Single methyl group determines prion propagation and protein degradation activities of yeast heat shock protein (Hsp)-70 chaperones Ssa1p and Ssa2p

Abstract

Organisms encode multiple homologous heat shock protein (Hsp)-70s, which are essential protein chaperones that play the major role in cellular protein “quality control.” Although Hsp70s are functionally redundant and highly homologous, many possess distinct functions. A regulatory motif underlying such distinctions, however, is unknown. The 98% identical cytoplasmic Hsp70s Ssa1p and Ssa2p function differently with regard to propagation of yeast [URE3] prions and in the vacuolar-mediated degradation of gluconeogenesis enzymes, such as FBPase. Here, we show that the Hsp70 nucleotide binding domain (NBD) regulates these functional specificities. We find little difference in ATPase, protein refolding, and amyloid inhibiting activities of purified Ssa1p and Ssa2p, but show that interchanging NBD residue alanine 83 (Ssa1p) and glycine 83 (Ssa2p) switched functions of Ssa1p and Ssa2p in [URE3] propagation and FBPase degradation. Disrupting the degradation pathway did not affect prion propagation, however, indicating these are two distinct processes where Ssa1/2p chaperones function differently. Our results suggest that the primary evolutionary pressure for Hsp70 functional distinctions is not to specify interactions of Hsp70 with substrate, but to specify the regulation of this activity. Our data suggest a rationale for maintaining multiple Hsp70s and suggest that subtle differences among Hsp70s evolved to provide functional specificity without affecting overall enzymatic activity.

Heat shock protein (Hsp)-70 molecular chaperones are among the most highly conserved proteins. They act through an ATP-regulated cycle of binding and release of hydrophobic surfaces on incompletely folded proteins. This interaction helps other proteins adopt and maintain native conformation and allows Hsp70 to play a central role in many essential processes including protein trafficking, signaling, and degradation (1–3). Due to its versatile roles, Hsp70 is a promising therapeutic target in several human diseases such as cancer and amyloid-based neurodegenerative diseases and diabetes. Not surprisingly, there is a broad interest in understanding the molecular basis of Hsp70 action.

Hsp70 activity is regulated by communication among its amino-terminal ATPase domain (nucleotide binding domain, NBD), an adjacent substrate binding domain (SBD), and a less conserved carboxyl-terminal region (CTD), which forms a lid over the binding pocket of the SBD and interacts with a variety of cochaperones. Binding of ATP to the NBD opens the lid and allows rapid reversible binding of substrate. ATP hydrolysis induces conformational changes that close the lid, trapping substrate within the binding pocket. Release of ADP and restoration of the ATP-bound “open” state promotes substrate release. Substrate binds with high affinity to the ADP bound state, whereas ATP binding decreases this affinity by approximately two orders of magnitude.

Most organisms encode multiple Hsp70 isoforms. Humans and Saccharomyces cerevisiae each have six canonical cytosolic Hsp70s. Their functions overlap to the extent that human Hsp70 (Hsc70) can support viability of yeast lacking its essential Ssa Hsp70 subfamily (4). Similarly, exogenously expressed heterologous Hsp70 protects animals from various stresses (5, 6). Nevertheless, even highly homologous isoforms display clearly distinct functions (4, 7–11). Whereas the amplification of Hsp70 genes has long been considered advantageous for allowing regulation of spatiotemporal abundance of Hsp70 according to cellular need for its chaperone function, the ability of Hsp70 to function distinctly implies that expansion of Hsp70 genes also provides an advantage as a source of functional diversity. Subtle differences in intrinsic or cochaperone-regulated activities likely allow certain Hsp70 isoforms to perform better than others in specific tasks. Despite intensive study, however, the presence of a regulatory element within Hsp70 that can confer such specialization has not been described.

The S. cerevisiae Ssa subfamily of cytosolic Hsp70 is composed of the constitutively expressed Ssa1p and Ssa2p and the stress-inducible Ssa3p and Ssa4p (12). Amino acid identity within this family is highest between isoforms expressed under similar conditions. Constitutively expressed Ssa1p and Ssa2p are 98% identical, inducible Ssa3p and Ssa4p are 88% identical, and Ssa3/4p are 80% identical to Ssa1/2p. Ssa family Hsp70 function is essential for viability, and any of these four isoforms can support cell growth if expressed abundantly enough (13, 14).

