Aggregation Formation in the Polyglutamine Diseases: Protection at a Cost?

Abstract

Mutant protein aggregation is a hallmark of many neurodegenerative diseases, including the polyglutamine disorders. Although the correlation between aggregation formation and disease pathology originally suggested that the visible inclusions seen in patient tissue might directly contribute to pathology, additional studies failed to confirm this hypothesis. Current opinion in the field of polyglutamine disease research now favors a model in which large inclusions are cytoprotective and smaller oligomers or misfolded monomers underlie pathogenesis. Nonetheless, therapies aimed at reducing or preventing aggregation show promise. This review outlines the debate about the role of aggregation in the polyglutamine diseases as it has unfolded in the literature and concludes with a brief discussion on the manipulation of aggregation formation and clearance mechanisms as a means of therapeutic intervention.

INTRODUCTION

The polyglutamine diseases are a family of nine genetically-similar progressive, neurodegenerative diseases that includes Huntington’s Disease (HD), Spinal and Bulbar Muscular Atrophy (SBMA), Dentatorubral-pallidoluysian atrophy (DRPLA), and the spinocerebellar ataxia (SCA) types 1, 2, 3, 6, 7 and 17 (Table 1). These diseases arise from the expansion of an unstable CAG triplet repeat within the coding region of a given gene. This results in the expansion of a glutamine stretch in the protein, which renders the host protein toxic mainly through unknown gain-of-function mechanisms. The host protein in each disease is distinct and many of their individual wildtype functions are still unknown. Nonetheless, these diseases share some similarities including progressive neurodegeneration in a disease-specific subset of neurons and the presence of insoluble protein aggregates (reviewed by Bauer and Nukina, 2009; Gusella and MacDonald, 2000; Orr and Zoghbi, 2007; Shao and Diamond, 2007; Takahashi et al., 2008; Watson et al., 2012).

Table 1.

The polyglutamine diseases

Disease Name	Protein	Wildtype Q length	Expanded Q length	Affected tissues	Inclusion Location	Clinical Features
Huntington's Disease (HD)	Huntingtin (HTT)	6–35	36–121	Striatum (caudate nucleus, putamen), globus pallidus, Cerebral cortex	Nucleus and Cytoplasm	Chorea, dystonia, cognitive deficits, psychiatric problems
Spinal and Bulbar Muscular Atrophy (SBMA)	Androgen Receptor (AR)	6–36	38–62	Motorneurons of the Anterior Horn and Bulbar Regions, Dorsal Root Ganglia, Skeletal Muscle	Nucleus and Cytoplasm	Proximal and bulbar muscle weakness, atrophy and fasticulation, mild androgen insensitivity
Dentatorubral-pallidoluysian Atrophy (DRPLA)	Atrophin-1 (ATN1)	3–38	49–88	Cerebellum (dentate nucleus), cerebral cortex, globus pallidus, basal ganglia	Nucleus	Ataxia, epilepsy/seizures, myoclonus, choreoathetosis, dementia
Spinocerebellar Ataxia Type 1 (SCA1)	ATAXIN-1 (ATXN1)	6–34	39–83	Cerebellum (Purkinje cells and dentate nucleus), inferior olive, pons, anterior horn cells and pyramidal tracts	Nucleus	Ataxia, dysarthia, cognitive impairments, progressive motor deterioration
SCA2	ATAXIN-2 (ATXN2)	15–32	32–200	Cerebellum (Purkinje cells), inferior olive, pons, substantianigra, frontotemporal lobes	Nucleus	Ataxia, decreased reflexes, dysarthia, Parkinsonian rigidity, sensory disturbance, mental deterioration
Machado-Joseph Disease (MJD)/SCA3	ATAXIN-3 (ATXN3)	12–40	61–86	Globus pallidus, cerebellum (molecular layer), pons, substantia nigra, anterior horn cells	Nucleus	Ataxia, dystonia. Parkinsonism, amotrophy, eyelid retraction and vision problems, faciolingual fasciculation, dysphagia, weight loss
SCA6	α1A subunit of the voltage-gated Ca2+ channel CACNA1A	4–19	21–33	Cerebellum (Purkinje cells, molecular and granular layers), inferior olive	Cytoplasm	Ataxia, dysarthria, oculomotor disorders, incontinence, peripheral neuropathy
SCA7	ATAXIN-7 (ATXN7)	4–35	37–306	Cerebellum (Purkinje cells, molecular and granular layers), pons, inferior olive, visual cortex	Nucleus	Ataxia, retinal degeneration, dysphagia, dysarthria, changes in reflexes or sensation
SCA17	TATA-Binding Protein (TBP)	25–43	45–63	Cerebelllum (Purkinje cells), inferior olive	Nucleus	Ataxia, cognitive decline/dementia, epilepsy/seizures, psychiatric problems

