DNA methylation and demethylation play a critical role in regulating gene expression as well as developmental and pathological processes1. For example, hyper-methylation in the promoter region typically represses a gene’s transcription, and aberrant DNA methylation has been demonstrated to be associated with certain tumors2. Till now, many enzymes have been demonstrated to be involved in DNA methylation or demethylation, including DNA methyltransferases, which use S-Adenosyl methionine as the methyl donor to catalyze the transfer of a methyl group to target DNA1. In mammalian cells, three active DNA methyltransferases have been characterized, including DNMT1, DNMT3A, and DNMT3B1. DNMT1, the most abundant methyltransferase, predominately methylates cytosines at hemi-methylated CpG dinucleotides1,3 and is responsible for the maintenance of DNA methylation, ensuring the fidelity of inherited epigenetic patterns during replication. To perform loci-specific demethylation modification in vivo, both Zinc fingers (ZFs)4 and transcription activator-like effectors (TALEs)5 have been used to fuse the ten-eleven translocation methylcytosine dioxygenase 1 (TET1) catalytic domain for DNA demethylation. However, both the design and the protein engineering of the ZFs (or TALEs) that are required for the binding of specific loci are time-consuming and costly, and their applications are therefore limited. Recently, with the improvement of the CRISPR (clustered regularly interspaced short palindromic repeats) technology, catalytically dead Cas9 (dCas9) was fused to the catalytic domain of TET1 (TET1CD) to hydroxylate specific loci and activate site-specific demethylation. Although this CRISPR system brings much convenience, it requires expression of a foreign TET1 enzyme6, and may therefore cause unexpected influences on cellular metabolisms. To minimize the introduction of foreign components, we here develop a new CRISPR/dCas9-based system (CRISPR/dCas9-R2) to reduce the DNA methylation level by blocking the DNMT1 activity in the targeted loci (Fig. 1a). Specifically, a short RNA sequence with the stem-loop structure (named R2-stemloop, Fig. 1b) is introduced to both positions of the tetraloop and the stemloop2 of the single-guide RNA (sgRNA) scaffold. Then, DNMT1 is specifically recruited by the R2-stemloop and its enzyme activity is inhibited upon binding to the R2-stemloop structure, therefore the methylation of the targeted DNA loci is blocked7.
The human RANKL gene has been reported to be repressed because of hyper-methylation in its promoter region in HEK-293T cells8, and demethylation in its promoter region could potentially activate the gene expression. To test whether the transcription of RANKL could be regulated by the CRISPR/dCas9-R2 system, five sgRNAs were designed to target the RANKL promoter region (Fig. 1c), and the impact of individual sgRNA on RANKL transcription was assayed by quantitative RT-PCR (qRT-PCR), using total RNA extraction prepared from the HEK-293T cells 5 days post transfection. Three of the five sgRNAs (i.e., RANKL-sgRNA2, 4, 5) were found to be able to activate the transcription of RANKL (Fig. 1d), among which RNAKL-sgRNA2 was the most effective. We then used the bisulfite-sequencing method to examine the methylation status of four CpG sites in the RANKL-sgRNA2 targeted locus, and found three of them (i.e., positions of 26, 126, and 276) had a decreased methylation level (Fig. 1e and Supplementary Table S1), indicating the close relationship between the upregulation of RANKL transcription and decreased methylation status in the promoter region. In addition, we also used the CRISPR/dCas9-R2 system to regulate the transcription of another gene (namely MAGEB2), which locates on the chromosome X and is also transcriptionally repressed by DNA methylation in its promoter region9. Similarly, five sgRNAs were designed to target the promoter region of MAGEB2 (Supplementary Fig. S1a), and four of them (i.e., MAGEB2-sgRNA2, 3, 4 and 5) showed effectiveness in upregulating MAGEB2 transcription (Supplementary Fig. S1b). In accordance with the transcriptional activation, the DNA methylation status in MAGEB2 promoter region was found to be decreased in cells transfected with MAGEB2-sgRNA2 and 3. Specifically, four of the eight CpG sites (i.e., positions of 119, 134, 189, and 257) had a decreased methylation level (Supplementary Fig. S1c and Supplementary Table S2).
We next examined the influence of the transfection time on demethylation efficiency by the CRISPR/dCas9-R2 system, and the RANKL transcription in HEK-293T cells was analyzed by qRT-PCR at different time points after transfection of RNAKL-sgRNA2. As shown in Fig. 1f, the transcription level of RANKL reached the highest at the 7th day post transfection (i.e., about threefold relative to the control group with no-sgRNA transfected). In accordance with the transcriptional results, three methylated CpG sites in RANKL promoter region (i.e., positions of 26, 126, and 276) also decreased gradually, and reached the lowest level at the 7th day (Supplementary Table S3). Similarly, transcriptional activation of MAGEB2 also became the most effective at the 7th day post transfection (Supplementary Fig. S1d). Moreover, we checked whether there existed additive effects on upregulating target gene expression by transfecting more than one sgRNAs targeting to the promoter region. For target RANKL gene, either individual or mixture of the three effective sgRNAs (i.e., RANKL-sgRNA2, 4 and 5) was transfected into the HEK-293T cell, and the transcription of RANKL was analyzed 5 days post transfection. According to the qRT-PCR results (Fig. 1g), either individual or mixture of sgRNAs led to similar level of transcriptional activation. Similarly, transfection of an sgRNA mixture did not lead to a higher transcriptional level of MAGEB2 (Supplementary Fig. S1e), therefore indicating the dCas9-R2 system had no additive effects for gene activation.
Here, we described a programmable approach for convenient demethylation of targeted DNA sequences in living cells. Considering the availability of multiple Cas9 mutants recognizing different PAM sites10, the CRISPR/dCas9-R2 system using either dCas9 or other dCas9 mutants may have the potential to target any DNA sequences. Compared with currently available CRISPR-based approaches for targeting DNA demethylation, our system showed similar editing efficiencies with the CRISPR/dCas9-TET1 system11 but does not require co-transfection of other epigenetic enzymes, therefore minimizing unexpected influences on cellular metabolisms. Besides, the editing window of the CRISPR/dCas9-R2 system is about 100-bp distance nearby the targeted site (Fig. 1e, Supplementary Fig. S1c and Supplementary Tables S1– S3), and could be more accurate than the present CRISPR/dCas9-TET1 system11, which is effective in regions of 100–300-bp distance from the targeted site. Moreover, advantage of CRISPR/dCas9-R2 system compared with other methods such as ZFs4 and TALEs5 which are challenging in the system construction is also summarized (Supplementary Table S4).
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This work was supported by the National Key Basic Research Program of China (973 Program; 2014CB745201), the National Natural Science Foundation of China (81772737), grants from the Strategic Priority Research Program of the Chinese Academy of Sciences (XDB19040200), the International Cooperation Program of the Chinese Academy of Sciences (153D31KYSB20170121), National Science Foundation Projects of Guangdong Province (2017B030301015) the Shenzhen Municipal Government of China (JCYJ20170413161749433 and JSGG20160301161836370), the Sanming Project of Shenzhen Health and Family Planning Commission (SZSM201412018 and SZSM201512037), and the High Level University Medical Discipline Construction (2016031638).