Despite their near identity, Ssa1p and Ssa2p have clear differences in physiologic activity. They were first shown to differ in the vacuolar import and degradation (Vid) pathway, which specifically requires Ssa2p for degradation of gluconeogenic enzymes after addition of glucose to starved cells (15). Ssa Hsp70 isoforms also function distinctly with regard to yeast prions. Prions are infectious amyloid-like aggregates of misfolded cellular proteins that propagate by inducing similar misfolding of the native form of protein. Yeast [URE3] prions, which propagate as amyloid forms of Ure2p, are unaffected by depletion of Ssa1p but are lost frequently from cells lacking Ssa2p (16, 17). Also, increasing abundance of Ssa1p in wild-type cells causes them to lose [URE3], whereas increasing Ssa2p does not (18).

Here, we used this distinction in prion phenotype to identify the structural elements of Hsp70 that regulate functional differences between Ssa1p and Ssa2p. After determining that the ATPase domain was responsible, we show a single amino acid in this domain is all that is necessary for the functional distinction between Ssa1p and Ssa2p with regard to propagation of [URE3]. Remarkably, this same residue specified Ssa2p function in protein degradation mediated by the Vid pathway. This pathway was not important for [URE3] propagation, however, indicating the same minimal change in Hsp70 can specify functional differences in different processes.

Results

Functional Difference Between Ssa1p and Ssa2p Is Not Apparent in Their Biochemical Activities.

Previously, we and others showed that [URE3] propagation is inhibited by increasing Ssa1p or depleting Ssa2p, but not by increasing Ssa2p or depleting Ssa1p (16–19). We assessed functions of purified Ssa1p and Ssa2p in vitro to determine whether these differences might be due to differences in their biochemical activities. The kinetics of binding and release of substrates by Hsp70 depend upon its ATP cycle. Intrinsic ATPase activity of various Hsp70s is very low, ranging from about 0.01 to 1 ATP hydrolyzed per minute (20). The ATPase activity of both Ssa1p and Ssa2p were found to be 0.02/min, which agrees with previous results (Fig. 1A) (21, 22).

Fig. 1.

Ssa1p and Ssa2p have comparable enzymatic activities. (A) ATPase activity of Ssa1p or Ssa2p (0.4 μM) in the absence or presence of Ydj1p (1.6 μM). ATP was 50 μM. Values are averages ± SD of two to three replicate experiments. (B) Representative surface plasmon resonance experiment of physical interaction of Ssa1p or Ssa2p (concentration indicated on Right) with immobilized Ydj1p. For each assay, background resonance from control cells lacking Ydj1p was subtracted. Highly reproducible results were obtained in two additional replicate experiments on separate machines using proteins from two different preparations and five concentrations of Ssa protein. (C) Denatured firefly luciferase (0.5 μM) was incubated with Ydj1p (0.2 μM) in the presence or absence of Ssa1p or Ssa2p (2 μM) as indicated. All reactions contained 20 μM ATP. Values are averages ± SD of two replicate experiments. (D) Purified Ure2p was dialyzed from buffer containing 1 M GdmCl into buffer without GdmCl and then centrifuged at 12,000 g for 1 h immediately before adding Ure2p (18 μM) to cuvettetes with or without Hsp70 (6 μM). Reactions were stirred at 25° and 10-μL aliquots were removed and added to 490 μL of 100 μM Thioflavin-T. Fluorescence was monitored at 485 nm with excitation at 450 nm.

Low intrinsic ATPase activity provides a key step for regulation and fine-tuning of Hsp70 function by cochaperones. Ydj1p, the major yeast cytosolic Hsp40, is an extensively studied cochaperone that stimulates Ssa1p ATPase. Ydj1p stimulated ATPase activity of Ssa1p and Ssa2p roughly 10-fold and 7-fold, respectively, which agrees well with previous data (Fig. 1A) (21, 22). We tested whether this small difference might be due to a difference in interaction of Ydj1p with Ssa1p and Ssa2p by using surface plasmon resonance. Ydj1p consistently interacted with Ssa2p slightly better than with Ssa1p (Fig. 1B), suggesting that the lower stimulation of Ssa2p ATPase by Ydj1p was not due to reduced physical interaction of Ydj1p and Ssa2p.

