Bisulfite Pcr Primer Design Tool

DNA methylation is the most extensively studied mechanism for epigenetic gene regulation (1). Recent studies have shown that DNA methylation plays an important role in a number of physiological processes as well as common diseases such as cancer and neurodegenerative disorders (2,3). In mammals, DNA methylation occurs at the C-5 position of cytosine in CpG dinucleotide sequences (Figure 1) (1), which are mainly concentrated in regions known as CpG islands. Methylation in CpG islands within gene promoters usually leads to gene silencing. More recently, DNA methylation in regions located up to 2 kb from known CpG islands (called CpG island shores) has also shown a strong correlation with gene expression (4).

At present, the vast array of platforms available to study DNA methylation present a challenge for scientists who wish to enter this field (5). Among the methods for studying DNA methylation in candidate regions, PCR-based approaches have several advantages (6). Here we provide a practical overview of experimental design and analysis for the most common PCR-based DNA methylation techniques: bisulfite sequencing PCR (BSP), methylation specific PCR (MSP), MethyLight, and methylation-sensitive high resolution melting (MS-HRM). These techniques do not need expensive specialized equipment and could be implemented in a typical molecular genetics laboratory.

Bisulfite conversion

The first step in almost all protocols for studying DNA methylation is bisulfite conversion of the DNA sequence of interest. Bisulfite conversion occurs through a number of chemical reactions (e.g., sulfonation, deamination, and desulfonation) on the DNA that transform non-methylated cytosines into uracils. Methylated cytosines remain unconverted (Figure 1). Classical DNA conversion protocols are time-consuming, often requiring more than 16 h to complete (7), and require multiple tube changing steps that increase the risk of contamination and human error. Classical protocols also risk losing more than 75% of the starting DNA (8,9) during purification and through single-strand breaks that occur during long incubation steps (7,9).

Commercially available bisulfite conversion kits improve recovery of the converted DNA by using shorter incubation steps and alternative purification procedures (9). These kits also facilitate efficient implementation of the conversion reaction, thereby improving downstream results with PCR-based techniques. Thus, kits are highly recommended, especially for those unfamiliar with this field of study. There are many considerations for selecting a kit, including cost, yield, efficiency, and time. A comparison of the main features of available DNA conversion and methylation control kits is included in Tables 1 and 2.

**Table 1.** **Comparison of commercially available kits for bisulfite conversion (single column format)**

**Table 2.** **Comparison of commercially available kits for DNA methylation controls**

Controls for DNA bisulfite conversion

Evaluation of the quality of converted DNA is recommended when beginning a DNA methylation study; this step is especially important for quantitative PCR-based methods such as MethyLight and MS-HRM. Since bisulfite-treatment can result in DNA fragmentation, thus reducing the number of molecules available for PCR amplification, it is best to test the bisulfite-converted DNA with primer sets that amplify a range of differently sized products. From these products, the ideal amplicon length for downstream analysis can be determined (10), providing information that will aid in primer design.

Incomplete bisulfite conversion will adversely affect the reliability and accuracy of DNA methylation measurements by PCR-based methods (11,12). Therefore, it is necessary to evaluate the efficiency of conversion using commercially available primer sets to amplify the converted DNA (e.g., DAPK1 Catalog #D5014–2, Zymo Research, Orange, CA) (Table 1). The resulting DNA product can be sequenced to verify the efficiency of conversion for all non-CpG cytosines. Alternatively, converted DNA may be amplified with primers designed for the non-converted DNA sequence. In this case, the absence of a PCR amplicon suggests a complete conversion reaction.

Converted DNA must also be quantified prior to downstream PCR applications. The amount of DNA may be determined by spectrophotometric measurements using the NanoDrop 2000 spectrophotometer (Thermo Scientific, Waltham, MA) (13) with settings for single-stranded DNA, or agarose gel electrophoresis and classical UV spectrometric analyses (5). Other more specific methods, such as qPCR (including MethyLight control assays) or PicoGreen may be more reliable and better suited for measuring limited amounts of DNA (14).

Designing primers for PCR-based DNA methylation analysis

Designing primers against a region of interest (ROI) is the most critical step in obtaining adequate DNA methylation results using PCR-based methods. Several software platforms such as Methyl Primer Express (Applied Biosystems, Foster City, CA), MethPrimer (15), BiSearch (16), MethMaker (17), and MSPprimer (18) have been developed for this purpose. All of these programs allow users to customize primer length, amplicon length, and Tm (melting temperature) differences, as well as enable searches for CpG islands in the input sequence, and identify possible stable primer-dimer or hairpin structures that should be avoided. The advantages and disadvantages of each program are compared in Table 3. Primers should not bind to regions containing common SNPs (19), which can be identified easily using the UCSC Genome Browser (http://genome.ucsc.edu).

**Table 3.** **Freely available software for primer design for methylation analysis**

Because bisulfite treatment decreases DNA sequence complexity, primers have an increased tendency to bind multiple target sequences in converted DNA (18). Therefore, in silico evaluation of primer specificity is a key step during primer design for bisulfite-converted DNA methods. BiSearch software is unique in terms of its ability to find the number of potential matches, including partial matches, for each individual primer in the bisulfite-converted methylated or unmethylated genome and to perform in silico PCR on the bisulfite-converted human genome using any primer pair (16). In our laboratory, we have observed greater success using primers with less than 3000 matches.

At present, the software available for primer design does not account for PCR bias (20–22). When faced with bias, it is important to use additional tools to review the ROI sequence, highlight the CpG and non-CpG cytosines, and design adequate primers. BioWord is a free Microsoft Word plugin that allows manipulation, editing, and processing of DNA sequences and has proven useful for working with sequences prone to PCR bias (23). Another available option is a shareware version of the licensed software FastPCR, which includes a tool for in silico bisulfite conversion of non-CpG cytosines (24).

PCR-based techniques

Bisulfite sequencing PCR

Bisulfite sequencing PCR (BSP) was the first technique described for analyzing DNA methylation status using PCR (25). The technique consists of PCR amplifying a bisulfite-converted DNA ROI, followed by Sanger sequencing of the product either directly or after cloning into a suitable vector.

Direct-BSP: By comparing sequencing results with the respective reference genomic DNA sequences, direct sequencing of PCR products provides information on the average methylation status for each CpG dinucleotide. Direct-BSP is the shortest form of BSP, but holds several technical challenges inherent in sequencing, such as poor signal quality and artifacts in cytosine signals that may affect electropherogram analysis; it also has a low sensitivity (26). Because of these difficulties, BSP with cloning is more common.

