The noncanonical small heat shock protein HSP-17 from Caenorhabditis elegans is a selective protein aggregase

Small heat shock proteins (sHsps) are conserved, ubiquitous members of the proteostasis network. Canonically, they act as “holdases” and buffer unfolded or misfolded proteins against aggregation in an ATP-independent manner. Whereas bacteria and yeast each have only two sHsps in their genomes, this number is higher in metazoan genomes, suggesting a spatiotemporal and functional specialization in higher eukaryotes. Here, using recombinantly expressed and purified proteins, static light-scattering analysis, and disaggregation assays, we report that the noncanonical sHsp HSP-17 of Caenorhabditis elegans facilitates aggregation of model substrates, such as malate dehydrogenase (MDH), and inhibits disaggregation of luciferase in vitro. Experiments with fluorescently tagged HSP-17 under the control of its endogenous promoter revealed that HSP-17 is expressed in the digestive and excretory organs, where its overexpression promotes the aggregation of polyQ proteins and of the endogenous kinase KIN-19. Systemic depletion of hsp-17 shortens C. elegans lifespan and severely reduces fecundity and survival upon prolonged heat stress. HSP-17 is an abundant protein exhibiting opposing chaperone activities on different substrates, indicating that it is a selective protein aggregase with physiological roles in development, digestion, and osmoregulation.

Proteins need to obtain and maintain a native folding state to be functional. To ensure correct protein folding, the cell is equipped with a complex protein homeostasis network composed of molecular chaperones and proteases (1)(2)(3)(4). Molecular chaperones maintain protein homeostasis (proteostasis) by guiding substrates to their correct folding state, by preventing or reversing protein misfolding, by disaggregating protein aggregates, and by targeting proteins to degradation. Failure of this proteostasis network due to advanced age, stress, or mutations can lead to an accumulation of misfolded and aggregated proteins and is associated with loss of function or pathological aggregation of proteins, including amyloid fibrils (1)(2)(3)(4)(5). In this proteostasis network, small heat shock proteins have been referred to as first line defenders, buffering unfolded proteins against aggregation in an ATP-independent fashion as "holdases" until other chaperones can further process these substrates (6,7). sHsps 4 may associate with protein aggregates, altering their properties, and thereby facilitate subsequent disaggregation and refolding (8). More recently, cell culture experiments have also demonstrated that the toxicity of preformed, aberrant protein oligomers can be alleviated by sHsps inducing segregation and further aggregation (9,10).
The sHsps are ubiquitous molecular chaperones and present in all domains of life. Whereas many unicellular organisms express no more than two distinct sHsps, their numbers are higher in metazoans with 10 sHsps in humans and 16 in Caenorhabditis elegans (11). Their expression is not necessarily separated to distinct organs, organelles, or developmental phases, raising the question of their unique chaperone functions and substrate spectrum (12,13). Besides binding their substrates, sHsps characteristically form oligomers as homooligomers or, together with other sHsps, as hetero-oligomers. This oligomerization is highly dynamic, sensitive to factors such as heat, pH, and phosphorylation, and is integral to their chaperone activity (6,7,14).
Human sHsps are involved in pathologies of muscles and neurons, such as cataract or myopathies, by mutations and consequent failure of their physiological function (15,16). Furthermore, sHsps are implied in amyloid pathology. HSPB5, or ␣B-crystallin, is the best-described human sHsp and was reported to act as chaperone for ␤-amyloid peptide and Huntingtin (17,18). Most other human sHsps are involved in at least one proteinopathy (19).
Whereas the inhibition of aggregation activity of sHsps such as HSPB5 has been established for a long time (20,21), an opposing effect was described only recently. Bacterial YocM and yeast Hsp42 were reported to act as a "molecular aggregase" and convert a metastable model substrate to an insoluble protein aggregate. This effect may depend on prion-like properties of the chaperone and is relevant in vivo for protein aggregation and viability in yeast and sets Hsp42 apart from the second sHsp in yeast, Hsp26 (22)(23)(24)(25).
Considering the functional specialization between even two sHsps in a single-cellular organism such as yeast, we set out to analyze noncanonical activity of the larger sHsp family in a metazoan. Here, we describe the C. elegans sHsp HSP-17, previously reported to be inducible by heat and heavy metal stress (26) and possibly localized to mitochondria (27). HSP-17 is expressed in the pharynx, the intestine, and the anus as well as in the excretory canal. In vitro, HSP-17 mirrors the molecular aggregase activity of Hsp42 for selected substrates, and we can demonstrate that HSP-17 promotes protein aggregation in vivo in intestinal and pharyngeal cells. Like Hsp42, HSP-17 can also inhibit the aggregation of some model substrates, making it a "selective protein aggregase." Animals that are depleted for HSP-17 show a delayed protein aggregation in the intestine, yet the animals are severely compromised in stress tolerance and exhibit a shortened lifespan and reduced progeny.

