Automated structure modeling of large protein assemblies using cross-links as distance restraints

Cross-linking mass spectrometry (XL-MS) is increasingly used for structural characterization of multi-subunit protein complexes. Chemical cross-linking captures conformational heterogeneity, which typically results in conflicting cross-links that cannot be satisfied in a single model, making detailed modeling a challenging task. Here we introduce an automated modeling method dedicated to large protein assemblies that (i) uses a form of spatial restraints that realistically reflects the distribution of experimentally observed cross-linked distances, (ii) automatically deals with ambiguous and / or conflicting cross-links and identifies alternative conformations within a Bayesian framework, (iii) allows subunit structures to be flexible during conformational sampling. We demonstrate the power of our method by testing it on known structures and available cross-linking data. We cross-linked and modeled  the 17-subunit yeast RNA polymerase III at atomic resolution. The resulting model agrees remarkably well with recently published cryo-electron microscopy structures and provides additional insights into the polymerase structure.