Cloning-based BSP: In cloning-based BSP, PCR products are cloned into a vector and transformed into competent E. coli cells. After expansion and purification of the plasmids, the PCR product inserts are sequenced. The CpG methylation status for each CpG dinucleotide in the ROI is determined by sequencing each expanded clone (27,28). The resulting averages are referred to as DNA-methylation haplotypes. Cloning-based BSP requires at least six sequencing reactions to obtain a sensitivity higher than direct BSP (29), making this an expensive and labor-intensive option that is especially cumbersome for population-based studies.

Digital (single-molecule) BSP: Another BSP option for producing DNA-methylation haplotypes is digital-BSP (22,30). This method requires serial dilution of a DNA template to optimize conditions for PCR amplification of a single converted DNA molecule per reaction tube (via the Poisson distribution), thus avoiding both PCR bias and cloning. Digital-BSP is considered the gold standard for detecting the methylation status of specific loci (22). However, this method is inefficient because 87% of the reactions cannot be analyzed, and 3% are control reactions; thus, useful information is only obtained from the remaining 10% (22). An alternative approach is to use MS-HRM to select the clones for sequencing (31).

Primer design considerations: Methylation independent PCR (MIP) primers should be designed to allow the amplification of bisulfite-converted DNA regardless of methylation status. Primers should also not bind regions containing CpG dinucleotides (Figure 2A) (25) and should flank a sequence of converted DNA containing as many thymines originating from the conversion of non-CpG cytosines as possible (25). Guidelines for designing BSP primers were initially published by Clark et al. in 1994 (25).

**Figure 2.** **Primer design and results f**or **bisulfite specific PCR (BSP) and methylation specific PCR (MSP).**
(A) BSP primer. The dashed line indicates the sequence of the primer binding region in this example. Note that CpGs are avoided in the design of this kind of primers. *For simplicity, we use the same binding sequence of the forward primer to illustrate a hypothetical reverse primer design. (B) Simplified electropherogram schema of possible results from BSP for two CpG cytosines. Left panel: Cloning-based BSP. Possible sequencing results for a single clone. The possibilities are reduced to methylated (remains C on the sequencing data) or unmethylated (T replaces C in comparison with non-treated sequence). Right panel: Direct BSP C denotes methylated status (100%), T denotes non-methylated (0%) and Y (C + T) denotes different methylation percentages. (C) Chart of an example of MSP primer design over a binding site specific for methylated DNA (top) and non-methylated DNA (bottom). The untreated sequence (left) is modified depending on its DNA methylation status (middle). This sequence is used to design the forward primer by substituting the Us with Ts (D) Diagram of the possible results of an MSP assay on an electrophoresis gel. N: non-affected nucleotides in bisulfite treatment. The Us and Ts in red represent the uracils derived from bisulfite conversion of CpG cytosines and the corresponding thymines in the primer sequence. The CpG dinucleotide and the corresponding UpG or TpG sequences are in bold and underlined.

Recently, several more techniques have been developed using primers based on the same principles (32). One variant of the direct-BSP method uses two rounds of nested PCR with primers designed by standard methods for the first round and primers with a GC-rich tag at the 5′ end for the second round. This primer modification is intended to reduce non-specific amplification during direct-BSP and compensate for the frequent artifacts seen in direct-BSP results (33,34).

Bias in BSP: Several studies using MIP primers have shown bias toward unmethylated or methylated alleles (8,20), likely due to sequence differences between methylated and unmethylated alleles (34). For example, Warnecke et al. found a 33-fold amplification bias toward the unmethylated allele when assaying a region of the RB1 gene promoter (20). In some cases, adjusting MgCl₂ concentrations or redesigning the primers to bind the opposite DNA strand may be sufficient to resolve this bias (20,34). Shen et al. found that, in some instances, adjusting the annealing temperature may correct this bias (35). Wojdacz et al. developed a new approach to primer design that allows the use of annealing temperature changes to adjust for amplification bias (36,37). These new guidelines for bias compensation are described in detail in the MS-HRM section of this article.

Another potential source of error occurs in cloning-based BSP methods. Cloning biases may skew the reliability of results generated from BSP assays (22,34). There is evidence that amplicons without cytosines may be more difficult to clone efficiently (8). Although direct-BSP has low sensitivity, it provides more accurate detection of differences as low as 20% in methylation status in a single CpG (29).

Data analysis: Quantification of methylation levels is determined by comparing the relative peak heights of cytosine and thymine (or adenine and guanine in cases of complementary strand sequences) (25) in each CpG position in the electropherogram (Figure 2B). A qualitative analysis of bisulfite sequencing results can be performed if a clear single peak is present for each CpG cytosine position. In that case, a thymine peak would be interpreted as a non-methylated CpG, and a cytosine peak would represent a methylated CpG. Analysis of raw sequence data from direct-BSP is often difficult, but correction algorithms aid data interpretation. The Epigenetic Sequencing Methylation Software (ESME) program includes an algorithm to analyze direct-BSP sequencing results and provides a quality control filter (Table 4) (29).

**Table 4.** **Multi-purpose software for DNA methylation test design and analysis of results.**

ESME may also be used to analyze cloning-based BSP electropherograms (29). In cloning-BSP or digital-BSP, satisfactory sequencing results belonging to the same sample should be averaged to determine the level of methylation for each CpG position. This task is facilitated by BiQ-Analyzer and BISMA (38,39) (Table 4).

BSP selection: Different BSP methodologies are optimal for different methylation studies, depending on the particular conditions of a study and other parameters, including cost, research question, and available samples. For example, the study initially validating digital MethyLight is a case where digital-BSP was the best choice, since it allowed accurate validation of another single-molecule technique and the use of automated PCR-processing for a large number of sequencing reactions (30). In many other cases, cloning-BSP is preferred because it is the only option for determining DNA-methylation haplotypes in general laboratories (34). Direct-BSP was selected to assess DNA methylation in BDNF to complement initial results from MSP screening experiments (40). Before deciding between direct-BSP and cloning-BSP for a particular application, we recommend testing and comparing previously validated primers and strategies (26,27,33).

Methylation Specific PCR

Methylation specific PCR (MSP), first described by Herman et al. in 1996, determines the methylation status of an ROI through selective amplification of methylated and unmethylated alleles. The two-tube approach employs two primer sets: one binding specifically to the methylated sequence and another binding to the unmethylated sequence (11,41) (Figure 2C). A two-round variant of MSP, referred to as nested-MSP (N-MSP), has been described and can be used in special cases (42).

