Extensive optimisation of a previously published bisulfite treatment method [6
] has led to the development of a fast bisulfite treatment with high recovery and potential for semi-automation. Optimisation involved the purification procedure after deamination in order to facilitate recovery of the fragmented DNA resulting from bisulfite treatment. This part of the procedure has previously been identified as a critical step [10
]. Optimisation of the purification method had a pronounced effect on recovery, especially replacement of lysis and extraction buffers with ethanol increased recovery. In addition, adjustments of the reaction time were made. During optimisation the dynamics of the reaction had to be measured in order to ensure optimal conversion of cytosine and limited conversion of methyl-cytosine.
In the discovery phase bisulfite treatment, cloning and sequencing of individual clones is the method of choice, because of the detailed information of methylation status of individual CpG sites achieved. This procedure, from primer design to presentation of results has been described in detail [22
]. When sequencing individual clones, complete deamination of unmethylated cytosine is extremely important. Incomplete deamination of unmethylated cytosine could lead to a false positive result. Typically more than one methylated CpG site is investigated when using methylated DNA as a tumour marker. Hence, this set up is less likely to yield a false positive result due to incomplete conversion of cytosine. However, if the reaction is poorly optimised and conversion of cytosine is far from complete, real-time detection will fail, leading to apparently low recovery. Therefore, while optimising recovery, dynamics of the reaction have been extensively monitored using both a real-time PCR based and a HPLC based method. The results of optimisation of the reaction time are shown in Figures and . After 5 min of deamination, no undeaminated product was detectable by real-time PCR (Figure ). The beacon and primers used in this experiment have been designed in order to give as comparable reaction conditions as possible. The overlap in the probe and primers used for the different reactions gives less stringent differentiation compared with methylation specific real-time PCR, where both primers and probes generally are specific for either the methylated or the unmethylated sequence. However, no cross-reactivity is observed in our experiment. The primers used to discriminate between undeaminated and deaminated product cover 5 cytosines (Figure ). Primer binding mainly depends on the conversion of the 2 cytosines in the 3' end of the primer binding site. Therefore complete conversion of cytosine appears to be achieved faster when measured by real-time PCR compared with HPLC, where a limited amount of cytosines (< 1%) is detectable after 10 min deamination, and none after 15 min (Figure ). Longer deamination time than 10 min results in reduced detection of both deaminated and undeaminated DNA, since the purpose is to optimise this protocol in order to achieve the best possible recovery of bisulfite treated DNA, the deamination time was set to 10 min.
Figure 2 Monitoring of reaction dynamics by HPLC. A) The distribution of nucleosides as measured by HPLC depicted as a function of deamination time. The levels of Gs, As and methylated Cs remain constant throughout the reaction. Cs are converted to Us as the deamination (more ...)
In addition to incomplete conversion of cytosine, inappropriate conversion of methylated cytosine could lead to reduced sensitivity when detecting methylated DNA. The real-time PCR based experiment (Figure ) does not address the problem of over-conversion, since the primers and the probe are designed to an unmethylated part of MLH1. Methylated cytosine is monitored by the HPLC procedure (Figure ). However, methylated cytosine constitutes only a small fraction of the nucleotides. Therefore, to further elucidate the reaction dynamics and recovery of both methylated and unmethylated DNA, the hemimethylated MEST promoter has been used as a naturally occurring model system. Comparable detection of methylated and un-methylated MEST was observed, when calculating detection frequency as the percentage of positive reactions out of the 24 replicates (Figure ).
Figure 3 Detection of the hemi-methylated MEST promoter. Recovery of methylated and unmethylated MEST, measured in a dilution series of cell-free DNA from plasma. Each reaction was run 24 times, and the detection frequency is calculated as the percentage positive (more ...)
Similar experiments monitoring the dynamics of the reaction were carried out when changing other parameters, such as reaction temperature and denaturation of DNA prior to bisulfite treatment (data not shown). Denaturation of DNA did not affect recovery or conversion in the current study, even though this step has been pointed out to be critical, and different denaturants such as NaOH, urea and formamide have been used. Formamide was shown to have a great effect when working on DNA from formalin-fixed paraffin-embedded tissue [23
]. The lack of need of denaturants in the protocol presented here possibly reflects the denaturating effect of the high concentration of bisulfite and high temperature used for deamination.
Further investigation of the analytical sensitivity of the newly optimised procedure was carried out using a dilution of universally methylated DNA in purified plasma DNA. RASSF1A was used as a model system. RASSF1A is fully methylated in the universally methylated DNA and unmethylated in the plasma DNA sample. Table illustrates that low copy numbers of methylated RASSF1A promotor region was reliably detected in a background of approximately 2000 copies of unmethylated DNA.
Detection of methylated DNA diluted in unmethylated DNA
Recovery of deaminated DNA using the optimised method was compared with recovery of DNA in a matrix of deamination reagents. Re-extraction of DNA in a matrix of deamination reagents results in a recovery of 83% whereas recovery of deaminated DNA is 60% (Table ). The likely explanation of the difference is the unavoidable degradation of DNA during deamination.
Markers such as septin 9 hold great promise of clinical usefulness. The improved recovery rate of the bisulfite reaction achieved by the presented protocol may further improve its potential in a clinical setting, and avoid some of the expensive and labour intensive steps such as the need to purify multiple plasma samples in parallel and subsequent pooling of DNA [16
]. Indeed, the initial promising reports on the potential of septin 9 have been strengthened by a recent report employing silica based magnetic DNA purification [25
]. This work by deVos et al.
address the problem of poor recovery, and the reported recovery rates are comparable to our results. However, the quantitation of DNA prior to and after bisulfite treatment is based on two different real-time PCR assays. We have employed a real-time PCR assay capable of detecting both treated and untreated DNA [11
] allowing the same assay to be used for measuring DNA levels both pre and post bisulfite treatment resulting in improved accuracy