The DNAJB6 and DNAJB8 Protein Chaperones Prevent Intracellular Aggregation of Polyglutamine Peptides*

Background: Intracellular aggregation of polyglutamine (polyQ) proteins can be prevented by the chaperones DNAJB6 and DNAJB8. Results: DNAJB6 and DNAJB8 prevent the aggregation of pure polyQ peptides. Conclusion: The polyQ tract is sufficient for DNAJB6 and DNAJB8 to prevent aggregation. Significance: By interacting with polyQ fragments, DNAJB6 and DNAJB8 reduce polyQ protein aggregation and may be potential therapeutic targets in polyQ disorders. Fragments of proteins containing an expanded polyglutamine (polyQ) tract are thought to initiate aggregation and toxicity in at least nine neurodegenerative diseases, including Huntington's disease. Because proteasomes appear unable to digest the polyQ tract, which can initiate intracellular protein aggregation, preventing polyQ peptide aggregation by chaperones should greatly improve polyQ clearance and prevent aggregate formation. Here we expressed polyQ peptides in cells and show that their intracellular aggregation is prevented by DNAJB6 and DNAJB8, members of the DNAJ (Hsp40) chaperone family. In contrast, HSPA/Hsp70 and DNAJB1, also members of the DNAJ chaperone family, did not prevent peptide-initiated aggregation. Intriguingly, DNAJB6 and DNAJB8 also affected the soluble levels of polyQ peptides, indicating that DNAJB6 and DNAJB8 inhibit polyQ peptide aggregation directly. Together with recent data showing that purified DNAJB6 can suppress fibrillation of polyQ peptides far more efficiently than polyQ expanded protein fragments in vitro, we conclude that the mechanism of DNAJB6 and DNAJB8 is suppression of polyQ protein aggregation by directly binding the polyQ tract.

Fragments of proteins containing an expanded polyglutamine (polyQ) tract are thought to initiate aggregation and toxicity in at least nine neurodegenerative diseases, including Huntington's disease. Because proteasomes appear unable to digest the polyQ tract, which can initiate intracellular protein aggregation, preventing polyQ peptide aggregation by chaperones should greatly improve polyQ clearance and prevent aggregate formation. Here we expressed polyQ peptides in cells and show that their intracellular aggregation is prevented by DNAJB6 and DNAJB8, members of the DNAJ (Hsp40) chaperone family. In contrast, HSPA/Hsp70 and DNAJB1, also members of the DNAJ chaperone family, did not prevent peptide-initiated aggregation. Intriguingly, DNAJB6 and DNAJB8 also affected the soluble levels of polyQ peptides, indicating that DNAJB6 and DNAJB8 inhibit polyQ peptide aggregation directly. Together with recent data showing that purified DNAJB6 can suppress fibrillation of polyQ peptides far more efficiently than polyQ expanded protein fragments in vitro, we conclude that the mechanism of DNAJB6 and DNAJB8 is suppression of polyQ protein aggregation by directly binding the polyQ tract.
Polyglutamine (polyQ) 3 disorders are a group of dominantly inherited, progressive neurodegenerative disorders. These disorders are caused by expansion of the polyQ tract within coding regions of unrelated proteins. At least nine different polyQ dis-orders are known, including Huntington´s disease (1). These disorders are characterized by atrophy of certain regions in the brain and the presence of intracellular aggregates that are thought to initiate disease, although neither the precise toxic species nor the cause of aggregation initiation are fully understood. Several studies showed that these aggregates contained not only the full-length disease-related protein but also shortened fragments, including the expanded polyQ tract (2)(3)(4). Furthermore, it was shown that truncated forms of huntingtin (Htt), ataxin 3, ataxin 7, atrophin 1, and the androgen receptor containing the expanded polyQ tract, are more aggregationprone and enhance toxicity (5)(6)(7)(8)(9)(10)(11). Although proteasomes are able to degrade polyQ-containing proteins (12)(13)(14)(15)), they may not be able to cleave within polyQ tracts in vitro (16). As a consequence, polyQ peptides may be released upon proteasomal degradation of expanded polyQ proteins. Similar polyQ peptides may be directly generated in related disorders, such as SCA8 and Huntington's disease like-2 (HDL2), which are thought to be initiated by polyQ peptides generated by antisense transcription that cause intranuclear inclusions (9,17). When mimicking polyQ peptide generation in living cells, we observed that expression of expanded polyQ peptides is sufficient to induce aggregation and toxicity in cells (18) and that expression of polyQ stretches alone causes toxicity in Drosophila (19).