There have been nine polyglutamine diseases identified to date. The human protein mutated in each disease is noted, along with a schematic of its relative size and glutamine (Q) expansion location (purple box and triangle). The approximate size of the wildtype protein is noted in kilodaltons (kDa). Polyglutamine-expanded protein size can vary depending on the length of the CAG repeat. The tissues principally affected in each disease, as well as the clinical output and the cellular location of the inclusions are listed.

Aggregation is a hallmark of many neurodegenerative diseases, but its role in the pathogenesis of these conditions is highly debated (Arrasate and Finkbeiner, 2012; Bates, 2003; Davies et al., 1999; Hammer et al., 2008; Hands and Wyttenbach, 2010; Michalik and Van Broeckhoven, 2003; Takahashi et al., 2010). While its correlation with the disease state has suggested that it could be an intrinsic player in pathogenesis, more recent studies have revealed that the role of aggregation in disease is not so simple. The large insoluble inclusion bodies seen via microscopy may actually represent a cytoprotective entity produced by the cell to help cope with the misfolded disease proteins (Arrasate et al., 2004; Doi et al., 2013; Saudou et al., 1998; Takahashi et al., 2008). Such studies suggest that pathogenesis could instead be dependent upon the misfolded monomer, small oligomers, or microaggregates that are formed during or as byproducts from the aggregation process (Miller et al., 2011; Nagai et al., 2007; Takahashi et al., 2008). This idea has led to a number of in vitro studies aimed at elucidating the aggregation mechanism and identifying and visualizing the toxic species in the polyglutamine diseases (Burke et al., 2011; Chen et al., 2001; Jochum et al., 2012; Kar et al., 2013; Legleiter et al., 2010; Nakano et al., 2013; Poirier et al., 2002; Sathasivam et al., 2010; Scherzinger et al., 1997; Wetzel, 2012). But verifying the existence of the various aggregation forms in mammalian mouse models and human autopsy samples and correlating them with disease phenotypes remain interesting avenues of research.

This review starts by outlining the debate about the role of aggregation in the polyglutamine diseases as it has unfolded in the literature and summarizes more recent attempts to identify the true toxic species. Finally, the manipulation of aggregation pathways and aggregate clearance mechanisms as potential therapeutic approaches and their success at the animal level will be discussed.

AGGREGATION AS THE BASIS OF PATHOLOGY

Early in vitro studies suggested that pure polyglutamine stretches were able to oligomerize (Perutz, 1994; 1995; Perutz et al., 1994). In addition, the incorporation of a polyglutamine repeat into a non-disease related protein was sufficient to trigger its oligomerization (Stott et al., 1995). Subsequent analyses of human samples identified the presence of mutant protein aggregates or “inclusions” in disease-affected regions of HD (Becher et al., 1998; DiFiglia et al., 1997), DRPLA (Becher et al., 1998), SCA1 (Skinner et al., 1997), SCA2 (Koyano et al., 1999), SCA3 (Paulson et al., 1997), and SCA7 (Holmberg et al., 1998) brains and in neuronal and non-neuronal regions in SBMA patient tissue (Li et al., 1998a; 1998b). This aggregation could be recapitulated in cell culture and mouse models of these diseases (Davies et al., 1997; Martindale et al., 1998; Merry et al., 1998; Paulson et al., 1997; Skinner et al., 1997). Cell culture models of HD using human mutant Huntingtin (HTT) exon 1 (HTT^ex1) protein fragments soon demonstrated that the ability to aggregate was dependent upon the length of the CAG repeat in a given protein and can correlate with cytotoxicity (Cooper et al., 1998; Li and Li, 1998; Martindale et al., 1998). Taken together, these data strongly suggested that mutant protein aggregation was a hallmark of the polyglutamine disorders and could represent the common, gain-of-function mechanism predicted to underlie their pathogenesis (Davies et al., 1998; Ross, 1997).

