Prion propagation by Hsp40 molecular chaperones

Abstract

Molecular chaperones regulate essential steps in the propagation of yeast prions. Yeast prions possess domains enriched in glutamines and asparagines that act as templates to drive the assembly of native proteins into beta-sheet-rich, amyloid-like fibrils. Several recent studies highlight a significant and complex function for Hsp40 co-chaperones in propagation of prion elements in yeast. Hsp40 co-chaperones bind non-native polypeptides and transfer these clients to Hsp70s for refolding or degradation. How Hsp40 co-chaperones bind amyloid-like prion conformers that are enriched in hydrophilic residues such as glutamines and asparagines is a significant question in the field. Interestingly, selective recognition of amyloid-like conformers by distinct Hsp40s appears to confer opposing actions on prion assembly. For example, the Type I Hsp40 Ydj1 and Type II Hsp40 Sis1 bind different regions within the prion protein Rnq1 and function respectively to inhibit or promote [RNQ⁺] prion assembly. Thus, substrate selectivity enables distinct Hsp40s to act at unique steps in prion propagation.

Molecular Chaperones and Yeast Prions

Proteins adopt a diverse and dynamic array of structural conformations. Prions are unique in that these proteins induce conversion of the soluble, native structure into the prion conformer with a high propensity to self-assemble into beta-sheet-rich, amyloid-like fibrils.1 Extensive investigation of prion biogenesis in the budding yeast Saccharomyces cerevisiae has uncovered some of the basic mechanisms underlying prion assembly into amyloid-like fibrils and inheritance of the prion state. One intriguing development in this story was the intimate role for heat shock protein (HSP) molecular chaperones in these pathways.2,3 Indeed, numerous yeast prions are dependent upon molecular chaperones for efficient maintenance and propagation of prion structures.4,5 On the other hand, overexpression of some molecular chaperones “cure” yeast of the heritable prion suggesting molecular chaperones antagonize prion assembly.4,6,7 How such opposing activities efficiently coordinate prion assembly into amyloid-like fibrils and propagation of the prion state inside the cell is an outstanding question in the field. Study of this process is significant because amyloid-like fibrils accumulate in numerous conformational disorders.8,9 However, the connection between amyloid assembly and neuronal cell death is still controversial as several recent studies implicate the assembly of amyloid-like fibrils as benign or even protective.10–12 In addition, prions found in S. cerevisiae possess domains enriched in glutamines (Gln) and asparagines (Asn),13 resembling proteins with expanded polyglutamine repeats (such as human huntingtin and several ataxins) that are very susceptible to aggregation.14,15 Many molecular chaperones are functionally conserved from yeast to humans, and as such, studying how molecular chaperones modulate prion propagation yields substantial mechanistic insight on the regulation of amyloid assembly in conformational disorders.

Several classes of molecular chaperone are implicated in prion propagation. For example, the AAA⁺ protein remodeling factor Hsp104 is required for propagation of several prions in yeast.4,16,17 Hsp104 is proposed to shear prion polymers to generate “seeds” that drive conversion of native protein into the prion conformation.18–21 Hsp70 molecular chaperones also regulate prion propagation although the particular function depends on the Hsp70 class and specific yeast prion. For example, mutations in the Hsp70 Ssa1 destabilize [PSI⁺] prion propagation while overexpression of Hsp70s from the Ssa family can stabilize the [PSI⁺] state.22–24 Interestingly, overexpression of Ssa1 has been shown to cure yeast of the prion [URE3].6,7 In contrast, Hsp70s of the Ssb family appear to antagonize [PSI⁺] prion propagation.24–26 As a result, protein flux through Hsp70 refolding pathways is a crucial step in prion biogenesis. Hsp70 chaperone activity is tightly coordinated by Hsp40 co-chaperones (also known as J-proteins). Additionally, Hsp40 and Hsp70 chaperones cooperate with Hsp104 to refold aggregated proteins.27 Several recent studies underscore a complex, yet fundamental role for Hsp40 co-chaperones in prion assembly and propagation.5,12,28–31 In this review we first describe the Hsp40 co-chaperone family and basic mechanisms underlying Hsp40:substrate recognition. Then, general roles for Hsp40s in propagation of the yeast prions [PSI⁺] and [URE3], and [RNQ⁺]/[PIN⁺] are discussed. Recent studies investigating the function of Hsp40s on assembly of the [RNQ⁺] prion are emphasized to highlight novel mechanisms in which Hsp40s bind Gln/Asn-rich prion proteins and modulate the accumulation of toxic or benign prion conformers. We propose that the distinct binding preferences of individual Hsp40s determine specific Hsp40 actions in prion assembly and propagation.

