|Home | About | Journals | Submit | Contact Us | Français|
Toxicity in amyloid diseases is intimately linked to the nature of aggregates, with early oligomeric species believed to be more cytotoxic than later fibrillar aggregates. Yet mechanistic understanding of how aggregating species evolve with time is currently lacking. We have explored the aggregation process of a chimera composed of a globular protein (cellular retinoic acid-binding protein, CRABP) and huntingtin exon 1 with polyglutamine tracts either above (Q53) or below (Q20) the pathological threshold using Escherichia coli cells as a model intracellular environment. Previously we showed that fusion of the huntingtin exon 1 sequence with >40Q led to structural perturbation and decreased stability of CRABP (Ignatova, Z., and Gierasch, L. M. (2006) J. Biol. Chem. 281, 12959–12967). Here we report that the Q53 chimera aggregates in cells via a multistep process: early stage aggregates are spherical and detergent-soluble, characteristics of prefibrillar aggregates, and appear to be dominated structurally by CRABP, in that they can promote aggregation of a CRABP variant but not oligoglutamine aggregation, and the CRABP domain is relatively sequestered based on its protection from proteolysis. Late stage aggregates appear to be dominated by polyGln; they are fibrillar, detergent-resistant, capable of seeding aggregation of oligoglutamine but not the CRABP variant, and show relative protection of the polyglutamine-exon1 domain from proteolysis. These results point to an evolution of the dominant sequences in intracellular aggregates and may provide molecular insight into origins of toxic prefibrillar aggregates.
Protein aggregation is associated with an expanding set of diseases, yet a clear correlation between the mechanism of aggregation and pathology is generally lacking. In the specific example of the polyglutamine (polyQ)3 disease family, to which Huntington disease belongs, pathology is almost invariably associated with mutations that increase the number of glutamine residues in the corresponding protein over a pathological threshold value of 30–40 repeats (1). But the direct role of polyQ extensions in toxicity has been questioned due to the poor correlation between the presence of visible aggregates and pathological symptoms in post-mortem studies and cell culture models (2–4). Complicating efforts to understand how aggregation correlates with pathology in glutamine repeat diseases is the compelling evidence from several systems that the nature of aggregates formed by polyQ-containing proteins evolves over time. In vitro studies showed that the fibrillization of expressed huntingtin (Htt) exon 1 containing a long polyQ stretch is a multistep process with early formation of globular oligomers (5–7), similar to the metastable oligomers and protofibrils identified in the aggregation of both Alzheimer amyloid plaque peptide Aβ (8, 9) and α-synuclein (10, 11). Of crucial importance to understanding the impact of the multistage aggregation events is the growing evidence, albeit largely from cell culture, that the most toxic species are in fact the prefibrillar aggregates that form early (9, 12–16).
Additionally, it is mechanistically significant that polyQ sequences and the sequences flanking them have an impact on each other, and that this context-dependent interplay depends in turn on the length of the polyQ segment. Indeed, profound effects of polyQ context have been reported on aggregation propensity (17–19) and cytotoxicity (20, 21). Based on their conclusion from in vitro results that polyQ extensions in ataxin-3 alter the conformation of the adjacent Josephin domain so as to favor its aggregation, Pastore and colleagues (19) proposed the “induced misfit” model, which argued for an indirect as well as a direct role of the polyQ sequence. Consistent with the earlier Pastore model, in vitro studies of ataxin-3 revealed a two-step aggregation process with the first step mediated by the globular Josephin domain and the second by the polyQ tract (22, 23). Our results using a chimera of an otherwise well folded and stable globular protein (cellular retinoic acid-binding protein, or CRABP) and the polyQ-containing Htt exon 1 indicated that extension of the polyQ tract above the pathological threshold perturbed the structural properties of the adjacent model protein and led to enhanced aggregation either in vitro or in Escherichia coli cells and suggested that the morphology of the resulting aggregates evolved over time (24).
