CpG Methylation Patterns Of Human Mitochondrial DNA

Objective: In our recent study funded by the NSFC, we found that DNA methylation levels of mitochondrial DNA(mt DNA) control region in human blood samples was 2% to 34%, while nearly 0% in saliva samples. Furthermore, it showed a significant difference within monozygotic twins(MZT) pairs and can identify two individuals up to 31.93% of the MZT. Tracking literatures on mt DNA methylation studies, we found little research in this area, and the results disputed between each other. The earliest studies reported on mt DNA methylation appeared 30 years ago, and the researches on frogs and Hela cells showed no mt DNA methylation was observed for both of them. Then studies using methylation-sensitive restriction found there were low levels of mt DNA methylation in mice, hamsters, and humans. Recent it was reported that mt DNA methylation was associated with many diseases, but other reports considered that mt DNA was almost not methylated. These discordant results may be due to different methods they used. Bisulfite pyrosequencing technology is widely recognized as the gold standard of DNA methylation analysis, which can get precise quantification of each Cp G site, and has higher accuracy through internal quality control monitoring bisulfite conversion efficiency. Therefore, this study applied bisulfite pyrosequencing technology to detect methylation of mt DNA including D-loop region and coding regions to map a human mitochondrial genome methylation profiling.Methods: According to the principle of informed consent, peripheral blood samples were collected from 154 healthy individuals, and saliva samples were collected from 24 healthy individuals. Bioinformatics analysis of mitochondrial genome infers a total of 435 Cp G sites unevenly distributed across the whole mt DNA, with the highest Cp G density in the light chain replication origin(OL)(5721 ~ 5798bp), and the minimum density in t RNA genes. Twenty pairs of primers for amplifying specific mt DNA segments andthe corresponding sequencing primers were designed using Assay Design Software. These 20 specific products named MT1~MT20 cover both D-loop region and coding region involving 83 Cp G sites. Genomic DNA was extracted from blood samples and saliva samples and treated with Bam HI to open loop structure of mt DNA followed by bisulfite conversion. PCR amplification was performed using converted DNA as template and the PCR products were pyrosequenced to get methylation values of each Cp G sites. DNA without Bam HI treatment was used to compare the bilsulfite conversion efficiency and methylation quantification values using MT1, MT2, MT3, MT4 four segments. SPSS13.0 software was applied for statistical analysis including paired t test, variance analysis, Wilcoxon test.Results: 1 The influence of circular structure of mt DNA on bisulfite conversion and pyrosequencing 1.1 Effect of circular structure of mt DNA on pyrosequencing pass ratio 1.1.1 Blood samples Through the detection of 65 to 154 blood samples, we found the sequencing pass ratio of four sequences MT1, MT2, MT3, MT4 in open-loop group were 78.07%, 97.32%, 97.73%, 90.26%, and in non open-loop group were 80.95%, 84.62%, 64.52%, 47.89%, respectively. After statistical analysis, except MT1, the pass ratio of the remaining three sequences between open-loop group and non open-loop group had a significant difference, and the pass ratio of non open-loop group were significantly lower than that of open-loop group. The mainly reasons leading to bisulfite pyrosequencing analysis failure include unpassed inner quality control due to incomplete bisulfite conversion, low signal-to-noise ratio or signal loss caused by insufficient PCR products, and uncertain reference sequence pattern due to stochastic effect. Further analyzing the causes for sequencing failure, we found that in the non open-loop group, the proportion of the unpassed inner quality control in all of the four sequences except MT2, were more than 80%. While in the open-loopgroup, the main reason for sequencing failure is low signal-to-noise ratio or signal loss, which are usually related to the experiment operation or caused by some stochastic effect. However, failing in quality control indicated that bisulfite conversion is not complete. Therefore, the above results showed that the circular structure of mt DNA in blood samples obviously affected