Protein Quality Control by Hsp40 Molecular Chaperones

Hsp40 co-chaperones are essential partners in Hsp70 function.32 Hsp40s share a highly conserved region called a J-domain that stimulates the intrinsic ATPase activity of its partner Hsp70.33 ATP hydrolysis causes a series of conformational changes that increase the affinity of client:Hsp70 interactions.34,35 Client release from Hsp70 is induced when ADP is replaced with ATP by a Hsp70 nucleotide exchange factor.36 The Hsp40 J-domain alone appears sufficient to maintain basic cellular processes required for physiological growth.37 However, based upon homology to the J-domain from the founder Hsp40 in Escherichia coli (DnaJ), there are twenty-two Hsp40s in budding yeast and forty-one Hsp40s in humans.32,38 Given the evolutionary expansion of the Hsp40 family, how these various Hsp40s specify Hsp70 function is an important unanswered question.

Importantly, Hsp40s utilize a variety of specialized domains outside of the J-domain to bind misfolded polypeptides and transfer these non-native clients to Hsp70 for refolding or degradation.39,40 Thus, Hsp40s select substrates for Hsp70 chaperone action and serve as the first line of defense in protein conformational disorders by recognizing non-native protein conformers. Hsp40s are classified based on the presence of several core domains found in DnaJ. Type I Hsp40s possess a J-domain, glycine/phenylaline(G/F)-rich region, and a zinc finger-like region (ZFLR) (Fig. 1). Type II Hsp40s possess the J-domain and G/F-rich region while Type III Hsp40s retain only the J-domain. The core Hsp40 domains described above influence Hsp40 quaternary structure and substrate selectivity (Fig. 1B).41,42 Specialized Hsp40s have further acquired unique domains and modifications that likewise influence substrate preferences as well as regulation of Hsp70 refolding activity.43-45 For example, the yeast Type I Hsp40 Ydj1 possesses a hydrophobic depression in its C-terminal domain that binds hydrophobic peptides46 as well as a CaaX motif that is modified by farnesylation.47 Interestingly, the Ydj1 ZFLR, hydrophobic polypeptide-binding pocket, and farnesyl modification all have been shown to participate in substrate binding.29,46,48 Yet specific features are either necessary or dispensable for binding to individual substrates. Thus, Hsp40s can utilize various domain combinations to bind a wide range of non-native clients. The yeast Type II Hsp40 Sis1 also possesses a hydrophobic polypeptide-binding pocket49 yet does not contain a ZFLR nor a CaaX motif. However, Sis1 does contain a G/M-rich region adjacent to the G/F-rich region,50 both of which appear to influence essential cellular functions of this Hsp40.51 While Ydj1 and Sis1 exhibit some overlapping physiological function,52,53 these Hsp40s also display distinct substrate preferences41 and as discussed below, exert very different activities on propagation of prions in yeast.

Hsp40 Activity in Propagation of [PSI⁺] and [URE3] Prions

Studies of the yeast prions [PSI⁺] and [URE3] have identified complex roles for Hsp40 co-chaperones in prion propagation and assembly into amyloid-like fibrils. The inheritable element [PSI⁺] is formed by the yeast translation termination factor Sup35.18,54 Both Ydj1 and Sis1 physically associate with large Sup35 aggregates,55 though propagation of the [PSI⁺] prion is specifically dependent upon Sis1.5,30 On the other hand, overexpression of Ydj1 in conjunction with its cognate Hsp70 destabilizes “weak” [PSI⁺] variants.6 Also noteworthy, overexpression of Apj1 (another Type I Hsp40 in yeast) cures cells of specific [PSI⁺] variants.56 Apj1 shares strong homology with Ydj1 yet its cellular functions are still unclear. Recent studies on Sup35 fibril assembly in vitro have demonstrated a direct role for Hsp40 molecular chaperones in regulating the assembly of amyloid-like fibrils.31,57 Interestingly, select Hsp40:Hsp70 pairs exert different actions on Sup35 assembly as well as the prion remodeling activity of Hsp104.58 Therefore, distinct chaperone complexes might selectively regulate prion assembly and propagation to alternate outcomes.

