Saccharomyces cerevisiae Hsp70 Mutations Affect [PSI+] Prion Propagation and Cell Growth Differently and Implicate Hsp40 and Tetratricopeptide Repeat Cochaperones in Impairment of [PSI+]
Gary W. Jones, Daniel C. Masison


We previously described an Hsp70 mutant (Ssa1-21p), altered in a conserved residue (L483W), that dominantly impairs yeast [PSI+] prion propagation without affecting growth. We generated new SSA1 mutations that impaired [PSI+] propagation and second-site mutations in SSA1-21 that restored normal propagation. Effects of mutations on growth did not correlate with [PSI+] phenotype, revealing differences in Hsp70 function required for growth and [PSI+] propagation and suggesting that Hsp70 interacts differently with [PSI+] prion aggregates than with other cellular substrates. Complementary suppression of altered activity between forward and suppressing mutations suggests that mutations that impair [PSI+] affect a similar Hsp70 function and that suppressing mutations similarly overcome this effect. All new mutations that impaired [PSI+] propagation were located in the ATPase domain. Locations and homology of several suppressing substitutions suggest that they weaken Hsp70's substrate-trapping conformation, implying that impairment of [PSI+] by forward mutations is due to altered ability of the ATPase domain to regulate substrate binding. Other suppressing mutations are in residues important for interactions with Hsp40 or TPR-containing cochaperones, suggesting that such interactions are necessary for the impairment of [PSI+] propagation caused by mutant Ssa1p.

Hsp70 is a highly conserved essential protein chaperone that assists protein folding in general and protects cells from stress by preventing aggregation of stress-denatured proteins. Hsp70 also aids translocation of polypeptides across membranes, such as secreted proteins into the endoplasmic reticulum and enzymes targeted for degradation into the vacuole (Beckeret al. 1996; Brownet al. 2000). Additionally, Hsp70 plays an important role in the dynamics of macromolecular assemblies such as microtubules and clathrin coats (Schlossmanet al. 1984; Okaet al. 1998). Hsp70 acts in these diverse roles through regulated binding and release of exposed hydrophobic surfaces of partially folded proteins (see Gething and Sambrook 1992 for review).

Hsp70 has conserved ATPase and substrate-binding domains, linked by a short stretch of hydrophobic amino acids, and a less conserved carboxy-terminal domain. The amino-terminal ATPase domain has a deep cleft between two lobes in which ATP binds (Flahertyet al. 1990). The adjacent substrate-binding domain consists of a β-sheet rich substrate-binding cavity and an α-helical region that forms a lid over this cavity (Zhuet al. 1996). Structures of the ATPase and substrate-binding domains were determined independently so their orientation with respect to each other is unknown. However, it is believed that ATP hydrolysis induces a conformational change in the protein that closes the lid, trapping substrate within the substrate-binding cavity and prolonging the substrate-bound state (Zhuet al. 1996; Bukau and Horwich 1998; Mayeret al. 2000b). Additionally, binding of substrate, which causes a conformational alteration of the substrate-binding domain, stimulates ATP hydrolysis in proportion to its binding affinity (Mayeret al. 2000b; Pellecchiaet al. 2000). This communication between the two domains coordinates substrate binding and ATP hydrolysis, which allows the proper binding and release of substrates that is necessary for optimal Hsp70 function. Details of the molecular mechanisms underlying this interdomain communication remain unclear.

Hsp70 function is also affected by interactions with cochaperones. Hsp40 interacts physically with Hsp70 and stimulates ATP hydrolysis (Cyret al. 1992). Hsp40 also binds to unfolded proteins and can present substrates to Hsp70 (Laufenet al. 1999). In mammals, Hop1 (Hsp organizing protein) interacts with Hsp70 and can simultaneously bind to Hsp90, forming a bridge between these two protein chaperones (Scheufleret al. 2000). Formation of this complex, which is conserved in Saccharomyces cerevisiae (Chang and Lindquist 1994), is mediated by the tetratricopeptide repeat (TPR) domains of Hop1. This tripartite interaction facilitates transfer of substrates from Hsp70 to Hsp90 and is important for proper folding maturation of Hsp90 substrates, many of which are involved in signaling. This maturation pathway can be initiated by presentation of substrate to Hsp70 by Hsp40 (Hernandezet al. 2002). Although physical interactions of Hop1 with Hsp70 and Hsp90 are well defined (Scheufleret al. 2000; Van Der Spuyet al. 2000; Brinkeret al. 2002), specific effects of this interaction on in vivo Hsp70 function are not.

S. cerevisiae cytosolic Hsp70's include the SSA and SSB (stress seventy subclass A and B) families. The two Ssb proteins associate with translating ribosomes and provide an important but nonessential function, possibly aiding translocation of proteins through the ribosome (Nelsonet al. 1992; Pfundet al. 1998). The SSA family includes four homologous and functionally redundant proteins (Ssa1p–4p), of which at least one is required for growth (Werner-Washburneet al. 1987; Boorsteinet al. 1994). Ssa1p and Ssa2p are constitutively expressed and 97% identical. In addition to its role in protein folding, Ssa1p negatively regulates its own promoter and those of other stress response proteins (Stone and Craig 1990). Because of this regulation, under normal growth conditions there is severalfold less Ssa1p than Ssa2p. Stress induces Ssa1p expression to levels similar to those of Ssa2p. SSA3 and SSA4 are expressed only under conditions of stress and in stationary phase and during sporulation (Werner-Washburneet al. 1987).

