Combined Bisulfite Restriction Analysis


Most molecular biological techniques used to analyze specific loci in complex genomic DNA involve some form of sequence-specific amplification, whether it is biological amplification by cloning in Escherichia coli, direct amplification by polymerase chain reaction (PCR), or signal amplification by hybridization with a probe that can be visualized. Since DNA methylation is added post replicatively by a dedicated maintenance DNA methyltransferase that is not present in either E. coli or in the PCR reaction, the methylation information is lost during molecular cloning or PCR amplification. Molecular hybridization does not discriminate between methylated and unmethylated DNA since the methyl group on the cytosine does not participate in base pairing. The lack of a facile way to amplify the methylation information in complex genomic DNA has been a significant impediment to DNA methylation research. The indirect methods that have been developed in the past decade to detect DNA methylation patterns at specific loci rely on techniques that alter the genomic DNA in a methylation-dependent manner before the amplification event. There are two main methods that have been utilized to achieve this methylation-dependent DNA alteration. The first is digestion by a restriction enzyme that is affected by its activity by 5-methylcytosine in a CpG sequence context. The cleavage or lack of it can subsequently be revealed by Southern blotting or by PCR. The other technique that has received recent widespread use is the treatment of genomic DNA with sodium bisulfite. This treatment converts all unmethylated cytosines in the DNA to uracil by deamination but leaves the methylated cytosine residues intact. Subsequent PCR amplification replaces the uracil residues with thymines and the 5-methylcytosine residues with cytosines. The resulting sequence difference can be detected using a variety of methods.

Sodium Bisulfite Techniques

All bisulfite-based methods are followed by a PCR reaction to analyze specific loci within the genome. There are two ways in which the sequence difference generated by the sodium bisulfite treatment can be revealed. The first is to design PCR primers that uniquely anneal with either methylated or unmethylated converted DNA. This technique is referred to as “methylation-specific PCR” or “MSP”. The method used by all other bisulfite-based techniques (such as bisulfite genomic sequencing, combined bisulfite restriction analysis [COBRA], and Ms-SNuPE) is to amplify the bisulfite-converted DNA using primers that anneal at locations that lack CpG dinucleotides in the original genomic sequence. In this way, the PCR primers can amplify the sequence in between the two primers, regardless of the DNA methylation status of that sequence in the original genomic DNA. This results in a pool of different PCR products, all with the same length and differing in their sequence only at the sites of potential DNA methylation at CpGs located in between the two primers. The difference between these methods of processing the bisulfite converted sequence is that in MSP, the methylation information is derived from the presence or absence of a PCR product, whereas in the other techniques a mix of products is always generated and the mixture is subsequently analyzed to yield quantitative information on the relative occurrence of the different methylation states. The best way to view this application is to envision that genomic DNA usually consists of a collection of many different methylation patterns. The PCR reaction will amplify each of these variants without affecting the relative ratio between them. Herein lies both the power and the Achilles heel of these techniques. Since all methylation profiles are amplified equally, quantitative information about DNA methylation patterns can be distilled from the resulting PCR pool. However, it is then essential that no bias occurs during the PCR reaction. If a particular sequence variant amplifies with different kinetics than another sequence variant, then incorrect quantitative data will result. Several methods of reducing or preventing PCR bias have been developed. An additional difficulty with all non-MSP variants of bisulfite-based DNA methylation analysis techniques is that it is not always easy to find suitable primers lacking CpGs in very dense CpG islands.