We compared the ability of Ssa1p and Ssa2p to refold misfolded protein by measuring reactivation of firefly luciferase after denaturing it to less than 1% activity by treatment with guanidinium chloride. Adding Ydj1p restored only 10% of the original luciferase activity (Fig. 1C). Coincubating denatured luciferase and Ydj1p with a 10-fold molar excess of either Ssa1p or Ssa2p, however, resulted in nearly complete restoration of luciferase activity. The rates of refolding assisted by Ssa1p and Ssa2p were nearly identical. Therefore, these Hsp70s did not differ in functional cooperation with Ydj1p to refold denatured protein.

When purified, Ure2p spontaneously assembles into amyloid filaments, and this assembly is inhibited by addition of purified Ssa1p (23, 24). In our hands assembly of purified Ure2p into amyloid showed a sigmoidal curve with a very short lag phase of about 5 min (Fig. 1D). Preincubation of Ure2p (18 μM) with either Ssa1p or Ssa2p (6 μM) increased the lag time of Ure2p amyloid formation and decreased the yield of amyloid formed to the same degree. These data suggest that the differences in the way Ssa1p and Ssa2p influence [URE3] propagation are not due to differences in the way they influence Ure2p amyloid formation. Together our data did not uncover any differences in the enzymatic activities of Ssa1p and Ssa2p that were substantial enough to explain their functional distinction in vivo.

[URE3] Is Unstable in Cells Expressing only Ssa1p.

[URE3] propagates as insoluble amyloid aggregates of Ure2p, which is a repressor of nitrogen catabolic genes such as the allantoate transporter Dal5p. When [URE3] is present, depletion of Ure2p into prion aggregates relieves this repression. In strains with ADE2 controlled by the DAL5 promoter, expression of Ade2p can be used to monitor the presence of [URE3] (25, 26). When [URE3] is present in these strains, they express Ade2p so they grow without adenine and are white. When they are [ure-o] (i.e., lack [URE3]) they require adenine to grow and are red on media containing limiting adenine (Materials and Methods).

Amino acid identity of the Ssa1p and Ssa2p NBD, SBD, and CTD is ∼99, 99, and 93% (Fig. 2A). Hsp70 amino acids conserved across species that are essential for Hsp70 function or known to interact with cochaperones or substrates are present in both Ssa1p and Ssa2p. For example, they both contain R169, which is crucial for interaction with Hsp40 cochaperones, linker residues 382–392 between NBD and SBD, which are required for interdomain communication, residues I400, A403, L410, F425, A428, V435, and I437, which form a hydrophobic substrate-interacting core, and the C-terminal GPT(I/V)EEVD motif of eukaryotic Hsp70s, which interacts with tetratricopeptide repeat (TPR) cochaperones. It was therefore difficult to predict a priori the residues critical for Hsp70 functional differences.

Fig. 2.

ATPase domain determines specificity of Hsp70 function. (A) Linear diagram of Ssa1/2p with structural domains indicated and locations of differences in amino acid residues between Ssa1p and Ssa2p highlighted as red bars. (B–D) Cells taken from white colonies grown nonselectively on 1/2 YPD were diluted in water, spread onto 1/2 YPD plates and grown for 3 d. (B) Leftmost panels show wild-type strain 1075 [URE3] and [ure-o] colonies (WT⁺ and WT^–, respectively), which display characteristic white and red color. Degree of mitotic instability of [URE3] in strain 1161 expressing Ssa1p (111), Ssa2p (222), or Ssa1/2p hybrid proteins (as indicated) is proportional to the frequency of pigmented colonies. (C and D) [URE3] phenotype of cells expressing Ssa1p (A1) and Ssa2p (A2) with the indicated single amino acid changes. (E) Quantified frequencies of loss of [URE3]. Cells taken from white colonies expressing the indicated Hsp70 were grown on 1/2 YPD, diluted in water, and spread onto 1/2 YPD plates. The percentage (±SD) of red [ure-o] colonies measured among a total of 900–1,500 colonies is shown.