Studies of polyglutamine proteins and fragments in vitro found that the inclusions were amyloid or fibrillar in nature, similar what is seen with prion proteins or the beta-amyloid fibrils associated with Alzheimer’s Disease (Scherzinger et al., 1997). Furthermore, while both wildtype and expanded mutant proteins could be added to growing fibrils, the expanded protein could reach the critical nucleus necessary for fibrillization much faster than its wildtype counterpart. This suggested that over the human lifespan, proteins with a wildtype glutamine repeat length do not reach the critical nucleus in time to form inclusions, and that the polyglutamine expansion speeds up the aggregation process so that inclusions do form (Chen et al., 2001). In fact, it was reported that the clinical age of onset in HD patients could be predicted by the intrinsic ability of the expanded protein to reach a critical nucleus and aggregate as a function of its glutamine repeat length (Chen et al., 2002).

While the kinetics of polyglutamine aggregation can be explored in vitro, it can be difficult to determine in vivo whether the correlation between aggregate formation and cytotoxicity is due primarily to the inclusions, monomers, or an intermediate form, as all are present within a single cell. One study found that globular oligomers, but not fibrils, of the non-disease associated but aggregatable proteins HypF-N and PI3-SH3 were cytotoxic when added to the media of cells, but it was unclear whether the cells took up the variously sized aggregates equally in this case (Bucciantini et al., 2002). To attempt to address this question, Yang et al. (2002) allowed pure polyglutamine peptides to aggregate in vitro, and then successfully delivered these pre-formed aggregates into the nuclei of cells. The pre-formed aggregates were able to induce cell death, suggesting that the nuclear localization of a polyglutamine aggregate alone can be toxic.

AGGREGATION MAY NOT BE TOXIC AFTER ALL

While initial studies suggested that the presence of inclusions correlated with regions of vulnerability in HD patient samples, other studies found that nuclear inclusions were actually quite rare in the striatum (the region most affected in HD) and did not correlate with degeneration patterns (Gutekunst et al., 1999; Kuemmerle et al., 1999). The authors did note that nuclear inclusions were larger in later stage samples compared to earlier stages, suggesting that aggregates grew with disease progression. Ubiquitination, a common feature of mutant protein inclusions, was also found to be more prevalent at late stages (Gutekunst et al., 1999). This finding is consistent with a recent study in C. elegans that suggested that ubiquitination plays a role in the secondary growth of aggregates rather than their initial formation (Skibinski and Boyd, 2012). The authors suggest, however, that these data may mean that neurons with aggregates are more likely to survive (Gutekunst et al., 1999; Kuemmerle et al., 1999). Indeed, a study with chimeric wildtype and R6/2 HD mice found that neurons with inclusions of the HTT^ex1 fragment could survive for long periods in vivo (Reiner et al., 2007).

Consistent with this idea, a cell culture study failed to find a correlation between mutant Androgen Receptor (AR; the protein affected in SBMA) aggregate formation and motor neuronal cell death (Simeoni et al., 2000). Similarly, transgenic mouse models expressing either pure polyglutamine under the control of the AR promotor (Adachi et al., 2001) or a truncated form of the expanded AR (Abel et al., 2001) showed robust aggregation and evidence of neuronal dysfunction, but failed to show any detectable neuronal cell death. Furthermore, addition of a polyglutamine repeat into the non-disease associated HPRT (hypoxanthine phosphoribosyltransferase) protein caused it to form inclusions and induced various late onset neurological phenotypes in transgenic mice, but again failed to induce any neuronal death (Ordway et al., 1997). While these studies suggest that aggregation can cause neurological defects, albeit without any cell death, some studies were able to show neurological defects in the absence of aggregation. Klement et al. (1998) developed a SCA1 transgenic model that recapitulated disease-related neurological phenotypes, but did not produce visible inclusions, suggesting that the nuclear expression of the polyglutamine-expanded mutant ATAXIN-1 protein is more important for SCA1 pathogenesis than its aggregation. Drosophila models expressing the full-length mutant HTT protein also show evidence of neurodegeneration in the absence of visible aggregation, leading the authors to suggest that the early stages of HD may depend upon neurological defects caused by the cytoplasmic expression of the mutant protein prior to its cleavage and nuclear accumulation (Romero et al., 2008). Overall, these studies shed doubt on the hypothesis that inclusions are the basis of polyglutamine-mediated toxicity.