Aggregates or inclusion bodies initiated by expanded polyQ peptides as well as expanded polyQ-containing proteins are decorated with various proteins, such as components of the ubiquitin-proteasome system and heat shock proteins (HSP) (20 -26). HSPs can function as molecular chaperones. Many of them are up-regulated under stress conditions, such as heat stress (27,28), and have been shown to protect cells against heat-induced protein aggregation (29,30). Heat shock proteins can be classified into a number of families on the basis of their approximate molecular mass and presence of conserved domains (31,32). The HSPA (Hsp70) family consists of 11 members, whereas the DNAJ (Hsp40) family consists of over 40 members in humans (32). In a recent comparative screen of all members of the HSPA, HSPH (Hsp110), and DNAJ (sub)families, DNAJB6 and DNAJB8 were identified as the two most potent suppressors of aggregation and related toxicity of expanded polyQ proteins (33). When examining the characteristics of these two chaperones in more detail, we now find that DNAJB6 and DNAJB8 can directly suppress aggregation of polyQ peptides generated in living cells. This action is largely dependent on the serine-rich region (SSF-SST) within the C terminus and less dependent on interaction with Hsp70. Intriguingly, expression of DNAJB6 in the nucleus also represses aggregation in the cytoplasm, suggesting that polyQ peptides are kept solubilized in one cellular compartment by these chaperones and remain small enough to travel to the other compartment for subsequent degradation. The ability of DNAJB6 and DNAJB8 to reduce aggregation of expanded polyQ peptides prevents these polyQ peptides from acting as inducers of aggregation and improves their clearance.

EXPERIMENTAL PROCEDURES
DNA Constructs-Generation of GFP-Ub-polyQ constructs was described before (18). However, the initial polyQ peptides started with a Leu residue and a Glu-Thr-Ser-Pro-Arg sequence at the C terminus. This Leu residue was changed into a Gln using QuikChange II site-directed mutagenesis (Stratagene). At the C terminus, a stop codon was introduced directly after the polyQ stretch using site-directed mutagenesis, resulting in GFP-Ub-Q16 and GFP-Ub-Q104, respectively. To insert a tetracysteine (C4) tag after the polyQ tract, a BamHI site was introduced at the N terminus of polyQ peptides using GFP-Ub-Q16/112, where Leu was already changed into a Gln. The C4 tag containing the sequence FLNCCPGCCMEP (34) was obtained by annealing two encoding oligo primers with overhangs compatible with BamHI at the N terminus and XbaI at the C terminus. A C4 tag was inserted into GFP-Ub-polyQ using BamHI and XbaI, thereby generating GFP-Ub-Q17-C4 and GFP-Ub-Q99-C4, respectively.