Growth of strain 1161, which lacks all four chromosomal SSA family genes (SSA1–SSA4), is supported equally well by Ssa1p or Ssa2p expressed from a plasmid (17), indicating that all essential Hsp70 functions of Ssa1p and Ssa2p are completely interchangeable. However, whereas [URE3] propagates stably in cells expressing only Ssa2p, it is unstable in cells expressing only Ssa1p (17). This instability is seen as the appearance of red [ure-o] colonies arising among a population of cells spread on a plate with limiting adenine (Fig. 2B, compare 111 with 222). Phenotypic differences between the cells expressing only Ssa1p or Ssa2p are not explained by differences in abundance of Ssa protein or the major Hsp70 cochaperones known to influence [URE3] propagation (Ydj1p, Sse1p, and Sis1p; Fig. S1A). We used the readily distinguishable prion phenotypes to determine the molecular basis of the functional differences between Ssa1p and Ssa2p.

ATPase Domain Determines Prion Stability.

We first examined the regulatory effect of each Ssa functional region by swapping the major Hsp70 structural domains. We constructed six Ssa1p–Ssa2p hybrid proteins representing all of the possible combinations of the three domains described above. Names of wild-type and hybrid proteins reflect their domain composition. For example, 111 is Ssa1p and 122 has the NBD of Ssa1p and the SBD and CTD of Ssa2p. All hybrids (and point mutants, see below) used here were expressed at a similar level (Fig. S1 B–D) and supported growth of cells like wild-type Ssa1p and Ssa2p (generation times of 104 ± 5 (SD) minutes in YPAD liquid medium at 30 °C), indicating that they function like the wild-type proteins with regard to essential Ssa functions.

Half of the hybrid proteins (122, 121, and 112) produced an Ssa1p-like unstable [URE3] phenotype (Fig. 2B). In contrast, cells expressing the remaining hybrids (211, 212, and 221) propagated [URE3] stably, like those expressing Ssa2p. Quantified frequencies of [URE3] loss are shown in Fig. 2E. Conspicuously, all hybrids that impaired [URE3] propagation had the NBD of Ssa1p and all those that supported stable [URE3] propagation had the NBD of Ssa2p. These results show that the ATPase domains of Ssa1p and Ssa2p were enough to confer the specific Hsp70 isoform character regarding prion stability.

Glycine Residue 83 Determines Ssa2p-Specific Hsp70 Function.

The NBDs of Ssa1p and Ssa2p differ at only four residues: 20 (A/S), 41 (A/G), 62 (S/A), and 83 (A/G) (Ssa1p residue listed first), and each is a relatively minor difference in structure (Fig. 2A). To determine the minimal difference that conferred the functional distinction, we began by switching individual residues of Ssa2p to the amino acid found in Ssa1p. Switching residues 20, 41, or 62 did not influence [URE3] stability (Fig. 2C). Remarkably, however, cells expressing Ssa2p with the G83A substitution (Ssa2^G83A) displayed an unstable Ssa1p-like [URE3] phenotype (Fig. 2 C and E). Thus, Ssa2^G83A behaves like Ssa1p even though it is identical to Ssa2p except in having a methyl group in place of the hydrogen as the side chain of residue 83.

We then switched glycine 83 to the bulkier residues leucine or phenylalanine to confirm the crucial role of glycine at this location. Both alterations similarly reduced [URE3] stability like G83A (Fig. 2C). To determine whether a glycine at residue 83 was enough to promote stable [URE3] propagation, we made the reciprocal A83G substitution in Ssa1p (Ssa1^A83G). Although Ssa1^A83G is identical to Ssa1p except for residue 83, cells expressing it as the only source of Ssa protein had a stable Ssa2p-like [URE3] phenotype. These data demonstrate that glycine residue 83 determines an Ssa2p-specific function important for [URE3] prion propagation.