Recombinant expression and purification of HSP-17
An analysis of the chaperone activity and substrate spectrum of HSP-17 necessitates the production of recombinant active HSP-17. A previous purification strategy of HSP-17 was based on a denaturing approach (28). Here, we established a protocol for the purification of HSP-17 from the soluble fraction of HSP-17-overexpressing E. coli (Fig. 1A). Analytical size-exclusion chromatography (SEC) was performed on the purified protein (Fig. 1A, fraction 17). The experiment shows that the purified HSP-17 forms an oligomer with a molecular mass of ϳ460 kDa, suggesting a 26-mer of HSP-17 (Fig. 1B), which is in accordance with a previous report (28). A smaller peak with a molecular mass of Ͻ43 kDa can be attributed to residual His-SUMO tag. To confirm the identity of the purified protein, we performed intact protein QTof-MS. The experiment yielded a major peak corresponding to a peptide with the mass of 17,417.5 Da (expected mass of HSP-17: 17,417.52 Da) (Fig. S1C).
For controls in the subsequent chaperone assays, we purified C. elegans HSP-12.6 using the same protocol (Fig. S1A). HSP-12.6 is a small heat shock protein from C. elegans, expressed in larvae and the gonads of adult animals, which was reported to have no molecular chaperone activity. HSP-12.6 is among the smallest sHsps, consists of a truncated ␣-crystallin domain, and forms low-order oligomers (29,30). Analytical SEC shows an approximate molecular mass of about 25 kDa, suggesting the formation of a dimer (Fig. S1B).

HSP-17 promotes aggregation of MDH
To assess potential chaperone activities of HSP-17, we performed static light scattering on a chaperone model substrate, malate dehydrogenase (MDH). At a concentration of 500 nM, a temperature of 47°C is permissive of MDH aggregation, leading to a sigmoidal signal curve in light scattering, whereas 41°C leads to very slow signal increase. At this concentration, MDH alone shows minimal aggregation (Fig. 1, C and D). Notably, HSP-17 did not inhibit, but instead led to an earlier onset of aggregation and higher light scattering signals in a concentra-tion-dependent manner at both temperatures (Figs. 1, C and D). Moreover, only the addition of HSP-17 leads to pronounced aggregation of MDH at 41°C (Fig. 1C). Using a ratio of MDH/ HSP-17 of 1:1, a sigmoidal signal curve, indicating aggregation, can be observed, which is even more pronounced at a 1:2 MDH/ HSP-17 ratio (Fig. 1, C and D). Importantly, even at the highest concentration, HSP-17 alone does not aggregate as assessed by light scattering experiments (Fig. 1, C and D).
To test whether HSP-17 co-aggregates with MDH, we analyzed the soluble and insoluble fraction of the samples of 500 nM MDH aggregating with a 2-fold excess of HSP-17 by SDS-PAGE (Fig. 1E). At 41°C, roughly 55% of MDH remains soluble after 1 h, which is reduced to 35% in the presence of a 2-fold excess of HSP-17. At 47°C, more than 90% of MDH can be found in the insoluble fraction regardless of the presence or absence of HSP-17. Interestingly, less than 25% of HSP-17 is present in the insoluble fraction with MDH at 41 and 47°C (Fig. 1C). After heat treatment of HSP-17 alone at 41 or 47°C, less than 10% of the protein could be detected in the insoluble fraction (Fig. 1F).
As a control, we used HSP-12.6, which was reported to neither inhibit nor enhance the aggregation of model substrates in titration experiments, at up to 225-fold excess (30). HSP-12.6 is employed here to demonstrate that HSP-17, but not any small heat shock protein or any peptide, can selectively enhance the aggregation of model substrates. Unlike for HSP-17, a 2-fold excess of HSP-12.6 had no effect on the aggregation of MDH at 41 or 47°C (Fig. S2, A and B). As a further control, we used human HSPB5, which canonically inhibits aggregation of model substrates in this assay (31). A 2-fold excess of HSPB5 significantly reduced the aggregation of MDH at 47°C, whereas no effect was observed at 41°C (Fig. S2, C and D). Notably, HSP-17 did not promote the aggregation of two other aggregation prone proteins, citrate synthase (CS) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Here, a 16-or 32-fold excess of HSP-17 partially inhibited the aggregation of both substrates (Fig. 2, A and B).
In conclusion, HSP-17 promotes the aggregation of a metastable protein like MDH even at 41°C, whereas MDH alone shows slow aggregation. Interestingly, only a fraction of 25% of HSP-17 co-sediments with the aggregated MDH. Importantly, the protein aggregase activity of HSP-17 is selective, as HSP-17 did not promote the aggregation of alternative chaperone substrates such as CS and GAPDH.