Confocal and FRET-FLIM Microscopy-HEK293T cells were cotransfected with GFP-Ub-polyQ and various chaperones, as indicated in the figure legends, and the percentage of aggregatepositive cells was determined after 72 h. The number of cells containing aggregates in the nucleus, cytoplasm, or both compartments was determined 48 and 72 h after transfection. Confocal microscopy images were obtained 48 h after transfection using a Leica SP2 confocal system equipped with an Ar/Kr laser with a ϫ63 objective. For fluorescence loss in photobleaching analysis, either a part of the cytoplasm or the whole cell, except for the aggregate, was repeatedly photobleached at full laser power. Fluorescence loss was measured either within the nucleus or within the non-bleached region containing the aggregate, respectively. For fluorescent recovery after photobleaching analysis, fluorescently tagged chaperones trapped in polyQ peptide-induced aggregates were bleached at full laser power, and fluorescence recovery was measured in time. FLIM was carried out after biarsenical labeling of living cells. To reduce interference of autofluorescence by DMEM, a special medium was added to cells (20 mM Hepes (pH 7.4), 137 mM NaCl, 5.4 mM KCl, 1.8 mM CaCl 2 , 0.8 mM MgCl 2 , and 20 mM glucose). FLIM was performed using the wide-field frequency domain approach on a home-built instrument (35) using a radio-frequency-modulated image intensifier (Lambert Instruments, model no. II18MD) coupled to a CCD camera (Photometrics HQ) as a detector. A ϫ 40 objective was used for all measurements. The modulation frequency was set to 75.1 MHz. Eighteen phase images with an exposure time of 20 -200 ms were acquired in a random recording order to minimize artifacts because of photobleaching (36). From the phase sequence, an intensity image and the phase and modulation lifetime image were calculated using Matlab macros. From this data, the lifetime of individual cells was determined. Subsequently, average phase and modulation lifetimes (Ϯ S.D.) were calculated. The FRET efficiency (E) was calculated according to the following: E ϭ (1 Ϫ ( DA / D )) ϫ 100%, in which DA is the fluorescence lifetime of the donor in the presence of the acceptor (i.e. samples labeled with both FlAsH or GFP and ReAsH) and D is the fluorescence lifetime of the donor (i.e. FlAsH only or GFP only) in the absence of the acceptor. Frequency domain FLIM yields a phase lifetime ( ) and a modulation lifetime ( M ). Because is more sensitive than M , FRET efficiency was calculated on the basis of .

RESULTS
DNAJB6b and DNAJB8 Reduce Aggregation of PolyQ Peptides-To determine the effect of chaperones on polyQ peptide aggregation in living cells, we used GFP-ubiquitin-polyQ (GFP-Ub-polyQ) constructs that are immediately cleaved upon expression into GFP-Ub and polyQ peptides by Ub C-terminal hydrolases (Fig. 1A) (39). The resulting polyQ peptides lack any flanking amino acids, such as a starting methionine, and only expanded polyQ peptides beyond 40 glutamines induced aggregation, as quantified by the sequestration of GFP-Ub into aggregates (18). Although ϳ60% of cells expressing Glu-104 peptides contained aggregates after 72 h, coexpression of HSPA1A (Hsp70) had no significant effect on polyQ peptide aggregation, whereas DNAJB1 (Hsp40) slightly reduced aggregation (Fig. 1B). In contrast, coexpression of DNAJB6b or DNAJB8 resulted in a dramatic decrease in polyQ peptide aggregation when scoring the number of polyQ-expressing cells having aggregates (Fig. 1B) or by filter trap analysis (C) when similar levels of the different chaperones were expressed (D). In contrast, when endogenous DNAJB6 levels were decreased by siRNA (Fig. 1E,  lower panel), aggregation of polyQ peptides increased, as indicated by filter trap analysis (upper panel).