Structural information for Ssa Hsp70 is not known. Hsp70 chaperones are highly conserved, however, and structural information is available for Hsp70 proteins from many species. Using a Web-based program (http://swissmodel.expasy.org/), we modeled Ssa1p structure. Residue 83 lies in a short α-helix in the ATPase domain (residues 80–88), which is conserved among Hsp70s whose structure is known (Fig. S2). None of these Hsp70s has a glycine at this homologous position. Because glycine is unfavorable for helix stability, Ssa2p might have an alternative structure due to the presence of G83. We tested whether making substitutions Q82G or P, and M85G or P in Ssa1p, which we expected would similarly disrupt a putative helix, would restore [URE3] stability like A83G. We also tested a deletion of residue 83 in both Ssa1p and Ssa2p (A83Δ and G83Δ, respectively) to determine the importance of this residue directly. None of these changes stabilized [URE3] like A83G, and most of them reduced [URE3] stability more than G83A (Fig. 2D). Moreover, the prion phenotypes of cells expressing Ssa1p or Ssa2p lacking residue 83 were indistinguishable, indicating the other nonconserved residues do not contribute functional specificity of Ssa1p and Ssa2p regarding [URE3] propagation. Therefore, G83 seems to provide a unique sequence element important for Ssa2p-specific Hsp70 activity.

Overexpressing Ssa1p, but not Ssa2p, cures cells of [URE3] (18). We tested whether interchanging residue 83 would also switch this specificity and found it did not (Fig. S3). This result shows that Hsp70 activities important for prion propagation differ from those important for prion elimination.

We also tested whether the minor G/A83 difference distinguishes Ssa1p and Ssa2p in another cellular process known to depend strictly on Ssa2p. Starving cells of glucose induces expression of gluconeogenic enzymes, such as fructose-1,6-bisphosphatase (FBPase), and restoring glucose after 3 d causes degradation of these enzymes through a vacuolar import and degradation (Vid) pathway (15). FBPase was degraded similarly in wild-type cells and in cells expressing only Ssa2p, but it was stabilized in cells expressing only Ssa1p (Fig. 3A). These results confirmed the specific requirement of Ssa2p for Vid pathway function. Amazingly, FBPase was stabilized in cells expressing Ssa2^G83A and was degraded in cells expressing Ssa1^A83G. Therefore, this minimal difference in structure also switched the roles of Ssa1p and Ssa2p in Vid pathway function.

Fig. 3.

Residue 83 specifies Hsp70 function in the Vid pathway. (A) Cultures of strain 1161 [ure-o] cells were starved of glucose for 3 d and then split in half. Glucose was added to one half where indicated and after 2 h the abundance of FBPase was monitored by Western analysis. Numbers 1 and 2 designate Ssa1p and Ssa2p, respectively; WT is wild type (strain 1075). Similar results were obtained with [URE3] cells. (B) The experiment was repeated using an isogenic sec28Δ strain. (C) [URE3] phenotypes of wild-type and sec28Δ cells grown on 1/2 YPD plates for 3 d at 30 °C.

Because these findings suggested a link between [URE3] propagation and Vid pathway function, we tested whether deleting SEC28 reduced [URE3] stability. Sec28p is a coatomer complex subunit that regulates Golgi to ER trafficking and facilitates import of Vid cargo into Vid vesicles (27). Deleting SEC28 stabilized FBPase (Fig. 3B), indicating it disrupted the Vid pathway, but it did not affect [URE3] stability (Fig. 3C). Thus, stable [URE3] propagation did not require a functional Vid pathway, which was somewhat unexpected because we supposed the minimal alanine/glycine difference might have a unique and limited effect on Hsp70 activity. Therefore, residue 83 appears to have differentiated to specify a functional distinction that is important for defined yet unrelated Hsp70 tasks in the cell.

Discussion

We identify a minimal amino acid change as a basis for the functional distinction between highly homologous Hsp70 family chaperones, a phenomenon that is conserved throughout evolution. This adaptability of the ATPase domain to regulate specificity of Hsp70 function in different cellular processes widens its versatility beyond a mere structural domain that enzymatically influences substrate binding. Our findings demonstrate that exquisitely defined variations of Hsp70 structure can strongly influence the actions of Hsp70 in specific and complex physiological processes. They also show that we could successfully switch functional specificity between Ssa1p and Ssa2p by coupling protein engineering with use of yeast prions to monitor Hsp70 activity, which shows potential for designing Hsp70 to perform in desired ways.