Yeast prions are cellular proteins that misfold and form self-replicating aggregates, thought to be amyloid in nature (Wickner 1994; Patinoet al. 1996; Gloveret al. 1997; Kushnirov and Ter-Avanesyan 1998; Edskeset al. 1999; Tayloret al. 1999). Prions propagate by recruiting the soluble form of the protein into the aggregates, which are transmitted between cells during cell division and cell fusion. The [PSI+] genetic element is a prion of Sup35p (eRF3), a component of the translation release factor (Wickner 1994; Stansfieldet al. 1995; Zhouravlevaet al. 1995). Aggregation of Sup35p in [PSI+] cells reduces the amount of functional Sup35p, which decreases the efficiency of translation termination and thus causes nonsense suppression.

Yeast prion propagation is affected by the activities of different protein chaperones. Overexpressing Hsp104, a protein chaperone required for propagation of yeast prions, causes loss of [PSI+] but not other prions (Chernoffet al. 1995; Derkatchet al. 1997; Moriyamaet al. 2000). Conversely, overexpressing Ssa1p causes loss of the [URE3] prion but does not cure [PSI+] (Schwimmer and Masison 2002). Excess Ssa1p also moderates the [PSI+]-curative effect of overexpressed Hsp104 (Newnamet al. 1999) and, depending upon Hsp40 abundance, can destabilize some prions (Kushnirovet al. 2000). Sis1p, an essential Hsp40, is necessary for propagation of the yeast [PIN+]/[RNQ1+] prion (Derkatchet al. 1997; Sondheimeret al. 2001) and overexpression of Sti1p, the yeast Hop1 homolog, was found to inhibit propagation of a hybrid form of [PSI+](Kryndushkinet al. 2002).

We previously described an SSA1 mutant (SSA1-21) that considerably impairs [PSI+] propagation (Junget al. 2000). SSA1-21 cells lacking Ssa2p cannot maintain [PSI+] at all. Despite these dramatic effects on [PSI+], the mutation in SSA1-21 (L483W) has little affect on growth, even when the mutant protein is the only Ssap in the cell. To better understand the nature of the altered Hsp70 function caused by L483W, we generated both new mutants of Ssa1p that impaired [PSI+] propagation and second-site suppressors of SSA1-21. We identified several alleles of both classes that shed light on Hsp70 interactions with [PSI+] and that implicate cochaperone involvement as an important contributing factor in this interaction.


Strains, plasmids, media, and genetic methods: Strains 628-3A (MATα kar1-1 SUQ5 ade2-1 leu2Δ1 trp1Δ63 ura3-52 [PSI+]; Junget al. 2000), JA1NC (MATα kar1-1 SUQ5 ade2-1 his3Δ202 leu2Δ1 trp1Δ63 ura3-52 ssa1::KanMX [PSI+]; this study), and G400-1C (MATa kar1-1 SUQ5 ade2-1 his3 leu2 lys2 trp1 ura3 ssa1:: KanMX ssa2::HIS3 ssa3::TRP1 ssa4::ura3-1f/pRDW10 [PSI+]; this study) were used. They are related but not isogenic. JA1NC was constructed by replacing chromosomal SSA1 of strain 779-6A (Jung and Masison 2001) with KanMX (Wachet al. 1994). The ura3-1f allele in G400-1C was obtained by selection on 5-fluoroorotic acid (5-FOA; Boekeet al. 1984). Because they depend on a plasmid-borne SSA1 gene for growth, G400-1C transformants can be grown in the absence of selection for the plasmid. Single-copy plasmids pRDW10 and pJ120 are YCp50 (URA3; Roseet al. 1987) and pRS315 (LEU2; Sikorski and Hieter 1989), respectively, carrying SSA1 with 500 bp of 5′ and 3′ flanking DNA on a BamHI-HindIII fragment. Plasmid pJ121 is pJ120 with the L483W substitution in SSA1. Remaining plasmids were derived from pJ120 (pG1 series) or pJ121 (pG21 series) and are designated by their SSA1 alterations. For example, pG1-34K and pG21-34K are pJ120 and pJ121, respectively, that contain the R34K substitution. Media and genetic methods were as described (Guthrie and Fink 1991; Junget al. 2000).

[PSI+] is a self-replicating aggregated form of the translation termination factor Sup35p (eRF3). Aggregation of Sup35p in [PSI+] cells causes nonsense suppression by reducing the amount of functional Sup35p. Nonsuppressed ade2-1 mutants are Ade and are red when grown on limiting amounts of adenine because of the accumulation of a pigmented substrate of Ade2p. Partial suppression of ade2-1 by [PSI+] allows growth without adenine and eliminates the pigmentation (Cox 1965). Weakening of [PSI+] propagation by Ssa1-21p increases the relative level of soluble Sup35p, which reduces nonsense suppression and causes cells to accumulate pigment (Junget al. 2000). These effects increase as temperature increases. Thus, wild-type [PSI+] cells are white and Ade+ when grown at 25° or 30°, and SSA1-21 [PSI+] cells are slightly pink and weakly Ade+ when grown at 25° and are darker pink and Ade at 30°. Strains with SSA1-21 as the only SSA gene grow well but cannot propagate [PSI+] at any temperature (Junget al. 2000).

