Aberrant DNA methylation of CpG site is among the earliest and most frequent alterations in developmental process and diseases including cancer. To elucidate the functional preferred site of DNMTs, we analyzed the feature of distinct methylated sequences and established the defined relationship between DNMTs and preference genomic DNA sequences. Small interfering RNA (siRNA) construct of DNTM1, DNMT3A, and DNMT3B was transfected into the human hepatocellular carcinoma cell line SMMC-7721, respectively. Distinguishing methylated fragments pool was enriched by SHH method in cells which is knocked down DNMT1, DNMT3A, DNMT3B, separately. The defined binding transcription factors (TFs) containing of 5'CpG islands were obtained with bioinformatics software and website. In SMMC-7721 hepatocellular carcinoma (HCC) cell line, DNMT1, DNMT3A, and DNMT3B were specific suppressed by their corresponding siRNA construct, separately. A 46, 42, 67 distinctive methylated fragments from three different DNMTs were evaluated according to genomic DNA database. Those separated fragments were distributed among genomic DNA regions of all chromosome complements, including coding genes, repeat sequences, and genes with unknown function. The majority of coding genes contain CpG islands in their promoter region. Cluster analysis demonstrated all of preference sequences identified by three DNMTs shares their own conserved sequences. In depleting of different DNMTs cells, 80 % of 103 upregulation genes induced by DNMT1 knock-down contain CpG sites; 76 % of 25 upregulation genes induced by DNMT3A knock-down contain CpG sites; 63 % of 126 upregulation genes induced by DNMT3B knock-down contain CpG sites. Our findings suggested that distinctive DNMTs targeted DNA methylation site to their preference sequences, and this targeting might be associated with diverse roles of DNMTs in tumorigenesis. Meanwhile, the analysis of preference sequences provides an alternative way to find out the individual function of DNMTs.
Keywords: 5′CpG site; DNA methyltranferases; Preference sequences.