The major Hsp70 activities in the cell depend on the same ATP-regulated cycle of binding and release of exposed hydrophobic surfaces on partially folded proteins. The necessity of possessing the same basic structural information to carry out this important reaction is evident in the conservation of both secondary and tertiary structures of even widely diverged Hsp70s. Gross structural information obtained by recombining domains of Hsp70 has shown that although the SBD can influence intrinsic activity of the adjacent NBD (28), the SBD, with its broad substrate specificity, can be interchanged without destroying functions associated with divergent Hsp70s (4, 7). The NBD, however, which regulates substrate binding by the SBD, is much more important for determining specific activities of different Hsp70s in vivo. For example, a majority of Hsp70 alterations that affect yeast prion propagation are in the NBD, and the capacity of Hsp70 to support yeast viability is governed primarily by the NBD (4, 29–31). These observations suggest that the regulation of Hsp70 activity, and not substrate binding per se, is the critical parameter for determining specificity of function.

In line with this conclusion, we now show that a minimal Gly/Ala difference in the NBD is enough to interconvert specificity of function of the Ssa1p and Ssa2p cytoplasmic Hsp70s in two independent cellular processes. Additionally, deleting this residue from Ssa1p and Ssa2p generated proteins that were functionally equivalent, yet behaved differently than either Ssa1p or Ssa2p. Because Hsp70s with much less homology still overlap functionally, it is a bit surprising that such nearly identical proteins act so differently and rather astonishing that such a small change is responsible for the distinction in function. Despite the known regulatory activity of the NBD, this finding was unexpected because the NBDs of Ssa1p and Ssa2p differ by only a few minimally different and spatially separated residues. We anticipated the C-terminal region was more likely to provide the specificity because it is the most divergent region among Hsp70s, including Ssa1p and Ssa2p, and it possesses elements that interact with a variety of cochaperones that regulate Hsp70 function. Our finding that residue 83 does not influence specificity of overexpressed Ssa1p and Ssa2p in curing cells of [URE3] indicates that these other amino acid differences do, in fact, contribute additional regulatory specificity.

The major human Hsp70s HspA1A and HspA1B differ by only two amino acids, one of which resides in the NBD, and these proteins are believed to be fully interchangeable (10). Our findings raise the alternative possibility that their differences determine specific functions in as yet unidentified tasks. They also raise concerns about commonly performed experiments using exogenously expressed Hsp70s. As mentioned already, the 98% identical Ssa1p and Ssa2p differ in their ability to cure yeast of prions when overexpressed, and similar differences in effects using different Hsp70s might be expected in other model systems.

There is strong allosteric coordination among the structural domains of Hsp70. Although substrate binds at the SBD, the kinetics of binding and release of substrate are regulated primarily through ATP hydrolysis and nucleotide exchange by the NBD. We have not ruled out the possibility that an intrinsic Hsp70 activity is altered, but among a large cohort of Hsp70 cochaperones many interact at the NBD to regulate these processes. Although residue 83 is not known to interact directly with cochaperones, the subtle Gly/Ala difference might influence the efficiency by which they regulate the ATP reaction cycle. Residue 83 lies in NBD lobe IB, which is near the nucleotide-binding cleft (Fig. S2). A structural implication of the Gly/Ala difference is that it could affect the movement of this lobe upon interaction with nucleotide exchange factors (NEFs) (32), possibly altering the way different NEFs are able to induce a structural transition that facilitates release of phosphate or ADP.

The Gly/Ala difference might also influence the efficiency of Hsp70 interaction with major chaperone machinery components such as Hsp104 and Hsp90 or other nonchaperone factors involved in specific processes that require a particular Hsp70 for the job. Alternatively, a modest change in the kinetics of either specific steps or the overall rate of the Hsp70 reaction cycle might have little affect on general functions of Hsp70 but be of importance for certain tasks.