For growth rate comparisons, [psi] G400-1C transformants were grown on YPAD plates at various temperatures and rate of colony formation was scored. All strains were first cured using guanidine to eliminate [PSI+] and any other known prions that may have been present. To ensure that [PSI+] was lost and the effects on suppression were not direct effects of the SSA1 alleles, similarly cured [psi] JA1NC transformants were grown nonselectively to allow plasmid loss. All strains lacking the plasmids retained the [psi] phenotype. None of the mutations had any overt effect on nonsense suppression in the absence of [PSI+].

Curing of [PSI+] by guanidine: For nonquantitative curing, cells were grown to colonies twice on YPD/G3. Quantitative curing assays were done as described (Junget al. 2000). Briefly, guanidine hydrochloride was added at a final concentration of 3 mm to log phase cultures of G400-1C transformants in liquid YPAD at 30°. Cultures were maintained in log phase by repeated dilution into fresh guanidine-containing medium. [PSI+] was monitored by transferring 300–500 cells onto duplicate YPD plates and scoring for white ([PSI+]) or red ([psi]) color of resulting colonies. Colonies with red and white sectors were scored as [PSI+].

Mutagenesis: Plasmids pJ120 and pJ121 were randomly mutagenized using hydroxylamine (Schatzet al. 1988). Ninetyminute treatments resulted in mutation frequencies of 4.2% for pJ120 and 6.6% for pJ121. Site-directed mutagenesis of pJ120 using the Quickchange kit (Stratagene, Burlingame, CA) and appropriate mismatched primers was done to construct remaining SSA1 alleles.

Isolation of SSA1 mutants that impair [PSI+] propagation: When expressed from a single-copy plasmid in a wild-type strain grown at 30°, SSA1-21 causes normally white [PSI+] cells to become pink. Among ∼3000 628-3A transformants of mutagenized pJ120, 10 caused such dominant accumulation of red pigment. Plasmids recovered from 7 of these reproduced the impaired [PSI+] phenotype upon transformation of G400-1C, which has SSA1 on a plasmid (pRDW10) as the only SSA gene. In a separate screen of 628-3A cells transformed by mutagenized pJ121, two second-site mutations that enhanced the pigment accumulation caused by SSA1-21 were identified.

Isolation of second-site suppressor mutations of SSA1-21: Strain G400-1C was transformed with mutagenized pJ121 and selection plates with 300–500 transformant colonies were replica plated onto plates lacking adenine and containing 5-FOA. FOA plates, which select for growth of cells that have lost pRDW10, were then incubated at 30°. These conditions select for mutant alleles that simultaneously restore [PSI+] propagation and provide essential Hsp70 activity. Among ∼3000 transformants screened, 29 grew. Plasmids from all 29 conferred the same restored [PSI+] phenotype upon retransformation. Among these plasmids, 10 new alleles and a direct revertant of L483W were identified.

A second screen was performed by selecting for loss of the dominant SSA1-21 phenotype. Strain 628-3A was transformed by mutagenized pJ121 and white (rather than pink) transformants were selected at 30° on medium lacking leucine with limiting adenine. From ∼3000 transformants, 11 plasmids, among which four new SSA1 alleles were identified, were isolated.


New SSA1 mutations that impair [PSI+] propagation: The HSP70 allele SSA1-21, which has a substitution of a conserved residue in the substrate-binding domain (L483W), dominantly impairs [PSI+] propagation leading to frequent mitotic loss of [PSI+] and accumulation of red pigment in [PSI+] cells. To identify new HSP70 mutations that impaired [PSI+] propagation we randomly mutagenized SSA1 and isolated alleles that caused a similar pink phenotype when expressed from plasmids in wild-type [PSI+] cells (see materials and methods). We also mutagenized SSA1-21 and selected alleles with second-site mutations that showed a further increase in pigmentation, which reflects enhanced ability of SSA-21 to impair [PSI+]. To determine if mutant proteins could support cell growth, strain G400-1C, which lacks all four chromosomal SSA genes and carries wild-type SSA1 on a plasmid (pRDW10), was first transformed by plasmids carrying the mutant alleles. These transformants were then tested for ability to grow on medium selecting against pRDW10. The extent to which the mutant Ssa proteins provide essential Hsp70 function is reflected in the rate of growth of cells having only the mutagenized plasmid.

Table 1 lists eight new forward mutations and their relative ability to impair [PSI+] and to support growth at optimal (30°) and elevated (37°) temperature. Assays were done using the mutant allele in place of SSA1 and, if the mutant protein supported cell growth, when it was the only Ssa protein present. Most forward mutations were isolated only once and we did not isolate L483W, indicating that the screen was not saturated and that there are likely to be other mutations in SSA1 that compromise [PSI+] propagation. Like L483W, all caused pigment accumulation, and [PSI+] cells expressing these alleles were unable to form colonies on plates lacking adenine at 30° (Figure 1A; data not shown). Unlike L483W, all new forward mutations were in the ATPase domain (residues 1–386), indicating that the activity of this domain is more sensitive than that of the substrate-binding domain (residues 395–599) to perturbations that affect [PSI+].

