Intercellular chaperone transmission via exosomes contributes to maintenance of protein homeostasis at the organismal level

Elevated Expression of HSPs in Cells Restores the Protein-Folding Environment in Other Cells.

To examine whether cellular proteostasis is affected by the expression levels of chaperones in other cells, we set up an in vitro coculture experiment in which Neuro2A cells with different levels of chaperone expression were incubated separately across cell culture inserts (Fig. 1A). An expanded polyglutamine stretch of 81 repeats fused with EGFP (Q81-EGFP), a model aggregation-prone protein (15), readily forms insoluble aggregates as inclusion bodies in these cells, which can be easily visualized under the fluorescence microscope, and overexpression of HSPs has been reported to suppress such polyglutamine inclusions (16). We found that coculturing Q81-EGFP–expressing cells with cells overexpressing Hsp40 surprisingly resulted in less inclusion bodies (Fig. 1C), which was calculated as a 26% decrease in the number of inclusion bodies formed relative to the mock-transfected control cells (Fig. 1D). In addition, coincubation with the Hsp40-expressing cells resulted in improvement of cell viability of the Q81-EGFP–expressing cells (21% increase relative to the mock-transfected cells) (Fig. 1E). In contrast, coincubation with cells expressing Hsp40 mutants that are deficient for chaperone function [i.e., deletion mutants of Hsp40 that lack either the C-terminal and Gly/Phe-rich domains (Hsp40-J) or the N-terminal J domain (Hsp40-GC) (Fig. 1B)] showed no suppression of inclusion body formation (Fig. 1 C and D) or improvement of cell viability (Fig. 1E). These results indicate that elevated expression of Hsp40 in the cells leads to a non–cell-autonomous beneficial effect on the other cells, and that the chaperone activity of Hsp40 is responsible for this effect. Likewise, coculturing with Hsp70-expressing cells also decreased inclusion body formation (32% decrease) (Fig. 1 F and G) and improved cell viability (39% increase) (Fig. 1H) in the Q81-EGFP–expressing cells, whereas Hsp70 mutants lacking either ATPase activity [Hsp70(K71E)] (17) or substrate-binding activity (Hsp70dS) (Fig. 1B) had no such effects (Fig. 1 F–H). These results reveal that elevated levels of HSPs in a group of cells can improve the protein-folding environment and suppress polyglutamine-mediated cytotoxicity in another group of cells.

Hsp40 Is Secreted from Cells via a Nonclassical Pathway.

Because the restoration of cellular proteostasis occurred without any physical contact between the two types of cells, we next asked whether extracellular factors released by the chaperone-expressing cells (i.e., activated signaling molecules or the overexpressed chaperones themselves) could be responsible for this non–cell-autonomous effect. We therefore focused our attention on the culture medium of Neuro2A cells in which Hsp40 was transiently transfected and investigated the responsible extracellular factors. To avoid contamination from dead cells, the culture medium was replaced with fresh medium at 24 h after transfection. Subsequent Western blot analysis of the culture medium revealed that Hsp40 itself was detected abundantly and accumulated over time in the culture medium (Fig. 2A), suggesting that cellular Hsp40 is released from cells into the extracellular space. Significant cell death was not detected by the lactate dehydrogenase release assay (Fig. S1A), which excludes the possibility of Hsp40 leakage from the dead cells under this condition. We further confirmed that endogenous Hsp40, as well as Hsc70 and Hsp90, is clearly detected in the culture medium, whereas the endoplasmic reticulum (ER)-resident chaperone Grp78 was not detected (Fig. 2B). Incubation of the cells at a low temperature, however, resulted in significant or almost complete suppression of Hsp40 release (64% and 95% decrease at 25 °C and 4 °C, respectively) compared with incubation at 37 °C (Fig. 2 C and D). These results reveal that Hsp40 is physiologically secreted from cells via a temperature-dependent cellular process, but not via cell death or passive diffusion across cell membranes.

