|Home | About | Journals | Submit | Contact Us | Français|
Epigenetic changes, or heritable alterations in gene function that do not affect DNA sequence, are rapidly gaining acceptance as co-conspirators in carcinogenesis. Although DNA methylation signature analysis by methylation-specific polymerase chain reaction has been a breakthrough method in speed and sensitivity for gene methylation studies, several factors still limit its application as a routine diagnostic and prognostic test.
DNA methylation is the main epigenetic modification of the human genome, among histone modifications and noncoding RNAs, with specific targeting to the C5 position of cytosine at the dinucleotides cytosine-phosphate-guanine (CpG). The covalent addition of a methyl group on the cytosine-phosphate transforms it into a methylated cytosine, resulting in a different global electric charge to the DNA, preventing binding of the transcription machinery and leading to subsequent transcriptional silencing of the associated genes.1
Aberrant DNA methylation plays an important role in both cancer initiation and progression, as well as in imprinting disorders, diseases with trinucleotide expansions and aging-related conditions.2,3 Unscheduled DNA hypermethylation in the promoter regions of tumor-suppressor genes is one of the earliest changes that can be detected along the malignancy process. Since it affects small regions of DNA (about 100-bp in size), it is detectable in fresh as well as fixed archived tissue samples.
Unlike genetic changes, epigenetic alterations are not recorded in the genome in a manner that can be directly amplified, cloned, and sequenced. Therefore, the DNA sample must be fixed in the methylated state before amplification and the application of other technologies. A widely used method for “typesetting” epigenetic changes is bisulfite conversion of unmethylated cytosines. Treatment with bisulfite transforms unmethylated cytosine into uracil, whereas methylated cytosine is left intact. Furthermore, it is important for a methylation assay to be quantitative. Often, a “yes or no” answer with regard to methylation status does not provide useful information. Different CpG sites within a promoter region may be methylated at different levels, hence an assay with single CpG resolution is desirable. Also, relatively small differences (e.g., 50% vs 70%) in the methylation status of a given CpG site in a tumor sample can have big consequences in terms of gene expression.
The preferred method for DNA methylation analysis, called methylation-specific polymerase chain reaction (PCR; MSP) takes advantage of the bisulfite-mediated chemical conversion of cytosine to uracil, followed by PCR using two different pairs of primers complementary to the methylated or unmethylated versions of the gene promoter under consideration. Each primer contains at least two CpG dinucleotides either unmethylated when they are TC or methylated when they are GC.4
The specificity of MSP is higher than that of other methylation detection techniques due to specific primer design. MSP consists of chemical modification followed by amplification. Chemical modification creates the sequence differences between the methylated and the unmethylated DNA. After chemical modification, a set of three primers (U, M, and W) should be designed to anneal to the DNA, based upon these sequence differences. One primer (U) will anneal to unmethylated DNA that has undergone a chemical modification. A second primer (M) will anneal to methylated DNA that has undergone a chemical modification. A third primer (W) will anneal to any DNA (unmethylated or methylated) that has not undergone chemical modification. This reaction serves as a control for the efficiency of chemical modification. Therefore, MSP can eliminate the false-positive results inherent to previous PCR-based approaches, which relied on differential restriction enzyme cleavage to distinguish methylated from unmethylated DNA.
Obtained amplicons are then usually analyzed separately by agarose gel electrophoresis or by specific TaqMan probes during real-time PCR (Methylight). Such probes are expensive to detect one per one short methylated versions of a limited region (about 200 bp). The sensitivities of MSP for various genes are very different,5 ranging from 10−5–10−2. Overall, the sensitivity of MSP for multiple genes is higher than that (0.5–10−2) for detecting micro-satellite alterations.6
Although MSP is a sensitive and specific technology for the detection of gene methylation using small amounts of DNA, it can be influenced by the bisulfitation reaction and efficient melting curve analysis.
Denatured DNA (either by heating or at high pH by NaOH) is treated with sodium bisulfite which deaminates all the cytosine-phosphate located on single-stranded DNA through formation of a 5,6-dihydrocytosine-6-sodium sulphonate intermediate at acidic pH.7 Sulfonated DNA is further adsorbed by chromatography on silica in presence of sodium iodide. Last, it is further desulfonated with NaOH and washed extensively with ethanol before elution in water. Sulfonated DNA can also be cleaned of unwanted salts by ultrafiltration in 96-well units being a real method improvement.
The change of environment to alkaline one causes the degradation of sodium bisulfite and the transformation of indirect product into uracil. 5-Methylcytosine may also undergo such a reaction (deamination to thymine). However, in this case, the process of indirect product creation is very slow and the time of reaction inhibits the formation of the final product.8 Methylated cytosines are unaffected by the treatment.
The bisulfite reaction is highly single-strand specific. The formation of the sulphonated cytosine derivative (cytosine-SO3) is reversible and the equilibrium depends on pH, bisulphite concentration and temperature. The reversible sulphonation reaction and the subsequent irreversible deamination step are both carried out at acidic pH. On the other hand, the last desulfonation step is favored by alkaline pH.
