The architecture of its complex with chromatin factor HMGN2 was derived on the

basis of CSP, PRE and mutagenesis data, demonstrating the feasibility of modeling nucleosome–protein complexes using solution NMR. Recently, for the first time a structural model for the read-out of an epigenetically modified nucleosome was determined in our lab [80], characterizing see more the interaction between the PSIP1-PWWP domain and a nucleosome trimethylated at H3K36 (H3K36me) (Fig. 6). Comparing the interactions of the PWWP domain with isolated H3K36me-peptides, DNA and H3K36me-nucleosomes, revealed that the nucleosomal DNA plays an important role in the specific recognition of this modification, boosting the affinity by more than 10,000-fold. The complex was modeled using HADDOCK and AIRs based on an extensive mutagenesis analysis and find more observed CSPs. Acknowledging the flexibility of the H3 N-terminal tail, the flexible multi-domain docking protocol was adapted [81]. First, the H3K36me3 peptide was docked to the aromatic cage of the PWWP domain on the basis of CSP and homology derived AIRs. Second, the resulting complex was docked back to nucleosome, guided by the identified DNA interaction surface and covalent restraints for the H3-tail. In this step, a threading approach was taken to systematically sample the binding

site of PWWP on the nucleosomal DNA. The DNA surrounding the H3 N-terminal tail exit point was divided in 10 patches of each 5 bp. For each docked structure one of these patches were defined as passive residues. The resulting structures were cross-validated against mutagenesis data, leaving a single cluster of solutions. The solutions show how the arrangement of aromatic cage and basic patches on the surface PWWP domain matches perfectly to its nucleosomal substrate. Particularly, the solutions reveal a clonidine network of extensive electrostatic

interactions between PWWP Lys and Arg residues and the DNA phosphate backbone. Subsequent modeling of other H3K36me3-readers showed that the relative configuration of aromatic cage and basic patches is conserved, suggesting conserved role of the nucleosomal DNA in H3K36me recognition. Modeling of non-symmetrical complexes with three or more subunits is especially challenging, because of the increase in degrees-of-freedom and the requirement of obtaining experimental restraints for all mutual interactions. NMR data can be used to determine binding interface on all subunits, thus positioning the subunits. Restraints on the overall shape of the whole or part of the complex can be extremely useful to improve the quality of the models. Recent work of the Sattler group on a ternary protein–protein–RNA complex [61], systematically explored how SAXS/SANS-derived molecular envelops could help to refine structural models obtained from CSP-driven HADDOCK-models (Fig. 7). First, the RNA binding surfaces on the two proteins were mapped using TROSY experiments on perdeuterated proteins.