It is critical to determine whether these behaviors are general to multidomain polyQ-containing proteins, to learn the nature of the species formed at different stages in aggregation, and to explore how the cellular environment may affect the aggregation mechanism. In our preliminary characterization of the CRABP-Htt exon 1 chimera containing a polyQ repeat in the pathological range (Q53), we observed that aggregates formed changed from detergent-soluble to detergent-insoluble as a function of time (24). In the present study, we have built on this earlier work with the CRABP-Htt exon 1 system and further characterized its multistep aggregation process in E. coli cells to better understand the nature of the different aggregates formed as a function of time and to assess the generality of the interplay between polyQ sequences and flanking domains. The E. coli system, although not a direct mimic of aggregation processes in mammalian cells, has been shown by several laboratories to support the formation of cellular aggregates (generally termed “inclusion bodies”) that are structurally homologous to those formed in eukaryotic cells (25, 26). The manipulability of E. coli as an in vivo system is a clear advantage for the types of experiments we are carrying out. Aggregates formed can be isolated and characterized. In this way, using limited proteolysis, antibody reactivity, and electron microscopy of the isolated aggregates, we show in the present study that different segments of the CRABP-Htt exon 1 chimera are involved in the aggregate core at different times. Thus, the role of the polyQ stretch in the structure of the aggregate core ranges from peripheral in the early spherical detergent-soluble prefibrillar species to central in the second stage, the formation of detergent-resistant fibrils. Strikingly, our results on in-cell aggregation of this polyQ-containing chimeric system are entirely analogous to those observed for ataxin-3 in vitro in the Bottomley laboratory (22, 23) and for a chimera of glutathione S-transferase and exon 1 of huntingtin (17). This emphasizes the potential general significance of the effect of polyQ segments on their neighboring domains and the resulting impact on the time-dependent character of aggregates and suggests that the complex aggregation mechanisms previously observed in vitro can also occur in the cellular environment. As pointed out by Pastore and co-workers (17), such mechanistic insights are critical to understanding pathology, because the early, potentially more toxic prefibrillar aggregates could form in a variety of systems via indirect effects of elongated polyQ tracts.
Exon 1 of huntingtin with 20 or 53 CAG repeats, termed Htt20 or Htt53, respectively, was fused C-terminally to CRABP carrying a tetra-Cys motif for binding to FlAsH, and the chimeras were expressed in E. coli BL21(DE3) cells and labeled with FlAsH as described previously (24, 27).
At different times after induction, 250 ml of culture was harvested at 2,060 × g for 5 min at 4 °C and used to isolate intact aggregates following the procedure described by Oberg et al. (28) with the following modifications. Pelleted cells were re-suspended in 10 ml of 20 mM Tris·HCl, pH 8.0, buffer containing 1mM phenylmethylsulfonyl fluoride, 5 μg/ml lysozyme, 50 ng/ml DNase, and 0.1% Triton X-100, and were sonicated on ice for 5 min (50% duty cycle). Cell debris was removed by low speed centrifugation at 376 × g for 3 min at 4 °C, and the supernatant suspension was centrifuged again at 15,870 × g for 5 min to separate the soluble cytoplasmic proteins. The pellet fraction, which primarily comprises intact aggregates, was re-suspended in 5 ml of the same buffer, aliquoted, frozen in liquid nitrogen, and stored at −80 °C. Small aliquots of each fraction of the in vivo aggregates were dissolved either in 10 mM Tris·HCl, pH 8.0, containing 8 M urea or in formic acid (for the later fibrillar species), and their concentration was determined spectrophotometrically using the ε280 value of tetra-Cys CRABP (21,750 M−1cm−1; the Htt exon 1 sequence contains no Trp or Cys). Buffer containing 8 M urea or formic acid served as a blank.