To examine whether expression of DNAJB6b and DNAJB8 also reduced the amount of SDS-insoluble polyQ aggregates, cells expressing GFP-Ub-polyQ in combination with DNAJB6b or DNAJB8 were separated into SDS-soluble and SDS-insoluble fractions. Expression of Glu-104 peptides for 24 h resulted in a specific polyQ-positive band in the soluble and various bands in the insoluble fractions ( Fig. 2A, arrows). Coexpression of DNAJB6b and DNAJB8 remarkably reduced the amount of polyQ peptides present in the SDS-insoluble fraction ( Fig. 2A, right panel), which is in agreement with the reduced number of aggregates visualized in cells. Interestingly, the soluble fraction of Glu-104 peptides increased, ( Fig. 2A, left panel), suggesting that DNAJB6b and DNAJB8 keep polyQ peptides soluble. To examine whether the SDS-soluble fraction was a precursor of one of the insoluble polyQ peptide fractions, we separated fractions at different time points after transfection. Both the soluble and insoluble fraction decreased when cells were treated with cycloheximide to block synthesis of new polyQ peptides during the last 16 h, suggesting that the soluble fraction was not a precursor to a particular insoluble fraction (data not shown). SDS-insoluble levels of Htt-exon1-Q103 were also decreased by coexpression of DNAJB6b or DNAJB8 (Fig. 2B,  right panel). In contrast to polyQ peptides, the SDS-soluble Htt-exon1-Q103 protein levels were not affected (Fig. 2B,  left panel). The observed decrease in insoluble Htt-exon-1 could, however, indicate that DNAJB6b and DNAJB8 can improve solubility and clearance of Htt-exon-1, as shown before (33,40).
DNAJ6b and DNAJB8 Are Recruited into PolyQ Peptide Aggregates-To examine the effect of DNAJB6b and DNAJB8 with polyQ peptides in more detail, polyQ peptides were tagged with the small tetracysteine (C4) motif FLNCCPGCCMEP at the C terminus for direct visualization. The membrane-permeable biarsenical dye ReAsH can bind this small tag, and only then does it becomes fluorescent (34). To examine whether the C4 tag affected the behavior of polyQ peptides, we transiently expressed GFP-Ub-polyQ and GFP-Ub-polyQ-C4 peptides. Expression of either C4-tagged or non-tagged polyQ peptides resulted in a similar percentage of fluorescent cells that contained aggregates after 72 h (Fig. 3A). Similar to untagged Q104 peptides, DNAJB6b and DNAJB8 kept Glu-99-C4 SDS-soluble (Fig. 3B), and similar SDS-insoluble fractions were detected between cells expressing untagged Glu-104 or C4-tagged Glu-99 peptides (C). In cells without aggregates, both GFP-Ub and expanded polyQ peptides were present throughout the cytoplasm and nucleus (Fig. 3D, upper panel). However, aggregation led to the redistribution of both GFP-Ub and ReAsHlabeled Glu-99-C4, with ReAsH-labeled Glu-99-C4 sequestered in the core of aggregates and GFP-Ub in a ring around the core (Fig. 3D, lower panel, also in the inset). Only C4-tagged polyQ peptides of disease-related lengths accumulated and initiated aggregation, whereas short C4-tagged polyQ peptides were degraded similarly to untagged polyQ peptides.
PolyQ Peptides Sequestered in Aggregates Strongly Interact with DNAJB6b and DNAJB8-To examine whether diffuse and aggregated polyQ peptides showed intermolecular interactions, we performed FRET experiments. FRET enables sensitive evaluation of protein-protein interactions in living cells. Only when proteins are in close proximity (Ͻ 10 nm) can FRET occur and be measured (41). FLIM was used to measure FRET-FLIM, which is detected as a decrease in donor fluorescence lifetime (42). HeLa cells expressing Ub-Q99-C4 were labeled simultaneously with equal amounts of FlAsH and ReAsH at 48 h after transfection, leading to individual labeling of polyQ peptides with either FlAsH or ReAsH. When polyQ peptides were not clustered in aggregates, no reduction in FlAsH lifetime was measured (Fig. 4A), suggesting that polyQ-polyQ peptide interactions hardly occurred when not sequestered in aggregates. In contrast, aggregated polyQ peptides showed a remarkable decrease in lifetime of 39.6% of FlAsH, which is indicative of FRET (Fig. 4A). Together, these data show that polyQ peptides are mainly monomeric when not present in aggregates, whereas aggregated polyQ peptides are immobile, and only then strong polyQ-polyQ peptide interactions are observed.