Hsp40s also regulate propagation of the yeast prion [URE3]. The [URE3] prion is formed by Ure2, a modulator of nitrogen catabolism in yeast.59 Similar to [PSI⁺], propagation of the [URE3] prion requires Sis15 while overexpression of the Ydj1 cures yeast of [URE3].16 Furthermore, Ydj1 inhibits the in vitro assembly of Ure2 into amyloid-like fibrils.60,61 In recent studies though, overexpression of J-domains from other yeast Hsp40s was shown to be sufficient to cure yeast of the [URE3] trait.5,62 These data indicate that modulating cycles of Hsp70 activity in the cell perturb [URE3] prion biogenesis and inheritance. This effect seems specific for the [URE3] prion,5 suggesting this yeast prion is particularly sensitive to aberrations in prion flux through Hsp70 refolding pathways. Altogether, Sis1 and Ydj1 drive [PSI⁺] and [URE3] propagation to completely different outcomes whereby Sis1 promotes the efficient propagation of prion elements yet Ydj1 antagonizes this pathway. However, studies on [PSI⁺] and [URE3] propagation have not revealed the molecular mechanisms underlying these divergent chaperone actions.

Selective Recognition of the [RNQ⁺] Prion by Opposing Hsp40 Activities

Studies of the yeast prion [RNQ⁺] have recently revealed novel mechanisms by which Hsp40 co-chaperones bind amyloid-like prion conformers and perhaps regulate prion propagation pathways to distinct endpoints. The yeast prion [RNQ⁺]/[PIN⁺] is formed by the yeast protein Rnq1 (rich in asparagines and glutamines) (Fig. 2A).63,64 The [RNQ⁺] state facilitates the conversion of other prions in yeast64,65 as well as seeding toxic conformers of an expanded glutamine form of human huntingtin.66 Rnq1 possesses a C-terminal Gln/Asn-rich prion domain that is sufficient to assemble into amyloid-like fibrils in vitro67,68 and induce prion formation when fused in place of the Gln/Asn-rich N-terminal domain of Sup35.63 The N-terminal non-prion domain of Rnq1 appears to regulate [RNQ⁺] prion propagation though the function of this domain is still unclear.69 Not long after [RNQ⁺] was first described, propagation of [RNQ⁺] prions was shown to be dependent upon Sis1.70 Deletion of other Hsp40s in yeast has no effect on the [RNQ⁺] state suggesting this dependency is specific for Sis1.5 In contrast to Sis1, overexpression of Ydj1 cures yeast of some [RNQ⁺] prion variants.71 Thus, similar to other yeast prions discussed above, Sis1 promotes [RNQ⁺] propagation while Ydj1 can inhibit [RNQ⁺] assembly perhaps reflecting distinct fundamental functions for these two Hsp40 co-chaperones in the prion assembly pathway (Fig. 2B).

What might account for the opposing functions of Sis1 and Ydj1 on [RNQ⁺] prion propagation? Recent studies demonstrated that Sis1 and Ydj1 bind to different regions within the Rnq1 protein. For example, Sis1 binds a short, hydrophobic motif in the non-prion domain12 while Ydj1 binds numerous motifs in the Gln/Asn-rich prion domain of Rnq1 (Fig. 2A).29 Interestingly, interaction between Rnq1 and either Hsp40 is dependent upon the [RNQ⁺] prion conformation.29,70 Conformational conversion of native Rnq1 into the [RNQ⁺] prion state might expose the Sis1-binding site in the non-prion domain. This provides a mechanism by which Sis1 action in [RNQ⁺] propagation is regulated through conformation-specific recognition by an Hsp40 co-chaperone. Binding between Ydj1 and the Gln/Asn-rich prion domain of Rnq1 is quite surprising because peptide array studies suggest Type I Hsp40s such as Ydj1 prefer substrates enriched in hydrophobic residues.41,72,73 Furthermore, select binding to the Rnq1 prion domain in the [RNQ⁺] prion state implies that Ydj1 recognizes the Gln/Asn-rich motifs in a conformation-specific manner. Altogether, two Hsp40s in the cell bind [RNQ⁺] prions yet target different regions in the Rnq1 protein. The outcome of such binding preferences might (at a rudimentary level) account for the disparate chaperone activities on [RNQ⁺] prion propagation.

What features in Ydj1 and Sis1 direct these Hsp40 chaperones to bind distinct domains within Rnq1? Interestingly, binding between Ydj1 and the Rnq1 prion domain is dependent upon the Ydj1 ZFLR and farnesylation at its C-terminal CaaX motif.29 The Ydj1 ZFLR is adjacent to two anti-parallel beta-strands that might bind the beta-rich Rnq1 prion domain through a beta-strand donor mechanism.74,75 How lipid modification of an Hsp40 co-chaperone contributes to substrate interaction is unclear, although farnesylation of Ydj1 has been implicated in binding to the kinase Ste11.48 Thus, farnesylation appears required for binding to numerous chaperone substrates including yeast prions. In contrast, Sis1-dependent maintenance of the [RNQ⁺] prion state requires unique extensions in the G/F-rich region of Sis1.76 These observations collectively suggest that Hsp40 co-chaperones rely on specialized modules to bind distinct domains in Rnq1 and regulate different aspects of [RNQ⁺] prion propagation.