The total fraction of aggregates isolated from cells were diluted to 3 μM in 200 μl of 20 mM Tris·HCl, pH 7.5, containing 150 mM NaCl (native buffer), and the total volume of the aggregate suspension was applied untreated to a nitrocellulose membrane (Schleicher and Schuell) through a slot-blot filtration unit and washed three times with 200 μl of the native buffer. Immunodetection was accomplished using the oligomer-specific A-11 antibody (diluted 1:5,000) (29) (Chemicon) and the anti-polyQ antibody MW1 (diluted 1:1,000) (30) (Developmental Studies Hybridoma Bank, Iowa City, IA). Visualization was via peroxidase-conjugated anti-rabbit or anti-mouse secondary antibody (Sigma-Aldrich). In parallel, the same amount of the aggregates was dissolved in SDS-containing buffer (20 mM Tris·HCl, pH 7.5, containing 150 mM NaCl, 4% SDS, and 1 mM dithiothreitol) and centrifuged at 15,870 × g at 4 °C for 10 min to separate the SDS-insoluble fraction from the total aggregate fraction. The detergent-insoluble pellet was re-suspended in the same amount of the SDS-buffer and applied onto the slot-blot membrane. Samples were washed three times with 200 μl of 1% SDS solution and probed with anti-polyQ antibody MW1.
Intact inclusion bodies isolated from tetra-Cys CRABP Htt53-expressing cells at different times after induction or from in vitro aggregation samples were sonicated for 2 min in 30-s bursts to improve their seeding potency.
15 μM P39A tetra-Cys CRABP expressed and purified as described elsewhere (31) was labeled with FlAsH-EDT2 for 60 min at 37 °C. In vitro aggregation was initiated by destabilization with 1.7 M urea (31) and/or addition of 5% (w/w) seeds of the in vivo isolated or in vitro grown tetra-Cys CRABP Htt53 aggregates. Aggregation kinetics was monitored at 37 °C by FlAsH fluorescence, which increases as aggregates form (27).
100 ng of sonicated in vivo or in vitro aggregates of CRABP Htt53 were diluted in 100 μl of phosphate-buffered saline, placed into each well of a microtiter plate and dried overnight at 37 °C. The wells were washed and blocked with 1% bovine serum albumin for 1 h and washed with 100 μl of phosphate-buffered saline. The elongation reaction was initiated by adding 100 μl of a 100 nM solution of biotinylated-Q29 peptide at 37 °C. The reaction was stopped by washing and addition of 100 μl of Eu-Streptavidin (1:1,000) for 1 h in the dark. After incubation with 100 μl of enhancement solution for 15 min, the europium counts were determined by fluorescence, converted into femtomoles of europium, and converted further to femtomoles of bound biotin-peptide (32).
Intact aggregates were re-suspended to 15 μM in buffer (10 mM Tris·HCl, pH 7.6, containing 50 mM NaCl and 2 mM CaCl2) to which freshly dissolved trypsin (Sigma-Aldrich; stock solution 1 mg/ml in 1 mM HCl/NaCl buffer, pH 3.0, was diluted in 10 mM Tris·HCl, pH 7.6, 2 mM CaCl2) was added to a final concentration of 0.75 μM (enzyme:protein ratio of 1:20 w/w, relative to the concentration of the tetra-Cys CRABP Htt53 monomer in the fibrils determined as described above), and incubated at room temperature for 45 min. Digestion was stopped by addition of formic acid, which deactivates the protease and converts the samples into soluble material, addition of SDS-loading buffer, and boiling. The digestion bands resolved on 15% SDS-gels were detected by immunoblotting after transfer onto polyvinylidene difluoride membranes. The membranes were probed with anti-CRABP antibodies (33), which recognize the CRABP domain in the chimeras (Abcam, UK), and anti-polyQ MW1 antibodies (each diluted to 1:1,000).
Under the same digestion conditions, monomeric soluble tetra-Cys CRABP Htt53 was fully digested to fragments smaller than 10 kDa, which were not distinguishable on the gel. Hence, limited proteolysis of this protein was performed for 45 min on ice using a lower trypsin:protein ratio (1:50, w/w).