MSP is a simple method that requires resources commonly available in a molecular genetics laboratory and, once standardized, is effective for detecting methylated or unmethylated alleles without quantification. Processing up to 24 samples for both primer sets using conventional MSP requires about 4 h. Commercially available PCR master mixes for MSP (EpiTect MSP Kit, Qiagen, Hilden, Germany) are available; however, only conventional PCR reagents, including Hot-Start Taq polymerase, are required for the setup of MSP (43). While MSP assay kits are not commercially available, the MethPrimerDB database is available for help in selecting MSP primers (44).

Several real-time PCR adaptations of MSP also have been developed, including MethylQuant, a common option based on the measurement of increased fluorescence from SYBR Green I (45), and a real-time MSP approach combining conventional qPCR measurements with an additional melting step to detect amplicons associated with incomplete DNA conversion (46) or to distinguish the methylation status of individual alleles by comparison with standards of known allelic methylation status for an SNP located in the amplicon region (47).

Primer design considerations: As described by Herman et al., both methylated and unmethylated MSP primer sets should be designed to anneal to the same CpG containing region. MSP primers should include abundant CpG sequences at the primer binding sites to provide maximal discrimination between the methylated and unmethylated alleles. For the same reason, these CpGs should be as close as possible to the 3′ region of the primer (11) (Figure 2C). Additionally, a high number of thymines derived from non-CpG cytosines should be included to ensure specificity for converted DNA. MSP primer design is facilitated by the software listed in Table 3.

Data analysis: An amplification product of the correct molecular weight on an electrophoresis gel can be interpreted as methylated or unmethylated, depending on the specific primers used (11). The presence of amplification products using both sets of primers indicates a sample with both methylated and unmethylated DNA in the ROI (Figure 2D). However, a band from a reaction with methylated-specific primers might be a false positive. To avoid misinterpretation, inclusion of unmethylated DNA, non-converted DNA, and no-template negative controls is required (46,48). Likewise, the absence of an amplicon could be due to issues with the PCR reaction and must be controlled for as well (49).

The primary limitation of this technique is that it is qualitative (11). In general, well-standardized MSP assays provide information restricted to three possible outcomes: (i) presence of a methylated allele, (ii) presence of an unmethylated allele, or (iii) presence of both alleles. In assays intended to test MSP sensitivity, several ratios of the methylated and unmethylated DNA were used as templates. The results showed no clear correspondence between band intensity and dilution ratio, with many cases exhibiting very similar bands even for disparate levels of DNA methylation (50). On the other hand, several MSP assays demonstrated high sensitivity, detecting methylation percentages (MP) as low as 0.1% (50 pg of methylated DNA out of 50 ng of total DNA) or 1% (0.1 ng of methylated DNA out of 10 ng of total DNA) in different studies (11,43,50).

Possible challenges: Low quality DNA is associated with a decrease in reproducibility (51). As mentioned above, it is critical to avoid amplification of non-converted DNA using MSP primers (11,52). Kristensen et al. identified false positive MSP results due to incomplete bisulfite-conversion, which is particularly problematic if only four or fewer non-CpG cytosines are included in the primer binding region (52). This issue has been associated with the apparent low reproducibility of numerous MSP assays (46,53). On the other hand, even after PCR amplification, MSP results can be validated by means of pyrosequencing to confirm the full conversion of every non-CpG cytosine (49). In MSP, PCR for methylated-specific or unmethylated-specific primer sets can frequently be standardized with non-identical PCR conditions (for example, different annealing temperatures) (11), possibly through inherent differences in sequence composition between primer sets. Therefore, identical PCR conditions for both MSP primer sets are not required for accuracy (11).

Real time PCR-based methods

MethyLight

Dual TaqMan labeled probes were developed for genotyping studies several years ago (54). Eads et al. subsequently introduced the use of TaqMan technology to determine DNA methylation status in specific genomic regions, a technique that was named MethyLight (55). Peter Laird's group defined four types of MethyLight reactions, depending on which oligonucleotides are designed to discriminate the methylation status: (i) only the primers, (ii) only the TaqMan probe, (iii) both primers and probe, or (iv) none (in cases where a control reaction is required to discriminate the converted DNA) (55) (Figure 3A). Using these reactions, several variations of MethyLight have been proposed to address different biological questions, such as the amount of methylated versus unmethylated alleles (56) or methylation status at the CpG dinucleotide level (which would be very expensive) (55). The most commonly used MethyLight methodology uses two primers and a TaqMan probe designed to bind the methylated allele specifically and requires a reference gene for normalization (55) (Figure 3A). It is important to note that MethyLight, depending on the method subtype, can assess the methylation status of all CpG sites covered by the TaqMan probes.

The quantitative aspect of MethyLight has been explored since its inception. Analyses using different ratios of methylated to unmethylated DNA have been employed to verify the linearity of this quantitative assay, with a high linear correlation found between the dilution ratios and the MethyLight MP measurements (55,57). MethyLight has shown higher levels of accuracy and lower rates of false negatives when compared with previously described techniques (45,55). For this reason, MethyLight is frequently used to validate other techniques for DNA methylation studies (36). However, unlike MS-HRM (36), the most commonly used MethyLight technique cannot detect heterogeneous methylation in a sample because the MethyLight primers and probes are designed to measure a specific methylation pattern (fully methylated).

Primers and probe design considerations: In a MethyLight assay, it is necessary to normalize each qPCR reaction using sets of primers and probes that bind a converted DNA region independent of its methylation status (Figure 3A, top). Therefore, each MethyLight assay should include both an ROI amplification reaction and an MIP reaction for the control region. Currently, commercially available Beacon Designer software (Primer Biosoft, Palo Alto, CA) is the method of choice for MethyLight primer design. This program is able to design MethyLight assays, but in practice it is restricted to CpG islands, possibly because CpG density in the island shores and in some promoters lacking islands is low.

Data analysis: Depending on the MethyLight design, relative fluorescence units (RFUs) should be used to calculate the methylation percentage. For the common MethyLight design described in this section, the broadly accepted determination formula for DNA methylation percentage is shown in Equation 1 (55).

In order to evaluate the methylation status of an ROI using MethyLight, it is best to select a previously validated primer/probe set for that ROI if a similar research question is to be addressed. Houshdaran et al. (58) have performed ∼300 MethyLight assays and have made the primer and probe sequences available.