Assuming that the PCR product is a faithful representation of the original collection of DNA sequences following sodium bisulfite treatment, then this resulting pool of PCR products can be analyzed by any technique capable of detecting sequence differences, preferably in a quantitative fashion. COBRA, which stands for Combined Bisulfite Restriction Analysis, is based on the restriction digestion of the PCR product with an enzyme for which the recognition sequence is affected by the methylation state in the original DNA. Accurate quantitation of the percent methylation can be obtained by subsequent quantitative hybridization. COBRA can be applied to DNA samples derived from cell lines and tissues. It has even been shown to be compatible with the analysis of microdissected paraffin-embedded sections. In addition, COBRA is not as labor-intensive as some of the other bisulfite te-based techniques. One interesting feature that COBRA shares with bisulfite genomic sequencing of subclones are that linkage of methylation patterns within individual molecules can be investigated. If two individual CpG sites are interrogated in a COBRA analysis, then a lack of cutting at both sites or complete cutting at both sites with little if any single cutting would suggest that the genomic DNA is comprised of a pool of fully methylated and fully unmethylated DNA molecules and not of mixed methylation. MS-SNuPE interrogates each CpG site independently of other CpGs and would not reveal this distinction.

Experimental Design

Sequence Manipulation

Sodium bisulfite treatment substantially alters the primary sequence of genomic DNA. Unmethylated cytosines, which comprise the vast majority of cytosine residues in vertebrate DNA, are chemically converted to uracil residues (which are later substituted by thymine residues during PCR amplification). The immediate consequence of this conversion is that the original DNA strands are no longer complementary. Throughout the genome, the two DNA strands now represent two different, noncomplementary, single-stranded sequences. To design a COBRA assay for a particular region of interest, the first step is the selection of one of the two DNA strand sequences for PCR amplification. For COBRA, the predominant issue in this choice of DNA strand is the location of suitable restriction sites in the final PCR products. One of the confusing concepts for researchers new to bisulfite-based methods is the fact that the amplification of the two opposing strands requires different PCR primers and that the resulting restriction maps of the PCR products will differ. Even the G+C content can differ between the top-strand amplicons and the bottom-strand amplicons if the distribution of G residues was unequal in the original DNA. To facilitate the design of PCR primers and the localization of restriction-enzyme sites in the subsequent PCR products, it is advisable to perform a theoretical bisulfite conversion for both a fully methylated and completely unmethylated version of both the top and the bottom strand of the sequence of interest. This will generate four new sequences, methylated and unmethylated for both top and bottom, respectively.

PCR Primer Design

Once the four sequences and their accompanying restriction maps have been generated, primers can be chosen that flank restriction sites that are differentially present based on the methylation status of the original genomic DNA. In the primer design, it is best to use the bisulfite converted methylated sequence versions (top or bottom) as a guide. This facilitates the avoidance of CpG dinucleotides within the primer sequences. It is essential that neither the PCR primers nor the hybridization probes encompass CpG dinucleotides to ensure equal recognition of DNA regardless of original methylation status. It is helpful to highlight the CpG dinucleotides in the sequence. This can be done by hand or by instructing the DNA analysis software to recognize CpG as a restriction site.

PCR amplification of bisulfite-treated DNA is more difficult than of native DNA. Due to the depletion of cytosine residues from the genomic DNA, the resulting PCR products contain an unequal distribution of bases. Each PCR product contains one C-poor/T-rich strand and one G-poor/A-rich strand. This reduced sequence complexity diminishes the discriminatory ability of the PCR primers and of the hybridization probes. Therefore, it is advisable to design primers slightly longer than for standard PCR. We prefer primers of at least 24 bases. It may be difficult to design such long primers in very dense CpG islands while simultaneously avoiding the inclusion of CpGs within the primer sequences. If possible, other standard criteria for PCR primer design should be 0adhered to. These include a primer G+C content of 40–60%, similar Tm values for the primer pairs and the avoidance of palindromic or repetitive sequences within the primers and of 32 complementary nucleotides between primer pairs to prevent primer-dimer formation. Due to the difficulties of PCR amplification of bisulfite-treated DNA, it is advisable to design relatively small amplicons. A maximum length of 150 bp is advised if paraffin-embedded samples are to be analyzed. In addition, PCR bias seems to occur less frequently for small amplicons.