AGGREGATION IS PROTECTIVE

While the studies outlined above suggest that the presence of mutant protein inclusions may not be able to fully explain the toxicity seen in human patients and mouse models of the polyglutamine diseases, they do not distinguish whether inclusions could be slightly toxic entities or passive benign formations. Nonetheless, some cell culture studies found that inclusions could actually be cytoprotective.

Saudou et al. (1998) developed a cellular model of HD that recapitulated the polyglutamine- and cell type-specific cytotoxicity observed in human patients. While cell death was specific to striatial cells in this model, aggregation could be detected in both striatal and hippocampal (unaffected in HD) cells and did not correlate with cell death. Furthermore, expressing a dominant negative form of an ubiquitin-conjugating enzyme was able to dramatically reduce aggregation, but increase cytotoxicity. This led the authors to conclude that inclusions could be protective for the cell. Kim et al. (1999) were also able to dissociate neuronal cytotoxicity from protein aggregation in HTT-expressing striatal cells by inhibiting caspases. Similarly, Taylor et al. (2003) reported that the formation of microtubule-dependent aggresomes, entities that are distinct from the large inclusions, are a mechanism by which the cell degrades mutant proteins and are therefore protective. As both aggresomes and inclusions can appear similar in immunofluorescence and immunohistochemical assays, the correlation between visible aggregates and disease can be complicated. The authors also suggest that toxicity is dependent more upon the mutant monomer or small microaggregates rather than the large inclusion bodies.

To further try and address this debate over the role of aggregation in cytotoxicity, Arrasate et al. (2004) developed an automated microscope system that was capable of visualizing individual cells over long time periods. They used rat primary striatal neurons that were co-transfected with monomeric red fluorescent protein (mRFP) and GFP-tagged HTT^ex1 fragments of varying glutamine lengths. By monitoring mRFP expression, they could determine when each individual cell died and whether or not it correlated with the visible aggregation of the green fluorescent HTT^ex1. They found that under these conditions, neuronal death depended only upon HTT^ex1 dose and the length of polyglutamine expansion. Both time and the presence of visible inclusions were independent of cytotoxicity. In fact, cells that produced large visible inclusions survived longer than those that did not. Cell death could instead be predicted by the levels of diffuse HTT^ex1 expression, which decreased upon inclusion formation. Similarly, Takahashi et al. (2008) used a Fluorescence Resonance Energy Transfer (FRET)-based method to visually distinguish monomers, soluble oligomers, and large inclusion bodies of a fluorescently tagged Atrophin-1 (the protein mutated in DRPLA). In this study, they found that cytotoxicity correlated with the expression of the mutant oligomers, but not the inclusions or monomers.

WHAT IS AN AGGREGATE?

Given the competing hypotheses for what constitutes a toxic versus protective aggregate, several studies have now been undertaken to attempt to visualize the aggregation process and have revealed that there are perhaps several on- and off-pathway intermediates formed within the cell at any given time (reviewed by Burke et al., 2011; Hands and Wyttenbach, 2010; Ross and Poirier, 2005).

