Examination of Ataxin-3 (atx-3) Aggregation by Structural Mass Spectrometry Techniques: A Rationale for Expedited Aggregation upon Polyglutamine (polyQ) Expansion

Expansion of polyglutamine stretches leads to the formation of polyglutamine-containing neuronal aggregates and neuronal death in nine diseases for which there currently are no treatments or cures. This is largely due to a lack in understanding of the mechanisms by which expanded polyglutamine regions contribute to aggregation and disease. To complicate matters further, several of the polyglutamine-disease related proteins, including ataxin-3, have a multistage aggregation mechanism in which flanking domain self-assembly precedes polyglutamine aggregation yet is influenced by polyglutamine expansion. How polyglutamine expansion influences flanking domain aggregation is poorly understood. Here, we use a combination of mass spectrometry and biophysical approaches to investigate this issue for ataxin-3. We show that the conformational dynamics of the flanking Josephin domain in ataxin-3 with an expanded polyglutamine tract are altered in comparison to those exhibited by its nonexpanded counterpart, specifically within the aggregation-prone region of the Josephin domain (amino acid residues 73–96). Expansion thus exposes this region more frequently in ataxin-3 containing an expanded polyglutamine tract, providing a molecular explanation of why aggregation is accelerated upon polyglutamine expansion. Here, harnessing the power of ion mobility spectrometry-mass spectrometry, oligomeric species formed during aggregation are characterized and a model for oligomer growth proposed. The results suggest that a conformational change occurs at the dimer level that initiates self-assembly. New insights into ataxin-3 fibril architecture are also described, revealing the region of the Josephin domain involved in protofibril formation and demonstrating that polyglutamine aggregation proceeds as a distinct second step after protofibril formation without requiring structural rearrangement of the protofibril core. Overall, the results enable the effect of polyglutamine expansion on every stage of ataxin-3 self-assembly, from monomer through to fibril, to be described and a rationale for expedited aggregation upon polyglutamine expansion to be provided.Polyglutamine (polyQ)¹ diseases comprise a group of hereditary neurodegenerative disorders in which expansion of polyQ stretches within their causative proteins induces protein aggregation and the formation of polyQ-containing neuronal aggregates (1). The mechanisms by which expanded polyQ regions contribute to aggregation and disease are not well understood. In all cases, polyQ length is negatively correlated with the age of onset of the disease (2), but the various polyQ disorders are associated with different neurodegenerative symptoms and affect different regions of the brain (3). Several of the polyQ proteins, including ataxin-3 (atx-3) (4) and huntingtin (5), have been shown to aggregate in vitro through a complex multidomain misfolding pathway (6) in which flanking domain aggregation precedes polyQ aggregation. Increasing evidence also suggests a key role for misfolding of flanking regions in the process of polyQ aggregation in vivo (7 –10). Thus, as the proteins have no sequence similarity other than in their polyQ regions, flanking domain content may be significant in determining the disease state and neuronal-specific selectivity. Given that there is growing support to suggest that the toxic entities in polyQ diseases are the soluble oligomers and assembly intermediates, rather than the fibrillar aggregates (11), effective therapeutics may be generated by targeting flanking domain interactions (12) rather than targeting the polyQ region itself. An enhanced understanding of the molecular mechanisms of assembly of polyQ proteins is required, as is a greater comprehension of the effects of polyQ length on the structure, dynamics, aggregation propensity, and oligomerisation pathway of the flanking domains. Here, we set out to determine the influence of an expanded polyQ tract on each stage of atx-3 aggregation by harnessing the power of mass-spectrometry-based approaches to identify and characterize assembly mechanisms (13, 14).Atx-3 consists of a structured N-terminal Josephin domain (JD), which has ubiquitin protease activity (15) and an intrinsically disordered C-terminal region, the latter containing several ubiquitin-interacting motifs (UIMs) followed by the polyQ tract and a variable region (16) (Fig. 2A). In vivo, expansion of the polyQ stretch beyond ca. 55 glutamine residues results in Machado–Joseph disease (17). Consistent with this, atx-3 with a polyQ tract beyond ca. 55 glutamine residues aggregates into amyloid-like fibrils rapidly in vitro (18). Aggregation proceeds by means of a two-stage pathway (4): the first stage resulting in the production of SDS-sensitive, short, curvilinear, protofibrils, and the second producing long-straight and SDS-resistant mature fibrils. The first stage involves self-association of the JD (19) and occurs in all atx-3 variants whether or not they contain a polyQ region of nonpathological length (nonexpanded, 12–40 glutamine residues (17)), an expanded polyQ region of disease length (polyQ-expanded, 55–84 glutamine residues (17)), or are devoid of a polyQ region (20). The second stage occurs only in polyQ-expanded atx-3 and involves hydrogen bonding between side-chains in the polyQ region (21), which renders aggregation irreversible.Open in a separate windowFig. 2.Limited proteolysis of protofibrils and mature fibrils. (A) Schematic illustrations of atx-3(14Q) (left) and atx-3(78Q) (right) with amino acid residue numbers for each domain shown. Mass spectra obtained following limited proteolysis with trypsin of (B) atx-3(14Q) protofibrils (C) atx-3(78Q) protofibrils and (D) atx-3(78Q) mature fibrils. Mass spectra of (left) the depolymerized fibrillar material are contrasted with those obtained from analysis of (right) the soluble products of proteolysis. Asterisks represent species observed in the pellet fraction that were also observed in supernatant samples ((-16)-45⁴⁺ and (-16)-47⁴⁺, respectively). Peaks identified as containing the polyQ tract are highlighted in pink, while those representing QBP-1 are highlighted in orange.Despite the fact that the first stage of atx-3 aggregation does not require the polyQ tract, aggregation of polyQ-expanded atx-3 occurs more rapidly (with a shorter lag time) than aggregation of nonexpanded atx-3 (4, 20). The precise molecular mechanism for this observation has yet to be elucidated. An initial hypothesis was that polyQ expansion destabilizes atx-3, allowing the JD to adopt misfolded, aggregation-prone conformations more readily (15, 18). However, a study comparing atx-3 constructs with polyQ regions of different lengths showed that polyQ expansion does not affect the folding/unfolding kinetics or thermodynamic stability of the JD (22). Consequently, it has been postulated that the expanded polyQ tract may perturb the structure of the JD without affecting its stability (20).We set out to address why aggregation occurs more rapidly in atx-3 with an expanded polyQ tract by studying monomeric, oligomeric, and fibril structures for atx-3 with a pathological length polyQ tract of 77 glutamines with a single, naturally occurring, lysine residue in the fourth position (named atx-3(78Q)); atx-3 with a nonpathological length polyQ tract of 13 glutamines (also with a single lysine residue in the fourth position) (atx-3(14Q)); and the isolated JD. Results from a combination of electrospray ionization-ion mobility spectrometry-mass spectrometry (ESI-IMS-MS), limited proteolysis, fluorescence spectroscopy, and transmission electron microscopy (TEM) analyses confirm that protofibrils of these three atx-3 constructs are formed through equivalent processes and reveal that the resulting protofibril cores are similar, if not identical. Limited proteolysis experiments combined with MS analyses provide evidence that an expanded polyQ tract alters the conformational dynamics of the JD, exposing its aggregation-prone region more frequently than in its nonexpanded counterparts, rationalizing the enhanced aggregation potential of the polyQ-expanded protein. Finally, oligomers populated en route to fibrils are examined by ESI-IMS-MS and a model for oligomer growth is provided. Together these results reveal how polyQ length affects each stage of atx-3 aggregation and demonstrate how different MS-based techniques can provide information about each stage of the aggregation mechanism.