Restriction-Enzyme Choice

Some restriction-enzyme sequences are innate in the initial sequence and retained after bisulfite treatment. However, new sites may be generated by the bisulfite conversion and subsequent PCR amplification. For example, the restriction-enzyme site for TaqI (TCGA) can be retained in a methylation-dependent manner. Sites that are created, rather than merely retained, are preferable since the use of these sites helps to verify complete bisulfite modification of the DNA. The site will not be created if the bisulfite treatment is insufficient. It is important to stress that the restriction-enzyme cleavage itself is not methylation-dependent. PCR products do not contain 5-methylcytosine. The methylation status is revealed by the presence or absence of a restriction enzyme site, not by inhibition of cleavage by methylation of the restriction site. In order to analyze a specific region, there must be at least one restriction site within the methylated bisulfite converted strand that is absent in the unmethylated bisulfite te-converted strand or vice versa. The easiest way to identify suitable restriction-enzyme sites is to use a DNA-analysis program to generate restriction maps for bisulfite-converted sequences representing the methylated and unmethylated versions of a sequence. An advantage of COBRA is that more than one restriction site can be tested on one PCR product given that additional sites are available. A single PCR amplification reaction can be analyzed for any number of restriction enzymes and hybridization probes.

Probe Design

Oligonucleotide probes should not cover either restriction enzyme-recognition sites of the enzymes used in the COBRA analysis, nor should they contain CpG dinucleotide sequences. Longer oligos are easier to use in hybridization reactions, although CpG-rich CpG islands sometimes necessitate the use of probes as short as 15 bases to avoid inclusion of a CpG dinucleotide within the probe.

Sodium Bisulfite Treatment

The sodium bisulfite conversion of cytosine proceeds through several steps. Sulfonation of cytosine at the C-6 position can only occur on single-stranded DNA. Therefore, it is essential that the genomic DNA is fully denatured and remains denatured until sulfonation is complete. Bisulfite induced deamination of both methylated and unmethylated cytosine residues occurs, but the reactivity of 5-methylcytosine is much lower than that of unmethylated cytosine residues. A competing reaction is the depurination of DNA, which can lead to severe degradation to the point of failure of the PCR reaction. The difficulty of sodium bisulfite conversion of genomic DNA is to find the best balance of complete denaturation of the DNA with complete conversion of unmethylated cytosine residues with minimal DNA loss, depurination, and conversion of 5-methylcytosine residues. Various improvements and modifi cations of the original protocol have been proposed in an attempt to achieve the best balance.

There has been some confusion about concentrations of sodium bisulfite solutions. Sodium bisulfite is sold as a mixture of sodium bisulfite (NaHSO3, FW = 104) and sodium metabisulfite (Na2S2O5, FW = 190). The ratio can be calculated from the SO2 assay listed on the bottle. The bottle usually contains mostly sodium metabisulfite. However, since sodium metabisulfite is converted to sodium bisulfite upon dissolution in water (one mole of sodium metabisulfite gives rise to two moles of sodium bisulfite), it is simpler to purchase pure (98.8%) sodium metabisulfite and dissolve it at half the specified sodium bisulfite concentration. For example, for a 5 M concentration of sodium bisulfite, prepare a 2.5 M solution of sodium metabisulfite. The preparation of high concentrations of sodium (meta) bisulfite require the addition of NaOH to allow the salt to go into solution. Sodium bisulfite oxidizes easily. Therefore, care should be taken not to excessively aerate during the dissolving process. Heating to 50°C may be necessary to obtain complete dissolution of the sodium bisulfite. Hydroquinone is added to prevent oxidation of the sodium bisulfite. The original protocol calls for a freshly prepared 3.6 M solution of sodium bisulfite to be added, along with the hydroquinone to the denatured DNA to give a final concentration of 3.1 M sodium bisulfite, followed by a 16-h incubation at 55°C. However, variations in the concentration of sodium bisulfite, the incubation temperature and time may yield a better balance of complete conversion, while maintaining the integrity of the DNA. Also, it is more efficient to combine the sodium bisulfite and hydroquinone solutions before adding them to the individual denatured DNA samples. Investigators are encouraged to experiment with time, temperature, and sodium bisulfite concentrations to find the best conditions for their particular application.

