In vivo site-specific DNA methylation with a designed sequence-enabled DNA methylase.

As an alternative to the continual expression of transcriptional repressors to turn off genes after they have served their purpose, nature has developed epigenetic strategies that result in the covalent modification of DNA itself to induce heritable gene silencing. Mounting evidence supports the notion that once a genomic region has been targeted for silencing by acquisition of one or more covalent epigenetic marks, mark can be propagated and may influence acquisition of others.1 If epigenetic modifications can be made specifically by the addition of targeted exogenous agents, new approaches to transcriptional therapy should result. Methyltransferases recognize specific DNA sequences and transfer a methyl group from the cofactor S-adenosyl-L-methionine (AdoMet) to an amine on either cytosine or adenine.2 The best characterized methyltransferase is HhaI DNA m5c-methyltransferase (M.HhaI) that recognizes the sequence 5′-GCGC-3′ and converts the internal cytosine (in bold) to 5-methylcytosine using a base-flipping mechanism.3 In eukaryotes, CpG methylation serves as a signal for the recruitment of proteins that ultimately act to silence transcription.4 Recently, fusions of zinc finger proteins (ZFPs) to full-length methyltransferases such as M.HpaII and fusions with the enzymatic domains derived from the murine enzymes Dnmt3a and 3b have been studied.5 These fusion proteins methylate native target sites as well as sites adjacent to the binding sites for the attached ZFPs.6 Although these proteins are biased in their activity, they do not direct the truly programmable site-specific methylation we aim to achieve. To address this challenge we sought to design methyltransferases that would act only at a targeted site by adopting a sequence-enabled reassembly strategy.7 Our aim was to reassemble a fragmented methylase on DNA as directed by zinc finger binding, thereby restoring its activity at a specific site and providing for site-specific cytosine methylation (Figure 1). We hypothesized that a functional and site-specific enzyme could be self-assembled on a particular DNA sequence using ZFPs appropriately fused to a recently described split M.HhaI enzyme.8 If correctly designed, such a fragmented protein would be active only at the site of assembly and not elsewhere. To explore this hypothesis, each domain of split M.HhaI (Nand C-terminal domain fragments) was fused to previously characterized three-finger ZFPs, HS1 and HS2.9 These ZFPs, HS1 and HS2, bind the DNA sequences 5′-GGGGCCGGA-3′ and 5′-GCCGCAGTG-3′, respectively. The N-terminal fragment was fused to HS2 to create MeNDHis and the C-terminal fragment was fused to HS1 to create MeCD proteins. A His6 tag was fused to MeNDHis and an HA epitope tag was fused to MeCD to facilitate protein detection. A fusion of intact M.HhaI with HS2 (HS2Me) was also constructed (Figure 1A). Genes for the fusion proteins were cloned downstream of a pBAD promoter with Shine-Dalgarno sequences for bacterial expression. A 26 bp DNA methylation target sequence, GCGC-ZFS, consisting of a GGCGCC site flanked by ZFP binding-sites was present on the plasmid (Figure 1B). To stringently test the specificity of the assembled enzyme, 18 native M.HhaI GCGC sites without flanking ZFP binding sites were also encoded on the plasmid. Expression of the fusion proteins in E. coli was confirmed by Coomassie blue staining of cell extracts and western blotting using antibodies against His6 and HA tags. DNA binding of fusion proteins was detected by ELISA as described previously.9 DNA binding affinities of ZFPs HS1 and HS2 are 35 and 25 nM, respectively.10 As our first methylation test, HhaI restriction enzyme cleavage was performed on plasmid DNA recovered from E. coli that expressed our designed enzymes (Figure 1C). DNA cleavage by the HhaI endonuclease is inhibited when the internal cytosine in the GCGC site is methylated. In our assay, DNA fragmentation patterns of plasmid following HhaI restriction enzyme treatment will differ in accord with the methylation status of the various HhaI sites.11a HhaI endonuclease cleavage of plasmid coding MeNDHis and MeCD produces 20 fragments sized 312, 353, 390, 1108, and 1710 bp together with 15 fragments smaller than 300 bp. Sitespecific methylation at GCGC-ZFS will inhibit endonuclease cleavage between the 353 and 1108 bp fragments, resulting in production of a novel 1461 bp fragment. Results of this study revealed that the MeNDHis fragment did not methylate any of the 19 HhaI sites when expressed alone (lane 1). To test ZFP directed self-assembly, methylation of the plasmid derived from cells expressing the combination of Nand C-terminal enzyme fragments, MeNDHis and MeCD, was evaluated. Here too, no band indicative of methylation was detected (lane 2). On the basis of the possibility that a His6 tag could inhibit the reassembly of split domains, we further tested expression of MeND (identical to MeNDHis but lacking the C-terminal tag) and MeCD. Reassembly of this Figure 1. (A) Schematic drawing of Split-M.HhaI and ZFP fusions. Blue segment shows linker sequences. (B) Cartoon illustrating ZFP directed assembly of a fragmented DNA methylase on the target DNA sequences. The DNA sequence bound by ZFPs, HS2, and HS1, are boxed. (C) HhaI digestion assays for methylation detection. Plasmid DNA expressing the various enzyme fragments was isolated and digested with HhaI: lane M, marker; lane 1, MeNDHis; lane 2, MeNDHis + MeCD; lane 3, MeND + MeCD; lane 4, HS2Me; lanes 5 and 6, M.HhaI only (C, no digestion control; D, HhaI digestion). White arrow in lane 3 indicates a band created by inhibition of HhaI digestion owing to ZFP-targeted methylation. Published on Web 06/21/2007