The effects of the A17V and R23H substitutions were identical to those of L483W. When expressed in place of Ssa1p these mutations caused frequent mitotic loss of [PSI+] as well as increased pigment accumulation. When expressed as the sole Ssap they supported cell growth as well as wild-type Ssa1p but failed to support [PSI+] propagation (Table 1; Figure 1B; data not shown). Thus, alteration of residues located in different domains of Ssa1p caused the same dramatic impairment of [PSI+] propagation without significantly affecting cell growth.

Proteins with the G32D, G32S, R34K, and C303Y substitutions weakened [PSI+] propagation but, unlike L483W, supported [PSI+] propagation when expressed as the sole Ssa protein (Table 1; Figure 1B; data not shown). Therefore, mutant Ssa1 proteins that remained capable of supporting [PSI+] could have a dominant negative effect on [PSI+] propagation in the presence of the severalfold more abundant Ssa2 protein. The G32D and C303Y mutant proteins were reduced in the capacity to support growth, indicating that these residues were important for both [PSI+] propagation and cell growth (Table 1). Interestingly, nonsense suppression caused by [PSI+], reflected as reduced pigment, was stronger when Ssa1p with the G32D substitution was the only Ssa protein than when it was expressed in place of Ssa1p (Figure 1, A and B). This mutation thus had a greater effect on [PSI+] when Ssa2p was present, suggesting that its effects may be indirect.

View this table:

Relative effects of SSA1 mutations on cell growth and [PSI+] propagation

We were unable to isolate cells expressing Ssa1p with the G336D substitution as the sole Ssa protein, which indicates that this mutant protein was unable to support growth. Thus, ability to provide essential Hsp70 function was not a requirement for mutant Ssa1p to dominantly impair [PSI+]. Moreover, although the forward mutations were selected to impair [PSI+] dominantly, only C303Y had a noticeable dominant negative effect on growth (data not shown), and most forward mutations supported growth at elevated temperature as well as wild type. These results reveal a distinction between ability of the mutant proteins to interfere with [PSI+] propagation and with cell growth or stress protection, suggesting that the effect of the mutations on Ssa1p activity was specific to [PSI+] propagation. It is also possible that the dominant effects on [PSI+] were due to alteration of the activity of a complex of which Ssa1p is a component. If so, then [PSI+] propagation was more sensitive to such disturbance than were essential processes that require Hsp70.

We also identified mutations that enhanced the ability of Ssa1-21p to impair [PSI+] propagation. R34K, which alone modestly impaired [PSI+], further weakened [PSI+] when combined with L483W (Table 1). R269K alone had no noticeable effect on [PSI+] or cell growth but similarly enhanced the L483W impairment of [PSI+] propagation (Table 1). Thus, the weakened [PSI+] propagation caused by L483W alone could be compromised further by altering residues in the ATPase domain.

Figure 1.

—[PSI+] phenotypes of cells expressing mutant SSA1 alleles. (A and B) Mutations that impair [PSI+]. (C and D) Mutations that restore [PSI+] propagation when combined with L483W. JA1NC transformants (ssa1 null, A and C) were streaked onto –leu plates and G400-1C transformants (ssa1,2,3,4 null, B and D) were streaked onto YPD plates. Plates were incubated for 2 days at 30° followed by 3 days at 25°. SSA1 alleles on plasmids either are wild type (WT) or contain codon substitutions as indicated. Those with asterisks also have the L483W substitution. The G336D allele did not support growth (see text) and therefore is not among the streaks in B.

Second-site suppressors of L483W: Selection for restored [PSI+] propagation was used to identify secondsite suppressing mutations of L483W (see materials and methods). Table 1 lists 14 such mutations and their relative effects on Hsp70 function and [PSI+] propagation, and Figure 1, C and D, illustrates some of these phenotypes. Unlike the forward mutations, the suppressing mutations were dispersed throughout the protein.

All alleles with suppressing mutations combined with L483W restored [PSI+] propagation when expressed in place of wild-type SSA1 (Figure 1C; data not shown). When cells expressed the double mutant alleles as the only SSA gene, all supported [PSI+] propagation but those with R444K and E540K had increased pigment accumulation and displayed mitotic loss of [PSI+] (Figure 1D; Table 1; data not shown). Additionally, although every double mutant allele supported cell growth, all except those with G404D, V435I, and P636S were clearly compromised in this ability, most showing pronounced effects at elevated temperature. These phenotypes revealed a connection between the ability of a substitution to suppress L483W and reduced Hsp70 function with respect to growth. However, the degree of support of [PSI+] propagation did not correlate directly with the capacity to support cell growth (Table 1). Thus, specific Hsp70 functions necessary for [PSI+] propagation may be separable from those required for cell growth.