Atomic Force Microscopy (AFM) provides a means to look at the ultrastructure of individual oligomers and inclusions in vitro and has been used to monitor HTT^ex1 aggregation over time (reviewed by Burke et al., 2011). Legleiter et al. (2010) found that purified HTT^ex1 initially formed spherical oligomers and this oligomer population dropped as fibrils began to form. By later stages fibrils were the predominant species. The kinetics of fibril formation depended on the length of the CAG repeat, with longer repeats forming fibrils faster than shorter repeats. Wildtype HTT^ex1, in contrast, formed oligomers, but did not form fibrils. In addition, the expanded mutant HTT^ex1 fragment, but not pure polyglutamine, occasionally formed annular aggregates. These structures were never seen with wildtype HTT fragments. This is in contrast to what has been described for AR, which forms annular structures when wildtype but forms small fibrils when polyglutamine-expanded. In this case, treatments or mutations that increased or diseased toxicity in Drosophila triggered the conversion of annular structures to fibrils or vise versa, suggesting that the fibrillar structures were more toxic (Jochum et al., 2012). Nonetheless, based on these studies, an aggregation mechanism was proposed for HTT in which the monomer samples different misfolded conformations that sometimes trigger different aggregation pathways. Sometimes they form the large annular structures observed, but more often they form small oligomers. These oligomers can then form “off-pathway” small annular aggregates or amorphous aggregates or they can undergo a structural transition that allows them to form fibrils. “Inclusions” are bundles of fibrils and amorphous aggregates (Fig. 1) (Burke et al., 2011; Hands and Wyttenbach, 2010; Legleiter et al., 2010). While these AFM studies do not exclusively demonstrate whether the observed oligomeric and fibrillar structures are toxic or benign, size exclusion chromatography has been used to isolate subsets of oligomeric forms from pure protein extracts and AFM analysis confirmed the homogeneity of these size exclusion chromatography fractions (Lotz et al., 2010). This has prompted some researchers to suggest that future studies may be able to definitively test whether a given oligomeric species is cytotoxic or not (Burke et al., 2011).

Fig. 1.

The predicted polyglutamine protein aggregation pathway and common detection methods. (A) Schematic of the proposed polyglutamine (polyQ) protein aggregation pathway. PolyQ expansion results in the misfolding of the protein monomer, which can lead to oligomerization. Small oligomers can be incorporated into protofibrils and fibrils, which, along with amorphous aggregates, contribute to inclusion formation. Misfolded monomers and small oligomers can also form other “off-pathway” structures that may or may not contribute to eventual inclusion formation. Detecting different aggregated forms can be accomplished in a number of ways as noted below the pathway diagram. EM, Electron Microscopy; AFM, Atomic Force Microscopy. (B, C) Examples of some of the most common aggregate detection methods are shown. (B) Soluble extracts from a polyglutamine disease mouse model are compared to those of a littermate control via western blot and a filter trap assay. Aggregation is only detected in extracts from the polyglutamine disease mouse model. The black arrow marks the accumulation of oligomerized mutant protein in the stacking gel of the western blot. (C) Large inclusions (white arrows) can be detected by immunofluorescence (IF) as shown, or by immunohistochemistry (IHC). An example of GFP-tagged polyQ-expanded mutant protein inclusions (green) in a transfected HeLa cell is shown. A nuclear marker, DAPI, is in blue and red marks a protein present in the cytoplasm and nucleus, allowing for the visualization of the cell.

While in vitro studies of the polyglutamine proteins can shed some light on the aggregation intermediates and pathways that a protein can undergo, identifying these intermediates in vivo in mammalian models or in human patient tissue is not so trivial. AFM is more applicable to pure protein extracts compared to more heterogeneous cell and tissue extracts. Sathasivam et al. (2010) were able to look at a subset of oligomers isolated from mouse tissue extracts via AFM, however, and found that these spherical oligomers were remarkably similar to those observed for pure HTT and polyglutamine fragments in vitro. Electron microscopy, agarose gel electrophoresis (“AGERA”), and FRET-based methods have also been used to look an aggregation in mouse tissue and cell extracts, and such studies suggest that many of the structures seen in vitro may exist in vivo as well (Baldo et al., 2012; Cowin et al., 2012; Legleiter et al., 2010; Sathasivam et al., 2010; Weiss et al., 2008). Nonetheless, many of these studies remain only correlative and it will be interesting to see whether any individual misfolded or oligomeric species are linked to toxicity in vivo at the mammalian level in the future.

WHAT IS RESPONSIBLE FOR TOXICITY?

The work by Arrasate et al. (2004) mentioned above represented something of a paradigm shift in the field of polyglutamine research. Studies now tend to consider large visible inclusions as protective. But then what is toxic? While several of the various oligomers that have been identified in vitro and in vivo have yet to be specifically correlated with cellular dysfunction and/or death, some theories have been put forth as to what constitutes the toxic species in polyglutamine diseases.