Example of MethyLight selection: MethyLight is the technique of choice when a study requires accurate quantitative assessment of DNA methylation. It has been used in a number of cancer associated DNA methylation studies, including the development of an assay to measure the presence of methylated alleles in three genes associated with colorectal cancer (59). Here, the assay was focused on clinical applications of cancer detection. It should be noted that the cost of using TaqMan probes can be higher than other real-time PCR methods that utilize cheaper intercalating dyes (46). This is of particular importance if the sample size of the proposed study is large, or if a significant number of ROIs is to be assessed.

Methylation-sensitive high resolution melting

In the DNA double helix, a cytosine and a guanine of complementary strands are linked by a triple hydrogen bond while a thymine and an adenine are joined by a double hydrogen bond (36). Therefore, base composition can directly influence the thermodynamic behavior of DNA in a melting analysis. Tm is defined as the temperature at which the PCR product dissociates into two single strands and a sharp drop in fluorescence of a DNA intercalating dye is observed (37). This basic principle can be used to discriminate between methylated and unmethylated alleles following bisulfite conversion. Distinction between alleles is achieved through Tm analysis of the MIP-PCR products in the ROI, in which the methylated allele usually has a higher Tm than the unmethylated allele (60) (Figure 3D).

Initially, Worm et al. (61) described an in-tube melting protocol for analyzing DNA methylation prior to the development of high resolution melting (HRM) technology. After technical improvements in melting assays, Wojdacz et al. developed a DNA methylation assay implementing HRM technology: the methylation-sensitive high-resolution melting (MS-HRM) techniques (36). The MS-HRM methodology consists of real-time PCR using bisulfite-converted DNA (regardless of the methylation status) and melting analysis of PCR products (HRM) to discriminate the ROI methylation status reflected in the thermodynamic behavior of the MS-HRM amplicon.

The MS-HRM method enables assessment of the percentage of the methylated allele present for a particular sample in an ROI. This is possible through comparison with melting standard curves created by different dilution ratios of methylated and unmethylated DNA controls (37). Since the technique analyzes the melting properties of the final PCR products, MS-HRM not only evaluates fully methylated alleles in proportion to fully unmethylated ones, but is also able to detect heterogeneously methylated samples (62).

PCR bias: As for other MIP based amplifications, potential PCR bias for MS-HRM was evaluated during development of the technique (21). MS-HRM showed a strong amplification bias toward unmethylated sequences when the classic recommendations for primer design stated by Clark et al. were followed (25). In contrast, using the recommendations of Wojdacz et al. for primer design (21), variations of the annealing temperature in the PCR cycling step allowed for control of PCR bias (21). Monitoring of real-time PCR amplification establishes an additional quality control step for MS-HRM experiments (13). Similar to digital-BSP, digital MS-HRM is also useful for reducing PCR bias (62). Considering the possibility of PCR bias, it is important to highlight that quantitative methylation analysis with MS-HRM is based on the assumption that methylation levels of CpG sites between the primers is the same as methylation levels of CpG sites covered by the primers.

Primer design considerations: MS-HRM primer design follows the same general principles of classic MIP design as previously detailed by Clark et al. (25). However, in order to compensate for PCR bias, there are new recommendations for MS-HRM primer design that advise inclusion of one or two CpG annealing sites (located as far as possible from the 3′end of the primers to avoid methylation specific amplification) (Figure 3C) (60). Currently, there are no programs for MS-HRM primer design that incorporate the new recommendations to compensate for PCR bias. Finally, several programs such as OligoCalc, Poland, and MELT (Table 4) can predict the melting curves of the PCR products.

Data Analysis: Wojdacz et al. (36,37) proposed a method for estimating methylation levels by comparing the melting patterns of standard templates with known proportions of methylated and unmethylated DNA controls to the melting patterns found in a sample. The semiquantitative estimate is based on similarities in HRM patterns without a mathematical approach for calculating the DNA methylation percentage. More recently, Tse et al. (2011) implemented an MS-HRM approach to quantify the methylation status of each sample with high reproducibility. Peak-height and area under-the-curve from the normalized, temperature-shifted difference curves were used to generate linear standard curves (13) (Figure 3D). Quantitative data were obtained by interpolation of the first derivative of the normalized melt curves, generated by the linear regression analysis of the standard curve (13). When heterogeneous DNA methylation patterns are present in a sample, HRM analysis will identify such heterogeneity by the complex shape of the melting curves; however, in such cases quantitative HRM measurement is not possible (62). The presence of SNPs in the amplicon region could generate additional variations in the melting profiles (37).

Examples of MS-HRM selection: MSP-based assays only evaluate DNA methylation for CpG sites present in the primer binding region (usually <25 bp per primer). In contrast, MS-HRM evaluates all of the CpGs banked by the primers (usually >80 bp), regardless of the methylation status of CpGs within the primer binding site (36). Therefore, MS-HRM provides the ability to evaluate a larger genomic region when compared with MSP-related techniques (Figure 4) (11). MS-HRM is a good choice for quantitative determination of DNA methylation levels, when sequence level detail is not required (63). A good example of the usage of MS-HRM is a colorectal cancer study where the authors distinguished different stages of the disease and their correlations with the quantity of DNA methylation (64).

**Figure 4.** **CpG covera**ge of PCR-based DNA methylation techniques.
A simplified lollipop schema shows the CpG dinucleotides as circles. The methylation status is not represented. Grey circles represent the non-evaluated CpGs in each corresponding technique. Red circles represent the CpG dinucleotides evaluated in each method. In the MethyLight assay, the green oval represents the fluorescent molecule and the black oval represents the quencher molecule.

Proper controls for PCR-based DNA methylation analysis

In addition to the controls used in conventional PCR assays, other steps should be taken to verify the accuracy of DNA methylation data generated in PCR-based assays. Unconverted genomic DNA is an essential control that should be included in all optimization processes for PCR-based DNA methylation assays; it provides information on the amplification of non-converted DNA with primers that are specific for the converted DNA. For specific assays, amplification from bisulfite-treated DNA should show a clear difference from any possible result using non-converted DNA.

In one of the pioneering MSP studies, Herman et al. verified primer specificity for the bisulfite modified p16 sequence using untreated DNA in reactions with either methylated-specific or unmethylated-specific primers (11). As expected, no amplification was found when non-converted DNA was used as a template. Nonetheless, several reports of MSP standardization did not include or report this kind of control (65,66).