Because Hsp40 is believed to be an intracellular protein, we then asked how Hsp40 gains access to the outside of cells. Most proteins targeted to the outside of cells have a signal sequence at their N terminus, which allows them to be secreted via the classical ER/Golgi pathway (18). However, Hsp40 lacks a distinct signal sequence for classical secretion, as analyzed by the signal peptide prediction program SignalP 4.1 (19). In agreement with this prediction, we found that Hsp40 secretion was insensitive to the treatment of cells with brefeldin A, an inhibitor of the ER/Golgi-dependent pathway (Fig. 2 C and E), whereas secretion of Metridia luciferase (MetLuc), a secretory protein containing an N-terminal signal peptide (18), was completely inhibited under the same condition (Fig. S1B). These results suggest that Hsp40 is secreted via a pathway that is different from the classical ER/Golgi-dependent pathway.

We then tested various chemical compounds that are reported to affect the nonclassical secretion pathways and examined their effects on the secretion levels of Hsp40. Treatment of Neuro2A cells with these compounds revealed that the chemicals that raise the intracellular Ca²⁺ level, such as thapsigargin, monensin, and ionomycin, significantly increased the secretion levels of endogenous Hsp40 (Fig. 2F). In addition, chloroquine and bafilomycin A, both of which perturb endosomal trafficking, affected Hsp40 secretion (Fig. 2F). Likewise, treatment of these chemicals resulted in an increase in secretion levels of other chaperones, including Hsp/Hsc70 and Hsp90 (Fig. S1C). These results strongly indicate that intracellular Ca²⁺ levels and endosomal transport act as key regulators of Hsp40 secretion.

Hsp40 Is Secreted via the Hsp/Hsc70-Dependent Exosome Pathway.

Because both intracellular Ca²⁺ levels and endosomal transport have been suggested to be involved in the generation of exosomes (20), we decided to examine whether an exosome-mediated pathway could be involved in Hsp40 secretion. Following the established centrifugation protocol for the isolation of exosomes (21), we separated the supernatant (S100) and the pellet (P100) fractions from the culture medium of Neuro2A cells (Fig. 3A). We confirmed that the isolated P100 fraction contained small vesicles with uniformly rounded and cup-shaped morphology with a 50- to 100-nm diameter (Fig. 3A), which is in good agreement with the characteristics of exosomes (22). Western blotting of these fractions revealed that Hsp40 is indeed present in the P100 fraction, together with exosome-specific proteins, such as Alix, Flotillin-1, and GAPDH (23), but is not present in the S100 fraction (Fig. 3B). We further confirmed that Hsp40 is also detected in the P100 fractions from a broad range of cell lines, including SH-SY5Y, U373MG, C6, and mouse embryonic fibroblast (MEF) cells (Fig. 3C), and also in the P100 fraction purified from mouse serum (Fig. 3D). The P100 fraction was further analyzed by sucrose-density gradient centrifugation, showing that Hsp40 cofractionates with Hsc70 and Alix at a density of ∼1.15 g/mL (Fig. 3E), which is consistent with the density of exosomes (1.13–1.19 g/mL) among various extracellular membrane vesicles (22). Immunoelectron microscopic analysis of this fraction revealed that Hsp40 is indeed detected inside the membrane vesicles (Fig. 3F). Furthermore, treatment of cells with GW4869, an exosome inhibitor (24), significantly suppressed the secretion of Hsp40, as well as the secretion of Hsc70 and Alix (Fig. 3G). Taken together, we conclude that Hsp40 is secreted via an exosome-mediated pathway.