Incomplete conversion of unmethylated cytosine residues is occasionally seen. Therefore, it is essential to check this for the conditions used in a particular application. Complete conversion of the DNA can be readily verified by restriction digestion with an enzyme that contains a cytosine in the recognition sequence that is not within a CpG sequence context. Such sites should be completely lost during bisulfite conversion since the unmethylated cytosine should be converted to thymine. Any cutting of the PCR product by such an enzyme indicates either non-CpG methylation or incomplete bisulfite conversion. Comparison of the restriction maps of the unconverted sequence with the maps generated by converted sequence should yield several choices of control enzymes. The addition of urea can improve the efficiency of conversion by maintaining the DNA in a denatured state.


As with any PCR reaction, initial optimization of the thermal-cycling parameters is advisable. The lower sequence complexity of the bisulfite converted DNA and the amplifi cation primers and potential degradation of the DNA by depurination contribute to the diffi culty of bisulfi te PCR reactions. Initial denaturation of the DNA for 2–4 min at 95°C in the fi rst cycle seems to be benefi cial. A 1-min denaturation can suffi ce for subsequent cycles. We obtain better results with mixtures of Taq polymerase and high-fi delity polymerases such as Roche Boehringer Mannheim Expand Hi Fidelity polymerase. The number of cycles needed to generate a product depends on the number of starting molecules. For cell-line and tissue DNA samples, the amount of DNA is often at the microgram level, in which case 30 cycles are more than sufficient to generate a robust PCR product. However, since paraffi n-embedded samples may have less than a nanogram of DNA initially and subsequent loss and degradation of DNA occurs during the bisulfi te treatment, it may be necessary to increase the number of cycles to 40. In extreme cases, nested PCR can be employed, but this increases the risk of PCR bias. Other parameters, such as MgCl2 and primer concentrations should be optimized as for any PCR assay.

Restriction Enzyme Digestion

Following PCR amplification the product must be cleaned up before further restriction-digestion analysis. The residual salts from the PCR buffers may inhibit complete enzyme digestion. In addition, some proprietary PCR buffers, such as those that are supplied with the Expand polymerase, contain components that are inhibitory to restriction-enzyme digestion. If the PCR produces a strong single band on an agarose gel, then the product can be simply purified by a commercial PCR clean-up kit or microfiltration spin column. However, if nonspecific PCR products result from amplification, then gel extraction of the desired product is recommended. Restriction digestion is performed according to manufacturers specifications.

Polyacrylamide Gel Electrophoresis, Electroblotting, and Hybridization

After restriction digestion of the purified PCR product, the sample is separated on an 8% denaturing polyacrylamide gel. The large volume of the restriction digestion can be a problem in achieving a fourfold volume of denaturing loading dye, which is essential to prevent secondary structures. A high concentration of pure PCR product will allow the use of fairly small restriction-digestion volumes. In addition, the use of small protein gels with thick spacers allows for larger loading volumes and provides sufficient separation of digestion products. The extra thickness of the gels also facilitates manipulation of the gel during the electroblot set up. High-concentration agarose gels do not blot efficiently. Once the gel electrophoresis is completed, the DNA is transferred to a positively charged membrane by electroblotting. Electroblotting provides a fast and efficient transfer of the DNA. Following electroblotting, the DNA is crosslinked to the membrane by exposing the membrane to 1200J of UV. At this stage, the membrane can be processed immediately or wrapped in Saran wrap and stored at 4°C for future use.

