A directed evolution design of a GCG-specific DNA hemimethylase

Abstract

DNA cytosine-5 methyltransferases (C5-MTases) are valuable models to study sequence-specific modification of DNA and are becoming increasingly important tools for biotechnology. Here we describe a structure-guided rational protein design combined with random mutagenesis and selection to change the specificity of the HhaI C5-MTase from GCGC to GCG. The specificity change was brought about by a five-residue deletion and introduction of two arginine residues within and nearby one of the target recognizing loops. DNA protection assays, bisulfite sequencing and enzyme kinetics showed that the best selected variant is comparable to wild-type M.HhaI in terms of sequence fidelity and methylation efficiency, and supersedes the parent enzyme in transalkylation of DNA using synthetic cofactor analogs. The designed C5-MTase can be used to produce hemimethylated CpG sites in DNA, which are valuable substrates for studies of mammalian maintenance MTases.

Figure 1.

Recognition of the GCGC target sequence by the HhaI methyltransferase. (A) Top—schematic representation of contacts between the target recognition loops (Loops 1 and 2) of M.HhaI and the DNA bases in the GCGC target site; DNA contacting residues are underlined; lines represent direct H-bonds to the DNA bases, dotted lines indicate nucleobase contacts through a water molecule; residues of the conserved TL dipeptide are boxed; target cytosine C₂ is shown in bold; residues in Loops 1 and 2 are colored red and blue, respectively; bottom—aligned Loop 2 sequences of from WT M.HhaI and its engineered variants with altered target specificity; ΔL2 represents the randomized library; randomized positions are bold; an additional mutation outside Loop 2 is bold underlined. (B) and (C)—stick models depicting interactions of M.HhaI with the fourth and third target G:C base pair, respectively, based on a crystal structure of the M.HhaI-DNA-AdoHcy complex (PDB code 3mht). Deleted residues are marked with Δ, other coding as in A.

Figure 2.

Bisulfite sequencing analysis of the in vivo sequence specificity of M.HhaI-L2Bsp. Plasmid DNA was isolated from E. coli cells overexpressing M.HhaI-L2Bsp and analyzed by bisulfite sequencing. The analyzed region spanned 1577 bases and contained 134 GCN sites. Methylation of each site was assayed by sequencing 16–32 independent clones and the methylation density at each position was determined as a ratio of methylated cytosines over the total number of sequence reads. An average methylation density at GCX (A) and GCNX (B) sequences is shown (the target cytosine residues are underlined) and the number of individual target sites is indicated underneath each sequence.

Figure 3.

Directed evolution of GCG-specific C5-MTases. A schematic representation of random library selection for active GCG-specific MTases. The MTase encoding gene is shown as a grey arrow, the recognition Loop 2 is shown in black; open and filled circles indicate nonmethylated and methylated GCG sites, respectively.

Figure 4.

DNA protection analysis of the in vitro specificity of the HhaI-ΔL2 MTases. (A) Apparent number of enzymatic turnovers executed at different target sites. Serial 2-fold dilutions starting with equimolar amounts of MTase and target sites were used to methylate bacteriophage λ DNA, which was then challenged with a set of 5-methylcytosine-sensitive restriction endonucleases (Figure S4). Enzymatic turnover rates (turnovers per hour) were estimated based on a minimal molar ratio of MTase to its target sites that is required for a complete protection of DNA in 1 h. (B) The sequence fidelity of the ΔL2-methyltransferases, expressed as the ratio of methylation turnover rates at GCGC to GC[A/T]GC sites.

Figure 5.

Bisulfite sequencing analysis of in vitro specificity of the L2Bsp and ΔL2–14 MTases. pBR322 DNA was methylated with L2Bsp (top) or ΔL2–14 (bottom) at a MTase to GCG target sites ratio of 1:64 and subjected to bisulfite modification. Methylation densities at individual 46 GCN sites were determined by sequencing of a 538 nt pBR322 fragment in individual clones obtained after cloning the bisulfite-converted DNA. The methylation density is expressed as a ratio of methylated cytosines observed to a total number of sequence reads.

Figure 6.

Enzymatic transalkylation of DNA using synthetic cofactor analogs. λ DNA was incubated with decreasing amounts (two-fold serial dilutions) of WT (A) or ΔL2–14 (B) M.HhaI in the presence of cofactor analog AdoPentyn for 1 h at 37°C, then digested with R.Hin6I or R.Bsh1236I and analyzed by agarose gel electrophoresis. Numbers above lanes indicate molar ratios of MTases to their target sites (GCGC or GCG, respectively). (C) Chemical structure of the AdoMet cofactor and its synthetic analog AdoPentyn.