Integrative Modeling of Nucleosomes and Supranucleosomal Structures
Nucleosomes are repeating elementary units of chromatin compaction. A con- ventional nucleosome consists of linker DNA and around 147 bp of DNA wrap- ped around an octamer of histones. It is well known that nucleosomes can interact with each other forming different supranucleosomal structures. As nu- cleosomes govern the accessibility of DNA they play a crucial role in genome organization, DNA transcription, and repair. Nucleosome structures are dy- namic with conformational plasticity at various levels and timescales, which is necessary to fulfill their functions. This dynamic behavior limits the applica- bility of conventional structural biology techniques. X-ray crystallography and Cryo-EM studies have consistently yielded similar atomistic structures but other biophysical and biochemical techniques suggest that nucleosomes exhibit substantial polymorphism with respect to DNA conformation. As methods of direct determination of large protein-DNA complexes are not readily available, we have developed a range of modeling and experimental data analysis methods to address this problem.Our integrative modeling approach is based on a popular flexible DNA fiber model (6 variables for each base pair step) combined with rigid protein structures. Such an approach allows us to (i) rapidly thread different DNA sequences through defined DNA conformations to find structurally beneficial sequences (ii) generate large supranucleosomal structures end ensembles of nucleosomal structures based on external distance constraints. We are building such models based on the known atomistic struc- tures and available experimental data, such as distance distributions accessed from the spFRET measurements or. pair distances derived from Micro-C maps. We also developed a computational method for integrating hydroxyl radical footprinting data for precise determination of DNA positioning in nu- cleosomes (this method is implemented in a software package called HY- DROID). The resulting toolkit is written in Python and will be released as open-source module.