Given such discrete binding preferences, how might Sis1 and Ydj1 exert their opposing activities on [RNQ⁺] prion propagation? Importantly, Sis1 binds to Rnq1 in a near 1:1 stoichoimetric complex while binding between Ydj1 and Rnq1 (or its Gln/Asn-rich prion domain) appears substoichiometric.29,76 Sis1 might coat [RNQ⁺] prion assemblies via binding the Rnq1 non-prion domain, direct Hsp70/Hsp104 molecular chaperones to [RNQ⁺] fibrils, and facilitate shearing to generate heritable [RNQ⁺] prion seeds.28,30 In addition, Sis1 might facilitate addition of new Rnq1 subunits into an elongating [RNQ⁺] fibril because overexpression of Sis1 increases the pool of amyloid-like [RNQ⁺] assemblies while overexpression of Hsp70 and Hsp104 does not result in such an increase.12

In contrast to Sis1, Ydj1 might cap the exposed ends of [RNQ⁺] prion assemblies and thereby inhibit fibril elongation by sterically hindering contacts between exposed Gln/Asn-rich motifs in the [RNQ⁺] prion conformer. In addition, Ydj1 might bind a [RNQ⁺] prion assembly intermediate and cooperate with Hsp70 to refold the Rnq1 protein into its native conformation or partition this protein conformer into an alternative off-pathway assembly that is subsequently remodeled by another chaperone complex.77 The net result of either mechanism would be solubilization of assembled Rnq1 and loss of the [RNQ⁺] prion trait. Importantly, Sis1 activity must normally out-compete Ydj1 to promote efficient [RNQ⁺] propagation. This might occur because Sis1-binding to Rnq1 is stoichiometric76 and Ydj1-binding motifs in the Rnq1 prion domain are buried within most [RNQ⁺] assemblies. Furthermore, some but not all [RNQ⁺] prion variants are sensitive to Ydj1 overexpression71 suggesting that Ydj1 may differentially recognize Rnq1 prion domain surfaces exposed in specific [RNQ⁺] prion variants.

These studies provide novel insight on the selectivity of Hsp40 interaction with amyloid-like substrates, yet several significant questions still remain. For example, if Sis1 and Ydj1 bind distinct domains in Rnq1 do these co-chaperones bind simultaneously to the same Rnq1 protein or do Sis1 and Ydj1 act on discrete [RNQ⁺] intermediates/assemblies? Future studies will be required to dissect this question conclusively, yet the answer will likely reveal significant insight on the basic mechanisms of chaperone recognition of amyloid-like protein species in conformational disorders.

Hsp40s Protect Cells from Toxic Prion Conformers

The study of Hsp40 action in [RNQ⁺] assembly has further revealed that Sis1 and Ydj1 protect the cell from the accumulation of cytotoxic protein conformers. Overexpression of Rnq1 is toxic to yeast in the presence of pre-existing [RNQ⁺] prion.12 Importantly, overexpression of Sis1 suppresses cytotoxicity caused by excess Rnq1, an effect that correlates with enhanced [RNQ⁺] prion assembly into SDS-insoluble aggregates and a decrease in the pool of SDS-soluble, Rnq1 protein species. Furthermore, mutating the Sis1-binding site in the Rnq1 non-prion domain decreases the efficiency of [RNQ⁺] prion assembly and exacerbates toxicity.12 Thus, chaperone-mediated [RNQ⁺] assembly appears protective although the specific nature of the cytotoxic protein conformer is still unclear.

In contrast to full length Rnq1, overexpression of the Rnq1 prion domain alone is not toxic to yeast.12 However, overexpression of this prion fragment becomes toxic in the absence of Ydj1.29 Unlike full-length Rnq1, whose toxicity correlated with the appearance of a low molecular weight, soluble protein species, the toxicity by the Rnq1 prion domain in a ydj1-null background correlated with an increase in the pool of large, SDS-insoluble assemblies.29 Thus, Ydj1 might be required for the cell to tolerate excessive levels of large, amyloid-like species, although this point requires further investigation. Altogether, the above observations suggest a model in which Sis1 and Ydj1 coordinate the flux of Rnq1 proteins through the [RNQ⁺] prion assembly pathway in order to maintain the accumulation of soluble versus amyloid-like particles within a tolerable threshold for the cell. Interestingly, Sis1 and Ydj1 were previously shown to have opposing effects on aggregation and toxicity of an expanded polyglutamine model in yeast.78 Altogether, Hsp40 co-chaperones selectively bind non-native protein species to either maintain protein solubility or drive aggregation.79 Even though amyloid assembly can be protective it is important to consider that excessive amyloid burden might also result in cell death.80 Susceptibility to various amyloid-like protein conformers is likely dependent upon the global expression pattern of environmental factors that buffer proteotoxicity.81