Tetra-Cys CRABP Htt53 was expressed in E. coli BL21(DE3) and purified from the insoluble SDS-labile fraction as already described (24). Aliquots of purified tetra-Cys CRABP Htt53 (10–15 nM) were labeled with FlAsH-EDT2 and concentrated to the desired concentration by ultrafiltration (cut-off 10 kDa) prior to use; the concentration was verified spectrophotometrically, and the solutions were used immediately in in vitro aggregation experiments (24). Aggregation was initiated by destabilization of the Htt chimera with 1 M urea (10 mM HEPES, pH 7.8) and monitored by FlAsH-fluorescence. The samples were incubated at 37 °C without stirring and gently vortexed prior to withdrawal of an aliquot. The fluorescence of controls without protein was used as a blank and subtracted from each time point (24).
2 μl of 3 μM suspensions of aggregates isolated either from cells or from in vitro preparations (as above) were deposited on carbon-coated copper grids, stained with 3% (w/v) ammonium molybdate (pH 6.8), and imaged with a Phillips CM12 electron microscope.
We designed chimeras of a well characterized and stably folded protein, CRABP, and exon 1 of the huntingtin protein, Htt, with a polyQ tract either in the non-pathological (tetra-Cys CRABP Htt20) or in the pathological range (tetra-Cys CRABP Htt53). By incorporating a specific motif (Cys-Cys-Gly-Pro-Cys-Cys) into CRABP, we can label the fusion protein in cells with the low molecular weight membrane-permeant fluorescent dye, FlAsH (34). Our previous work has established that the fluorescence signal from FlAsH-labeled CRABP-Htt chimeras reports with fidelity on in-cell aggregation (35). Moreover, in our E. coli expression system, aggregates can be isolated at different times after induction of synthesis of the chimera for physical and biochemical characterization.
The tetra-Cys CRABP Htt chimera retained the polyQ length-dependent aggregation behavior characteristic of polyQ-dependent neuropathies (1, 24). The time course of expression of tetra-Cys CRABP Htt exon 1 fusion with a polyQ repeat in the non-pathological range (20 glutamines) resembles the fluorescence time course of soluble tetra-Cys CRABP (Fig. 1A), and the solubility of the tetra-Cys CRABP Htt20 chimera over the entire period of expression was confirmed by fractionation studies (Ref. 24 and data not shown). In contrast, the fluorescence of the tetra-Cys CRABP Htt53-expressing cells rises steeply from time of induction and at all times exceeds that of cells expressing tetra-Cys CRABP Htt20 or the tetra-Cys CRABP construct alone (without Htt exon 1) (Fig. 1A), indicating the formation of non-native states, including aggregates (27). In our previous work, we observed a time-dependent change in the solubility of the intracellular deposits of tetra-Cys CRABP Htt53: purified intact aggregates from tetra-Cys CRABP Htt53-expressing cells were isolated at different times after induction and subjected to electron microscopy analysis. Aggregates isolated at early time points (60 min), which we previously demonstrated were SDS-soluble (24), were exclusively small spherical oligomers (Fig. 1B). At 120 min after induction, short curvilinear structures appear and co-exist with the small oligomeric spherical aggregates, whose size and morphology did not change (Fig. 1B). At later time points after induction (180 and 240 min, Fig. 1B), corresponding to the plateau in the fluorescence curve and to the development of detergent insolubility (24), spherical species were no longer apparent, and the aggregates show a more uniform fibrillar-like morphology by electron microscopy. All of the aggregates isolated from the E. coli cells reacted under non-denaturing conditions with the anti-polyQ monoclonal antibody MW1 (30) (Fig. 1C).
An antibody (A-11) raised against the Aβ (1–40) peptide has been reported to recognize a generic epitope present on early oligomeric structures and prefibrillar species of several amyloidogenic polypeptides, independent of the amino acid sequence of the aggregating protein (see positive reactions of the A-11 antibody with known prefibrillar sample) (29). The A-11 antibody bound only to the early aggregates of tetra-Cys CRABP Htt52, isolated 60 min and 120 min after induction (Fig. 1C). These results support the electron microscopy data in suggesting that the early tetra-Cys CRABP Htt53-aggregated species display a generic conformation, characteristic of an oligomeric prefibrillar state, which corresponds to neither the monomer nor the mature fibers of these proteins.