The membrane is prepared for hybridization by first prewetting it in 6X SSC and then soaking it in prehybridization buffer at 42°C for at least 30 min. The membranes are then hybridized with a 52-end labeled oligonucleotide (15–25 nucleotides) overnight. The hybridization reaction depends on the length, concentration, base composition, and Tm of the probe. In general, the hybridization should be performed at 5°C below the Tm of the probe. However, if the probe is less than 15–20 nucleotides, then less stringent conditions are recommended. Specialized hybridization buffers and conditions that have been optimized for oligo hybridization can also be employed. For short probes (less than 15–20 nucleotides), the membrane is washed at low stringency using 2X SSC/1% SDS at 25–37°C. Longer probes allow more stringent conditions, 0.5–2X SSC/1% SDS at 42–65°C. The membranes are washed in a volume of 200 mL with replacement of the washing solution at 10–15 min intervals. Oligo probes do not require very long washing. We usually wash less than an hour. The filter is then exposed in a phosphoimager system which allows quantitation of the individual bands. DNA methylation levels are calculated according to the formula.

%Methylation=100 * (B/(A+B))

B-> amount of sample basepairs cut by TaqI

A->amount of sample basepairs with no methylation by TaqI

Materials & Methods

DNA Isolation

  1. Phosphate-buffered saline (PBS).
  2. Lysis Buffer: 100 mM Tris-HCl, pH 8.5, 10 mM ethylenediaminetetraacetic acid (EDTA), 200 mM NaCl, 1% SDS.
  3. Proteinase K.
  4. RNase A.
  5. Phenol/chloroform/isoamyl (25:24:1).
  6. Ethanol.
  7. TE Buffer: 10 mM Tris-HCl, pH 7.5, 0.1 mM EDTA.
  8. For Paraffin-embedded sections: Buffer K: 10 mM Tris-HCl, pH 8.0, 5 mM EDTA, 10 μg/μL Proteinase K.

Bisulfite Treatment

  1. 1–10 μg DNA.
  2. Salmon sperm DNA (only if initial DNA is < 1 μg).
  3. 2 M NaOH, a 5 M sodium bisulfite, pH 5.0, made from sodium metabisulfite, which also contains 125 mM Hydroquinone.
  4. Mineral oil.
  5. Wizard DNA clean-up System.
  6. 3cc syringes.
  7. 80% isopropanol.
  8. 3 M NaOH
  9. 5 M Ammonium Acetate, pH 7.4).
  10. Ethanol (100% and 70%).
  11. TE Buffer: 10 mM Tris-HCl, pH 7.5, 0.1 mM EDTA.


  1. 10X Expand buffer.
  2. 25 mM Expand MgCl.
  3. 4 mM dNTPs.
  4. 50 μM of each primer.
  5. 2 units of Expand HF Enzyme.
  6. Sterile ddH20.

DNA Extraction

  1. Commercial PCR clean-up kit or gel extraction kit.

Restriction Enzyme Digestion

  1. 10 ng of purified PCR product.
  2. 1 unit of enzyme.
  3. 1X enzyme buffer
  4. Sterile ddH2O.
  5. Mineral oil.

Polyacrylamide Gel Electrophoresis

  1. 8% denaturing polyacrylamide gel (7 M Urea).
  2. 10% Ammonium Persulfate.
  3. TEMED.
  4. Loading buffer: 98% formamide, 10 mM EDTA, 0.0025% xylene cyanol, 0.0025% bromophenol blue.
  5. 1X TBE Buffer.


  1. Zetabind charged membrane.
  2. Blotting paper.
  3. Fiber pads.
  4. 0.5X TBE.


  1. End-labeling probe: 10 pmoles oligonucleotide, 10X T4 polynucleotide kinase buffer, 20 units T4 polynucleotide kinase, γ-32P-ATP (10 mCi/mL), sterile ddH2O, G-50 sephadex, 1 cc syringe.
  2. Prehybridization: 6X SSC, 10 mL hybridization buffer: 500 mM phosphate buffer, pH 6.8, 1 mM EDTA, 7% SDS.