Concluding Remarks and Future Directions

Hsp40 molecular chaperones have recently emerged as critical regulators of prion propagation in yeast. Interestingly, individual Hsp40s modulate discrete steps in the prion assembly pathway and in the case of Ydj1 and Sis1, execute opposing activities on propagation of several yeast prions. Such disparate activities might originate from the unique binding preferences exhibited by the Hsp40 co-chaperone family, including recognition of Gln/Asn-rich regions responsible for assembly into beta-rich amyloid-like fibrils. Sis1 and Ydj1 possess unique structural domains that might account for such differential binding preferences.41 For example, Ydj1 utilizes its ZFLR and a farnesyl moiety to bind the prion domain of Rnq1.29 These features are conserved in various human Hsp40s and might contribute to the recognition of expanded polyglutamine conformers in various human diseases.82–84 Studies of Hsp40 action in [RNQ⁺] assembly have also demonstrated that Sis1-mediated acceleration of [RNQ⁺] prion assembly is cytoprotective.12 Such a role for an Hsp40 co-chaperone in cytoprotection is consistent with other recent observations that facilitated protein aggregation protects against cell death mediated by Aβ(1–42)85,86 and the expanded polyglutamine huntingtin77,87,88 altogether suggesting that multiple protein quality control pathways might exist to cope with the accumulation of toxic protein conformers. Thus, dissecting how Hsp40s selectively recognize toxic protein species and recruit other chaperone complexes (such as Hsp70 and Hsp104) will yield significant insight on how aggregation pathways are regulated in human conformational disorders.

Abbreviations

Hsp	=	heat shock protein
Gln	=	glutamine
Asn	=	asparagines
G/F-rich	=	glycine/phenylaline-rich
ZFLR	=	zinc finger-like region
G/M-rich	=	glycine/methionine-rich

Figures and Tables

Figure 1 Domain structures of the Type I Hsp40 Ydj1 and Type II Hsp40 Sis1 (A). Both Hsp40s possess N-terminal J-domains and adjacent G/F-rich regions. Ydj1 also contains a zinc finger-like region with two zinc-binding domains (denoted as I and II). Sis1 contains a G/M-rich region in addition to the G/F-rich region. C-terminal domains (CTD) in these Hsp40s share limited homology although both possess hydrophobic polypeptide-binding pockets (PBP) and dimerization domains (DD). Furthermore, Ydj1 is farnesylated at a C-terminal CaaX motif. (B) Quaternary structures of Ydj1 and Sis1 homodimers (for details see ref. 42). The polypeptide-binding pockets of each Hsp40 are noted by a small grove in the C-terminus. While the J-domains are highly conserved between Ydj1 and Sis1, the orientations of these domains with respect to the polypeptide-binding pocket are very different.

Figure 2 Domain structure of Rnq1 from S. cerevisiae (A). Rnq1 contains a N-terminal non-prion domain (aa1–152) and a C-terminal, Gln/Asn-rich prion domain (aa153–405). Sis1 binds a short, hydrophobic motif in the non-prion domain while Ydj1 binds numerous motifs in prion domain. (B) Model for Hsp40 action in [RNQ⁺] prion assembly pathway. Native Rnq1 is converted to the [RNQ⁺] prion state and assembles into large, [RNQ⁺] prion aggregates that are sheared to generate heritable [RNQ⁺] prion seeds. Sis1 might facilitate shearing to maintain a pool of [RNQ⁺] prion seeds thereby propagating the [RNQ⁺] prion state. In addition Sis1 might promote the elongation of [RNQ⁺] prion particles. In contrast, Ydj1 might antagonize [RNQ⁺] prion assembly through binding an early [RNQ⁺] prion conformer thus hindering [RNQ⁺] assembly or by accelerating the disassembly of large, [RNQ⁺] prions such that Rnq1 is eventually converted to its native state.

Acknowledgements

D.W.S. is supported by a pre-doctoral training grant from the National Institutes of Health (5T32GM008581-09). P.M.D. is supported by a pre-doctoral fellowship from the American Heart Association. D.M.C. is supported by funds from the National Institutes of Health (5R01GM067785-06).