We next examined the mechanism by which Hsp40 is secreted via the exosome pathway. To identify the domain within Hsp40 that is required for its secretion, we transfected plasmid vectors expressing deletion mutants of Hsp40 (Fig. 1B) and examined their secretion levels in the culture medium of Neuro2A cells. We found that Hsp40-J, which consists of the J domain alone, showed high levels of secretion, whereas Hsp40-GC, which lacks the J domain, was undetectable in the culture medium (Fig. 4A), suggesting an essential role of the J domain for Hsp40 secretion. To confirm this observation further, we designed an artificial chimeric protein, J-MetLucΔSP, in which the N-terminal signal peptide of the model secretory protein MetLuc was replaced with the Hsp40 J domain (Fig. 4B). Whereas the original MetLuc was detected in the S100 fraction (Fig. 4C) and its classical secretion was completely inhibited by treatment with brefeldin A (Fig. 4D, Left), we found that J-MetLucΔSP was detected in the P100 fraction (Fig. 4C) and its secretion was not affected by brefeldin A (Fig. 4D, Right). These results strongly suggest that domain swapping of the signal peptide to the J domain resulted in a pathway switch from classical to nonclassical secretion, confirming that the J domain is responsible for Hsp40 secretion via an exosome-mediated pathway. We also examined whether other Hsp40 family proteins that also have the highly conserved J domain (4) are secreted via an exosome-mediated pathway. Among those Hsp40 family proteins tested, the proteins that are localized in the cytosol, such as DnaJA1, DnaJA2, and DnaJB6a, were detected in the P100 fraction (Fig. 4E), whereas the proteins that are localized in the mitochondria (DnaJA3) or ER (DnaJC1) were not detected in either the P100 or S100 fraction (Fig. 4E). Thus, these results indicate that the cytosolic Hsp40s share the characteristic of exosomal secretion from cells.

Because the J domain acts as an interaction domain with Hsp/Hsc70 (4), we then examined whether Hsp/Hsc70 is involved in the exosome-mediated secretion of Hsp40. As expected, Hsc70 knockdown using RNAi resulted in significant suppression of Hsp40 secretion in the P100 fraction (51% decrease relative to control siRNA) (Fig. 4F), despite the increased cellular levels of Hsp40 that probably resulted from compensatory mechanisms in response to the decreased Hsc70 levels. This finding indicates that the exosomal secretion of Hsp40 is affected by the intracellular levels of Hsc70. In support of this finding, overexpression of Hsp70 resulted in a 122% increase in Hsp40 secretion relative to the mock-transfected cells (Fig. 4G). Thus, these results suggest that exosomal secretion of Hsp40 is regulated by Hsp/Hsc70 levels, probably by its interaction with Hsp/Hsc70 via its conserved J domain.

Exosomal Hsp40 Is Transmitted Intercellularly and Restores the Protein-Folding Environment in Recipient Cells.

It has been reported that exosomes secreted from donor cells are internalized by recipient cells in an endocytosis-dependent manner (23). Accordingly, we either coincubated cells with the P100 fraction containing Hsp40-enhanced yellow fluorescent protein (EYFP) (Fig. 5 A and B) or cocultured them with other cells expressing Hsp40-EYFP (Fig. S2A), and found that Hsp40-EYFP fluorescent signals were clearly detected in the recipient cells, confirming that Hsp40 is also transmitted intercellularly using exosomes. Likewise, exosomal Hsp70 and Hsp90 are internalized by other cells (Fig. S2B). The transmitted HSPs in these experiments showed almost identical punctate fluorescent patterns that were distributed around the nucleus, which are in good agreement with the endocytosis-mediated internalization of exosomes. Because exosomes are known to deliver various biomolecules transcellularly, such as mRNAs, microRNAs, and proteins, which function to exert their bioactivities in recipient cells after their internalization (25, 26), we hypothesized that Hsp40 transmitted intercellularly in an exosome-dependent manner also exerts its function as a molecular chaperone in recipient cells to restore the protein-folding environment.

