Unique physical properties and interactions of the domains of methylated DNA binding protein 2

Abstract

Methylated DNA binding protein 2 (MeCP2) is a methyl CpG binding protein whose key role is the recognition of epigenetic information encoded in DNA methylation patterns. Mutation or misregulation of MeCP2 function leads to Rett syndrome as well as a variety of other autism spectrum disorders. Here, we have analyzed in detail the properties of six individually expressed human MeCP2 domains spanning the entire protein with emphasis on their interactions with each other, with DNA, and with nucleosomal arrays. Each domain contributes uniquely to the structure and function of the full-length protein. MeCP2 is approximately 60% unstructured, with nine interspersed alpha-molecular recognition features (alpha-MoRFs), which are polypeptide segments predicted to acquire secondary structure upon forming complexes with binding partners. Large increases in secondary structure content are induced in some of the isolated MeCP2 domains and in the full-length protein by binding to DNA. Interactions between some MeCP2 domains in cis and trans seen in our assays likely contribute to the structure and function of the intact protein. We also show that MeCP2 has two functional halves. The N-terminal portion contains the methylated DNA binding domain (MBD) and two highly disordered flanking domains that modulate MBD-mediated DNA binding. One of these flanking domains is also capable of autonomous DNA binding. In contrast, the C-terminal portion of the protein that harbors at least two independent DNA binding domains and a chromatin-specific binding domain is largely responsible for mediating nucleosomal array compaction and oligomerization. These findings led to new mechanistic and biochemical insights regarding the conformational modulations of this intrinsically disordered protein, and its context-dependent in vivo roles.

Figure 1. Organization of MeCP2 and relation to disorder predictions

(a) Upper panel – map of MeCP2 showing the six major domains identified by partial proteolysis (11). The graph shows the order-disorder score of MeCP2 predicted by PONDR VLXT, a neural network predictor of native disorder (39). Grey bars denote predicted molecular recognition features (MoRFs) – see discussion. (b) The amino acid composition of MeCP2 is characteristic of a highly unstructured protein. Bar chart (filled bars) shows differences in amino acid composition between MeCP2 and the average composition of a set of ordered proteins for each amino acid. Positive values and negative values correspond to greater and lesser abundance of an amino acid in MeCP2 compared to ordered proteins. Clear bars show the differences in average composition for each amino acid between disordered proteins from the DisProt database (54) and the same set of ordered proteins. The amino acid residues are arranged in an increasing order of disorder promoting potential (54). For explanation regarding calculation of fractional difference in composition see materials and methods.

Figure 2. Circular dichroism spectra of MeCP2 domains reveal marked differences in secondary structure content

CD spectra are representative of two - four separate acquisitions. (a) Compared to the 195nm peak indicative of β-sheet structure within the MBD (black squares), all the other domains show a negative band in the 195nm–198nm region indicative of disorder. NTD (black circles), TRD (white circles) and CTD-α (black rhombi) have lower structure content than ID (stars) and CTD-β (half filled circles) (see Table 1 for quantitation). (b) Addition of DNA (methylated as well as unmethylated) to the ID induces changes typical of the formation of β-structure, namely a marked increase in positive ellipticity at 195nm and negative in the 220–225nm range. (c) Addition of DNA (methylated as well as unmethylated) to the TRD results in an increase in order irrespective of the methylation status of the DNA.

Figure 3. MeCP2 domains induce electrophoretic mobility shifts upon addition to DNA or chromatin

Gel images for each experiment are representative of two to three separate trials. (a) Interaction between individual MeCP2 domains and DNA. Domains (NTD, MBD, ID, TRD, CTD-α, CTD-β, TRD-CTDα-CTDβ) were incubated with methylated 601-12 DNA at molar input ratios of 0 to 8, and the products are displayed on 1% agarose gels. The ID, TRD and CTD-α induce substantial retardation of the DNA. In contrast, the MBD shows only minor shifts and the NTD appears to have virtually no interaction with DNA. (b) To examine methylation specificity, MBD and constructs that include its flanking domains were incubated with unmethylated (−) or methylated (+) DNA in the presence of two-fold excess of 208-1 DNA competitor at molar input ratios of 0 to 10. A distinct methylation-dependent enhancement of the gel shifts is seen in all constructs containing the MBD. Of particular interest is the large shift shown by the NTD-MBD construct, which suggests a synergism between these two domains. Full length MeCP2 produces pronounced gel shift at much lower input than the MBD containing contiguous domain fusions. (c) The ID and TRD-CTD polypeptides produce strong shifts, but there is no methylation-dependent enhancement. (d, e) as (a, b) but with 601-12 nucleosomal arrays (NAs) as substrate and 208-1 mononucleosomes as competitor. With the exception of CTD-β which induces a moderate but consistent mobility shift with chromatin but not with naked DNA, the patterns of electrophoretic shift with DNA and NAs are similar. M denotes molecular weight marker lanes.