In addition to its defined functions in essential housekeeping processes, Hsp70 protects a large number of proteins that are prone to aggregation under various kinds of stresses. During evolution there would have been a clear advantage to possess different Hsp70s that allow interaction with many protein sequence elements while maintaining specificity for performing certain tasks. Our findings suggest that the NBD has been evolutionarily optimized for conferring such functional specificity. It is likely that the SBD is more optimized for interacting with many proteins that collectively have high diversity in amino acid sequence. The unique combination of functional specificity by the NBD and broader range to bind various substrates at the SBD allows Hsp70 to perform a wide variety of cellular tasks not known for any other cellular protein. The genomic amplification and differentiation of Hsp70 genes magnifies and diversifies these benefits. The observation that such a minimal tweaking of structure determines major distinctions of function points to Hsp70 as a powerfully malleable object of selection.

Materials and Methods

Yeast Strains and Plasmids.

Strain 1075 is MATa, kar1-1, SUQ5, P_DAL5::ADE2, his3Δ202, leu2Δ1, trp1Δ63, ura3-52 (17). Strain 1161 has in addition ssa1::KanMX, ssa2::HIS3, ssa3::TRP1, ssa4::ura3-2f and carries plasmid pJ401 (see below). Strain 1366 is 1075 with sec28::KanMX. Strain 628–4B (18), used for monitoring curing of [URE3], is MATα, kar1-1, SUQ5, ade2-1, his3Δ202, ura2. Plasmid pJ401 is a URA3-based plasmid with SSA2. Plasmid pC210 is a single-copy LEU2-based plasmid containing SSA1 with an NdeI site at the initiator ATG codon, under control of the SSA2 promoter (18). Plasmid p111 is pC210 with XbaI and XhoI sites created without altering amino acid coding information at SSA1 codons 391 and 540, respectively (4). Plasmid p222 is p111 with the coding region of SSA1 replaced by that of SSA2. Plasmids with SSA1/SSA2 hybrid genes were constructed by replacing the NdeI–XbaI (NBD), XbaI–XhoI (SBD), and XhoI–SphI (CTD) fragments of p111 with PCR-amplified fragments of SSA2. Substitutions of single amino acids were done using the Stratagene Quick-Change kit. All HSP70 alleles were verified by DNA sequencing. Plasmids pA1-83G and pA2-83A are pC210 encoding SSA1^A83G and SSA2^G83A, respectively. These alleles were cloned into pRS313 (single-copy HIS3 vector) to generate pRS313A1-83G and pRS313A2-83A for monitoring curing of [URE3].

To assess effects of hybrid Hsp70 proteins, plasmid pJ401 in strain 1161 was replaced by the LEU2-based plasmids carrying hybrid alleles. The only Ssa protein in these cells is the one on the plasmid. Because Ssa protein is essential, these cells must retain the plasmid to grow, even on rich media.

Media, Growth Conditions, and Monitoring Prions.

Media and growth conditions were as described (29) or as indicated in Results. [URE3] was monitored by using the wild-type ADE2 gene regulated by the DAL5 promoter (P_DAL5::ADE2) (25, 26). On standard ammonium-containing media Ure2p represses expression of the DAL5 promoter so Ade2p is not produced and cells are Ade^– and red. When [URE3] is present the depletion of functional Ure2p into insoluble prion aggregates activates the DAL5 promoter and ADE2 expression, making cells Ade⁺ and white. Weakened [URE3] propagation can be seen as intermediate colony color (pink) or increased frequency of appearance of [ure-o] cells in a population. The presence of [URE3] was confirmed by its dominant infectious phenotype and ability to be cured when cells are grown in the presence of 3 mM guanidine hydrochloride, which inactivates Hsp104 (33–37).

Protein Purification.

Plasmid pPROEX-HTb-SSA1 for expressing His6–Ssa1 fusion protein was obtained from Addgene (38). The plasmid encodes from 5′ to 3′ direction hexa-His-tag, TEV cleavage site, and Ssa1p. Protein was purified from Escherichia coli strain Rosetta 2(DE3) (Invitrogen) grown in LB medium with 300 mM NaCl at 30 °C until OD₆₀₀ = 0.6–0.8. The culture was induced with 0.5 mM IPTG and grown overnight at 18 °C. Cells were harvested and lysed in buffer A (20 mM Hepes, 150 mM NaCl, 20 mM MgCl₂, mM KCl 20, complete protease inhibitor mixture) using lysozyme. Cell debris was cleared by centrifugation (12,000 g) and the supernatant loaded on cobalt-based Talon metal affinity resin. After washing, protein was eluted with buffer A containing 250 mM imidazole. Purified protein was incubated with His-TEV (Invitrogen; catalog no. 12575-023) protease at 30° for 1 h. The sample was extensively dialyzed at 4° and again passed through Talon metal affinity resin to remove the cleaved His tag and His-TEV protease. Protein purity was more than 99% as determined by SDS/PAGE and Coomassie staining. Protein identity was confirmed using mass spectrometry.