Changes in the morphology of the aggregates of the polyQ-containing chimera and in their physicochemical properties suggest a complex aggregation pathway in vivo. A sensitive and remarkably discriminating assay to explore the structural features of distinct aggregates is their seeding ability (36, 37). We tested the ability of tetra-Cys CRABP Htt53 aggregates formed in E. coli cells to enhance aggregation of two distinct species comprising different parts of the tetra-Cys CRABP Htt53 chimera: polyQ peptides (36, 38) or an aggregation-prone point mutant of CRABP (P39A tetra-Cys CRABP (31)). The former assay was developed to assess the seeding of elongation of polyQ (32, 36, 38). The latter assay is based on our characterization of the slow folding, aggregation-prone variant of CRABP, which normally displays a lag time before aggregation proceeds under defined conditions, and this lag time can be reduced by adding pre-formed aggregates (31). Our logic was that the ability to seed elongation of polyQ peptides is a measure of the involvement of the Htt-extended polyQ sequence in ordered regions of the chimera aggregates. Conversely, the ability of the isolated aggregates to accelerate the aggregation of P39A tetra-Cys CRABP would reveal participation of the CRABP sequence in the ordered regions of the chimera aggregates. Aggregates from the first two time points after induction (60 and 120 min) showed almost no ability to stimulate elongation of polyQ aggregates, as monitored by enhanced immobilization of biotinylated Q29-polypeptides (36, 38) (Fig. 2A). In contrast, aggregates of tetra-Cys CRABP Htt53 isolated at later times (180 and 240 min) significantly enhanced the deposition of Q29 peptides (Fig. 2A). The ability of the aggregates of tetra-Cys CRABP-Htt53 to accelerate aggregation of P39A CRABP exhibits the opposite trend: aggregates isolated at the earliest time points significantly reduced the lag phase in P39A tetra-Cys CRABP aggregation, whereas aggregates from later time points were unable to alter the time course of P39A tetra-Cys CRABP aggregation (Fig. 2B).
These cross-seeding experiments suggest that different regions of the chimera are transiently involved in the aggregate core, which may in turn account for the distinct morphology of the aggregates at different times after induction. We then used limited trypsinolysis to map the relative flexibility and accessibility of various regions of the aggregates. Limited proteolysis has recently been shown to be effective in identifying those peptide/protein segments that are not protected in the amyloidogenic core in fibrils of different amyloidogenic proteins (39–41). Trypsinolysis of tetra-Cys CRABP Htt53 should lead to a maximum of 12 cleavages, with three of these in Htt exon 1 (Fig. 3C). To map the trypsin accessibility in the aggregated species of tetra-Cys CRABP Htt53 isolated at different times after induction, we visualized the same digestion products with either polyQ antibodies (Fig. 3A) or anti-CRABP antibodies (Fig. 3B). The digestion patterns of the earlier time point aggregates with a spherical phenotype in the electron microscopy images (60 and 120 min) were clearly different from the patterns for in the 180 and 240 min fibrillar aggregates (compare lanes 2 and 3 with lanes 4 and 5 of Fig. 3A or lanes 1 and 2 with lanes 3 and 4 of Fig. 3B). In the earlier species, the C-terminal part of the chimera, comprising Htt exon 1, was more susceptible to trypsinolysis leading to bands a–d that reacted with the MW1 antibody (Fig. 3A). These bands can be attributed to cleavages near the boundary between CRABP and Htt exon 1 (band a) and within the N-terminal sequence of Htt exon 1 (bands b, c, and d on Fig. 3A and shown schematically in Fig. 3C), although caution must be exercised in specific assignment of bands due to the well known reduced mobility and consequent poor molecular weight correlation of polyQ-containing sequences. In contrast to the early lability of the polyQ-containing Htt domain, the CRABP domain remained more protected in the 60- and 120-min species, where only one dominant cleavage occurred yielding a fragment of ~18-kDa molecular mass that cross-reacted with the anti-CRABP antibody (band e in Fig. 3B, shown schematically in Fig. 3C). The proteolytic products from later time aggregates (180 and 240 min) showed a reversed pattern: the C-terminal polyQ domain was more protected; the proteolytic pattern was dominated by the 18-kDa proteolytic band marked as b on Fig. 3 (A and C), whereas the CRABP domain was more accessible, yielding fragments with molecular masses of 14, 10 and 6 kDa (marked as f, g, and h on Fig. 3, B and C). Together, these results are consistent with the formation of two different protease-resistant aggregate cores over the time course of aggregation of tetra-Cys CRABP Htt53 in cells: the model emerging is that the CRABP-domain dominates the aggregate core of the initial detergent-labile aggregates, whereas the polyQ-containing Htt-domain comprises the core of the later time aggregates. Notably, the formation of the polyQ aggregate core coincides with the appearance of fibrillar, detergent-resistant species that have the ability to seed aggregation of polyQ peptide monomers.