The presence of GFP-Ub in aggregates most likely represents sequestered proteins that become ubiquitinated in time, whereas the polyQ peptides are not likely ubiquitinated because no lysine residues are present. Because expression of GFP-Ub-Q99-C4 results in an efficient separation of GFP-Ub and Glu-99-C4, cells expressing either GFP-Ub-Q99-C4 or GFP-Ub combined with Ub-Q99-C4 should hardly show interactions between polyQ peptides and GFP-Ub (Fig. 4B). The recruitment of proteasomes into the core of polyQ peptide-initiated aggregates suggests that proteasomes interact with the polyQ peptides. To substantiate this idea, cells were cotransfected with Ub-Q99-C4 and the GFP-tagged proteasomal component ␤7. Coexpression of ReAsH-labeled Glu-99-C4 with the GFPtagged proteasomal component ␤7 caused a decreased lifetime of GFP (Fig. 4B), indicating a direct interaction with the polyQ peptides. In the few cells that showed aggregates upon DNAJB6b or DNAJB8 coexpression, these chaperones were present in the core of aggregates (Fig. 4C). Therefore, we examined whether these chaperones also interact with aggregated polyQ peptides. Both GFP-tagged DNAJB6b and DNAJB8 interacted strongly with polyQ peptides, as shown by the reduction in lifetime (Fig. 4D). In contrast, the recruitment of Hsp70 into polyQ peptide-induced aggregates did not result in a significantly reduced lifetime of GFP-tagged Hsp70 (Fig. 4D), suggesting that Hsp70 does not interact with polyQ peptides. Thus, most proteins that sequester in the core of aggregates, such as ␤7, DNAJB6b, and DNAJB8, interacted with aggregated polyQ peptides. This suggests that in the rare event of aggregation when DNAJB6b and DNAJB8 are coexpressed, they are trapped into aggregates induced by polyQ peptides in an early stage, which may reflect a failed function of these chaperones.
The Serine-rich Region in DNAJB6 and DNAJB8 Is Essential for Reduction of Aggregation-The DNAJ family is defined by the presence of a J domain that can regulate chaperone activity of the HSPA family by stimulating ATP hydrolysis (Fig. 5A) (43,44). The HPD sequence in the J domain is essential for interaction with and accelerating the ATPase activity of HSPA (45). The other domains of DNAJ family members are involved in recognition and binding of clients and are considered to be a main factor in driving the specificity of Hsp70 machines (46). Mutating His to Glu (H31Q) in the HPD motif results in an inactive J domain (45), but this only partly impaired the ability of DNAJB6 and DNAJB8 to reduce polyQ peptide aggregation as scored by the amount of fluorescent cells having aggregates  's t test). ***, p Ͻ 0.001; ns, not significant. Scale bar ϭ 10 m. (Fig. 5B). The H31Q mutants also still reduced the appearance of polyQ peptides on a filter trap (Fig. 5C) as well as in the SDS-insoluble fraction (D). However, the SDS-soluble fraction was also decreased (Fig. 5E), indicating that the J domain is somewhat affecting DNAJB6 and DNAJB8 function, albeit that their main antiaggregation function is intact, consistent with was found before for preventing the aggregation of htt-exon-1 fragments (33).
In the C-terminal region of DNAJB6b and DNAJB8, a conserved serine-rich region (SSF-TST and SSF-SST, respectively) is present, which is absent in DNAJB1 or its related subfamily members (Fig. 5A) (47). Deletion of this serine-rich region was found to be crucial for DNAJB6 and DNAJB8 to interact with histone deacetylases that regulate their function as suppressors of expanded polyQ protein aggregation (33). Deletion of this serine-rich region also severely affected the ability of DNAJB6 and DNAJB8 to reduce polyQ peptide aggregation (Fig. 5B), resulting in more SDS-insoluble material on a filter trap (Fig.  5C) and Western blot analysis (D) and a reduction in SDSsoluble levels (E). The defective chaperones were still recruited into polyQ peptide aggregates, but these sequestered chaperones lost their direct interaction with the labeled polyQ peptides once the serine-rich region was disabled (Fig. 5F). This suggests that the sequestration of both functional and disabled DNAJB6 and DNAJB8 may not be their primary way of interfering with aggregation but, rather, reflect an inefficient attempt to prevent aggregate initiation and growth.
