Engineering a high-affinity methyl-CpG-binding protein

Abstract

Core members of the MBD protein family (MeCP2, MBD1, MBD2 and MBD4) share a methyl-CpG-binding domain that has a specific affinity for methylated CpG sites in double-stranded DNA. By multimerizing the MDB domain of Mbd1, we engineered a poly-MBD protein that displays methyl-CpG-specific binding in vitro with a dissociation constant that is >50-fold higher than that of a monomeric MBD. Poly-MBD proteins also localize to methylated foci in cells and can deliver a functional domain to reporter constructs in vivo. We propose that poly-MBD proteins are sensitive reagents for the detection of DNA methylation levels in isolated native DNA and for cytological detection of chromosomal CpG methylation.

Figure 1

A multimerized-MBD protein binds to methylated DNA. (A) Diagram of engineered wild-type (NxMBD) and mutant (NxMBD–R22A) proteins with 1 or 4 MBDs. Asterisks mark the position of the inactivating point mutation (R22A). The NLS is in grey, the MBD in white and the HA-tag in black. An N-terminal His₆-tag, which is present in bacterially expressed proteins, is hatched. The insertion point of the VP16-activation domain (VP16) is indicated. (B) A coomassie-stained polyacrylamide gel showing purified 4xMBD (4×) and 1xMBD (1×). (C) Increasing amounts (2.5, 5, 10, 25, 50 ng) of wild-type (4xMBD) or negative control (4xMBD–R22A) tetrameric MBD proteins were incubated with M.SssI-methylated (M) or non-methylated (U) CG11 probe that contains 27 CpG sites (22). The complexes were resolved on 1.3% agarose gels. The ladder of bands is due to complexes that contain multiple protein molecules on a single DNA molecule.

Figure 2

Multimerization of the MBD increases the affinity for methylated DNA. Increasing amounts (2.5, 5, 10, 15, 25, 50, 75, 100 ng) of wild-type 1xMBD (A) or 4xMBD protein (B) were incubated with duplex oligonucleotide probes containing either 0, 1, 2 or 3 methyl-CpGs (0ME, 1ME, 2ME and 3ME, respectively). The complexes were resolved on 6% polyacrylamide gels. (C) Multimeric MBD protein can displace the methyl-CpG-binding protein MeCP2 from methylated DNA in vitro. Increasing amounts of wild-type 4xMBD (2.5, 10, 25, 100, 250 ng) or control 4xMBD–R22A (2.5, 10, 25, 100, 250 ng) tetrameric protein were added to reactions containing 150 ng recombinant GST–MeCP2 fusion protein (rMeCP2) and incubated with the 2ME oligonucleotide probe (see above). Both MBD proteins and MeCP2 were present in the reactions before the addition of the labelled probe. The absence of multiple bands as seen with the CG11 probe (Figure 1) may be explained by the small number of CpGs in 2ME and 3ME probes (2 or 3 compared to 27 in CG11) and the use of polyacrylamide rather than agarose gel for fractionation of complexes.

Figure 3

Recombinant 4xMBD proteins is a sensitive detection reagent for methylated CpG in immobilized DNA. CpG methylated (M) or unmethylated (U) lambda phage DNA was immobilized on membranes and detected using the indicated NxMBD proteins (10 µg/ml) followed by anti-HA and HRP-anti-mouse antibody, or with anti-m⁵C antibody (1:500) followed by HRP-anti-mouse antibody. The amount of DNA in each slot is indicated at the left. Native DNA (native) was blotted on to nitrocellulose membrane and denatured (denat) DNA was blotted on to nylon membranes. The bottom right slot on each filter contained 50 ng of HA-tagged 1xMBD as a positive control.

Figure 4

The 4xMBD protein is a sensitive stain for methyl-CpG-rich regions in fixed cells. Permeabilized and fixed mouse fibroblasts were incubated with 50 µg/ml 1xMBD (A) or 4 µg/ml 4xMBD (B), or their R22A mutant forms, and visualized with anti-HA plus Alexa-594-conjugated anti-mouse antibodies. Specific staining is seen at heterochromatic DAPI bright spots with the wild-type probe but not with their mutant counterparts. Weak staining by the 4xMBD–R22A mutant protein is sometimes detected, but does not co-localize with DAPI bright spots. Staining shows improved resolution with 4xMBD compared to 1xMBD and requires >10-fold less protein. (C) Metaphase spreads from mouse fibroblasts were incubated with 4 µg/ml 4xMBD and visualized with anti-HA plus Alexa-594-conjugated anti-mouse antibodies. DNA is counterstained with DAPI. Intense staining of pericentromeric satellite DNA is apparent plus weaker staining of the chromosome arms.

Figure 5

Wild-type NxMBD proteins localize to CpG-rich heterochromatin in a DNA methylation dependent manner in vivo and can deliver a functional domain to a methylated reporter gene. The indicated NxMBD proteins were expressed in wild-type (A) or DNA methylation deficient Dnmt1-null (B) mouse fibroblasts that were fixed and stained using an anti-HA antibody. Left panels show anti-HA signals and right panels show DNA counterstained with DAPI. (A) Co-localization of wild-type 1xMBD, 4xMBD and the 4xMBD–VP16 fusion proteins with heterochromatic foci in interphase cells. (B) Dispersed nuclear localization of 1xMBD and 4xMBD in Dnmt1-null cells that lack DNA methylation. (C) The 4xMBD–VP16 fusion protein specifically activates a methylated reporter construct in vivo. Non-methylated or methylated reporter constructs were co-transfected alone, or with 4xMBD (500 ng), 4xMBD–R22A (500 ng) or a 4xMBD–VP16 fusion construct (250 ng). The average relative reporter activity for a triplicate experiment is shown. The error bars represent ±SD.