Several problems hinder the conventional bisulfitation technology, the major being the lack of proper DNA bisulfitation control (Table 1). Indeed, some cytosine-phosphates will appear as bona fide methylated cytosine because the DNA bisulfitation steps were incomplete. In promoter regions which are very rich in CG, it is difficult to fully denature the DNA and bisulfitation is only efficient on denatured DNA. Nondenatured DNA will leave unconverted cytosine-phosphate which will appear further as methylated cytosine. In addition, when desulfonation is improperly conducted false positives are also obtained. The only control for an efficient bisulfitation is DNA sequencing to verify on both strands that all Cytosine-phosphate have been converted to T. Bisulfited DNA is at the end of its harsh conditioning process broken in about 500-bp fragments, which are further unable to form base pairs since uracil is not complementary to the guanine on the other DNA strand and sequencing template is halved, therefore sequencing bisulfited DNA without cloning is not easy. Pyrosequencing being quantitative is an improvement but is not easier and cheaper.
As mentioned before, ultrafiltration cleaning of DNA is better than silica chromatography since bisulfited DNA recovery by adsorption chromatography on silica is poor, because at the end of bisulfitation DNA is single-stranded and broken (50% loss). Moreover, this full procedure is not automated and requires a lot of manual work. However, high salt levels may clog membrane pores especially when the membrane surfaces are small.
A long incubation with sodium bisulfite (≥16 hr) damages about 60% of purine bases and phosphodiester bonds in the DNA molecule and destroys pyrimidine bases. The change of pH into alkaline generates purine-free and pyrimidine-free sites by means of breaking N-glycoside bonds. The number of these sites increases when the reaction with sodium bisulfite is longer and they are more often the effect of depyrimidination than depurination.5 Long incubation, high temperature, and high molar concentration of sodium bisulfite degrades even 84–96% DNA, while using less aggressive media may limit cytosine conversion.9
A method developed by Worm et al.10 exploits differences in DNA melting temperature between methylated and unmethylated alleles after bisulfite conversion. Double-stranded DNA under gradual heating or increasing concentration of denaturants, melts in a series of steps, in which each step represents the melting of a discrete segment, a so-called “melting domain.” Melting temperature (Tm), which is defined as the temperature required to convert the double helix into random coils, depends on GC content and length of DNA sequence.11 G·C base pair contains three H-bonds and is more stable than an A-T base pair containing only two H-bonds. For this reason, the Tm of a melting domain increases with an increase in G-C content. After the conversion of unmethylated cytosines to uracil by bisulfite treatment and amplification of uracil to thymine by PCR, methylated and unmethylated alleles differ in Tm due to their different contents of GC base pairs and thermal stability.12
DNA-melting curves are acquired by measuring the fluorescence of a double-stranded DNA-binding dye (SYBR Green I) during a linear temperature transition. Melting curves are then converted into melting peaks by plotting the negative derivative of fluorescence over temperature versus temperature (−dF/dT vs T). The Tm of the PCR amplification product is determined by the content of methylated and unmethylated alleles in the original DNA template. If all alleles are unmethylated, all cytosines will be converted to thymine, so the amplification product will have a relatively low Tm. By contrast, methylated alleles in CpG dinucleotides will give a significantly higher Tm.10 Melting-curve analysis can be used for the detection of methylated and unmethylated alleles in the same reaction, avoiding further post-PCR experimental steps and the risk of PCR contamination. Methylated positive and negative controls are required, whereas the use of enzymatically methylated DNA is not appropriate since it does not represent the de novo methylation observed in vivo. Cell lines DNA with known methylated regions are better.
The melting-curve analysis of DNA is a challenge since amplicons resolution on Tm is not exquisite on most real-time PCR machines when using SYBR Green. At least 1°C difference between unmethylated and methylated amplicons is required for a duplex MS-PCR. Indeed, because SYBR Green is not used at a saturating concentration that would inhibit PCR, fluorescent melt curves obtained are small and poorly resolved. On some real-time PCR machines, high resolution melt-curve analysis can be carried out using saturating cyto 9 as the DNA-intercalating dye, but in that case resolution is too exquisite and PCR unfinished amplicons give shoulder peaks that are undistinguishable from misprimed amplicons.
In order to acquire correct methylation detection by melting-curve analysis, certain parameters should be taken into consideration. Firstly, the bisulfite reaction needs long incubation, high temperature, and high molar concentration of sodium bisulfite for complete conversion of cytosine residues to uracil.13 Otherwise incomplete conversion could lead to false positive results with appropriate design of PCR primers also being crucial.14
There are currently many different approaches to generate DNA methylation data. Methylation-specific real-time quantitative PCR has proven the most effective in detecting the methylation level of a single cytosine. In combination with sequencing and array-based approaches, elucidation of epigenetic mechanisms will comprise a valuable tool for identification of biomarkers to assist in cancer diagnosis and treatment.