Because DNAJB6 and DNAJB8 strongly reduce aggregation of polyQ peptides and interact with polyQ peptides in the core of aggregates, these interactions may affect aggregation of polyQ peptides directly. To examine whether chaperones were able to reduce the strong interactions observed between polyQ peptides, we coexpressed Glu-99-C4 peptides in combination with DNAJB6, DNAJB8, Hsp40, or Hsp70. Double-labeling of C4-tagged Glu-99 peptides with FlAsH and ReAsH resulted in a high FRET-FLIM efficiency (Fig. 5G). Coexpression of DNAJB6b or DNAJB8 hardly affected interactions between aggregated polyQ peptides, indicating that the sequestration of DNAJB 6 or DNAJB8 does not interfere with peptide interactions within the aggregate. Similarly, Hsp40 and Hsp70 did also not affect the interactions between aggregated polyQ peptides.
Nuclear DNAJB6a Reduces Aggregation in the Cytoplasm and Nucleus-The effect of DNAJB6b on SDS-soluble levels of polyQ peptides suggests that it may counteract very early olig-omerization steps of these aggregation-prone peptides. To examine this phenomenon in more detail, we reasoned that only monomeric or small oligomers can shuttle freely through the nuclear pore by diffusion. If correct, the presence of DNAJB6 in the nucleus or cytoplasm only should then reduce aggregation in both compartments by keeping the polyQ peptides soluble. DNAJB6 has two isoforms. DNAJB6b is present in both the cytoplasm and nucleus, whereas DNAJB6a contains a putative nuclear localization signal and is, therefore, localized in the nucleus only (Fig. 6A) (33). Despite the exclusive localization of DNAJB6a in the nucleus, the expression of DNAJB6a was as effective as DNAJB6b and DNAJB8 in reducing polyQpeptide aggregation (Fig. 1B). Interestingly, the reduction in aggregate formation was not limited to the compartment were these chaperones were residing. Coexpression of DNAJB8 did not alter the ratio of nuclear and cytoplasmic aggregates after 48 h, whereas coexpression of DNAJB6b resulted in a slight change in this ratio in favor of nuclear aggregates (Fig. 6B, upper  panel). Expression of DNAJB6b and DNAJB8 resulted in an even higher percentage of aggregates present in nuclei after 72 h (Fig. 6B, lower panel). The nuclear localization of DNAJB6a did not lead to the exclusive presence of aggregates in the cytoplasm because the ratio of nuclear and cytoplasmic aggregates remained unaltered after expression of DNAJB6a (Fig. 6B). It is concluded that although DNAJB6a is present in the nucleus only, it does not exclusively reduce aggregation in the nuclear compartment.
Together, these data suggest that DNAJB6 and DNAJB8 are efficient in targeting earlier stages of aggregation, whereas the observed recruitment of DNAJB6 and DNAJB8 into aggregates reflects sequestration of these chaperones without any effect on improved clearance. To examine whether DNAJB6a, DNAJB6b, and DNAJB8 were able to dissociate from the core of aggregates, we studied the mobility of the sequestered chaperones using fluorescent recovery after photobleaching and fluorescence loss in photobleaching analysis (48,49). Fluorescent recovery was observed of HSPA1A but not of DNAJB6a, DNAJB6b, and DNAJB8 (Fig. 6C). When the entire cell was bleached repeatedly, with the exception of one fluorescent aggregate, fluorescence loss in the aggregate was measured of HSPA1A, whereas DNAJB6a, DNAJB6b, and DNAJB8 present in the core of aggregates were hardly affected by bleaching (Fig.  6D). These data show that DNAJB6a, DNAJB6b, and DNAJB8 were sequestered irreversibly in the core of polyQ-induced aggregates, indicating that once the activity of these chaperones has failed to prevent aggregation, they are irreversibly trapped in the inert core of aggregates.