To test this hypothesis, exosomes purified from the culture medium of donor cells were added to recipient cells expressing polyglutamine protein, and aggregation was analyzed (Fig. 5A). Incubation of the P100 fraction purified from Hsp40-expressing donor cells significantly suppressed inclusion body formation in the Q81-EGFP–expressing recipient cells (Fig. 5C), resulting in a 57% decrease in the number of inclusions relative to the untreated control recipient cells (Fig. 5D). The filter trap assay confirmed that insoluble aggregates in the Q81-EGFP–expressing cells were effectively reduced by treatment with the P100 fraction derived from the Hsp40-expressing cells (Fig. 5E). The P100 fraction purified from the Hsp70-expressing cells also showed similar suppressive effects (Fig. S3 A and B). These results suggest that extracellularly added exosomes that are secreted from the chaperone-expressing donor cells improve the protein-folding environment in the recipient cells. It is noted that the P100 fraction, but not the S100 fraction, suppressed Q81-EGFP aggregation (Fig. S3 C and D) in a dose-dependent manner (Fig. S3E), indicating that the suppressive effects may be due to endogenous molecules within the exosomes. Because the amount of Hsp40 secreted in the P100 fraction was significantly increased by overexpression of Hsp40 in the donor cells (Fig. S3F, Left), which resulted in enhanced suppressive effects in the recipient cells (Fig. 5 C–E), the Hsp40 that is transmitted by exosomes should be one of the factors responsible for these effects. In support of this idea, the P100 fraction with a lower level of Hsp40, which was purified from donor cells transfected with an Hsp40-specific siRNA (Fig. S3F, Right), showed less ability to suppress inclusion body formation (Fig. 5 F and G) and to reduce the insoluble aggregates of Q81-EGFP (Fig. 5H) than the P100 fraction from the control siRNA-transfected cells. Thus, these results indicate that exosome-mediated intercellular transmission of Hsp40 restores the cellular protein-folding environment in a non–cell-autonomous manner.

Exosome-Mediated Intercellular Transmission of HSPs Is Activated Under Stressed Conditions and Compensates for the Protein-Folding Environment in HSR-Deficient Cells.

When challenged by proteotoxic stresses, cells transiently activate the HSR, which is a cell-autonomous mechanism to maintain cellular proteostasis. Therefore, we next asked whether the exosome-mediated non–cell-autonomous mechanism that improves the protein-folding environment in surrounding cells could also be activated under stressed conditions. Cells were heat-shocked at 42 °C, and the contents of the exosomes produced by these cells were examined as above. We found that heat stress indeed resulted in an increase in the levels of HSPs, including Hsp40, Hsp70, and Hsp90, not only in the cytosol but also in the P100 fraction from the Neuro2A cells (Fig. 5I). Addition of the P100 fraction purified from the heat-shocked donor cells resulted in more effective suppression of inclusion body formation in the Q81-EGFP–expressing recipient cells compared with the P100 fraction from the unstressed cells (Fig. 5 J and K). These results indicate that, under stressed conditions, not only the expression of intracellular HSPs but also the exosome-mediated intercellular transmission of HSPs is activated to non-cell autonomously improve the cellular protein-folding environment.

The above findings led us to hypothesize that cells with only limited levels of chaperone induction even under stressed conditions might be protected by the exosome-mediated transmission of HSPs from other chaperone-expressing cells. To test this hypothesis, we used HSF1-null MEF cells, an HSR-deficient cell line that lacks the heat shock-mediated induction of HSPs (27), and repeated the above experiments. In contrast to Neuro2A cells, incubation of HSF1-null MEF cells at 42 °C resulted in no significant increase in the expression levels of HSPs or their secretion levels in the P100 fraction (Fig. S3G). As expected, addition of the P100 fraction purified from the heat-shocked HSF1-null MEF cells resulted in no enhancement of the suppressive effects on Q81-EGFP aggregation in the recipient HSF1-null MEF cells compared with the P100 fraction from the unstressed cells (Fig. 5L and Fig. S3H). These results indicate that HSR-deficient cells lack not only the cell-autonomous HSR but also the exosome-mediated intercellular protective response against proteotoxic stresses, both of which may contribute to their vulnerability to stress. However, the extracellular addition of the P100 fraction purified from the heat-shocked WT MEF cells resulted in efficient suppression of inclusion body formation even in the Q81-EGFP–expressing HSF1-null MEF cells (Fig. 5M and Fig. S3I), suggesting that the lack of the protective responses in HSR-deficient cells could be functionally compensated for by exosomes secreted from other HSR-inducible cells. Thus, these results indicate that the proteostasis in cells that lack the cell-autonomous HSR could be non-cell autonomously maintained by the surrounding chaperone-expressing cells, via the enhanced exosomal release and intercellular transmission of HSPs.

