Mapping the Structural Topology of the Yeast 19S Proteasomal Regulatory Particle Using Chemical Cross-linking and Probabilistic Modeling

Structural characterization of proteasome complexes is an essential step toward understanding the ubiquitin-proteasome system. Currently, high resolution structures are not available for the 26S proteasome holocomplex as well as its subcomplex, the 19S regulatory particle (RP). Here we have employed a novel integrated strategy combining chemical cross-linking with multistage tandem mass spectrometry to define the proximity of subunits within the yeast 19S RP to elucidate its topology. This has resulted in the identification of 174 cross-linked peptides of the yeast 19S RP, representing 43 unique lysine-lysine linkages within 24 nonredundant pair-wise subunit interactions. To map the spatial organization of the 19S RP, we have developed and utilized a rigorous probabilistic framework to derive maximum likelihood (ML) topologies based on cross-linked peptides determined from our analysis. Probabilistic modeling of the yeast 19S AAA-ATPase ring (i.e., Rpt1–6) has produced an ML topology that is in excellent agreement with known topologies of its orthologs. In addition, similar analysis was carried out on the 19S lid subcomplex, whose predicted ML topology corroborates recently reported electron microscopy studies. Together, we have demonstrated the effectiveness and potential of probabilistic modeling for unraveling topologies of protein complexes using cross-linking data. This report describes the first study of the 19S RP topology using a new integrated strategy combining chemical cross-linking, mass spectrometry, and probabilistic modeling. Our results have provided a solid foundation to advance our understanding of the 19S RP architecture at peptide level resolution. Furthermore, our methodology developed here is a valuable proteomic tool that can be generalized for elucidating the structures of protein complexes.Basic cellular homeostasis depends on the regulated protein degradation and turnover by the ubiquitin-proteasome system (1, 2). Central to this pathway is the 26S proteasome complex, which is responsible for ubiquitin/ATP-dependent protein degradation (3–5). The 26S holocomplex is a megadalton-sized protein assembly consisting of the 20S catalytic core particle (CP)¹ and the 19S regulatory particle (RP). The eukaryotic 20S CP is composed of two copies of 14 nonidentical subunits (α_1–7 and β_1–7) arranged into four stacked heptameric rings in an order of α₇β₇β₇α₇. The crystal structure and topology of the highly ordered 20S CP has been resolved and is evolutionarily conserved (6). Although α subunits of the 20S CP are essential for the assembly of the complex and its interactions with the regulatory complex, three catalytic β subunits (β1, β2, and β5) harbor various catalytic activities responsible for regulated proteasomal degradation. The 19S RP is composed of 19 subunits, which forms two subcomplexes, the base consisting of six related AAA-ATPase (Rpt1–6) and four non-ATPase (Rpn1, Rpn2, Rpn10, and Rpn13) subunits and the lid containing nine non-ATPase subunits (Rpn3, Rpn5–9, Rpn11, Rpn12, and Rpn15/Sem1) (7, 8). In comparison with the 20S core, the function and structure of the 19S RP is much less well understood. Nevertheless, it is believed that the 19S RP is involved in multiple functions including recognition of polyubiquitinated substrates (9, 10), cleavage of the polyubiquitin chains to recycle ubiquitin (11), unfolding of substrates, assisting in opening the gate of the 20S chamber, and subsequently translocating the unfolded substrates into the catalytic chamber (4, 12–14). The six AAA-ATPase subunits (Rpt1–6), which directly interact with the 20S α-ring, function as a molecular chaperone responsible for protein unfolding and are involved in substrate translocation and modulating gating of the CP (5, 15). Although detailed functions for most of the 19S non-ATPase subunits remain elusive, Rpn11 is known to carry an Mpr1p and Pad1p N-termini (MPN) domain, which harbors an essential deubiquitination activity responsible for cleaving polyubiquitin chains from proteasomal substrates (11, 16). In addition, two proteasome subunits, Rpn10 and Rpn13, have been identified as ubiquitin receptors, which are important in docking ubiquitinated substrates to the proteasome for degradation (4). Moreover, the two largest proteasome subunits, Rpn1 and Rpn2, interact with a variety of proteins including ubiquitin receptors and deubiquitinases and thus may function as scaffolding proteins to assist proteasomal degradation. Thus far, no atomic resolution structures are available for either the 19S RP or the 26S holocomplex. New insights of the overall topology of the 19S RP will illuminate protein interactions within, thus providing evidence for its otherwise unknown functions.Although many studies have been performed to characterize the 19S structure utilizing various techniques including cryo-EM (17, 18) and native mass spectrometry (19), details on spatial interfaces and subunit interconnectivity of the 19S RP remain to be unraveled. During the course of our study, the rough topology of the 19S RP was determined by cryo-EM alone (20) or coupled with other approaches (21); nevertheless more detailed information at the peptide or atomic level is still required. In addition to technological limitations in current approaches, the highly dynamic and heterogeneous nature of the 19S RP may attribute to the difficulty in obtaining its high resolution structure. In recent years, chemical cross-linking coupled with mass spectrometry (XL-MS) has become an attractive alternative for structure analysis of proteins and protein complexes (22, 23). The ability of XL-MS to identify interaction interfaces between proteins allows us to define low resolution protein topology. In addition to protein interaction networks and the site of protein interactions at binding interfaces, cross-linking analysis can reveal information about the spatial distance between cross-linked amino acids on the surface of folded proteins. Although such knowledge only reveals the maximum distance given by the length of the cross-linker and can be influenced by protein conformational flexibility, it can be used as the distance constraint for molecular modeling of protein folds and complex topologies, i.e., the arrangement of the constituents of a complex in space. A recent study by Chen et al. (24) on yeast RNA polymerase II (RNAPII) complex has exemplified the power of XL-MS in elucidating the architecture of large multisubunit complexes. Although effective, cross-linking studies have been challenging because of the low abundance of cross-linked products and the inherent complexity of sequencing interlinked peptides by MS for unambiguous identification. To facilitate MS detection and identification of cross-linked products, we have recently developed a novel homobifunctional amine reactive, low energy MS-cleavable cross-linker, disuccinimidyl sulfoxide (DSSO), and successfully applied it to cross-link the yeast 20S proteasome for rapid, accurate, and simplified determination of protein interaction interfaces within the complex (25). The unique functionality of our cross-linking reagent and specialized bioinformatics tools significantly increase our confidence and speed in the identification of cross-linked products when compared with cross-linking studies using traditional noncleavable reagents. Current cross-linking studies have been focused on protein complexes with known crystal structures, but topological structures of protein complexes based primarily on cross-linking data have not yet been reported. This is due to the lack of computational tools that use cross-linking data to deduce the spatial organization of subunits in a given complex. To define the architecture of the yeast 19S RP, we have characterized the proximity and interconnectivity of the subunits by employing our newly developed cross-linking strategy. The resulting cross-linking information serves as a basis for a rigorous probabilistic analysis to obtain the maximum likelihood (ML) topology. This strategy is developed by first analyzing our cross-linking data for the 19S six-member AAA-ATPase base ring, as the topology ordering of yeast orthologs has been recently determined (14, 26–28). The effectiveness of this new probabilistic platform is supported by the agreement between our derived ML topology of the AAA-ATPase base ring and previous reports. When the same probabilistic approach is applied to the 19S lid subcomplex, the resulting topology is also in agreement with recently proposed models (20, 21). This work represents the first application of probabilistic modeling of protein complexes based solely on cross-link data, establishing a new workflow for future structural analysis of large protein complexes using XL-MS.