DISCUSSION
Intracellular polyQ clearance may be facilitated by HSPs that prevent the formation of insoluble aggregation, allowing clearance pathways involving proteases and autophagic pathways to remove hazardous fragments. In this study, we examined mechanisms to prevent aggregation of polyQ fragments, focusing on expanded polyQ peptides that can initiate aggregation and toxicity. We found that the DNAJ family members DNAJB6 and DNAJB8 were efficient suppressors of polyQ peptide aggregation. Interestingly, detergent-soluble levels of mutant Htt-exon-1 were not affected by DNAJB6, whereas the level of soluble polyQ peptides was increased. Apparently, these chaperones prevent aggregation of polyQ peptides, leading to improved clearance of polyQ peptides, whereas soluble Httexon-1 is not affected. The solubilized polyQ peptides might be more efficiently cleared than Htt-exon-1 fragments. Alternatively, aggregation of Htt-exon-1 might be initiated by smaller fragments derived from Htt-exon-1 that are subsequently tar-FIGURE 6. DNAJB6 and DNAJB8 prevent aggregation of Glu-104 peptides by keeping them soluble. A, confocal microscopy images of HeLa cells transfected with DNAJB6a, DNAJB6b, or DNAJB8, respectively, stained with Hoechst, and fixed 72 h after transfection. Scale bar ϭ 10 m. B, the ratio of aggregates present in the cytoplasm, nucleus, or both compartments 72 h after transfection with GFP-Ub-Q104 in combination with DNAJB6a, DNAJB6b, or DNAJB8, respectively, in Mel JuSo cells. Data are mean Ϯ S.E. of three independent experiments. Chaperones did not specifically reduce aggregation in the compartment where they reside. C, fluorescent recovery after photobleaching analysis shows that DNAJB6a, DNAJB6b, and DNAJB8 did not recover after photobleaching, whereas HSPA1A fluorescent levels recovered. D, fluorescence loss of aggregates was measured using fluorescence loss in photobleaching analysis by repetitive photobleaching of another part of the cytoplasm. Fluorescence loss of HSPA1A was observed, whereas fluorescent levels of DNAJB6a, DNAJB6b, and DNAJB8 were not affected. geted by DNAJB6. In contrast to polyQ peptides, fluorescently tagged, expanded polyQ proteins, such as Htt-exon1-GFP, were not present in the core of aggregates but sequestered in a ring around the aggregate (18, 25, 50 -52). This suggests that smaller fragments containing the polyQ tract initiate aggregation and that the original GFP-tagged polyQ proteins are sequestered in a later stage. Still, it remains to be resolved whether polyQ peptides are generated in cells that express polyQ-expanded proteins, similar to in vitro conditions (16). This will, however, be a challenging task because the generated polyQ peptides will either be captured by DNAJB6 or DNAJB8 and/or degraded or initiate aggregation and subsequent sequestration of the original polyQ protein. To address the first option, we purified DNAJB6 from polyQ fragment-expressing cells to detect putatively associated polyQ peptides. Although we did find several peptide fragments associated with them, suggesting they indeed could be "peptide chaperones," polyQ peptides could unfortunately not be detected because of the limitations of polyQ analysis by mass spectrometry (data not shown).