Alleles with the suppressing substitutions but lacking L483W were also made and analyzed. None of them significantly affected [PSI+] propagation when expressed in place of Ssa1p (Table 1). For those that supported growth, when present as the sole Ssap none affected [PSI+] propagation and, in general, there was little or no difference in growth between cells with the single mutations and those with the mutations combined with L483W (Table 1). Three alleles with single substitutions (R169H, A367V, and E540K) were unable to support growth, which indicates that L483W acts as a secondsite suppressor of these inactivating mutations. Interestingly, none of these three inactive proteins dominantly impaired [PSI+] propagation like the nonfunctional G336D forward mutant. Therefore, loss of essential Ssap function was not enough to confer a dominant weakening effect on [PSI+]. This result is consistent with the interpretation that the G336D protein retained or gained an activity, absent in these three proteins, that is incompatible with [PSI+] propagation.

View this table:

Relative effects of suppressing mutations on new forward mutations

Suppressors of SSA1-21 also suppress the new SSA1 mutations: To further assess the relationship between the forward and suppressing mutations, alleles with the A17V and R34K mutations were combined with secondsite substitutions of G404D, A519T, E540K, or P636S. The dominant impairment of [PSI+] caused by both A17V and R34K was suppressed by all four second-site substitutions (Table 2). All eight double mutant alleles also supported growth showing that, like L483W, A17V and R34K suppressed the lethal E540K substitution. These results suggest that mutations in the different domains that impair [PSI+] propagation alter a similar Hsp70 function.

Suppressors G404D and A519T restore [PSI+] seed number: Replication of inheritable [PSI+] particles or “seeds” is arrested when cells are grown in the presence of millimolar amounts of guanidine (Eaglestoneet al. 2000). The nonreplicating seeds then become diluted as they are randomly distributed among dividing cells. Typical [PSI+] cells have ∼60 [PSI+] seeds (McCreadyet al. 1977; Eaglestoneet al. 2000; Junget al. 2000) and thus after the addition of guanidine show a lag of four to five cell divisions before [psi] cells appear in the population. Slight differences in the lag would reflect large differences in number of seeds per cell. For example, a lag extending a single additional division would reflect an increase from 60 to 120 cells per cell. As we showed previously (Junget al. 2000), when grown in 3 mm guanidine, cells expressing wild-type Ssa1p divided four to five times before [psi] cells began to appear, but no lag was observed for SSA1-21 cultures (Figure 2). This difference means that ∼15-fold fewer inheritable [PSI+] particles are in the mutant cells, which is consistent with the frequent spontaneous mitotic loss of [PSI+] from these cells.

The profile of [PSI+] loss from guanidine-treated cells expressing alleles containing G404D, with or without L483W, was the same as that of wild type (Figure 2). This shows that G404D alone had no effect on seed number but completely restored the ability of Ssa1p with L483W to maintain a normal number of [PSI+] seeds. The lag before appearance of [psi] cells in strains expressing alleles with A519T was slightly longer than that of wild type, indicating that A519T completely suppressed the seed number deficiency caused by L483W and that more [PSI+] seeds than normal may be in cells expressing alleles with this substitution. Consistent with this “stronger” [PSI+] phenotype, cells expressing A519T alleles also displayed stronger nonsense suppression, seen as reduced pigment accumulation (Figure 1, C and D).

Abundance of mutant proteins: Because [PSI+] increases expression of Ssa1p and we cannot distinguish Ssa1p from other Ssa proteins, abundance of mutant proteins was measured in [psi] cells lacking other Ssa proteins. The G336D protein did not support growth so we could not directly measure its abundance. However, its strong dominant inhibition of [PSI+] propagation indicated that it was expressed stably at some level. The amount of Ssa1p was not considerably different in the mutant strains (Figure 3; data not shown). This result was expected since the SSA1 promoter should be derepressed when no other Ssa proteins are present.

Figure 2.

—Curing of [PSI+] by guanidine. Guanidine was added to a concentration of 3 mm to log phase [PSI+] cells and the percentage of [PSI+] cells was monitored as a function of cell doublings.

Figure 3.

—Abundance of Ssa1p and Hsp104. Western analysis was done as described (Junget al. 2000) by using whole cell lysates of G400-1C transformants separated in denaturing 10% polyacrylamide gels and probing with antiserum raised against Hsp70 (Ssa1p, top) and Hsp104 (middle). Identical aliquots from the same boiled samples were used for Hsp70 and Hsp104 blots. A portion of the membranes from the Hsp104 blots, which were stained by amido black to compare protein amounts, is also shown (bottom). Codon substitutions of SSA1 alleles, shown at top, are as described in Figure 1 (asterisks indicate that the L483W substitution is also present) except that Q607X is Q607Ochre and the (–) and (+) superscripts indicate strains were [psi] or [PSI+], respectively.

As expected, L483W protein with the Q607Ochre mutation migrated faster in gels than the others, and we estimate that antigen migrating at the same size as wild-type Ssa1p represented at most 5% of the total (Figure 3). In [PSI+] cells ∼20–30% of the protein was full length, which is likely due to partial suppression of the ochre mutation caused by the presence of [PSI+]. Similar results were seen for the Q607Ochre protein without L483W. Thus, suppression of the L483W effects on [PSI+] by Q607Ochre may be partly due to production of a full-length protein. However, [PSI+] cells expressing Q607Ochre alleles with or without L483W grew significantly more slowly than their [psi] counterparts (doubling times of 230 and 160 min for [PSI+] and [psi] cells, respectively, for cells with both alleles). Because readthrough of this nonsense mutation was likely mediated by the SUQ5 ochre suppressing tRNA in our strains, the glutamine at residue 607 would be replaced by serine (Waldronet al. 1981), which may adversely affect Hsp70 function. Thus, while these data alone are not proof, they strongly suggest that the C-terminal 35 amino acids of Ssa protein are dispensable for viability.