Similarly, the use of non-converted DNA is also recommended when using the MS-HRM technique during assay optimization (37). This type of control is the easiest to include but, paradoxically, is the controal most commonly omitted or not reported (63). It allows experimental verification of the specificity of the assay for converted DNA. In most cases, there should be no amplification products; however, in some instances products will be amplified that can be easily identified when compared with the converted DNA (37).

Use of fully methylated and unmethylated DNA is a critical experimental control as well. It should be noted that DNA considered to be fully unmethylated comes from a variety of different sources. The practice of using DNA obtained from peripheral blood mononuclear cells (PBMC) as a fully unmethylated DNA control is valid in cases where the samples are indeed completely unmethylated at the loci of interest. Several reports have focused on detecting DNA methylation status in peripheral blood, showing biologically important methylation levels for multiple genes (52,67). For example, low level methylation of many cancer-relevant genes may be found in the PBMCs from normal individuals. Therefore, the indiscriminate use of DNA from PBMCs as a negative control in sensitive assays for DNA methylation detection may be particularly problematic (52).

Manufacturers of commercially available DNA controls have different strategies for providing fully methylated and unmethylated DNA. For example, fully non-methylated kits from Zymo and Millipore use DNA from cells that contain genetic knockouts of 2 key DNA methyltransferases, thus reducing methylation levels by more than 95% (68).

Fully methylated DNA can be obtained from M.SssI-methylated DNA from, among many sources, double knockout cells for DNMT1 and DNMT3b (Table 2). Another alternative is to use the product from whole genome amplification (WGA) with kits such as REPLI-g (Qiagen), which does not reproduce the DNA methylation pattern and has a theoretical methylation level of less than 10⁻⁶. However, this amplification approach may carry the risk of reduced representation of the loci of interest (69). Therefore, the use of identical amounts of methylated and unmethylated controls derived from the same class of template (genomic DNA or WGA-products) could guarantee that equivalent amounts of effective templates are included.

Discussion

Studying DNA methylation for a candidate ROI using PCR-based methods is a topic of present and future importance. There are many advantages of genome-wide platforms; however, PCR-based techniques permit detailed analysis of specific regions of the genome, including CpG island shores. In addition, the associated costs of implementing and executing PCR-based techniques are lower, allowing the initial study of several candidate ROIs. PCR-based approaches also offer the advantage of a lower burden of false discoveries and the ability to confirm a large number of ROIs identified in genome-wide screening of a few samples (70).

DNA methylation analysis using pyrosequencing is a quantitative approach that does not require a cloning step, but presents the risk of PCR bias, similar to BSP. More importantly, pyrosequencing instrumentation is not commonly available in a general laboratory. For interested readers, comparisons and discussions of pyrosequencing techniques are available elsewhere (71).

This article highlights several considerations for PCR-based DNA methylation studies. The approaches reviewed here have different advantages and disadvantages that should be evaluated before starting any DNA methylation study. Similarly, it is clear that the different PCR-based techniques discussed here have differences in CpG coverage and possibility for quantitation (Figure 4). A comparison of all PCR-based DNA methylation techniques is presented in Table 5.

**Table 5.** **Comparison of PCR-based DNA methylation techniques**

Several considerations concerning PCR-based methods for DNA methylation analysis will be crucial for the consolidation of the field of molecular epigenetics. Current needs in this field include (i) detailed experimental comparisons of results obtained with different PCR-based techniques (72), (ii) the availability of a large number of predesigned PCR-based DNA methylation assays to facilitate broad use, (iii) the implementation of minimum reporting guidelines for manuscripts describing results of PCR-based analyses of DNA methylation, including details of experimental conditions such as controls, primer sequences, and programs used for primer design (73), (iv) the further development of additional PCR-based techniques that allow DNA methylation measurements in a more quantitative and reproducible way (5), and (v) the implementation of automatic and multiplexed protocols for DNA methylation using currently available techniques to improve efficiency and reduce costs (59). For readers interested in genome-wide DNA methylation analysis, we recommend two available review articles (74,75). Finally, it is critical to keep in mind that the results of PCR-based DNA methylation methodologies are reliable only in an experimental setting with adequate methodological controls.

Acknowledgments

This work was supported by grants from Colciencias (Contract # 401-2011), UAN-VCTI, and UNAL-DIB. HGH is a recipient of a PhD fellowship from Colciencias. The authors would like to thank the anonymous reviewers for their important comments and suggestions.

Competing interests

The authors declare no competing interests.