Elevated Expression of HSPs Non-Cell Autonomously Suppresses Polyglutamine-Mediated Neurodegeneration in Remote Tissues in Vivo.

We next asked whether this non–cell-autonomous restoration of proteostasis is also conserved at the multicellular organismal level in vivo. We used Drosophila melanogaster models, in which the conditional expression of transgenes in a tissue-specific manner can be easily achieved using the Gal4-upstream activation sequence (UAS) system. To monitor the non–cell-autonomous effect of HSPs, flies conditionally expressing HSPs under the control of various tissue-specific Gal4 drivers were crossed with GMR-HttQ120 flies, a polyglutamine disease model fly line that constitutively expresses a mutant huntingtin protein with 120 polyglutamine repeats, under the compound eye-specific GMR promoter (28). We then examined whether progression of polyglutamine-mediated degeneration in photoreceptor neurons could be affected by the elevated expression of HSPs in other tissues.

In the control GMR-HttQ120 flies, the compound eyes progressively degenerated due to expression of the polyglutamine protein (Fig. 6A, i–iv) (28). Increased expression of Hsp40 in the same tissue using the GMR-Gal4 driver resulted in cell-autonomous suppression of photoreceptor degeneration (Fig. 6D, vi) compared with the control flies expressing GFP (Fig. 6D, ii), as reported previously (29). Surprisingly, Hsp40 overexpression in the pigment cells, which are located next to the photoreceptor neurons, using the 54C-Gal4 driver (30) also resulted in morphological improvement of photoreceptor degeneration during the 7–21 d after eclosion (Fig. 6A, vi–viii). Upon quantitative evaluation by counting the number of remaining rhabdomeres in each ommatidium, we found that elevated expression of Hsp40 in the pigment cells resulted in a significant increase in the number of rhabdomeres per ommatidium, calculated as 64%, 66%, and 45% rescue in the GMR-HttQ120 flies at 7, 14, and 21 d of age relative to the GFP-expressing control flies, respectively (Fig. 6B). Consistently, Hsp40 expression using another pigment cell-specific driver, rdhB-Gal4, also ameliorated the photoreceptor degeneration in the GMR-HttQ120 flies (Fig. 6E and Fig. S4A, vi vs. ii). These results indicate that pigment cell-specific expression of Hsp40 non-cell autonomously suppresses the degeneration of photoreceptor neurons in neurotoxic HttQ120-expressing flies. Similarly, Hsp70 expression in the pigment cells also resulted in non–cell-autonomous suppression of photoreceptor degeneration in these flies (Fig. 6B and Fig. S4B). In contrast, GMR-HttQ120 flies bearing the UAS-Hsp40/Hsp70 transgenes, but not the pigment cell-specific Gal4 drivers, showed no improvement in rhabdomere number or photoreceptor degeneration (Fig. 6C and Fig. S4C), suggesting that possible leak expression of these chaperones from the UAS-Hsp40/Hsp70 transgenes is at a negligible level in this study. These data suggest that the increased level of Hsp40/Hsp70 in the pigment cells leads to non–cell-autonomous suppression of polyglutamine-dependent degeneration in the neighboring photoreceptor neurons in vivo, although we cannot exclude the possibility that the rescue of photoreceptor degeneration may be attributed to the improved appearance of the overall structures of the eyes by pigment-specific expression of HSPs.

