were separated using ion-pair, reverse-phase, high-performance liquid chromatography (75). The methylated and unmethylated CpGs were differentiated and quantified based on the different masses and hydrophobic-ities of the two extended primer products.
2.7. ENZYMATIC REGIONAL METHYLATION ASSAY Enzymatic regional methylation assay (ERMA) is a quantitative method for determining the methylation density of any DNA region of interest (76). This technique is particularly useful for genes like p15INK4B in which methylation density is more important than methylation of particular sites (64,77). Bisulfite modified DNA is amplified with primers containing two dam sites (GATC). The PCR products are purified and incubated with 14C-labeled S-adenosyl-methionine (SAM) and dam methyltransferase, which methylates the adenine residues as a control for the quantity of DNA. A second incubation uses 3H-labeled SAM and SssI methyltransferase to which methylates the CpG cytosines to quantify the CpG sites. The ratio of the 3H/14C signals is directly related to the methylation density of the amplified sequence.
2.8. METHYLATION-SENSITIVE SINGLE-STRAND CONFORMATION ANALYSIS Methylation-sensitive singlestrand conformation analysis (MS-SSCA) uses single-strand conformation analysis to screen an amplified region of bisul-fite-modified DNA for methylation changes (56,78-81). The amplified products are denatured and electrophoresed on a non-denaturing polyacrylamide gel. The sequence differences between unmethylated and methylated sequences lead to the formation of different secondary structures (conformers) with different mobilities. Once the normal (usually unmethylated) pattern is established, any variation would indicate some degree of methylation. Methylated and unmethylated sequences will frequently have multiple-nucleotide differences that will favor the adoption of new conformers. However, there is still the possibility that the conformers might comigrate. Thus, it is advisable to try more than one set of running conditions. Most commonly, two different temperatures—room temperature and 4oC—are used, although there are several other running conditions that can be useful (82).
Methylation-sensitive SSCA is a convenient method with a very low false-positive rate for screening large numbers of samples for methylation. However, polymorphisms within the region being amplified also might give rise to variant bands. Sequencing is necessary to show whether a variant band is the result of methylation or polymorphism.
MS-SSCA was developed to screen a relatively large number of CpG sites occurring close together, such as those in a CpG island. Whereas all bisulfite methodologies are capable of being used with formalin-fixed paraffin sections, the optimum resolution of MS-SSCA of 150-250 bp is particularly suited to the analysis of DNA derived from this source. All of the CpG sites within a large island can be assessed by developing pairs of primers that will enable the whole island to be amplified in 200-300 bp overlapping fragments. Generally, however, one pair of primers is designed to analyze the sequence flanking the transcriptional start site.
Analysis of tumor material is complex because the specimen can contain substantial amounts of normal cells with normal methylation patterns or there might be intratumor heterogeneity of methylation pattern. MS-SSCA can be used on unfrac-tionated tumor material because it can detect methylation when cells with a methylated DNA sequence comprise less than 10% of the total cells (56). An important advantage of MS-SSCA is that individual bands can be directly sequenced. MS-SSCA can thus be used as the first step for genomic-sequencing studies of methylation. Samples with no identified methylation do not need to be sequenced.
2.9. DNA MELTING ANALYSIS Several methylation screening methods are based on the "melting" properties of DNA in solution. DNA denatures in discrete segments called melting domains as the concentration of a denaturant increases. The melting temperature of a domain is determined by its sequence
(83). In denaturing gradient gel electrophoresis (DGGE), PCR products are electrophoresed through a linear gradient of increasing denaturant concentration. When a DNA fragment enters the concentration of denaturant where its lowest temperature domain melts, the molecule changes its structure and migrates more slowly through the gel. This results in separation of different sequences according to their melting properties. The attachment of a GC-rich segment, called a GC clamp, which never denatures at the conditions chosen for the experiment, is necessary to allow for the detection of mutations in the most stable melting domain. DGGE allows detection of virtually all single-base changes in the PCR product.
"Bisulfite-DGGE" uses the same principle to identify methylation (67). Because unmethylated sequences are less cytosine rich than methylated sequences, and partially methylated sequences are intermediate, the fragments separate according to their degree of methylation. This allows clear visualization of heterogeneity of methylation in a way that can be directly related to the density of methylation. By contrast, SSCA will also allow visualization of methylation heterogeneity, but this is not related to methylation density. The DGGE approach is an unusually powerful one but has been adopted by few laboratories.