References

1. Portela, A. and M. Esteller. 2010. Epigenetic modifications and human disease. Nat. Biotechnol. 28:1057–1068.Crossref, Medline, CAS, Google Scholar
2. Iraola-Guzmán, S., X. Estivill, and R. Rabionet. 2011. DNA methylation in neurodegenerative disorders: a missing link between genome and environment? Clin. Genet. 80:1–14.Crossref, Medline, CAS, Google Scholar
3. Heyn, H. and M. Esteller. 2012. DNA methylation profiling in the clinic: applications and challenges. Nat. Rev. Genet. 13:679–692.Crossref, Medline, CAS, Google Scholar
4. Irizarry, R.A., C. Ladd-Acosta, B. Wen, Z. Wu, C. Montano, P. Onyango, H. Cui, K. Gabo, et al. . 2009. The human colon cancer methylome shows similar hypo- and hypermethylation at conserved tissue-specific CpG island shores. Nat. Genet. 41:178–186.Crossref, Medline, CAS, Google Scholar
5. Kristensen, L.S., M.B. Treppendahl, and K. Gronbaek. 2013. Analysis of epigenetic modifications of DNA in human cells. Curr. Prot. Hum. Genet. 20:Unit20 22.Google Scholar
6. Candiloro, I.L., T. Mikeska, and A. Dobrovic. 2011. Closed-tube PCR methods for locus-specific DNA methylation analysis. Methods Mol. Biol. 791:55–71.Crossref, Medline, CAS, Google Scholar
7. Rother, K.I., J. Silke, O. Georgiev, W. Schaffner, and K. Matsuo. 1995. Influence of DNA sequence and methylation status on bisulfite conversion of cytosine residues. Anal. Biochem. 231:263–265.Crossref, Medline, CAS, Google Scholar
8. Grunau, C., S.J. Clark, and A. Rosenthal. 2001. Bisulfite genomic sequencing: systematic investigation of critical experimental parameters. Nucleic Acids Res. 29:E65.Crossref, Medline, CAS, Google Scholar
9. Munson, K., J. Clark, K. Lamparska-Kupsik, and S.S. Smith. 2007. Recovery of bisulfite-converted genomic sequences in the methylation-sensitive QPCR. Nucleic Acids Res. 35:2893–2903.Crossref, Medline, CAS, Google Scholar
10. Ehrich, M., S. Zoll, S. Sur, and D. van den Boom. 2007. A new method for accurate assessment of DNA quality after bisulfite treatment. Nucleic Acids Res. 35:e29.Crossref, Medline, Google Scholar
11. Herman, J.G., J.R. Graff, S. Myohanen, B.D. Nelkin, and S.B. Baylin. 1996. Methylation-specific PCR: a novel PCR assay for methylation status of CpG islands. Proc. Natl. Acad. Sci. USA 93:9821–9826.Crossref, Medline, CAS, Google Scholar
12. Sriraksa, R., P. Chaopatchayakul, P. Jearanaikoon, C. Leelayuwat, and T. Limpaiboon. 2010. Verification of complete bisulfite modification using Calponin-specific primer sets. Clin. Biochem. 43:528–530.Crossref, Medline, CAS, Google Scholar
13. Tse, M.Y., J.E. Ashbury, N. Zwingerman, W.D. King, S.A. Taylor, and S.C. Pang. 2011. A refined, rapid and reproducible high resolution melt (HRM)-based method suitable for quantification of global LINE-1 repetitive element methylation. BMC Res Notes. 4:565.Crossref, Medline, CAS, Google Scholar
14. Ahn, S.J., J. Costa, and J.R. Emanuel. 1996. PicoGreen quantitation of DNA: effective evaluation of samples pre- or post-PCR. Nucleic Acids Res. 24:2623–2625.Crossref, Medline, CAS, Google Scholar
15. Li, L.C. and R. Dahiya. 2002. MethPrimer: designing primers for methylation PCRs. Bioinformatics 18:1427–1431.Crossref, Medline, CAS, Google Scholar
16. Tusnády, G.E., I. Simon, A. Varadi, and T. Aranyi. 2005. BiSearch: primer-design and search tool for PCR on bisulfite-treated genomes. Nucleic Acids Res. 33:e9.Crossref, Medline, Google Scholar
17. Schüffler, P., T. Mikeska, A. Waha, T. Lengauer, and C. Bock. 2009. MethMarker: user-friendly design and optimization of gene-specific DNA methylation assays. Genome Biol. 10:R105.Crossref, Medline, Google Scholar
18. Brandes, J.C., H. Carraway, and J.G. Herman. 2007. Optimal primer design using the novel primer design program: MSPprimer provides accurate methylation analysis of the ATM promoter. Oncogene 26:6229–6237.Crossref, Medline, CAS, Google Scholar
19. Sherry, S.T., M.H. Ward, M. Kholodov, J. Baker, L. Phan, E.M. Smigielski, and K. Sirotkin. 2001. dbSNP: the NCBI database of genetic variation. Nucleic Acids Res. 29:308–311.Crossref, Medline, CAS, Google Scholar
20. Warnecke, P.M., C. Stirzaker, J.R. Melki, D.S. Millar, C.L. Paul, and S.J. Clark. 1997. Detection and measurement of PCR bias in quantitative methylation analysis of bisulphite-treated DNA. Nucleic Acids Res. 25:4422–4426.Crossref, Medline, CAS, Google Scholar
21. Wojdacz, T.K. and L.L. Hansen. 2006. Reversal of PCR bias for improved sensitivity of the DNA methylation melting curve assay. Biotechniques 41:274–278.Link, CAS, Google Scholar
22. Chhibber, A. and B.G. Schroeder. 2008. Single-molecule polymerase chain reaction reduces bias: application to DNA methylation analysis by bisulfite sequencing. Anal. Biochem. 377:46–54.Crossref, Medline, CAS, Google Scholar
23. Anzaldi, L.J., D. Muñoz-Fernández, and I. Erill. 2012. BioWord: a sequence manipulation suite for Microsoft Word. BMC Bioinformatics 13:124.Crossref, Medline, Google Scholar
24. Kalendar, R., D. Lee, and A.H. Schulman. 2011. Java web tools for PCR, in silico PCR, and oligonucleotide assembly and analysis. Genomics 98:137–144.Crossref, Medline, CAS, Google Scholar
25. Clark, S.J., J. Harrison, C.L. Paul, and M. Frommer. 1994. High sensitivity mapping of methylated cytosines. Nucleic Acids Res. 22:2990–2997.Crossref, Medline, CAS, Google Scholar
26. Jiang, M., Y. Zhang, J. Fei, X. Chang, W. Fan, X. Qian, T. Zhang, and D. Lu. 2010. Rapid quantification of DNA methylation by measuring relative peak heights in direct bisulfite-PCR sequencing traces. Lab. Invest. 90:282–290.Crossref, Medline, CAS, Google Scholar
27. Clark, S.J., A. Statham, C. Stirzaker, P.L. Molloy, and M. Frommer. 2006. DNA methylation: bisulphite modification and analysis. Nat. Protoc. 1:2353–2364.Crossref, Medline, CAS, Google Scholar