In summary, the above results establish that the aggregation of tetra-Cys CRABP Htt53 in vivo in E. coli cells is a multistep process involving a transient aggregated species exhibiting core and exposed peptide elements that differ from those of the mature, end-point aggregates. To assess the extent to which the environment of the cell influenced the observation of multistep aggregation, we examined the mechanism of tetra-Cys CRABP Htt53 aggregation in vitro. Aliquots were removed at different time points during the aggregation time course of tetra-Cys CRABP Htt53 at physiological temperature, 37 °C (Fig. 4A), and their abilities, on a constant mass basis, to favor either polyQ (Q29) (Fig. 4B) or P39A tetra-Cys CRABP (Fig. 4C) aggregation were tested. Interestingly, aggregates of tetra-Cys CRABP Htt53 formed in vitro show less distinct time-dependent changes in seeding ability: the seeding of polyQ-mediated aggregation by tetra-Cys CRABP Htt53 aggregates was more efficient in the later time point samples (from the mid-logarithmic to the plateau phase, Fig. 4A), although aggregates isolated from the lag phase (30 min) and early logarithmic phase (100 min) showed a higher ability to seed polyQ aggregate elongation than did the comparable aggregates isolated from E. coli cells (Fig. 2A). Additionally, the same early time points of in vitro grown tetra-Cys CRABP Htt53 aggregates slightly accelerated the aggregation of P39A tetra-Cys CRABP, but none of the tested species was able to completely by-pass the lag phase (Fig. 4C). Similar to the later aggregates isolated from cells, the in vitro grown aggregates from the later time points (150 and 200 min) were completely unable to accelerate P39A tetra-Cys CRABP aggregation.
A converging body of evidence challenges the direct role of insoluble fibrillar aggregates of amyloidogenic proteins in cytotoxicity, implicating instead either the aggregation process itself and/or one or more types of aggregation intermediates (12, 13). The fibrillization pathway has been studied in the greatest detail for the two prominent amyloidogenic proteins, Aβ and α-synuclein. These studies have revealed multiple transient precursor structures prior to the appearance of mature fibrils (9, 13, 42). Much remains to be learned about the roles of these intermediates both in amyloid assembly (43) and in pathology (44).