One contribution to the toxic versus protective nature of a given aggregate may simply be its location. It has long been known that the nuclear localization of the mutant protein is essential for disease progression in many of the polyglutamine diseases and it has been argued that nuclear aggregates are able to induce cell death, while cytoplasmic aggregates are more benign (Bates, 2003; Chen et al., 2001; 2002; Yang et al., 2002). Recent work in the R6/2 model of HD has shown that highly expanded polyglutamine lengths (> 300 glutamines) are actually protective compared to shorter pathogenic glutamine lengths (Cowin et al., 2012; Dragatsis et al., 2009; Morton et al., 2009). Cowin et al. (2012) found that while inclusion formation still correlated with onset in these mice, lines with longer glutamine lengths tended to develop cytoplasmic inclusions while shorter pathogenic lengths were associated with intranuclear aggregation. This further supports the idea that the location of an inclusion can greatly affect its toxicity. It is worth noting, however, that there are also a number of studies that suggest that the cytoplasmic or neuropil aggregation of HTT in particular can correlate with at least early disease progression and induce neurodegenerative effects (Gutekunst et al., 1999; Lee et al., 2004; Li et al., 1999; Sapp et al., 1999).

Some studies support the hypothesis that the kinetics by which the mutant protein aggregates is the basis of its toxicity. Gong et al. (2008) carried out a time-lapse study of HTT aggregation in PC12 cells and found a less conclusive relationship between inclusions and cytotoxicity than that observed by Arrasate et al. (2004). Instead, they found that cells with any size aggregates could undergo cell death, but cells that formed inclusions faster tended to die earlier and those that slowly formed large inclusions tended to live longer. These data support the idea that some larger, slow-growing inclusions are protective, but also suggest that perhaps the kinetics of aggregation is important in disease pathogenesis (Gong et al., 2008). Support for this idea comes from a study by Kvam et al. (2009) in which intrabodies specific to either the polyglutamine region or other regions of HTT^ex1 could be used to speed up and slow down aggregation, respectively. Slowing down aggregation decreased cell death, while the intrabody that promoted aggregation on to the fibrillar state slightly increased cell death. This intrabody had a similar effect in ATAXIN-3 expressing cells.

But perhaps the most prevalent idea in the field is that a specific conformer, be it a misfolded monomer or part of an oligomer or fibril, is responsible for polyglutamine-mediated toxicity. The automated microscope studies performed by Arrasate et al. (2004) suggested that cell death could be predicted by the levels of diffuse HTT^ex1 in the cells, suggesting that the toxicity arose from either the monomer or a small oligomer. This group went on to identify an antibody that is capable of identifying the species that predicted cell death and support a model in which a specific misfolded conformation of the polyglutamine monomer is responsible for cytotoxicity (Miller et al., 2011). This is consistent with other studies that use conformation-specific antibodies to attempt to identify the toxic species in disease models (Bennett et al., 2002; Legleiter et al., 2009; Trottier et al., 1995). Nagai et al. (2007) also argued for a toxic monomeric conformation based on their work in cultured cells. In contrast, other groups have argued that the toxic species is an oligomer and that the expression of the polyglutamine monomer does not correlate with toxicity (Behrends et al., 2006; Poirier et al., 2002; Ross and Poirier, 2005; Sánchez et al., 2003; Takahashi et al., 2008; 2010). Based on their AFM studies, Legleiter et al. (2010) also conclude that cytotoxicity arises from either a misfolded monomer or oligomer, but more specifically, they predict that the ability of the polyglutamine-expanded monomer to sample different conformations underlies its toxicity. They suggest that each misfolded state could produce different gain-of-function interactions and the level of these detrimental interactions would be proportional to the number of different conformations that the monomer sampled. But they also acknowledge that their data do not suggest whether all or only some of the misfolded states are toxic.

Support for a toxic conformer model can also be found in a study by Nekooki-Machida et al. (2009). In this study the authors were able to form two different forms of fibrillar aggregates in vitro - one with more compact beta sheets and one with a more labile organization. They then delivered these preformed inclusions into mammalian cells that were expressing a GFP-tagged mutant HTT^ex1 and found that while both inclusion structures induced equivalent levels of overall aggregation, the more compact inclusion structure was less toxic than the labile structure. They were also able to purify aggregated structures from various brain regions of R6/2 HD mice and found that the aggregated material from the more vulnerable striatum region was more toxic and resembled the labile conformation while material from less vulnerable regions was more compact and less toxic. These data strongly argue that a specific conformation of the polyglutamine region, be it in a monomer, oligomer, or fibrillar structure, is responsible for toxicity, and that inclusions with more compact, buried polyglutamine regions could efficiently sequester the mutant proteins into a more benign or protective inclusion.