28. Frommer, M., L.E. McDonald, D.S. Millar, C.M. Collis, F. Watt, G.W. Grigg, P.L. Molloy, and C.L. Paul. 1992. A genomic sequencing protocol that yields a positive display of 5-methylcytosine residues in individual DNA strands. Proc. Natl. Acad. Sci. USA 89:1827–1831.Crossref, Medline, CAS, Google Scholar
29. Lewin, J., A.O. Schmitt, P. Adorjan, T. Hildmann, and C. Piepenbrock. 2004. Quantitative DNA methylation analysis based on four-dye trace data from direct sequencing of PCR amplificates. Bioinformatics 20:3005–3012.Crossref, Medline, CAS, Google Scholar
30. Weisenberger, D.J., B.N. Trinh, M. Campan, S. Sharma, T.I. Long, S. Ananthnarayan, G. Liang, F.J. Esteva, et al. . 2008. DNA methylation analysis by digital bisulfite genomic sequencing and digital MethyLight. Nucleic Acids Res. 36:4689–4698.Crossref, Medline, CAS, Google Scholar
31. Snell, C., M. Krypuy, E.M. Wong, kConFab investigators, M.B. Loughrey, and A. Dobrovic. 2008. BRCA1 promoter methylation in peripheral blood DNA of mutation negative familial breast cancer patients with a BRCA1 tumour phenotype. Breast Cancer Res. 10:R12.Crossref, Medline, Google Scholar
32. Wojdacz, T.K., T. Borgbo, and L.L. Hansen. 2009. Primer design versus PCR bias in methylation independent PCR amplifications. Epigenetics 4:231–234.Crossref, Medline, CAS, Google Scholar
33. Han, W., S. Cauchi, J.G. Herman, and S.D. Spivack. 2006. DNA methylation mapping by tag-modified bisulfite genomic sequencing. Anal. Biochem. 355:50–61.Crossref, Medline, CAS, Google Scholar
34. Warnecke, P.M., C. Stirzaker, J. Song, C. Grunau, J.R. Melki, and S.J. Clark. 2002. Identification and resolution of artifacts in bisulfite sequencing. Methods 27:101–107.Crossref, Medline, CAS, Google Scholar
35. Shen, L., Y. Guo, X. Chen, S. Ahmed, and J.P. Issa. 2007. Optimizing annealing temperature overcomes bias in bisulfite PCR methylation analysis. Biotechniques 42:48–58.Link, CAS, Google Scholar
36. Wojdacz, T.K. and A. Dobrovic. 2007. Methylation-sensitive high resolution melting (MS-HRM): a new approach for sensitive and high-throughput assessment of methylation. Nucleic Acids Res. 35:e41.Crossref, Medline, Google Scholar
37. Wojdacz, T.K., A. Dobrovic, and L.L. Hansen. 2008. Methylation-sensitive high-resolution melting. Nat. Protoc. 3:1903–1908.Crossref, Medline, CAS, Google Scholar
38. Bock, C., S. Reither, T. Mikeska, M. Paulsen, J. Walter, and T. Lengauer. 2005. BiQ Analyzer: visualization and quality control for DNA methylation data from bisulfite sequencing. Bioinformatics 21:4067–4068.Crossref, Medline, CAS, Google Scholar
39. Rohde, C., Y. Zhang, R. Reinhardt, and A. Jeltsch. 2010. BISMA--fast and accurate bisulfite sequencing data analysis of individual clones from unique and repetitive sequences. BMC Bioinformatics 11:230.Crossref, Medline, Google Scholar
40. Roth, T.L., F.D. Lubin, A.J. Funk, and J.D. Sweatt. 2009. Lasting epigenetic influence of early-life adversity on the BDNF gene. Biol. Psychiatry 65:760–769.Crossref, Medline, CAS, Google Scholar
41. Fraga, M.F. and M. Esteller. 2002. DNA methylation: a profile of methods and applications. Biotechniques 33:632–649.Link, CAS, Google Scholar
42. Palmisano, W.A., K.K. Divine, G. Saccomanno, F.D. Gilliland, S.B. Baylin, J.G. Herman, and S.A. Belinsky. 2000. Predicting lung cancer by detecting aberrant promoter methylation in sputum. Cancer Res. 60:5954–5958.Medline, CAS, Google Scholar
43. Derks, S., M.H. Lentjes, D.M. Hellebrekers, A.P. de Bruine, J.G. Herman, and M. van Engeland. 2004. Methylation-specific PCR unraveled. Cell. Oncol. 26:291–299.Medline, CAS, Google Scholar
44. Pattyn, F., J. Hoebeeck, P. Robbrecht, E. Michels, A. De Paepe, G. Bottu, D. Coornaert, R. Herzog, et al. . 2006. methBLAST and methPrimerDB: web-tools for PCR based methylation analysis. BMC Bioinformatics 7:496.Crossref, Medline, Google Scholar
45. Thomassin, H., C. Kress, and T. Grange. 2004. MethylQuant: a sensitive method for quantifying methylation of specific cytosines within the genome. Nucleic Acids Res. 32:e168.Crossref, Medline, Google Scholar
46. Kristensen, L.S., T. Mikeska, M. Krypuy, and A. Dobrovic. 2008. Sensitive Melting Analysis after Real Time- Methylation Specific PCR (SMART-MSP): high-throughput and probe-free quantitative DNA methylation detection. Nucleic Acids Res. 36:e42.Crossref, Medline, Google Scholar
47. Kristensen, L.S., H.M. Nielsen, H. Hager, and L.L. Hansen. 2011. Methylation of MGMT in malignant pleural mesothelioma occurs in a subset of patients and is associated with the T allele of the rs16906252 MGMT promoter SNP. Lung Cancer 71:130–136.Crossref, Medline, Google Scholar
48. Rand, K., W. Qu, T. Ho, S.J. Clark, and P. Molloy. 2002. Conversion-specific detection of DNA methylation using real-time polymerase chain reaction (ConLight-MSP) to avoid false positives. Methods 27:114–120.Crossref, Medline, CAS, Google Scholar
49. Shaw, R.J., E.K. Akufo-Tetteh, J.M. Risk, J.K. Field, and T. Liloglou. 2006. Methylation enrichment pyrosequencing: combining the specificity of MSP with validation by pyrosequencing. Nucleic Acids Res. 34:e78.Crossref, Medline, Google Scholar
50. Häfner, N., H. Diebolder, L. Jansen, I. Hoppe, M. Durst, and I.B. Runnebaum. 2011. Hypermethylated DAPK in serum DNA of women with uterine leiomyoma is a biomarker not restricted to cancer. Gynecol. Oncol. 121:224–229.Crossref, Medline, Google Scholar
51. Licchesi, J.D. and J.G. Herman. 2009. Methylation-specific PCR. Methods Mol. Biol. 507:305–323.Crossref, Medline, CAS, Google Scholar
52. Kristensen, L.S., M.P. Raynor, I. Candiloro, and A. Dobrovic. 2012. Methylation profiling of normal individuals reveals mosaic promoter methylation of cancer-associated genes. Oncotarget. 3:450–461.Crossref, Medline, Google Scholar