HSP-17 inhibits disaggregation and refolding of aggregated luciferase
Co-aggregation of sHsps with aggregation-prone proteins can lead to the formation of aggregate species that can be recovered by ATP-dependent chaperones more efficiently (32,33). To test whether co-aggregation with HSP-17 can alter protein aggregates in a similar fashion, we performed luciferase disaggregation assays in the presence or absence of HSP-17.

C. elegans HSP-17 facilitates aggregation in vivo and in vitro
luciferase over 2 h (32). Prior to heat treatment, we added increasing concentrations of HSP-17 to luciferase. Stoichiometric concentrations or excess of HSP-17 (relative to HSP-1 (Hsc70)) led to a significant inhibition of luciferase recovery by the disaggregating chaperones (Fig. 2C). In parallel, we added HSP-17 after heat treatment of luciferase, and the disaggrega-
As control, we repeated both experiments with HSP-12.6 and observed no inhibitory effect on the disaggregation reaction regardless of whether HSP-12.6 was added before or after heat treatment of luciferase. However, we noticed a more efficient recovery of luciferase when HSP-12.6 was added after heat treatment of luciferase (Fig. S2, E and F). Taken together, the presence of HSP-17 during the heat-mediated aggregation of luciferase likely renders the luciferase aggregates more inert for remodeling by the disaggregating chaperones. Notably, HSP-17 exerts no effect on the disaggregation reaction when added after the aggregation step.

HSP-17 is expressed in the alimentary system and the excretory canal
The in vitro data showing that HSP-17 promotes the formation of selected protein aggregates and in addition also inhibits disaggregation are intriguing and raise the question of the biological role of HSP-17. We first wanted to gain insight into the expression profile and generated transgenic animals that express fluorescently tagged HSP-17 under the control of the endogenous promoter. HSP-17 was fused to either wrmScarlet (34,35) or GFP (phsp-17::hsp-17::wrmScarlet and phsp-17::hsp-17::gfp). As depicted in Fig. 3A, HSP-17 is expressed in the pharynx (i), in the excretory tract (v), in the intestine (ii, notably accumulating around the intestinal lumen), the pseudocoelom (iii), around the anus (iv), accumulating to spots inside the intestine and excretory tract (v), and the vulva (vi) (Fig. 3A and Fig. S3A). The observed foci formation of HSP-17 ( Fig. 3A (v) and Fig. S3 (v and vi)) is not due to a potential aggregation of the fluorescent fusion protein, as a transcriptional reporter (phsp-17::gfp) does not show foci formation, whereas phsp-17::hsp-17::gfp does (Fig. S4E). Moreover, localization of HSP-17 to foci appears to be distinct from stress granules, as we did not observe any co-localization to stress granule markers, such as PAB-1 (Fig.  S3D). We elected to confirm the observations in Fig. 3A and to distinguish the localization of HSP-17 to the tube-like structures of the intestinal tube and the excretory canal by genetic crosses of our HSP-17::wrmScarlet/GFP lines. First, we generated a cross with a GFP marker for the excretory canal cell (Pvha-1::gfp) and could demonstrate a clear co-localization of the GFP marker and HSP-17::wrmScarlet (Fig.  4A, i-iii). In addition, we also observed co-localization of HSP-17 with the intermediary filament B2 (ifb-2::cfp) around the intestinal lumen (Fig. 4A, iv-vi), which is part of a dense cytoplasmic network known as the endotube. These imaging analyses confirm our initial expression analyses ( Fig. 3A and Fig. S3).
To further investigate the localization of HSP-17 in the intestine, we performed correlative light and EM (CLEM) using the HSP-17::GFP strain. As depicted in Fig. 4B, HSP-17::GFP shows a distinct localization around the lumen of the intestinal cells ( Fig. 4B; the cross-section shows the intestine of the nematode). The longitudinal expression pattern around the lumen is distinct from the excretory canal, suggesting that HSP-17 also plays a role in digestion in addition to its function in osmoregulation (Figs. 3G and 5A).
Furthermore, HSP-17 appears to be distributed in the cytosol and does not show a nuclear localization (Fig. S4F).
The reporter strains express tagged HSP-17 in addition to the endogenous protein and thereby represent an overexpression condition. We thus wanted to determine the protein levels of the endogenous as well as tagged HSP-17 and performed a relative quantification of HSP-17 in normalized nematode lysates of synchronized 4-day-old adult hermaphrodites by Western blotting. The analysis revealed an almost equal expression for both transgenic constructs relative to the endogenous HSP-17 (endogenous HSP-17:HSP-17-wrm-Scarlet 1:1.35 and endogenous HSP-17:HSP-17-GFP 1:1.18) and thus an ϳ2-fold overexpression for both compared with the WT N2 (Fig. 3B).
To compare the levels of HSP-17 in the WT N2 with those of other chaperones (e.g. disaggregating chaperones), we performed absolute quantification of HSP-17 in C. elegans lysates of synchronized 4-day-old animals (young adults). The in vivo concentration of HSP-17 is 477 Ϯ 154 nmol/g total lysate protein and is thus comparable with the level of the constitutive Hsp70 homolog HSP-1 (525 Ϯ 136 nmol/g lysate (36)). We conclude that HSP-17 is an abundant protein, and the distinct expression pattern of HSP-17 may even lead to higher local concentrations in selected tissues, such as pharynx, intestine, and excretory tract ( Fig. S3C and Fig. 3A).