Plasmids expressing Ssa2p (pPROEX-HTb-SSA2) or Ydj1p (pPROEX-HTb-YDJ1) were obtained by replacing SSA1 of plasmid pPROEX-HTb-SSA1 using standard molecular biology techniques. The Ure2p expression vector (pKT41-Ure2) was a gift from Dr. Reed Wickner (National Institutes of Health, Bethesda, MD) (23). The vector encodes Ure2p with an N-terminal hexa-His tag. All proteins were purified using Talon metal affinity resin. Except for Ure2p, the polyhistidine tag was removed before assaying proteins for their activity.

ATPase Assay.

ATP hydrolysis was measured essentially as described (22). Briefly, ATPase activity of 0.4 μM Ssa1p or Ssa2p together with 1.6 μM Ydj1 was measured in 200 μL assay buffer (25 mM Tris-Cl pH 7.4, 100 mM KCl, 5 mM MgCl₂, 1 mM DTT) at 25 °C. Reactions were incubated for 10 min and the assay started by addition of [γ-³²P] ATP to a final concentration of 50 μM. Aliquots were removed at 10, 20, and 30 min and the amount of ATP hydrolysis determined as described. ATP turnover rates were linear over the course of the assay and were highly reproducible over multiple separate protein preparations.

Luciferase Refolding.

Promega (catalog no. E1701) recombinant firefly luciferase (0.5 μM) was denatured in 50 μL of denaturation buffer (6 M GdmCl, 30 mM Hepes, 50 mM KCl, 5 mM MgCl₂, 5 mM β-mecaptoethanol, pH 7.3) for 1 h at 30 °C. Luciferase was diluted in two sequential steps. First, 75 μL of refolding buffer (denaturation buffer without GdmCl and β-mecaptoethanol) was added and the reaction incubated at room temperature for 1 min. Second, 1 μL of the diluted solution was added to 50 μL refolding buffer containing 0.2 uM Ydj1p with or without 2μM Ssa1p or Ssa2p and further incubated at room temperature for 5 min. Luciferase refolding was initiated by adding 5 mM ATP. Aliquots were removed at the indicated intervals and luciferase refolding was recorded using a Zylux FB15 luminometer and luciferase assay reagent (Promega; catalog no. E1501). Spontaneous refolding of luciferase without added chaperones was subtracted as background.

Ure2p Amyloid Formation.

Purified Ure2p (120 μL of 30 μM stock) freshly dialyzed in PBS was added to 76 μL of PBS reaction buffer. Final concentrations were 1 mM ATP, 20 mM MgCl₂, 1 mM KCl, 18 μM Ure2p and 6 μM Ssa1p or Ssa2p. Samples were incubated with stirring in a 500-μL cuvettete. At indicated times, 10 μL samples were removed and added to 490 μL of 100 μM Thioflavin-T. Fluorescence was recorded at 485 nm with excitation at 450 nm.

Western Analysis.

Yeast were harvested by centrifugation and suspended in lysis buffer (PBS with 0.2% Triton X-100 and protease inhibitor mixture; Roche). Cells were broken by vortexing with glass beads, and cell lysate was centrifuged at 3,000 rpm for 2 min. Supernatant was boiled in loading dye for 15 min at 95 °C and 5 μg of protein was separated in 4–12% gradient SDS gels and then transferred onto PVDF membranes. Hsp70 antibodies were from Stressgen (catalog no. SPA-822), FBPase antibodies were a gift from C. Randell Brown (Pennsylvania State College of Medicine, Hershey, PA) (27), and Sec28p antibodies were a gift from R. Duden (University of Lübeck, Lübeck, Germany) (39).