the bisulfite conversion process, and the conversion efficiency was significantly decreased compared with that of linear mt DNA treated by Bam HI. 1.1.2 Saliva samples We detected 14 to 24 saliva samples. The sequencing pass ratio of four MT1, MT2, MT3, MT4 in open-loop group were 87.50%, 82.35%, 94.12%, and 87.50%, while that in non open-loop group were 90.00%, 87.50%, 87.50%, 94.44%, respectively. There was no significant difference between two groups for any of the four sequences. And the pyrosequencing pass ratio of non open-loop group in saliva samples was significantly higher than that of blood samples. Even for the non open-loop mt DNA in saliva samples that was failure to be pyroseqenced, not any of them were caused by incomplete bisulfite conversion. The above results showed that the circular structure of mt DNA has little effect on bisulfite conversion for saliva samples. 1.2 Effect of circular structure of mt DNA on methylation quantification values 1.2.1 Blood samples Although the sequencing pass ratio of the non open-loop group in blood samples is greatly lower, more than 47% of the samples still succeeded in pyrosequencing and got methylation quantification data. Therefore, we compared the methylation quantification values between the two groups. The results showed that the average methylation levels of four sequences were 11.65%±0.29%, 3.58%±0.27%, 5.68%±0.20%, 7.21%±0.27% in non open-loop group, while they were 1.16%±0.09%, 1.68%±0.10%, 0.32%±0.05%, 0.78%±0.09%, respectively in open-loop group. Wilcoxon signed-rank test showed that significant differences exist between the two groups, and the methylation degree in open-loop group was significantly lower than that of non open-loop group.For the same samples tested by both methods and got methylation values, paired statistic test was performed and the results demonstrated that the D values within two methods of all the 17 Cp G sites except Cp G39 were statistically significant, which further confirmed that the methylation values got from the non open-loop mt DNA were significantly higher than that of the open-loop ones. 1.2.2 Saliva samples For saliva samples, the average methylation values of four sequences were 1.17%±0.25%, 1.79%±0.35%, 0.47%±0.17%, 1.06%±0.28% in non open-loop group, and 1.38%±0.21%, 2.14%±0.29%, 0.69%±0.18%, 0.93%±0.22% in open-loop group, respectively. Wilcoxon signed-rank test results showed that no significant differences exist between the two groups. For the same samples measured by both methods and got methylation data, paired comparison was conducted and no statistic difference was found in most of the Cp G sites except for Cp G8, Cp G39, Cp G109 sites. These results further confirmed that the circular structure of mt DNA in saliva samples has little effect on bisulfite conversion and the subsequent methylation quantification by pyrosequencing. 2 Human mitochondrial genome DNA methylation profiling Applying the above established Bam H Ⅰtreatment and bisulfite pyrosequencing methods, we detected the mitochondrial genome methylation of blood samples and saliva samples from 14 unrelated individuals. This detection system can measure 83 Cp G sites including 16 Cp G sites in D-loop region, 21 Cp G sites in 2 r RNA gene regions, and 46 Cp G sites in six protein gene regions. The results showed that, regardless of blood samples or saliva samples, whether D loop region or the coding region, the average methylation levels of the 83 Cp G sites were from 0% to 6%, and most of them were 0% ~ 2%, indicating that there was no methylation occurred in mt DNA.Conclusions: To our knowledge, it is the first study to detect the human mitochondrial genome DNA methylation using bisulfite pyrosequencing technique. The findings show that the circular structure of mt DNA has greatimpact on bisulfite conversion efficiency, and thereby affecting subsequent pyrosequencing pass rate. Even though the sequencing is passed, the quantitative methylation value is significantly higher than the actual value, leading to overestimation of mt DNA methylation. This effect caused by loop structure is only occurred in blood samples but not in saliva samples. Using above developed Bam HI treatment and bisulfite pyrosequencing technique, methylation profiling of 83 Cp G sites across mt DNA was mapped and the average methylation level is 0% to 6%, which means almost no methylation occurred in mt DNA.