Depletion of HSP-17 affects lifespan, fecundity, and viability upon stress
To gain insight into the biological role of HSP-17 in vivo, we depleted HSP-17 by RNAi that reduces HSP-17 levels in the nematode by a factor of 2 (37). This effect can also be observed by microscopy (Fig. S3B). First, we analyzed how depletion of HSP-17 affects the viability and fecundity as readouts for the organismal fitness of the animal. RNAi of hsp-17 reduced the median lifespan by 1.5 days (from 18.2 to 16.8 days; Fig. 3C) and also led to a reduction of viable offspring by about 20%, suggesting a systemic effect of HSP-17

C. elegans HSP-17 facilitates aggregation in vivo and in vitro
depletion on the animal (Fig. 3D). Next, we assessed how the animals cope with prolonged heat shock (6 h at 35°C) conditions that severely challenge protein folding. Notably, we observed a pronounced reduction of the survival rate for the HSP-17-depleted animals, whereas the overexpression lines coped similarly to WT animals (Fig. 3E).

C. elegans HSP-17 facilitates aggregation in vivo and in vitro
To test whether HSP-17 co-localizes with misfolded and aggregated proteins that accumulate in the insoluble fraction, we analyzed the amount of HSP-17 in the soluble and insoluble fractions by Western blotting. We observed a transition of HSP-17 from the soluble into the insoluble fraction with the progression of aging. HSP-17 remained mainly in the soluble fraction in developing nematodes (day 2; L2/3 larvae) and fertile adults (day 4 and day 6). As aging progresses (day 8), the vast majority of HSP-17 resides in the insoluble fraction (Fig. 3F). Our in vitro data of the co-sedimentation analysis of HSP-17 with aggregated MDH argue against a co-aggregation of HSP-17 and hence suggest an active recruitment to the aggregated proteins (Fig. 1E).
HSP-17 is expressed in the excretory system, and one important physiological role of the excretory system is the maintenance and regulation of osmotic pressure. To assess a possible function of HSP-17 in osmoregulation, we analyzed how animals exhibiting higher (overexpression) or lower (depletion by RNAi) HSP-17 levels can tolerate osmotic stress. For that, we exposed young adult animals to 5 times higher salt concentrations (250 mM NaCl) and monitored their survival. As depicted in Fig. 3G, an overexpression of HSP-17 does not exhibit a significant effect on the survival rate, but depletion of HSP-17 (red) renders the nematodes more sensitive toward higher salt concentrations, leading to earlier death. To conclude, reduced HSP-17 levels sensitize animals to osmotic stress, suggesting a physiological role for HSP-17 in osmoregulation.

Overexpression of HSP-17 promotes polyQ aggregation in vivo
Next, we wanted to test whether the aggregation-promoting activity of HSP-17 ( Fig. 1) can be observed in vivo in C. elegans. For that, we selected an aggregation-prone protein that is expressed in the same tissue as HSP-17, the intestinal polyQ model Q 44 -YFP (iQ44). iQ 44 exhibits a diffuse distribution during development, and the first foci appear between day 4 and day 6 of life (38), serving as a scorable measure of protein aggregation in vivo. We hypothesized that the modulation of the levels of HSP-17 by either RNAi or 2-fold overexpression in the transgenic lines could affect the onset and severity of iQ 44 foci formation. We crossed the iQ 44 strain with the HSP-17overexpressing strain and confirmed that the level of HSP-17 overexpression remained 2-fold in this genetic background (Fig. S4D). Intriguingly, we observed that HSP-17overexpressing animals show an earlier onset of iQ 44 foci formation, whereas animals that are depleted for HSP-17 exhibit a delayed polyQ foci formation compared with the untreated iQ 44 model (Figs. 5, A and B). The severity of polyQ foci formation was also affected, as both the number of nematodes that had at least one polyQ aggregate and the number of nematodes that exhibited a higher load of larger foci were increased for the HSP-17 overexpression line ( Fig. S4A and Table 1). Notably, the aggregation propensity of polyQ proteins in a distal tissue such as muscle, where HSP-17 is not expressed, was not affected by depleting HSP-17 (Fig. S4B). Next, we analyzed whether the increased and earlier onset of iQ44 aggregation upon overexpression of HSP-17 correlates also with increased toxicity. For that, we analyzed the fecundity of the nematodes as a readout for organismal fitness. Importantly, in WT animals, HSP-17 overexpression does not affect the number of offspring (Fig.  3D). iQ44 animals produce fewer offspring (251 Ϯ 22 on average), and this number is further reduced to an average of 210 Ϯ 23 upon overexpression of HSP-17 in this genetic background. We conclude that the overexpression of HSP-17 can facilitate the aggregation of aggregation-prone proteins such as polyQ, and that compromises the fitness of the animal (Fig. S4C).
Subsequently, we analyzed a potential co-localization of HSP-17 with polyQ aggregates, as observed for MDH in vitro (Fig. 1C) within the animal. For that, we crossed the phsp-17::hsp-17::wrmScarlet line with a reporter strain expressing pvha-6::Q 85 ::yfp (iQ 85 ) that shows a robust polyQ aggregation already at a younger age. Indeed, a pronounced number of iQ 85 foci also show wrmScarlet fluorescence, implying co-localization of HSP-17 with polyQ foci (Fig. 5C). These data validate the aggregate-promoting effect of HSP-17 as well as its co-localization with a substrate in vivo. To further confirm the specificity of the association between polyQ aggregates and HSP-17, we switched to an in vitro set up. We co-incubated purified HSP-17 in vitro with HTTExon1Q 48 , generating fibrils (36), and utilized immunogold labeling to analyze the interaction of HSP-17 with the HTTExon1Q 48 fibrils by EM. We could indeed detect specific immunogold-labeling signals on the fibrils, confirming a direct interaction of HSP-17 with polyQ proteins (Fig.  5D).