To exclude such a possibility, we next examined whether non–cell-autonomous restoration of proteostasis is maintained between not only adjacent tissues but also physically remote tissues. We therefore generated GMR-HttQ120 flies with elevated Hsp40 expression in remote tissues, such as muscle and fat body. The expression patterns of the tissue-specific Gal4 drivers used in this experiment were confirmed using GFP as a reporter (Fig. S5A). Surprisingly, muscle-specific expression of Hsp40 using either the Mef2-Gal4 or Mhc-Gal4 driver resulted in effective suppression of photoreceptor degeneration in the GMR-HttQ120 flies (Fig. 6D, vii vs. iii and Fig. S4A, vii vs. iii), calculated as 56% or 47% rescue in the number of rhabdomeres, respectively (Fig. 6E). Similarly, fat body-specific expression of Hsp40 using either the Lsp2-Gal4 or r4-Gal4 driver also improved photoreceptor degeneration (49% and 43% rescue, respectively) (Fig. 6 D, viii vs. iv and E and Fig. S4A, viii vs. iv). Moreover, tissue-specific expression of Hsp70 in muscle and fat body also ameliorated the photoreceptor degeneration in the GMR-HttQ120 flies, whereas expression of a deficient mutant of Hsp70 lacking a substrate-binding domain (Hsp70dS) did not show such beneficial effects under the same condition (Fig. S6 A and B). The leak expression of UAS transgenes was excluded by immunohistochemistry and quantitative RT-PCR analyses using GFP as a reporter, showing that no GFP expression was detected in the compound eyes of the UAS-GFP flies bearing the Mef2-Gal4 or Lsp2-Gal4 driver (Fig. S5 B and C). Thus, these results reveal that elevated levels of HSPs in one tissue can non-cell autonomously suppress polyglutamine-mediated degeneration in another remote tissue without any direct cell-to-cell contact.

Non–Cell-Autonomous Suppression of Polyglutamine-Mediated Neurodegeneration by Hsp40 Occurs via Ykt6-Dependent Exosomal Secretion.

We then asked whether exosomal secretion from Hsp40-expressing tissues is responsible for the non–cell-autonomous restoration of proteostasis observed in our Drosophila models. Ykt6, which is one of the R-SNARE proteins, has been reported to be necessary for exosome secretion in Drosophila, because depletion of this protein caused the accumulation of exosome proteins inside cells, resulting in a corresponding decrease in the extracellular levels of those proteins (31). Therefore, we reduced the level of Ykt6 in the Hsp40-expressing tissues by RNAi-mediated knockdown and examined its effect on polyglutamine-mediated degeneration in photoreceptor neurons in the GMR-HttQ120 flies, as above.

The GMR-HttQ120 flies coexpressing Hsp40 and an inverted repeat RNA (IR) against Ykt6 in pigment cells under the rdhB-Gal4 driver showed a similar level of severe photoreceptor degeneration to the control GMR-HttQ120 flies coexpressing GFP (Fig. 7A, iv vs. iii), demonstrating that Ykt6 knockdown almost completely cancels the non–cell-autonomous improvement observed in the Hsp40-expressing GMR-HttQ120 flies without Ykt6 knockdown (Fig. 7A, ii vs. i). Quantitative analyses revealed that GMR-HttQ120 flies with pigment-specific expression of Hsp40 and Ykt6 knockdown failed to show a significant increase in the number of rhabdomeres per ommatidium relative to the corresponding GFP-expressing GMR-HttQ120 flies, whereas significant rescue (42%) was observed in the Hsp40-expressing GMR-HttQ120 flies without Ykt6 knockdown (Fig. 7B, Left). Expression of the Ykt6 IR using the rdhB-Gal4 driver had no significant effects on photoreceptor degeneration in the control GFP-expressing GMR-HttQ120 flies (Fig. 7 A, iii vs. i and B, Left), suggesting that pigment-specific knockdown of Ykt6 itself has no deleterious effects in photoreceptor neurons. These results indicate that Ykt6 expression in the Hsp40-expressing tissues plays a critical role in the non–cell-autonomous improvement of photoreceptor degeneration in the GMR-HttQ120 flies. Consistent with this finding, Ykt6 knockdown in the fat body using the Lsp2-Gal4 driver in the Hsp40-expressing GMR-HttQ120 flies also largely cancelled the non–cell-autonomous improvement of photoreceptor degeneration (Fig. 7A, v–viii), whereas 49% rescue was observed in the corresponding Hsp40-expressing GMR-HttQ120 flies without Ykt6 knockdown (Fig. 7B, Right). Thus, we conclude that non–cell-autonomous suppression of polyglutamine-mediated neurodegeneration by remote tissue-specific expression of Hsp40 in Drosophila depends on Ykt6-mediated exosomal secretion.