Figure 4. Sedimentation velocity reveals differences in the ability of MeCP2 domains to compact nucleosomal arrays

Methylated 601-12 nucleosomal arrays were incubated with a two-fold molar input of MeCP2 constructs in 50mM NaCl, 10mM Hepes, 0.25 mM EDTA and analyzed by sedimentation velocity. NAs alone (circles) MBD (squares), NTD-MBD (triangles), MBD-ID (diamonds), NTD -MBD-ID (circles with cross), TRD-CTD (stars) and full-length MeCP2 (hexagons). For characterization of the full-length MeCP2 an equimolar input of protein was used since a two-fold input causes extensive self-association and oligomerization. Data were consistent over two to three separate trials for each experiment.

Figure 5. Direct EM observation reveals differences in conformational changes induced in undersaturated nucleosomal arrays by MeCP2 domains

Subsaturated NAs were mixed with different MeCP2 fragments at input ratios of 8 molecules of protein per 208 bp DNA, fixed, and imaged using darkfield EM. (a–i) Representative images of NAs showing the range of conformational changes from none for the MBD and NTD, to extensive compaction and self-association for the TRD-CTD fusion. (k) Mean array diameters with standard errors.

Figure 6. Interactions between the MBD and flanking domains revealed by tryptophan accessibility

Fluorescence quenching by acrylamide analyzed using Stern Volmer plots shows that the fluorescence of Trp 104 in the MBD is differentially accessible depending on the flanking domains present. MBD only (circles). NTD-MBD (stars), MBD-ID (diamonds), MeCP2 1-294 (circles with cross), full-length MeCP2 (squares). Plots are linear up to ~250 mM acrylamide. Error bars represent standard errors of mean.

Figure 7. In trans interactions between MeCP2 domains revealed by fluorescence anisotropy and CD

(a) Fluorescence anisotropy of fluorescently-labeled MBD upon mixing with other MeCP2 domain constructs. In the presence of ID (open circles) and TRD (filled circles), the MBD shows marked increase in anisotropy whereas addition of NTD (open squares) caused no change in anisotropy. CTDβ (filled squares) gave a small increase at higher input ratios. Error bars denote standard errors of mean. Insert shows the robust emission spectrum of the MBD-tetraCys bound FlAsH complex. (b, c) Fragment complementation was also detected by using CD to monitor interactions between domain pairs caused by changes in secondary structure. For each pairwise comparison (A:B), data were acquired separately for the two different domains (A) and (B) and also for their mixture (A+B). Plots show the difference spectra at each wavelength expressed as a percent of the spectrum obtained by addition of the individual spectra [((A)+(B))−(A+B)]/((A)+(B)). (b) MBD+ID (squares) and MBD+TRD (circles) spectra show strong differences from the composite spectrum of the individual domains while the MBD + CTD-β pair shows only a minor change. (c) The NTD + MBD pair (circles) show a distinct difference at ~198nm while NTD+ID (triangles) and ID + CTD-β (squares) show no differences.

Figure 8. Quantitation of DNA binding affinity of MeCP2 fragments

(a) Normalized fluorescence anisotropy, r_norm= (r_n−r₀)/(r_max−r₀), of a 5′-fluorescein labeled 22bp fragment of BDNF promoter DNA with a single methylated CpG was measured in the presence of increasing concentrations of MeCP2 fragments: MBD (black squares), NTD-MBD (white circles), MBD-ID (black circles) and TRD-CTD (white squares). Error bars denote standard errors of mean. X axis (protein concentration) and Y axis (normalized fluorescence anisotropy) are linear normal. (b) as in (a) but with different MeCP2 domains: ID (black triangle), TRD (black circle), CTD-α (black squares). Error bars denote standard errors of mean. X axis (protein concentration) is log decimal and Y axis (normalized fluorescence anisotropy) is linear normal. (c) Normalized fluorescence anisotropy, r_norm= (r_n/r₀), of the same DNA substrate as in (a) and (b) in the presence of increasing concentrations of NTD (black circle), ID (black triangle), CTD-β (black square). ID is included both in 8b and 8c to provide a reference scale for the two different types of normalizations used in 8b and 8c. Error bars denote standard errors of mean. r₀ = raw anisotropy at 0 protein input, r_max = raw anisotropy at maximum protein input, r_n = raw anisotropy at each protein concentration and r_norm is the corresponding normalized anisotropy.

Figure 9. Structure of MBD bound to DNA suggests that MoRFs flank interaction surfaces

Model of the MBD (tan, light blue and green) of hMeCP2 bound to 20bp of BDNF promoter DNA (gray), PDB file 3c2i (6). The MBD α-MoRFs are located in residues 87–104 (light blue) and 133–150 (green). The two MoRFs form a contiguous surface that is predominantly hydrophilic, winding across the MBD opposite the DNA interaction surface. Arrows point to the solvent accessible surface area of Trp104 (black), and the surface exposed regions (blue) of Arg106, Arg 133, and Phe155 where Rett syndrome-causing point mutations result in significant changes in the local surface properties (13). These all contribute to a MoRF surface, suggesting a role in inter- and intra-protein interactions related to MoRF disorder-to-order transitions.