Cytotoxicity of polyglutamine structures

Polyglutamine monomers

The conformational change of the molecular structure of native polyglutamine proteins into β-sheet-rich monomeric proteins is an essential step in the toxification of these gene products [17, 18]. Due to the obstacles that arise while trying to observe the structure of these β-strands in the actual disease protein, most of the studies with this focus have been conducted utilising artificial proteins [19-21]. Numerous investigations proposed cylindrical, hairpin and intramolecular β-sheet models, however it is not clear which of these might be the predominant form in affected cells. It has been shown that polyQ monomers are cytotoxic in cultured cells [22]. Despite these findings it remains elusive whether toxicity is conferred directly by the monomers themselves or if the transition into oligomers is responsible for the cytotoxicity. It is noteworthy that the monomer-oligomer transition propagates rapidly throughout the cell and can also take place in the reverse direction [23].

Polyglutamine oligomers

Oligomers of disease proteins have been proposed as the toxic species leading to cell death in a variety of neurodegenerative disorders, including Alzheimer’s disease [24]. There are several lines of evidence for oligomeric intermolecular structures of expanded polyglutamine proteins. For instance an anti-parallel β-sheet structure with intermolecular hydrogen bonds called “polar zipper” [25], a parallel β-sheet conformation [18] or a cylindrical assembly designated “nanotube” [20], all analysed in vitro, have been described.

Nevertheless, the predominance of any one of these species in living cells could not be verified. Studies investigating the formation of expanded polyQ oligomers out of monomers revealed a bidirectional transition of these species and the predominant cytotoxic potential of the oligomer fraction towards neuronal cells [23, 26, 27]. Furthermore, polyQ oligomers are more toxic than inclusion bodies [26] and heat shock proteins 40 and 70 (HSP40/70) are capable of ameliorating the deleterious effects of expanded polyQ proteins without influencing the formation of inclusions [28-30]. In a mouse model of SBMA, the presence of oligomers exhibited a close correlation to disease symptoms [31]. From these findings and other studies concerning polyQ oligomers [32-35], a pivotal role of these structures in polyQ pathogenesis can be deduced (toxic oligomer hypothesis). Some reports even favour a common toxic structure hypothesis [33, 36] based upon the cross-reactivity of antibodies against Aβ oligomers with other amyloidogenic proteins (like α-synuclein and polyglutamine proteins). Accordingly, amyloidogenic proteins causing neurodegenerative diseases would share a common toxic structure regardless of their amino acid sequence.

However, these findings have not yet been verified in tissue of polyQ disease patients.

Polyglutamine inclusions

The formation of intranuclear inclusion bodies composed of expanded polyQ proteins has for a long time been considered to be the toxic event underlying the pathogenesis of the respective disorders [37-41]. Apart from the polyQ gene products themselves, a variety of other proteins like ubiquitin and heat shock proteins have been shown to be present in nuclear inclusions. Deprivation of these proteins from other cell compartments may result in dysfunction of neuronal cells [37, 42] concomitantly with disruption of axonal transport and nuclear function [43]. Despite these findings, results of more recent studies have established a rather cell-protective role of polyQ inclusion bodies. In addition to the lacking

correlation between inclusion body formation on the one hand and cellular imbalance and death on the other [44, 45], polyQ inclusion bodies proved to be beneficial in rat striatal neurons exposed to mutant Huntingtin (Htt) [46]. Furthermore, cells with inclusions survived significantly longer than those with soluble oligomers [23]. Although this hypothesis is not yet fully verified in vivo, formation of polyQ inclusions appears to mitigate detrimental effects of the mutated proteins rather than being the initial molecular step of polyQ disease emergence.

Figure 2. Model of conformational change, oligomerisation and aggregation as underlying pathogenic mechanism for polyQ diseases.

PolyQ pathogenesis requires an expanded polyQ tract in the disease protein and a cellular environment promoting the accumulation of conformationally altered polyQ monomers. Cytotoxic effects are exerted in the course of oligomerisation of aggregate precursors and the formation of different aggregation states and species with varying impact on cellular dysfunction. Subsequent cellular impairment renders the environment even more aggregation-prone. Eventually, the toxic effects exceed the cell’s coping capability and lead to death of the dysfunctional cell and to disease onset.

Adapted from [1, 2].

Influence of residues adjacent to the polyglutamine tract

Although the expansion of the polyQ stretch in disease proteins is the molecular basis of cytotoxiciy and pathogenicity in polyQ diseases, it does not explain the selectivity for distinct neuronal populations and tissues in the respective disorders. The different disease proteins exhibit a widespread distribution throughout the central nervous system (CNS) and are not confined to the especially vulnerable cell types. For instance, Huntington’s disease mainly affects striatal GABAergic medium spiny neurons (MSNs) [47] whereas Ataxin-1 in SCA1 is most detrimental in Purkinje cells of the cerebellum [48]. In contrast, toxicity of Ataxin-3 in SCA3 affects a wide range of cell types in pons, substantia nigra, thalamus and diverse brain stem nuclei [49, 50]. An explanation for this discrepancy may be found in the disease protein portions apart from the polyQ stretch. Mutation in the CAG tract may also alter the protein-protein interactions of the non-polyQ parts of the protein.

The association of mutated Htt for instance is more tightly with Htt-associated protein 1 (HAP1) and less strong with Htt-interacting protein 1 (HIP1) compared to wild-type Htt [51]. The modified interaction properties lead to the disruption of axonal transport of brain-derived neurotrophic factor (BDNF) and disturbances of clathrin-mediated endocytosis respectively. The correlation of Ataxin-1 mutation and Purkinje cell demise probably arises from a complex the disease protein forms with the neurotoxic RNA-binding motif protein 17 (RBM17). RBM17 is highly expressed in Purkinje cells and opposes another interactor of Ataxin-1, the neuroprotective Capicua [52]. Mutation of Ataxin-1 shifts the interaction balance towards a stronger association with RBM17 and results in cerebellar cell loss [53, 54].

Posttranslational modifications of amino acid residues outside the polyQ stretch have a remarkable impact on the toxicity of the disease proteins by influencing protein-protein interactions as well as by determining processing of the respective gene products.

For example, phosphorylation of distinct amino acids of Htt, Ataxin-1 and the androgen receptor (AR) alters the affinity properties to ligands [55] and is capable of either reducing [56, 57] or increasing [58] the formation of inclusion bodies and toxicity.

Ubiquitination of polyQ-containing proteins subjects them to degradation by the ubiquitin-proteasomal system (UPS) and therefore represents a toxicity-ameliorating mechanism. On the contrary, the competing sumoylation renders the proteins more stable and promotes cell death via aberrant transcription and an increase in the amount of toxic oligomers [59, 60]. Selective expression of cofactors influencing posttranslational

modifications of polyQ proteins adds to the specificity of toxicity to certain cell populations [61].

According to the toxic fragment hypothesis, proteolytical processing of polyQ proteins is the initial step in rendering them toxic, leading to an increase in aggregation and to nuclear translocation [62]. Htt, Ataxin-3 and AR have all been described to be susceptible to cleavage by caspases at specific amino acid sites [31, 63-65]. Mutation or phosphorylation of these sites is sufficient to decrease inclusion body formation as a result of reduced proteolytical cleavage and hence toxicity [66, 67].