Abundance of Hsp104 varied slightly between strains expressing the different mutant proteins, and increased abundance correlated with reduced ability to support growth (Figure 3; data not shown). This may reflect differences in ability of mutant Hsp70's to suppress protein aggregation, which lead to accumulation of Hsp104 substrates, or, not exclusively, in ability to negatively regulate the heat-shock response. Since [PSI+] propagation is sensitive to elevated Hsp104 abundance, these differences in Hsp104 levels may have contributed to some of the observed effects on [PSI+]. We note, however, that in cells with reduced Ssap function, for example, in ssa1 ssa2 cells, Hsp104 levels are very high but [PSI+] propagation is normal (Junget al. 2000).

Sites of mutations and implications for their effects on Hsp70 function: Hsp70's are very highly conserved essential protein chaperones. Given the high degree of structural homology of Hsp70's across genera and the universal conservation of residues critical for function, it is reasonable to assume that the locations of our mutations on the structure of DnaK, the Escherichia coli Hsp70 homolog, will provide insight into how some of them affect Ssa1p function. Structures of the ATPase and substrate-binding domains of DnaK were determined independently but the C terminus of the ATPase domain can logically be placed near the N terminus of the substrate-binding domain. The panels in Figure 4 showing the structures of the two domains are presented in a conventional orientation.

ATPase domain mutations: The ATPase domain mutations that impair [PSI+] may directly affect ATP hydrolysis or nucleotide exchange. Because the ATP- and ADP-bound states of Hsp70 differ in affinity for substrates, alterations in ATPase activity will affect substrate interactions. However, most ATPase domain mutations cluster in the subregion near where this domain adjoins the substrate-binding domain and are not near the active site (Figure 4). Such a location evokes the possibility that, rather than directly affecting ATP hydrolysis or exchange, they affect conformational changes induced by ATP hydrolysis that regulate substrate domain function.

The R169H-suppressing substitution is homologous to a mutation in the DnaK ATPase domain (R167H) that was identified as an allele-specific suppressor of the Hsp40 mutant dnaJD35N (Suhet al. 1998). The D35N substitution of DnaJ, the E. coli Hsp40 homolog, disrupts binding of DnaJ to DnaK, which causes a growth defect and inability to propagate bacteriophage-λ. The homologous R169H substitution likely alters the interaction of Ssa1p with Hsp40, suggesting that efficient interaction with Hsp40 is necessary for Ssa1-21p to impair [PSI+]. Since the R169H mutant did not support growth, Hsp70-Hsp40 interaction presumably is essential for growth of yeast. The ability of L483W to partially overcome the growth defect of the R169H mutation would then suggest that this double mutant does not require a normal Hsp40 interaction to provide essential Hsp70 function. Alternatively, the L483W substitution restores Hsp40 interaction.

Figure 4.

—Locations of SSA1 mutations on the DnaK structure. Backbones of the Protein Data Bank structures 1DKG (ATPase domain, left) and 1DKZ (substrate-binding domain in closed conformation, right) are shown. The E. coli amino acid residue numbers homologous to those of the SSA1 mutations (see Table 1) are indicated. Residues identified as forward mutations are blue and those identified as suppressing mutations are green. Residues in the substrate-binding domain are represented as sticks to show orientation of side chains. Backbone of bound substrate, seen end-on, is yellow. The wild-type R467, which makes a salt bridge with D540, is shown in red. See text for additional details.

We tested an Hsp40 interaction on [PSI+] propagation in SSA1-21 cells by deleting the DnaJ homolog YDJ1, which encodes a known Hsp40 cochaperone of Ssa1p. In wild-type cells, Ydj1p deficiency caused cells to grow very slowly but did not affect [PSI+] propagation, showing that while Ydj1p is important for growth, its interaction with Ssa1p is not important for [PSI+] propagation (Figure 5). Deleting YDJ1 in SSA1-21 cells did not improve [PSI+] propagation as anticipated, but appeared to weaken it, as reflected in increased pigment accumulation (Figure 5). No effect of Ydj1p depletion on nonsense suppression, with or without [PSI+], was seen for the wild-type strain (Figure 5; data not shown). Thus, another unidentified Hsp40 protein(s) likely is critical for the Ssa1p interaction with [PSI+]. One candidate would be Sis1p, an essential Hsp40 known to interact with Ssa1p and to be involved in propagation of the yeast [RNQ1] prion (Sondheimeret al. 2001).

Figure 5.

—Effect of YDJ1 deletion on [PSI+]. Isogenic [PSI+] wild-type and SSA1-21 cells with and without YDJ1 deletion were streaked onto YPD and incubated at 25° for 10 days.