HSP-17 overexpression leads to increased aggregation of KIN-19
To further explore a chaperone function of HSP-17, we aimed to identify an endogenous protein substrate that could be utilized as a reporter. The criteria for such an endogenous substrate were proneness to misfolding and aggregation in response to proteotoxic challenges and expression in the alimentary or excretory system. The ortholog of human CSNK1A1 (casein kinase 1 ␣1), KIN-19, has been identified previously as a protein that becomes aggregation-prone with the progression of aging in one of the first proteomic aging studies of C. elegans (24,39). KIN-19 is expressed in the pharynx and is therefore an ideal candidate for an endogenous reporter. Crossing a strain overexpressing our HSP-17-GFP fusion with the KIN-19-tagRFP-overexpressing strain (double

C. elegans HSP-17 facilitates aggregation in vivo and in vitro
OE) revealed co-localization of both proteins in the excretory tract (Fig. 6A, i-iii) and in foci in the pharynx (Fig. 6A, iv-vii). We confirmed the co-localization of HSP-17 and KIN-19 by software analysis of the images with the Fiji plugin EzColocalization (Fig. S5A). Mirroring our findings for intestinal polyQ (Fig. 5 (A and B) and Fig. S4A), overexpression of HSP-17 significantly increased the average total of KIN-19-tagRFP foci in the pharynx (Fig. 6B). KIN-19 overexpression has deleterious effects on the animals, and only 60% of their offspring develop into viable adults by day 4 of life (compared with 100% for N2, HSP-17 overexpression, or N2 nematodes subjected to control or hsp-17 RNAi) (Fig. 6C). This effect is exacerbated by overexpressing HSP-17 in the KIN-19 overexpression background, with only 30% of the double OE animals reaching adulthood on day 4. Corroborating this finding, RNAi of hsp-17, compared with control RNAi, can alleviate the deleterious effect of KIN-19 overexpression, increasing the fraction of nematodes reaching adulthood (Fig. 6C). This confirms our hypothesis that KIN-19 is a bona fide substrate of HSP-17, which mediates its activity as well as toxicity upon overexpression. We note that depletion of hsp-17 could not reduce the average total of KIN-19 foci in adult worms, implying that KIN-19 aggregation is enhanced by HSP-17 but not dependent on it (Fig. 6B).