DNA Isolation

Genomic DNA can be collected from cultured cells, tissue, or paraffin-embedded sections by standard techniques. The quality of the DNA retrieved from this isolation procedure is appropriate for COBRA analysis. It is important to remove all traces of RNA from the sample in order to prevent RNA amplification during PCR.

Bisulfite Treatment

Carrier DNA needs to be added to samples that contain less than 1 μg of DNA since there can be substantial loss and degradation of DNA during the bisulfite conversion. We use 1–2 μg of salmon sperm DNA as a carrier. Embedding the DNA in agarose beads provides a good alternative to prevent loss of the DNA during the procedure.

  1. Add 1–10 μg of DNA to sterile ddH2O for a final volume of 18 μL and heat at 95°C for 20 min. Remove the sample from the heat and immediately place on ice.
  2. Add 2 μL of 3 M NaOH to denature the DNA and incubate for 20 min at 42°C.
  3. Prepare a fresh solution of 5 M sodium bisulfite solution/hydroquinone solution, pH 5.0, immediately prior to use. Calculate 0.5 mL of solution for each DNA sample. For 4 mL of solution (8 samples), add 1.9 g sodium metabisulfite to 2.5 mL sterile ddH2O. Add 0.7 mL 2 M NaOH. Add 0.5 mL 1 M hydroquinone (0.11 g in 1 mL H2O). Heat to 50°C. Invert frequently until fully dissolved.
  4. Add 380 μL of the 5 M sodium bisulfite/hydroquinone solution to the 20 μL of denatured DNA. Mix.
  5. Overlay sample with 5–6 drops of mineral oil to prevent evaporation and incubate 12–16 h at 50°C in the dark. Significant degradation may occur at longer incubation times.
  6. Desalt the sample with Wizard DNA clean-up system.
  7. Transfer to a clean tube and elute the DNA with 45 μL of sterile ddH2O.
  8. Add 5 μL of 3 M NaOH and incubate for 15 min at 37°C for desulfonation.
  9. Add 75 μL of 5 M Ammonium Acetate pH 7.4. Mix.
  10. Add 2.5 volumes of 100% ethanol and precipitate for 0.5–1 h at –70°C.
  11. Spin at maximum speed in the microcentrifuge for 20 min.
  12. Wash DNA pellet with 70% ethanol and resuspend DNA in 30–50 μL of TE, pH 7.5.
  13. Store bisulfite the samples in the –20°C freezer. The life-span of bisulfite-treated DNAs can exceed 12 months when stored properly.

PCR Amplification

PCR amplification is performed using standard PCR protocols

DNA Clean Up

Clean PCR product as described in Restriction enzyme digestion section.

Restriction-Enzyme Digestion

  1. Digest 10 ng of purified PCR product with 1 U of the enzyme in a 10–15 μL reaction for at least 4 h to ensure complete digestion.
  2. Overlay the sample with mineral oil to prevent evaporation.

Polyacrylamide Gel Electrophoresis

  1. Add 4 volumes of loading buffer to each digested sample. The mineral oil does not need to be removed.
  2. Denature for 2–3 min at 95°C.
  3. Load sample onto a 55°C prewarmed 8% denaturing polyacrylamide gel (7 M urea).
  4. Run gel in 1X TBE at 50–60 milliAmps.

Electroblotting, Prehybridization, Hybridization, and Washing

  1. Transfer the DNA to Zetabind Membrane by electroblotting for at least 1 h at 20 volts.
  2. UV crosslink the DNA to the membrane at 1200 Joules.
  3. Prewet the nylon membrane in 6X SSC and place membrane into a hybridization bottle.
  4. Add 10 mL of Church prehybridization buffer and rotate for 0.5–1 h at 42°C.
  5. Add 52 end-labeled probe (10 pmoles) and hybridize overnight at an appropriate temperature.
  6. Wash the membrane as described earlier.