Substrate-binding domain mutations: Except for V435I, all second-site suppressors in the substrate-binding domain are at residues in the loops of the binding pocket subregion that face the helices forming the lid or on the surfaces of the lid-forming helices that face these loops (Figure 4). These locations suggest that the substitutions affect interactions between the lid and substrate-binding subregions. Accordingly, the inactivating E540K substitution of Ssa1p is homologous to E543K of bovine Hsp70, which causes a large decrease in affinity for substrate (Haet al. 1997). These acidic residues correspond to D540 in DnaK (Figure 4), which forms hydrogen bonds with R467 in an outer loop of the substrate-binding cavity. This interaction forms part of a “latch” that stabilizes the lid in a conformation that closes the substrate-binding cavity (Zhuet al. 1996; see discussion). The basic residue in the outer loop is conserved in Ssa1p (R466), bovine Hsp70 (R469), and other Hsp70's. Thus, the E540K substitution likely reduces affinity of Ssa1p for substrate by disrupting the function of the latch.

The V435I mutation, which was the most frequently isolated suppressor, is at a conserved residue at the base of the substrate-binding cavity (Figure 4). A substitution of the homologous residue of DnaK, V436F, designed to occupy and block part of the substrate-binding pocket, considerably decreases affinity of DnaK for substrates and renders the protein unable to support growth (Montgomeryet al. 1999; Mayeret al. 2000b). Ssa1p with the V435F mutation is also deficient at substrate binding and has a dominant inhibitory effect on growth (Pfundet al. 2001). V435I also would be expected to weaken substrate interaction, but the effect of the additional single methyl group might not be expected to be as great. Accordingly (Table 1), Ssa1p with V435I alone or in combination with L483W functioned well at optimal growth temperature (30°) but was severely compromised at supporting growth at elevated temperature (37°).


Our results suggest that efficient propagation of [PSI+] is critically dependent upon regulation of substrate interactions by the ATPase domain of Hsp70 and that L483W may participate in this regulation. All of the new mutations that weaken [PSI+] are in the ATPase domain and therefore cannot affect substrate interactions directly. Two of them (A17V and R23H) impaired [PSI+] without affecting growth in a manner identical to that of L483W. Additionally, suppressors of L483W also suppressed A17V, and the inability of Ssa1p with one of these suppressors (E540K) to support growth was suppressed by both L483W and A17V. These results suggest that L483W and A17V, which are in different protein domains, affect the same Hsp70 function. The similarity in location, phenotypes, and suppressibility of other forward mutations suggests that they also affect a similar function.

The Hsp70 substrate-binding domain switches between an “open” conformation that allows rapid association and disassociation of substrate and a “closed” conformation that is inaccessible to binding of free substrate but prolongs binding of, or traps, previously bound substrate (Zhuet al. 1996). The frequency of transition between these conformations is determined by the bound nucleotide so that a much higher proportion of molecules will be in the open conformation when ATP is bound and in the closed conformation when ADP is bound. ATP hydrolysis and nucleotide exchange thus regulate Hsp70 chaperone function by affecting the frequency of transition between the open and closed conformations (see Mayeret al. 2000a for a review).

The ATPase domain mutations that impair [PSI+] propagation may directly alter intrinsic ATPase activity or nucleotide exchange or may alter the transition frequency between open and closed conformations of the ATP- or ADP-bound states. Either way, the effect would be an altered ability of the ATPase domain to regulate opening and closing of the substrate-binding cavity. All new forward mutations were in the ATPase domain, suggesting that there are likely fewer ways to disrupt this regulation in a way that impairs [PSI+] by altering residues in the substrate-binding domain. Since binding of substrate stimulates Hsp70 ATPase activity, another possibility is that L483W and the other forward mutations enhance this stimulation or cause a stimulatory effect in the absence of substrate.

The interpretation that the forward mutations all affect a similar function implies that the suppressing mutations all similarly oppose this affect. The locations of suppressing mutations in the substrate-binding domain and their homology with characterized Hsp70 mutations suggest that they reduce the stability of the closed conformation or directly reduce affinity for substrate. The E540K substitution is homologous to a mutation that lowers the energy barrier for the transition to the open conformation (Haet al. 1997; Mayeret al. 2001), and the V435I substitution, which is similar to V435F (Mayeret al. 2000b; Pfundet al. 2001), likely has reduced affinity for most substrates. Since stimulation of ATP hydrolysis by substrates is determined by the affinity of Hsp70 for substrate (Mayeret al. 2000b), V435I should cause a general reduction in this stimulation. The suppressive effect of the substrate-binding domain mutations therefore suggests that Ssa1p mutations that interfere with [PSI+] propagation promote conversion to or increase the stability of the closed conformation.

Suppressing mutations in the ATPase domain are clustered in the same region as the forward mutations and may oppose effects of the forward mutations directly or by altering cochaperone interactions. The predicted loss of Hsp40 interaction caused by the R169H substitution is expected to eliminate or considerably reduce the ability of Hsp40 to stimulate ATP hydrolysis. Consistent with the interpretation that L483W may enhance ATPase-mediated effects, suppression of the growth defect of R169H by L483W may be due to a reduced requirement for ATPase stimulation by Hsp40. Alternatively, the L483W substitution may partially overcome a requirement for efficient presentation of substrates to Hsp70 by Hsp40.