Discussion
In this study, we characterized the small heat shock protein HSP-17 and could demonstrate that it promotes the aggregation for selected substrates, such as MDH, polyQ proteins, and KIN-19, and that HSP-17 co-localizes with these protein aggregates in vivo. Yet HSP-17 can also inhibit the aggregation of other metastable proteins, such as CS and GAPDH. Thus, HSP-17 exhibits opposing chaperone functions on different substrates and could be labeled a "selective protein aggregase." A similar aggregase activity has recently been observed for the bacterial sHsp YocM and the S. cerevisiae sHsp Hsp42 (24,25). Hsp42 is necessary for the formation of cytosolic aggregates, CytoQ (35). Thus, in vivo a more appropriate term to describe the aggregate-promoting activity would be "sequestrase." This sequestrase activity of Hsp42 has been linked to a prion-like N-terminal domain, which is not present in the significantly smaller HSP-17 (40), suggesting a different mechanism for HSP-17.
A sequestration of metastable proteins forming proteotoxic oligomers can be cytoprotective (9, 10) and might be in partic-ular a defense strategy for organisms such as C. elegans that consist only of post-mitotic cells that do not have the possibility to eliminate aggregates by cell division (22,(41)(42)(43)(44). The sequestration of metastable proteins into dynamic deposits such as Q-bodies could be beneficial if aggregates represent either an inert structure or if the aggregate is subsequently resolubilized by other chaperones (45). Yet in our setup, HSP-17 prevented a disaggregation by the Hsp70/J/110 system when incubated together with luciferase during the heat-mediated aggregation. It is possible that HSP-17 changed the aggregate properties, as HSP-17 had no effect on the disaggregation reaction when added after the aggregation step. This observation also excludes the possibility that HSP-17 competes with the aggregates for binding to any of the disaggregating chaperones. An alternative explanation could be that an additional player such as a cochaperone is missing in the in vitro system that remains to be identified, as has been demonstrated for the cellular sequestrases in yeast (46). Yet, we observed an aggregation-promoting activity also in vivo for polyQ proteins and the endogenous protein KIN-19. Notably, the increased aggregation of the bona fide substrate KIN-19 upon overexpression of HSP-17 is very detrimental for the fitness and development of the nematode. This observation suggests that the cell and the animal need to tightly regulate the abundance and activity of HSP-17. Whereas an overexpression can promote the aggregation of specific substrates, a depletion of hsp-17 severely compromises the fitness of the whole animal, resulting in a reduced lifespan and fecundity and defects in thermotolerance. It remains to be shown whether HSP-17 could facilitate the aggregation of toxic protein oligomers in a cytoprotective manner as demonstrated for other sHsps (9, 10).
The identification of additional endogenous substrates of HSP-17 will be essential to gain more insight into its chaperone functions, particularly in vivo as the aggregation-promoting activity of HSP-17 appears to be substrate-specific.
By employing translational reporters and confocal microscopy as well as CLEM, we could demonstrate that HSP-17 is expressed in the digestive and excretory system. Notably, the aggregation-promoting effect of HSP-17 is limited to the tissues of its expression and thus excluding a trans-cellular chaperone signaling as observed for HSP-90 (47).
In summary, we have identified HSP-17 as the first metazoan-selective protein aggregase among the sHsps. HSP-17 differs

C. elegans HSP-17 facilitates aggregation in vivo and in vitro
from Hsp42 and YocM in its inhibitory effect of the disaggregation reaction by the ATP-dependent chaperones. This suggests a different mechanism and also a different role within the proteostasis network. We could establish a role for HSP-17 in the protein quality control by facilitating the formation of protein aggregates that likely exhibit a different morphology or structure. How the chaperone activity of HSP-17 manifests its role in a diverse functional spectrum ranging from development, osmoregulation, and digestion to mitochondrial function will be the subject of further research.

Molecular cloning
For cloning of shsps, mRNA was prepared from heatshocked C. elegans using a NucleoSpin RNA isolation kit (Macherey Nagel) to subsequently generate cDNA using the Maxima First Strand cDNA synthesis kit (Thermo Fisher Scientific). The cDNAs of the respective shsps were amplified and inserted into a pSumo vector, which was described previously (48). For hsp-17, the primers ccagtgggtctcaggtgg-tATGGATCGTCGTTTTCCACC (forward; cleavage site for BsaI) and ataagaatgcggccgcTCAGTTTCTCTTTGGCA-CAATTGTG (reverse; cleavage site for NotI) were used. For hsp-12.6, the primers ccagtgggtctcaggtggtATGATGAGCGT-TCCAGTGATGG (forward; cleavage site for BsaI) and ataa-gaatgcggccgcATGTCACTTTACCACTATTTCC (reverse; cleavage site for NotI) were used. To generate RNAi constructs for in vivo knockdown experiments, the E-RNAi web application (49) has been utilized to determine suitable dsRNA constructs. These have been amplified from genomic DNA and cloned into the vector L4440, which was described previously (50). For hsp-17, the primers ataagaatggtaccatGGACAC-CGAGTAGGAGATGC (forward; cleavage site for KpnI) and cagtcaaagcttTCAGTCTTCTCGTTATGCTTTCC (reverse; cleavage site for HindIII) were used. The correct sequence of all generated plasmids was confirmed by sequencing (Source BioScience (Nottingham, UK) and LGC Genomics (Berlin, Germany)).

