In practice, complexes with molecular weight above 50–100 kDa are too large for conventional, de novo NMR structure determination relying on an extensive network of short-range inter-proton distance. However, in many cases it is still possible to determine 3D-structures of isolated subunits

GSI-IX chemical structure either by NMR or crystallography, and to acquire structural information on their organization in the complex, although less complete and precise. In addition, complementary information might be available from other types of biochemical and biophysical experiments. The resulting collections of sparse data, of different experimental origins and information content, call for integrative computational tools to judiciously combine and translate them into meaningful atomic structures or models. These can be interrogated to test existing hypothesis or generate new ones, which can then be probed experimentally. In this Perspective, we briefly review NMR-based approaches for the integrative modeling of large and multi-subunit complexes. We warn the reader that the goal here is not to be comprehensive, nor to provide a thorough review of the current literature. We describe the NMR techniques available to characterize soluble high

molecular weight complexes, the types of data that can be extracted from these, and the sources of complementary data. We then outline the general procedure for integrative modeling and illustrate all this with a number of challenging cases from the literature. Finally, this website we dissect current bottlenecks and present an outlook to the future of integrative modeling of large multi-subunit complexes and the role of NMR in it. Both the sensitivity and resolution of solution NMR spectra deteriorate significantly with increasing molecular weight due to the line broadening of peaks. This broadening is due to long rotational tumbling correlation times τc, which enhance transverse relaxation. The key break-through

to circumvent these deleterious relaxation effects has been the development of transverse relaxation-optimized spectroscopy (TROSY [1]), in which slowly relaxing multiplet components are selectively observed in highly deuterated proteins. In the context of the characterization Sulfite dehydrogenase of large multi-subunit protein complexes, TROSY comes in basically two flavors ( Table 1). The first type is aimed at the sensitive detection of backbone amide signals (TROSY, CRIPT/CRINEPT-TROSY [2]), while the second aims specifically at the detection of methyl groups (MeTROSY [3]). Backbone-amide detection allows monitoring of all non-proline residues, making it an excellent tool for identifying binding surfaces. However, for single-chain proteins beyond 50–100 kDa the sheer amount of backbone signals complicates the spectra, and assignment becomes increasingly difficult. In such systems, methyl-based experiments offer a very attractive alternative.