The simplest methodology based on melting analysis monitors the fluorescence of the double- stranded DNA-binding dye SYBR Green 1 as the double-stranded PCR product is slowly denatured (84). This is readily performed on all realtime PCR machines. The analysis is done in the same tube or well as the PCR, which allows high throughput and reduces problems resulting from PCR product cross-contamination. Standard primers can be used but analysis might be facilitated by designing primers with GC clamps similar to those used in DGGE (83).
Methylation screening of bisulfite-modified PCR products using denaturing high-performance liquid chromatography (DHPLC) is only just beginning to be used but is likely to become an important screening methodology. As with SSCA and DGGE, DHPLC screening does not distinguish CpG methyla-tion and single-nucleotide polymorphisms, and sequencing needs to be used to distinguish between them.
The optimum temperature for a DHPLC run can be predicted using the sequence of the fully methylated product. The temperature then needs to be verified so that tight peaks without much spread are obtained. The retention time of the peak should correlate with methylation status, because the more unmethylated the target is, the less GC rich the PCR product is and the lower the retention time is. DHPLC has been used to differentiate the methylated and unmethylated alleles of imprinted loci (85) and MLH1 promoter region methylation (86).
2.10. REAL-TIME PCR METHODOLOGIES The major disadvantage of MSP is that although it is the most sensitive technique for detecting methylation, it is not quantitative. It can, however, be adapted to a real-time PCR approach that will allow quantitation. Importantly, quantitation allows discrimination of signal from low-level background, which might occur in normal tissue. Real-time PCR analysis also requires no further manipulation after the PCR step, which allows high-throughput analysis and eliminates problems resulting from cross-contamination of PCR products.
In the MethyLight methodology (87,88), MSP is combined with the TaqMan methodology, which uses a fluorescent-labeled probe to monitor amplification. Bisulfite-modified DNA is amplified using MSP primers flanking an oligonu-cleotide probe containing a 5' fluorescent reporter and a 3' quencher. The 5' to 3' nuclease activity of Taq DNA poly-merase cleaves the probe, separating the fluorescent reporter from the quencher each time an amplification occurs. The intensity of the resultant fluorescent signal is proportional to the amount of PCR product generated, thus allowing quantification of the PCR reaction.
MethyLight has been called a quantitative MSP assay. However, the most commonly used TaqMan probes contain several CpG sites in their sequence. The effect of the CpG containing probe is to make MethylLight more specific for hyper-methylation. However, this might make MethyLight data complex to interpret. According to its design and the PCR conditions used, the TaqMan probe might bind only if all the probe CpG sites are also methylated or might bind when only some of the sites are methylated.
MethyLight is capable of detecting methylated alleles in the presence of a 10,000-fold excess of unmethylated alleles (87,88). Methylation of candidate genes needs to be scored relative to standard curves constructed from dilutions of DNA prepared from a cell line that is methylated for the gene in question. These dilutions can be made in DNA prepared from a cell line that is correspondingly unmethylated.
Sequences that have been incompletely converted during bisulfite treatment might be coamplified during MethyLight, resulting in an overestimation of the level of DNA methylation. In ConLight-MSP, an additional fluorescent probe directed against unconverted DNA is used in the same reaction as the specific TaqMan probe (89).
There are many potential variations involving the confluence of methylation analysis and real-time methodology. For example, nonselective amplification could be used with probes for both methylated and unmethylated DNA in the same tube as a type of allelic discrimination assay. However, the results might not be all that easy to interpret where there is heterogeneity of methylation patterns. The simplest real-time approach would be to convert an established MSP protocol to a quantitative protocol using SYBR Green in the reaction mix. It is essential that the MSP reaction is highly specific, as SYBR Green-based methods cannot readily distinguish between specific and nonspecific reactions.
2.11. PROFILING AND ARRAYS It would be clearly advantageous to analyze a cancer specimen for the methyla-tion of dozens if not hundreds or thousands of CpG islands at the same time. Restriction landmark genomic sequencing is one such methodology (90,91) but is technically difficult and is more suited to the research laboratory at its current stage of development. Differential methylation hybridization (DMH) can be used to study genomewide methylation patterns in CpG islands in DNA from cancer cell lines, tumor samples, and normal tissue, but it is also more suitable for gene discovery because the methodology is technically complex (92). Currently, the most promising array-based approaches are oligonucleotide based. Bisulfite-treated DNA samples are amplified and hybridized to arrayed oligonucleotides that are specific for methylated and unmethylated sequences at chosen CpG islands (93,94).
2.12. PRIMER DESIGN Primer design is a critical aspect of PCR-based analysis of bisulfite-modified DNA. Primers can be designed to analyze any region but are most often designed so that the amplified region overlaps the tran-scriptional start site and consists mainly of sequence prior to the start site. The DNA strands are no longer complementary after bisulfite treatment. Primers are generally designed to amplify from the modified sense strand, but if it is difficult to design effective primers, the modified antisense strand should be examined.