Polyglutamine fibrillization has not been as extensively examined as some other aggregation systems and presents an even more complex picture than initially thought. The aggregation mechanism of simple polyQ peptides in vitro appears not to involve oligomeric or protofibrillar intermediates but rather follows a nucleated growth mechanism with monomer addition, which features only monomers and mature amyloid fibrils (32, 45). In contrast, studies on full-length Htt exon 1 fragments have identified non-amyloid aggregates (5–7) in addition to mature amyloid. Moreover, studies of natural multidomain polyQ-containing proteins (22, 23), as well as an artificial chimera like the one studied here (17), also reveal multistage aggregation with early prefibrillar species. In particular, the extensive recent studies of ataxin-3 aggregation in vitro have established that this multidomain protein follows a two-stage aggregation process with the first step dominated by the Josephin domain, and not the polyQ-rich domain (18, 22, 23). Our results for the CRABP-Htt exon 1 chimera are entirely consistent, demonstrating that the aggregation of polyQ-containing proteins is also highly complex when it occurs in the cellular environment (here in E. coli). In analogy to in vitro results, the polypeptide segments involved in the core aggregate structure can actually change over the course of the aggregation reaction, as the morphology of the aggregates changes from prefibrillar to fibrillar. Together with our previous results showing structural perturbation and destabilization of an adjacent domain by a polyQ-rich domain (24), the present study raises the possibility suggested in the induced misfit model of Pastore and coworkers (19) that the polyQ segments play an indirect role in the formation of the early aggregates by virtue of perturbing the adjacent globular domain (24) and only later becoming involved in the core structure of the mature, amyloid-like aggregates (see schematic model in Fig. 5). The structural perturbations presumably arise from as yet poorly understood conformational propensities of an extended polyQ tract (46–53) and in turn lead to a dramatic increase in the likelihood of spontaneous association of the tetra-Cys CRABP domains forming the core structure of metastable small spherical prefibrillar assemblies (Fig. 5). In these early oligomers, at least some polyQ segments are excluded from the amyloid core, whereas in mature aggregates they appear to form the aggregate core. Although the mechanism of this transformation is unknown, it is possible that early aggregation of the CRABP domain acts to orient and increase the local concentration of the polyQ elements, which might facilitate the nucleation of a polyQ-based amyloid structure. It would be of interest in the future to compare kinetics of aggregation of polyQ-containing constructs in vivo with and without the CRABP domain.
We observed some modest differences in the aggregation pathway of the CRABP Htt exon 1 chimera in vitro versus in vivo. In the in vitro aggregation pathway, even the earlier aggregates were able to recruit Q29 peptide, albeit to a lesser extent, consistent with aggregate core structures involving polyQ. The ability of aggregates formed in vitro to accelerate aggregation of the CRABP domain is negligible throughout the time course of aggregation, suggesting that the aggregates formed involve amyloid-like polyQ cores from the beginning of the process. This more direct progress toward polyQ amyloid core in vitro may be due either to a higher concentration of peptides present at all times in the in vitro reaction (as opposed to the progressive increase in concentration as a consequence of biosynthesis in vivo), or to the abilities of cellular components such as chaperones to suppress initial nucleation-dependent polyQ aggregation and thus allow the more indirect approach via CRABP-core aggregation intermediates. We are currently pursuing experiments to distinguish these possibilities and thus better understand the impact of the cellular environment.
The results we describe here provide evidence that aggregation pathways of multidomain polyQ proteins may in general be complex, which might shed light on the behavior of disease-relevant amyloidogenic polyQ proteins. Like the previous picture from in vitro studies, largely in the ataxin-3 system, aggregation in the cell emerges as a multistep process with distinct types of aggregates formed en route to the mature fibrils: 1) a spherical aggregate that can accelerate CRABP aggregation, suggesting a structure with a CRABP core and 2) fibrillar aggregate with polyQ seeding activity suggesting an amyloid-like structure with a polyQ core. The evidence for formation of various aggregated species, some of which may exert a cytotoxic effect, and the improved understanding of their amyloidogenic properties open new avenues for designing diagnostic approaches to ameliorate misfolding and aggregation diseases.
We thank Carola Langer for providing an aliquot of pure aggregates of Htt53 and purified soluble Htt30 used as standards in the slot-blot analysis. We thank Charles Glabe for giving us a slot-blot membrane with positive and negative controls to test the A-11 antibody. We appreciate Bharati Vasudeva Rao's help with the electron microscopy, we are grateful to Joanna Swain for critical reading of the manuscript, and we thank Manajit Hayer-Hartl and Ulrich Hartl for their generous support.
This work was supported in part by National Institutes of Health (NIH) Grant GM027616 and an NIH Director's Pioneer Award (to L. M. G.), by NIH Grant AG019322 (to R. W.), by the Deutsche Forschungsgemeinschaft (Grant IG 73/4-1 to Z. I.), and by a Heisenberg fellowship (IG 73/1-1 to Z. I.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
3The abbreviations used are: polyQ, polyglutamine; FlAsH, fluorescein arsenical helix binder; EDT, ethanedithiol; Htt, huntingtin; CRABP, cellular retinoic acid-binding protein.