The ability of a mutant polyglutamine protein to form a toxic versus benign conformation may depend upon the cellular context, perhaps due to varying expression levels of protein folding machinery or polyglutamine protein interactors (Nekooki-Machida et al., 2009). This idea is consistent with a more recent study by Kondo et al. (2013) on the distribution of inclusions in a transgenic model of SBMA. The authors of this study found that altering the expression pattern of heat shock factor-1 (HSF1), and its downstream target heat shock protein 70/72 (Hsp70/72), was able to affect the pattern of AR inclusion formation so that regions that expressed low levels of HSF1 formed inclusions more readily than regions with higher levels of the protein. Furthermore, ectopic aggregation of the mutant AR in additional neuronal and non-neuronal regions (such as the liver) induced cell death and dysfunction. The authors therefore also conclude that the distribution of inclusions and the neuronal specificity seen in each polyglutamine disease may depend upon the endogenous expression pattern of other factors such as Hsps.

MODULATING AGGREGATION PATHWAYS FOR THERAPY

Work on the possible protective role of inclusion formation has prompted the hypothesis that a compound capable of decreasing soluble oligomeric forms in favor of larger cytoprotective inclusions may represent a viable therapeutic avenue for the polyglutamine diseases. In this vein, the compound B2 was identified as being capable of promoting aggregate formation and decreasing cytotoxicity in an HD cell model, despite the fact that it also raised overall HTT^ex1 expression (Bodner et al., 2006). The anti-cytotoxic effects of this compound were also demonstrated in a cell model of SBMA, in which B2 increased visible aggregate formation in the presence of the AR ligand, but did not affect the presence of high molecular weight microaggregates seen in the stacking gels of western blots under the same condition. This study also found that treatment with B2 improved the rough eye phenotype of a full-length polyglutamine-expanded AR Drosophila model, again suggesting that it could have therapeutic potential (Palazzolo et al., 2010). As a similar therapeutic approach, Ehrnhoefer et al. (2006) found that Green tea polyphenol (-)-epigallocatechin-gallate (EGCG) could promote the formation of large spherical oligomers as detected by AFM and decrease the presence of more soluble oligomers, as detected by a nitrocellulose dot blot. HTT^ex1 is known to lose immunoreactivity with the MW1 antibody over time as it aggregates, and EGCG was able to prevent this, suggesting that it inhibits an early conformational switch in the aggregation pathway of the mutant protein. Treatment with EGCG also reduced visible aggregation and toxicity in a yeast model and improved phenotypes in a fly model of HD (Ehrnhoefer et al., 2006). Nonetheless, while such studies as those outlined above show promise, other studies such as those using the aggregation-promoting intrabody described by Kvam et al. (2009) suggest that caution should be taken when considering these aggregation-promoting therapeutic strategies, particularly until the true toxic species can be identified. As yet such approaches have not been tested at the mammalian level.

Agents that specifically prevent the aggregation of polyglutamine proteins are an active area of research and remain a viable therapeutic avenue (Lansbury and Lashuel, 2006). The ability of the azo-dye Congo red to inhibit aggregation and relieve disease-related phenotypes is a classic example of this approach (Sánchez et al., 2003), although its effects in mouse models have been debated (Wood et al., 2007). Trehalose is also known to prevent aggregation and disease symptoms in a mouse model of HD (Tanaka et al., 2004). More recent studies have shown that methylene blue can prevent the formation of mutant oligomers and inclusions and improve phenotypes in HD mouse models (Sontag et al., 2012). Methylene blue also modulates the degradation of the polyglutamine-expanded AR in SBMA cell models (Wang et al., 2010). Lastly, the polyglutamine binding peptide QBP1 has also been found to selectively bind polyglutamine proteins and prevent their misfolding and aggregation (Popiel et al., 2013). While the delivery of QBP1 to vulnerable regions in mammalian disease models has been complicated, the peptide remains a potentially viable therapeutic option.