For a new sequence, identifying the promoter region CpG island is the crucial first part of the primer design process. The transcriptional start region of the gene must first be located. Many cDNA sequences include very little 5' untranslated region sequence. The most recent 5' cDNA sequence available should be used to Blast search GenBank (http://www.ncbi. nlm.nih.gov/blast) to identify mRNAs or expressed sequence tags that have extra 5' sequence. Using the TRASER website facilitates this process (http://genome-www6.stanford.edu/cgi-bin/Traser/traser). The most 5' sequence identified is then used to identify genomic sequence that is likely to be at the tran-scriptional start site. This genomic sequence can then be examined for the presence of a CpG island. CpG islands can be identified by visually identifying clusters of CpGs as outlined next or using the Web-based program CpG Island Searcher (http://www.uscnorris.com/cpgislands/).
For MSP, regions that contain frequent CpG sites are chosen and many investigators try to fit as many CpG sequences into the 3' end of the primer as possible. It is desirable to have the 3' end corresponding to a cytosine of a CpG dinucleotide, but this is not universally done. MSP is best suited for screening genes where the methylation patterns are already known and the 3' ends can be placed at the most frequently methylated sites.
Quite different requirements are necessary for MIP primers that amplify all converted sequences prior to analysis, as is required for genomic sequencing or MS-SSCA. MIP primers are designed to amplify both bisulfite-modified methylated and unmethylated DNA by either avoiding CpG sites or including as few as possible and placing them as far as possible to the 5' end of the primer. The primer sequence at the cytosine of a CpG residue can either be (1) C (G on the antisense primer), which biases amplification in favor of methylated sequences, or (2) a C/T degeneracy (G/A on the antisense primer), or (3) an inosine. Because methylated sequences can have an amplification disadvantage, the first option is often acceptable, particularly when techniques screening for methylation are used.
Careful primer choice is important to minimize the amplification of incompletely bisulfite-modified DNA. Primers are chosen that have a T derived from a non-CpG cytosine at or as near as possible to the 3' end of each primer. A website for designing methylation primers (http://itsa.ucsf.edu/~urolab/ methprimer/index1.html) has recently become available (95). Choosing primers that meet the constraints of CpG placement and modified T placement can also be done by visual scanning of the modified sequence that is facilitated by most word processing programs.
The following approach has proven to be effective for visual scanning for suitable primers. The sense or antisense sequence is converted to lowercase and all spaces and carriage returns are removed. The font is changed to a proportional font such as Courier and the number of characters is adjusted to 50 or 100 per line. The color of the Cs and Gs is changed to green. The color of the CG dinucleotide is changed to red. The colors chosen are arbitrary, but the above combination allows ready visualization of the CpG island both in terms of high G + C content and CpG dinucleotide density. The sequence is copied twice. The first sequence is denoted the (bisulfite) modified unmethy-lated sequence and all lowercase C's are changed to an uppercase T. The second sequence is denoted the modified methylated sequence and all C's are changed to an uppercase T except for the C's in CpGs. This makes scanning for potential primers very easy because the CpGs are readily visible and the modified C's are readily seen as uppercase T's. This approach can be automated by making a word processing macro (96).
It is often preferable to use primers with high melting temperatures to help with specificity, particularly because the DNA sequence is less complex after bisulfite modification because of the depletion of cytosines. The last five 3' nucleotides preferably should contain two or three (but not more) G's (C's on the antisense strand) to stabilize the 3' end of the primer. We use the Web program Oligonucleotide Properties Calculator to check the melting temperatures (http://www.basic.nwu.edu/biotools/ oligocalc. html). We find that the most reliable calculation to set the initial annealing temperatures is 5oC below the salt-adjusted melting temperature. The primers can be further checked using the Macintosh freeware program Amplify (version 1.2) to identify nonspecific primer binding within the region of interest as well as potential primer dimers. We also use Amplify to ensure that the chosen primers do not amplify the unmodified sequence and that the modified methylated and modified unmethylated are both amplified.
The use of nesting is not recommended because of the increased potential for carryover of PCR amplicons. It is unnecessary, in most cases, with primers designed to the above criteria but might be necessary when working from micro-dissected material in order to generate sufficient PCR product.
Was this article helpful?
Prior to planning pregnancy, you should learn more about the things involved in getting pregnant. It involves carrying a baby inside you for nine months, caring for a child for a number of years, and many more. Consider these things, so that you can properly assess if you are ready for pregnancy. Get all these very important tips about pregnancy that you need to know.