Deleting YDJ1, the gene encoding an Hsp40 cochaperone of Ssa1p, was expected to suppress the L483W impairment of [PSI+] by eliminating the interaction between these two proteins. However, the opposite result was seen. An explanation for this result is that in the absence of Ydj1p, more of the Ssa1p is available to associate with another Hsp40 that mediates the effect on [PSI+]. A promising candidate is Sis1p, one of several S. cerevisiae Hsp40 homologs whose activity is known to be crucial for propagation of the [PIN+]/[RNQ1] prion (Sondheimeret al. 2001).

Among the suppressors, the C-terminal mutations would not be expected to affect substrate binding directly but to disrupt interactions with TPR-containing cochaperones. The P636S, E639K, and E640K substitutions are all within a conserved octapeptide at the extreme C terminus of Hsp70, and the truncation at residue 607 removes this region. This motif in human Hsp70 physically interacts with a TPR domain of the Hsp90 cochaperone Hop1 (Scheufleret al. 2000), which suggests that these four mutations suppress L483W by interfering with the ability of Ssa1-21p to interact with Sti1p, the S. cerevisiae Hop1 homolog, or other TPR cochaperones. Our mutant Ssa1 proteins therefore appear to require an association with both Hsp40 and a TPR cochaperone to impair [PSI+] propagation. The finding that overexpression of Sti1p can destabilize yeast prions (Kryndushkinet al. 2002) supports the notion that [PSI+] propagation may be influenced by the activity of TPR cochaperones. Additionally, Ssa1p with any of the four C-terminal mutations supported growth of [psi] cells lacking other Ssap, which raises the possibility that the essential function of Hsp70 or TPR cochaperone proteins in yeast does not require TPR interaction with the C terminus of Hsp70.

The reduced number of [PSI+] seeds in SSA1-21 cells (Junget al. 2000) may mean that enhanced or prolonged trapping of substrates, presumably caused by the forward mutations, interferes with the ability of Ssa1p to promote regeneration of [PSI+] seeds. Second-site substitutions restored ability of Ssa1-21p to maintain a normal number of seeds, and one of them (A519T) may actually increase the number of seeds per cell, even when present as the only substitution on Ssa1p. Together our results suggest that a function of Hsp70 in [PSI+] seed replication is critically dependent upon its interaction with Sup35p aggregates in [PSI+] cells.

Our hypothesis for how the SSA1 mutations cause impairment of [PSI+] includes the role of Hsp104 in [PSI+] seed replication. Hsp104 is believed to break Sup35p aggregates into smaller, more numerous aggregates that function as seeds (Paushkinet al. 1996). We propose that the mutant Ssa1 proteins associate with the Sup35p aggregates in [PSI+] cells more avidly than does wild-type Ssa1p, which restricts access of the aggregates to the activity of Hsp104. This simple explanation accounts for all of the characterized and predicted effects of the L483W mutation on [PSI+] propagation (Junget al. 2000). Restricted access of Hsp104 to Sup35p aggregates would result in an increase in size and a decrease in number of the aggregates, at least some of which would be expected to function as seeds. The reduced seed number corresponds to reduced mitotic stability of [PSI+]. Additionally, fewer seeds correspond to fewer ends of the presumed fibrous aggregates that are available to recruit the soluble form of Sup35p, which results in an overall increase in Sup35p solubility and is reflected as reduced nonsense suppression. The R169H and C-terminal mutations further suggest that these effects require efficient interaction of Ssa1-21p with Hsp40 and TPR cochaperones.

The impairment of [PSI+] propagation by mutant Ssa1p is more severe when Ssa2p is absent, which may be explained by the increased abundance of Ssa1p due to derepression of the SSA1 promoter in cells lacking Ssa2p or by competition of these two chaperones for similar cofactors. The lack of a significant growth phenotype of many of the forward mutations, even when they were the only Ssa protein in the cell, suggests that the effects of the mutations are more critical for propagation of amyloid than for processes required for growth that depend upon Hsp70 function. This effect is reminiscent of that observed for cells expressing the G/F deletion mutant of the essential Sis1p, which grow well but are unable to propagate the [RNQ1] prion (Sondheimeret al. 2001).

Our study uncovers several new sites in Hsp70 that are important for its function, a highlight of which is that, although several of the mutations that impair [PSI+] are in conserved residues, none have been identified in earlier screens for mutations affecting other Hsp70 activities. In addition to demonstrating that normal Hsp70 function is crucial for [PSI+] propagation, these novel effects are consistent with the idea that aggregated Sup35p in [PSI+] cells represents a unique type of Hsp70 substrate. Likewise, Hsp104 appears to interact differently with Sup35p aggregates in [PSI+] cells than with aggregates caused by thermal denaturation (Junget al. 2002) and together these observations point to another link between the activities of Hsp70 and Hsp104. The combined phenotypic effects of this collection of mutants on growth and [PSI+] propagation raise defined and testable predictions of how the mutations alter Hsp70 enzymatic activity. Further biochemical and genetic analyses of these mutants should provide significant new insight into how Hsp70 functions not only in the propagation of cytosolic amyloid but also in its many important roles in the cell.


We thank Andy Golden for helpful comments on the manuscript.


  • Communicating editor: A. P. Mitchell

  • Received September 17, 2002.
  • Accepted October 31, 2002.


View Abstract