53. Preusser, M., L. Elezi, and J.A. Hainfellner. 2008. Reliability and reproducibility of PCR-based testing of O6-methylguanine- DNA methyltransferase gene (MGMT) promoter methylation status in formalin-fixed and paraffin-embedded neurosurgical biopsy specimens. Clin. Neuropathol. 27:388–390.Crossref, Medline, CAS, Google Scholar
54. Livak, K.J., J. Marmaro, and J.A. Todd. 1995. Towards fully automated genome-wide polymorphism screening. Nat. Genet. 9:341–342.Crossref, Medline, CAS, Google Scholar
55. Eads, C.A., K.D. Danenberg, K. Kawakami, L.B. Saltz, C. Blake, D. Shibata, P.V. Danenberg, and P.W. Laird. 2000. MethyLight: a high-throughput assay to measure DNA methylation. Nucleic Acids Res. 28:E32.Crossref, Medline, CAS, Google Scholar
56. Zeschnigk, M., S. Bohringer, E.A. Price, Z. Onadim, L. Masshofer, and D.R. Lohmann. 2004. A novel real-time PCR assay for quantitative analysis of methylated alleles (QAMA): analysis of the retinoblastoma locus. Nucleic Acids Res. 32:e125.Crossref, Medline, Google Scholar
57. Campan, M., D.J. Weisenberger, B. Trinh, and P.W. Laird. 2009. MethyLight. Methods Mol. Biol. 507:325–337.Crossref, Medline, CAS, Google Scholar
58. Houshdaran, S., V.K. Cortessis, K. Siegmund, A. Yang, P.W. Laird, and R.Z. Sokol. 2007. Widespread epigenetic abnormalities suggest a broad DNA methylation erasure defect in abnormal human sperm. PLoS ONE 2:e1289.Crossref, Medline, Google Scholar
59. He, Q., H.Y. Chen, E.Q. Bai, Y.X. Luo, R.J. Fu, Y.S. He, J. Jiang, and H.Q. Wang. 2010. Development of a multiplex MethyLight assay for the detection of multigene methylation in human colorectal cancer. Cancer Genet. Cytogenet. 202:1–10.Crossref, Medline, CAS, Google Scholar
60. Wojdacz, T.K., L.L. Hansen, and A. Dobrovic. 2008. A new approach to primer design for the control of PCR bias in methylation studies. BMC Res Notes. 1:54.Crossref, Medline, Google Scholar
61. Worm, J., A. Aggerholm, and P. Guldberg. 2001. In-tube DNA methylation profiling by fluorescence melting curve analysis. Clin. Chem. 47:1183–1189.Crossref, Medline, CAS, Google Scholar
62. Candiloro, I.L., T. Mikeska, P. Hokland, and A. Dobrovic. 2008. Rapid analysis of heterogeneously methylated DNA using digital methylation-sensitive high resolution melting: application to the CDKN2B (p15) gene. Epigenetics Chromatin. 1:7.Crossref, Medline, Google Scholar
63. Balic, M., M. Pichler, J. Strutz, E. Heitzer, C. Ausch, H. Samonigg, R.J. Cote, and N. Dandachi. 2009. High quality assessment of DNA methylation in archival tissues from colorectal cancer patients using quantitative high-resolution melting analysis. J. Mol. Diagn. 11:102–108.Crossref, Medline, CAS, Google Scholar
64. Liu, W., M. Guan, B. Su, C. Ye, J. Li, X. Zhang, C. Liu, M. Li, et al. . 2010. Quantitative assessment of AKAP12 promoter methylation in colorectal cancer using methylation-sensitive high resolution melting: Correlation with Duke's stage. Cancer Biol. Ther. 9:862–871.Crossref, Medline, CAS, Google Scholar
65. Fox, B.P. and R.P. Kandpal. 2006. Transcriptional silencing of EphB6 receptor tyrosine kinase in invasive breast carcinoma cells and detection of methylated promoter by methylation specific PCR. Biochem. Biophys. Res. Commun. 340:268–276.Crossref, Medline, CAS, Google Scholar
66. Brock, M.V., C.M. Hooker, E. Ota-Machida, Y. Han, M. Guo, S. Ames, S. Glockner, S. Piantadosi, et al. . 2008. DNA methylation markers and early recurrence in stage I lung cancer. N. Engl. J. Med. 358:1118–1128.Crossref, Medline, CAS, Google Scholar
67. Langevin, S.M., D.C. Koestler, B.C. Christensen, R.A. Butler, J.K. Wiencke, H.H. Nelson, E.A. Houseman, C.J. Marsit, and K.T. Kelsey. 2012. Peripheral blood DNA methylation profiles are indicative of head and neck squamous cell carcinoma: an epigenome-wide association study. Epigenetics 7:291–299.Crossref, Medline, CAS, Google Scholar
68. Rhee, I., K.E. Bachman, B.H. Park, K.W. Jair, R.W. Yen, K.E. Schuebel, H. Cui, A.P. Feinberg, et al. . 2002. DNMT1 and DNMT3b cooperate to silence genes in human cancer cells. Nature 416:552–556.Crossref, Medline, CAS, Google Scholar
69. Dean, F.B., S. Hosono, L. Fang, X. Wu, A.F. Faruqi, P. Bray-Ward, Z. Sun, Q. Zong, et al. . 2002. Comprehensive human genome amplification using multiple displacement amplification. Proc. Natl. Acad. Sci. USA 99:5261–5266.Crossref, Medline, CAS, Google Scholar
70. Oh, T., N. Kim, Y. Moon, M.S. Kim, B.D. Hoehn, C.H. Park, T.S. Kim, N.K. Kim, et al. . 2013. Genome-Wide Identification and Validation of a Novel Methylation Biomarker, SDC2, for Blood-Based Detection of Colorectal Cancer. J. Mol. Diagn. 15:498–507.Crossref, Medline, CAS, Google Scholar
71. Tost, J. and I.G. Gut. 2007. DNA methylation analysis by pyrosequencing. Nat. Protoc. 2:2265–2275.Crossref, Medline, CAS, Google Scholar
72. Quillien, V., A. Lavenu, L. Karayan-Tapon, C. Carpentier, M. Labussiere, T. Lesimple, O. Chinot, M. Wager, et al. . 2012. Comparative assessment of 5 methods (methylation-specific polymerase chain reaction, methyl ight, pyrosequencing, methylation-sensitive high-resolution melting, and immunohistochemistry) to analyze O6-methylguanine-DNA-methyltranferase in a series of 100 glioblastoma patients. Cancer 118:4201–4211.Medline, CAS, Google Scholar
73. Taylor, C.F., D. Field, S.A. Sansone, J. Aerts, R. Apweiler, M. Ashburner, C.A. Ball, P.A. Binz, et al. . 2008. Promoting coherent minimum reporting guidelines for biological and biomedical investigations: the MIBBI project. Nat. Biotechnol. 26:889–896.Crossref, Medline, CAS, Google Scholar
74. Gupta, R., A. Nagarajan, and N. Wajapeyee. 2010. Advances in genome-wide DNA methylation analysis. Biotechniques 49:iii–vi.Link, CAS, Google Scholar
75. Laird, P.W. 2010. Principles and challenges of genomewide DNA methylation analysis. Nat. Rev. Genet. 11:191–203.Crossref, Medline, CAS, Google Scholar