DNAJB6 and DNAJB8 are the first examples of HSPs that reduce aggregation of pure polyQ peptides. Our data are supported by recent observations by the group of Cecilia Emanuelsson, 4 who showed that purified DNAJB6 and DNAJB8 can directly bind to and highly efficiently suppress polyQ peptide fibrillation initiated by soluble, pure polyQ peptides. In fact, DNAJB6 and DNAJB8 were better at suppressing fibrillation when initiated by polyQ peptides than when initiated by Httexon-1 fragments. 4 Given that in cells the two DNAJB proteins can prevent aggregation of Htt-exon-1 fragments (33) (this work), this implies that although some HSPs (HSPA1 or DNAJB1) might prevent sequestration of polyQ proteins into aggregates, DNAJB6 and DNAJB8 reduce aggregation of expanded polyQ peptides. Thereby they prevent that polyQ peptides can act as nucleators of aggregation and subsequent sequestering of larger polyQ-containing fragments, such as Htt-exon-1. Consistent with this model, DNAJB6 and DNAJB8 were trapped in the core of aggregates together with polyQ peptides, whereas expanded polyQ proteins and HSPA1A or DNAJB1 were absent from the core. The recruitment of HSPA1A to the outer ring of aggregates may, in fact, be due to its recognition of sequestered, possibly unfolded proteins and its attempt to solubilize these proteins. Indeed, HSPA1A, unlike DNAJB6 and DNAJB8, was not trapped in these aggregates (reflecting a failed function to prevent aggregate seeding) but showed a high degree of dynamics at the aggregate ring (Fig. 6C) (24). Together, this indicates that only DNAJB6 and DNAJB8 may be able to directly prevent aggregation of polyQ proteins by directly interacting with the expanded polyQ repeat, whereas most other chaperones are mostly preventing the sequestration of other proteins in the aggregate. As a result, improving the activity or protein levels of these DNAJB chaperones might delay or even prevent the aggregation of polyQ fragments, enabling intracellular clearance pathways to target and degrade these fragments. DNAJB6 and DNAJB8 may keep polyQ fragments soluble, whereby they may be maintained in a degradation-competent state. It remains to be established whether these chaperones can also control their degradation by targeting them to particular proteases, such as the proteasome or the puromycin-sensitive peptidase (PSA), or toward autophagosomes. Intriguingly, the ability of nuclear DNAJB6a to reduce polyQ peptide aggregation was not restricted to the nuclear compartment because aggregation within the cytoplasm was reduced similarly. This suggests that DNAJB6a keeps polyQ peptides in a soluble intermediate, which allows these peptides to freely translocate between the cytoplasm and nucleus. This is of importance because the autophagic machinery is present in the cytoplasmic compartment only (53,54). This may also explain our observation that the ratio of cytoplasmic versus nuclear aggregates decreased in time because cytoplasmic aggregates may be more efficiently cleared as compared with aggregates in the nucleus.
Deletion of the serine-rich region within DNAJB6 and DNAJB8 abolished the antiaggregation properties of these chaperones. This suggests that this region is essential to recognize polyQ peptides and to inhibit their aggregation. An active J domain was less important for DNAJB6b and DNAJB8 antiaggregation properties, although it affected the soluble levels of polyQ peptides. Because the J domain is known to interact with HSPA1A and stimulates its ATPase activity (43,44), DNAJB6 and DNAJB8 do not seem to require HSPA1A to reduce polyQ peptide aggregation. Little is known about the function of the serine-rich region in DNAJB6 and DNAJB8, although this region is suggested to be involved in DNAJB6/8 oligomerization and histone deacetylase binding (33). Also, purified DNAJB6 and DNAJB8 form large polydispersed oligomers, 4 which might be crucial for efficient polyQ peptide binding. Previous studies showed that DNAJB6 is highly enriched in the central nervous system (40,47) and colocalizes to Lewy bodies and aggregates induced by Htt-exon-1 (40,55). However, expression of DNAJB8 is restricted to the testes (47). Therefore, DNAJB6 may be the best candidate for potential therapeutic approaches in the fight against polyQ disorders, for example by inducing endogenous DNAJB6 expression. A combination of solubilizing polyQ peptides by expression of DNAJB6 and stimulating their subsequent degradation seems to be the most attractive alternative.