Purification of Exosomes.

Exosomes were prepared as described previously (21). Briefly, the culture medium of cells harvested in a 100-mm dish was replaced with exosome-depleted medium in which serum-derived exosomes were removed by ultracentrifugation at 100,000 × g for 16 h. After 24-h incubation, culture supernatants were collected and centrifuged sequentially at 2,000 × g for 15 min, 10,000 × g for 30 min, and 100,000 × g for 120 min. The pellet (P100) containing exosomes was resuspended in appropriate buffers. All centrifugations were performed at 4 °C.

Analysis of Inclusion Body Formation of Polyglutamine Proteins.

Cells were transfected with a Q81-EGFP–encoding plasmid vector using Lipofectamine LTX and PLUS reagent (Invitrogen) and incubated for 4 h. The medium was then replaced with fresh medium with or without P100 fractions. The cells were further incubated for 20 h and then subjected to microscopic analysis using a confocal microscope (FV1000; Olympus). Inclusion body formation was calculated as the ratio of the number of cells with inclusion bodies to the total number of transfected cells. For each sample, 700–1,300 transfected cells were analyzed, and the experiments were independently repeated at least three times.

EM.

Membrane vesicles, including exosomes, were prepared as described above. Vesicles were deposited on collodion-carbon–coated grids and fixed with 2% (wt/vol) paraformaldehyde. For immunoelectron microscopy, the vesicles were permeabilized with 0.1% saponin, followed by immunolabeling with an anti-Hsp40 (DnaJB1) antibody (Stressgen) and a secondary antibody conjugated with 10-nm gold particles (Sigma). The vesicles were negatively stained with uranyl acetate and analyzed with a transmission electron microscope (Tecnai Spirit; FEI).

Fly Stocks.

Flies were cultured and crossed under standard conditions at 25 °C. The transgenic fly lines bearing the GMR-HttQ120, UAS-GFP, UAS-Hsp70, UAS-Ykt6 IR, or tissue-specific GAL4 (GMR, 54C, rdhB, Mef2, Mhc, Lsp2, and r4) transgene were obtained from the Bloomington Drosophila Stock Center. For the generation of a transgenic fly line bearing the UAS-Hsp40 transgene, a DNA fragment coding for Drosophila Hsp40 (DnaJB1) was amplified from pOT2 cDNA clone GH26396 (obtained from the Drosophila Genetic Resource Center, Kyoto Institute of Technology) by PCR and was inserted into the pUAST vector. The resultant vector was injected into fly embryos by standard procedures to establish Hsp40 fly lines.

Histology.

Preparation of fly thin sections and evaluation of photoreceptor degeneration were performed as previously reported (43). Briefly, heads of GMR-HttQ120 flies were fixed in 2% paraformamide and 2.5% (wt/vol) glutaraldehyde and embedded in Epon. Compound eyes were then sectioned at 1 μm and stained with 1% toluidine blue. Microscopic images were obtained using a BX51 microscope with a DP71 CCD camera (Olympus), and the average number of rhabdomeres in each ommatidium was calculated. In all cases, the number of rhabdomeres was counted in 16 ommatidia in each section and at least six different sections from each eye were analyzed and averaged. The data are shown as the mean ± SEM of five to eight flies. The values of “% rescue” were calculated as follows:

in which I_HSP and I_GFP are the numbers of rhabdomeres per ommatidium in flies expressing HSPs and GFP, respectively, and I_total is the total number of rhabdomeres per ommatidium (I_total = 7).