MODULATING AGGREGATION CLEARANCE MECHANISMS FOR THERAPY

In general, therapeutic strategies that lead to reduced aggregation are beneficial in polyglutamine disease models, particularly if they decrease both large inclusions and more soluble forms and/or if they increase the degradation of the mutant protein monomer as well. Exploring ways to prevent the formation or to trigger the clearance of aggregates, such as modulating chaperone activity or promoting protein degradation activity, represent promising avenues for therapeutic intervention in the polyglutamine and other protein misfolding disorders.

It has been known for some time that upregulating the chaperone or heat shock response in cells has a beneficial effect in polyglutamine disease models, presumably by allowing for the refolding of mutant proteins and/or their targeted degradation via the ubiquitin-proteasome system (UPS). Both more general chaperone inducing agents (Katsuno et al., 2005; Malik et al., 2013) and the overexpression of specific chaperones or their interactors (Adachi et al., 2003; 2007; Al-Ramahi et al., 2006; Behrends et al., 2006; Cummings et al., 1998; Howarth et al., 2007; 2009; Jana et al., 2005; Miller et al., 2005) have been able to reduce various disease-related phenotypes, including protein aggregation, while their inhibition can be detrimental (Wacker et al., 2009; Williams et al., 2009). Whether the beneficial effects of modulating chaperone expression are due to a direct effect on protein aggregation is debated, however. While some studies have found that chaperones Hsp40 and Hsp70 act to refold and degrade misfolded oligomers or monomers (Lotz et al., 2010; Muchowski et al., 2000), others have suggested that these proteins exert their beneficial effects on polyglutamine toxicity via mechanisms that are independent of any effect on aggregated proteins (Wacker et al., 2009; Zhou et al., 2001). Still another study found that pharmacologically activating the ATPase activity of Hsp70 can stimulate the solublization of aggregated polyglutamine proteins, and that this effect is toxic, presumably due to the removal of protective inclusion bodies. In this case inhibiting Hsp70 activity was protective (Chafekar et al., 2012).

Stimulating classic macroautophagy has also been shown to be beneficial in polyglutamine disease models by promoting the degradation of the mutant protein and reducing aggregation (Montie and Merry, 2009; Montie et al., 2009; Ravikumar et al., 2004; Williams et al., 2006), although one group found that overstimulation of the pathway can have detrimental effects in an SBMA model (Yu et al., 2011). Inhibition of Hsp90 by the compound 17-AAG or 17-DMAG is beneficial in SBMA mouse models, and it has been suggested that the increased degradation of the mutant AR seen in these mice is also via autophagy or the UPS (Rusmini et al., 2011; Tokui et al., 2009; Waza et al., 2005). On the other hand, Doi et al. (2013) found that inhibiting autophagy by knocking down the receptor p62 increases mutant AR aggregation and has detrimental effects in an SBMA mouse model. This study also found that overexpression of p62 has beneficial effects, although the underlying mechanism appears to be a promotion of protective inclusion formation as opposed to an effect on macroautophagy per se.

Chaperone-mediated autophagy (CMA) has also been exploited as a means of removing misfolded polyglutamine proteins. Bauer et al. (2010) attached HSC70 binding sites to the QBP1 peptide to target polyglutamine proteins for degradation by CMA. When the construct was virally injected into two different HD mouse models, it both inhibited aggregation (similar to QBP1 alone) and reduced overall soluble HTT levels, resulting in an amelioration of the disease phenotypes.

CONCLUSION

In conclusion, the role of aggregation in the polyglutamine diseases remains a topic of debate within the field. While the presence of aggregates or inclusions correlates with disease in human tissue and various models of the diseases, it is no longer immediately assumed to be the driving force in the disease pathogenesis. It is now generally accepted that large inclusions may be a means by which the cell isolates and buries the toxic disease-causing proteins and are therefore protective. But prior to mature inclusion formation, misfolded monomers or small oligomers are generated and these intermediates may represent the true toxic species. Attempting to modulate the aggregation pathway may therefore allow for therapeutic intervention in these diseases, but could also promote the formation of toxic species and therefore exacerbate the condition. As future studies elucidate whether the various conformations seen in vitro and in vivo contribute to toxicity, such approaches may prove more feasible. All the same, therapeutic interventions that reduce the overall aggregate load, particularly by preventing or clearing both early and late stages of aggregation, are often beneficial in disease models and represent an interesting avenue of future research.