Recombinant expression and purification of sHsps
The generated plasmids were transformed into E. coli BL21 DE3 carrying the pRARE plasmid for recombinant expression. Cells were grown at 37°C to an A 600 between 0.5 and 0.8 and then shifted to 30°C. Expression was induced by the addition of 0.5 mM IPTG. Cells were harvested after 4 h and flash-frozen before resuspension in lysis buffer (500 mM KCl, 30 mM HEPES, 5 mM MgCl 2 , and 10% (v/v) glycerol, cOmplete protease inhib-itor tablets, DNase I, 20 mM imidazole, 1 mM phenylmethylsulfonyl fluoride, and 0.01% (v/v) Triton X-100, pH 7.4) and lysis using the LM10 Microfluidizer (Microfluidics, Westwood, MA). Lysates were cleared by centrifugation, and the supernatant was incubated with nickel-loaded agarose beads for 90 min. Beads were washed in a gravity-flow column by high-salt buffer (1000 mM KCl, 30 mM HEPES, 20 mM imidazole, 0.05% Tween 20, pH 7.4) and low-salt buffer (50 mM KCl, 30 mM HEPES, 5 mM MgCl 2 , 20 mM imidazole, pH 7.4) before elution in elution buffer (50 mM KCl, 30 mM HEPES, 5 mM MgCl 2 300 mM imidazole, pH 7.4). The His-SUMO tag was removed by ULP1 protease over 2 h. For large sample volumes, a concentration step was performed. Samples were diluted into low-salt buffer 2 (25 mM KCl, 30 mM HEPES, 5 mM MgCl 2 , pH 7.4) to a total volume of 50 ml. After 10 min of incubation with Ni-NTA-agarose beads, the protein sample was filtered and loaded onto a Resource Q column (6 ml). Elution was performed by an increasing gradient of high-salt buffer, and sample fractions containing the protein of interest were aliquoted and flash-frozen in liquid nitrogen. HSP-1, HSP-110, DNJ-12, and DNJ-13 and ULP1 were recombinantly expressed and purified as described previously (32). For subsequent assays, aliquots were thawed and centrifuged, and concentrations were measured in triplicates by Bradford assay or absorption at 280 nm using a NanoVue Plus spectrophotometer (GE Healthcare, Chalfont St. Giles, UK).

Analytical size-exclusion chromatography
Experiments were carried out with an ÄKTA Explorer (GE Healthcare). A Superose 6 10/300 GL column was calibrated with a high-molecular weight calibration kit (GE Healthcare) before 100 l of small heat shock proteins were applied to the column and separated at room temperature using the running buffer: 140 mM KCl, 30 mM HEPES, 5 mM MgCl 2 , pH 7.4. Samples were collected for subsequent SDS-PAGE analysis.

Intact protein QTof-MS
Intact proteins were analyzed using a Waters H-class instrument equipped with a quaternary solvent manager, a Waters sample manager-FTN, a Waters PDA detector, and a Waters column manager with an Acquity UPLC protein BEH C4 column (300 Å, 1.7 m, 2.1 ϫ 50 mm). Proteins were eluted at a column temperature of 80°C with a flow rate of 0.3 ml/min.

C. elegans HSP-17 facilitates aggregation in vivo and in vitro
The following gradient was used (A ϭ H 2 O ϩ 0.01% formic acid, B ϭ MeCN ϩ 0.01% formic acid): 5-95% B 0 -6 min at 40°C. Mass analysis was conducted with a Waters XEVO G2-XS QTof analyzer. Proteins were ionized in positive ion mode, applying a cone voltage of 40 kV. Raw data were deconvoluted with MaxEnt 1.

C. elegans strains
All C. elegans strains used in or generated for this study are listed in Table 2.

Generation of transgenic C. elegans strains
Reporter strains of hsp-17 were generated by microinjection (51). Co-injection of a high concentration of oligonucleotides was utilized to generate chromosomally integrated translational reporters as described previously (52).

Maintenance of C. elegans
C. elegans nematodes were maintained on NGM plates seeded with E. coli strain OP50 as a food source at 20°C for experiments and maintenance at 15°C by standard protocols (53). For RNAi-mediated knockdown of hsp-17, RNAi constructs for hsp-17 or an empty vector were transformed into E. coli strain HT115. Transformed bacteria were grown overnight before induction with 1 mM IPTG for 2 h and seeded onto NGM plates supplied with IPTG and ampicillin. For RNAi experiments, nematodes were placed as L1 on these RNAi plates. For second-generation RNAi experiments, their offspring was transferred to fresh RNAi plates to repeat the procedure.

Lifespan assay
A total of 150 hermaphrodites each for control and RNAi conditions were placed as L1 larvae on RNAi or control plates and checked for viability each day by gentle prodding of their heads. Nematodes were passaged every day for the first week and later as required to separate them from their progeny and maintain enough food. The Oasis online tool was used to analyze the resulting data (54).

Fecundity assay
At least 20 adult hermaphrodites per condition were separated onto individual plates, and the number of eggs laid was counted for each nematode. The assay was performed in triplicates. The average number of eggs laid was plotted.

Thermotolerance assay
To assay thermotolerance, young adult (4-day-old) hermaphrodites were exposed to heat shock at 35°C for 6 h and then transferred to 20°C for 24 h before survivors were counted. Three independent repeats were performed with 25 nematodes for each condition. The percentage of total nematodes surviving was plotted.

Developmental assay
10 L1 nematodes per condition were transferred onto fresh NGM plates, maintained at 20°C, and scored on day 4 if they reached adulthood. The experiment was performed in triplicates.

Quantification of HSP-17 levels
Absolute levels of HSP-17 in C. elegans lysates were quantified in three biological replicates, as described previously (36). In brief, C. elegans lysates were generated from populations of the same age in triplicates, adjusted to the same protein level, and labeled with Cy5-NHS (GE Healthcare). The samples were then subjected to Western blotting in parallel with calibration samples of purified HSP-17 of a known concentration. The HSP-17 concentration in the samples was then calculated based on the calibration; Cy3 fluorescence was used for normalization.

C. elegans imaging
For imaging, suitable specimens were mounted on pads of 2% agarose on glass slides, in a drop of 2 mM levamisole. Imaging

C. elegans HSP-17 facilitates aggregation in vivo and in vitro
was performed on a Zeiss LSM780 microscope. Parameters such as aperture and intensity of laser were chosen to provide adequate imaging and kept constant for each experiment.

DAPI staining of C. elegans
For nuclear staining of C. elegans, nematodes were fixed in formaldehyde solution (4% formaldehyde in 100 mM K 2 HPO 4 , pH 7.4) for 30 min before washing in PBST, pH 7.4. Animals were then incubated for 2 h in DAPI solution (0.5 mg/ml DAPI in PBS including 0.5% Triton X-100 and 10 mg/ml BSA, pH 7.4). Nematodes were imaged after washing in the same solution excluding DAPI.

Correlative light and EM
Nematodes were cryofixed by an HPM100 high-pressure freezing machine in bacteria (OP50) for cryoprotection. Nematodes were then subjected to freeze substituted (FS) in acetone plus 0.1% uranyl acetate, 0.1% tannic acid, 0.05% glutaraldehyde, 0.5% H 2 O (at Ϫ90°C). Following FS, animals were washed in pure acetone and infiltrated with HM20 resin/acetone mixes (1:2, 1:1, and 2:1) and pure HM20 changes for 2 h each plus pure HM20 for the last change overnight (everything at Ϫ45°C). UV-induced polymerization was carried out for 72 h (Ϫ45 to Ϫ5°C). The FS program used was as follows: 5 h at Ϫ90°C; 9-h increase from Ϫ90 to Ϫ45°C (5°C/h); 24 h at Ϫ45°C; 48 h at Ϫ45 plus UV light; 8.5-h increase from Ϫ45 to Ϫ5°C plus UV light; Ϫ5°C for 16 h plus UV light. Polymerized blocks where trimmed and ultrathin-sectioned onto the Finder grids (Plano). The grids were positioned into the imaging chamber in PBS/glycerol (10%) solution. Sections were imaged with LSM800 equipped with an AIRYscan detector. Retrieved grids were contrasted with uranyl acetate and lead citrate and imaged with a Zeiss 900 microscope.

Transmission EM
TEM and immunogold labeling analysis were performed as described previously (36).

Image analysis
Image data were evaluated and prepared using Fiji (55,56) and Adobe Photoshop. Quantification of co-localization of fusion proteins was carried out using the Fiji plugin EzColocalization (57) using default settings and no cell identification.

Light scattering
Light scattering was performed as described previously (23). Light scattering was measured with a Jasco FP8300 fluorescence spectrometer at 360 nm, medium sensitivity settings, and a slit width of 2.5 nm in volumes of 1 ml in semi-micro cuvettes (Hellma) under constant stirring at 120 rpm in a Peltier-heated chamber. After obtaining a stable baseline for the buffer, the sHsp of interest was diluted into the buffer and measured for control or at least for 5 min. Subsequently, substrate was diluted into the cuvette for measurement, and for MDH, 2 mM DTT was supplied. For MDH, the aggregation at 41°C was analyzed for 1 h and at 47°C for 30 min. For CS, aggregation was carried out at 43°C and for GAPDH at 45°C.

Sedimentation of MDH aggregates
To separate soluble and insoluble fractions after aggregation experiments with MDH, samples were centrifuged at 20,000 ϫ g for 30 min. The supernatant was carefully removed, and pellets were completely resolubilized before samples were analyzed on SDS-PAGE.

Luciferase assay
Luciferase assays were carried out as described before (32,58) at temperatures of 20°C using 4 M HSP-1, 2 M HSP-110, 2 M DNJ-12, and DNJ-13. 15 nM luciferase was denatured for 15 min at 45°C in the absence or presence of the specified amount (0.1-10 M) of sHsp before being mixed with the disaggregating chaperones. As a control, sHsps were heated separately before being added to denatured luciferase.

Data availability
Z-stack images of the HSP-17 transgenic animals (phsp-17::hsp-17::wrmScarlet) have been deposited on the Zenodo platform to provide the complete data set on the expression of HSP-17 in the whole body, the head region, and the tail (DOI: 10.5281/zenodo.3594412).