Sis1 is a multi-domain protein with an N-terminal J-domain followed by 2 unstructured regions of high glycine content, termed the glycine/phenylalanine-rich (G/F) and glycine/methionine-rich (G/M) regions (Fig. 1).²⁷ The C-terminal half of the protein contains a domain known to bind model hydrophobic peptides (CTD1), followed by a second putative peptide-binding domain of similar structure (CTD2) and a dimerization domain (DD).²⁷ Initially we focused our recent investigation on 2 well-characterized constructs of Sis1: Sis1–121, which possesses only the J-domain and G/F region, and Sis1-ΔG/F, which lacks only the G/F region (Fig. 1).²⁴ Both of these constructs propagate strong [PSI⁺] variants, while Sis1–121, but not Sis1-ΔG/F, maintains multiple variants of [RNQ⁺].^10,20,24,25 On the basis of these prior observations, we wondered whether [PSI⁺] prion variation would affect Sis1 requirements and, if so, would weak [PSI⁺] variants resemble [RNQ⁺] in their requirements for Sis1 function? We found that while all 4 strong [PSI⁺] variants that we examined were similarly maintained by both Sis1 constructs, the distinctions between weak and strong variants were significant.¹ Weak [PSI⁺] variants were stable when Sis1-ΔG/F is expressed but, surprisingly, could not be maintained by Sis1–121, indicating that these variants have markedly different requirements for Sis1 than either strong variants or the [RNQ⁺] variant used in our study.¹ One important question when evaluating these experimental results is whether or not they represent an inability of the chaperone to propagate the prion, or rather, an induced selection for a small population of prion-free cells which then overtake the culture due to induced prion toxicity. For example, others have observed that [PSI⁺] becomes cytotoxic when Sis1 activity is severely compromised in some strain backgrounds.^19,20,26 Likewise we observed [PSI⁺]-dependent toxicity in our strains when Sis1 was repressed.¹ By examining growth rates of cells with weak and strong variants during Sis1 repression, we were able to confirm the previous assertion made by Kirkland et al. that [PSI⁺]-dependent toxicity is positively correlated to [PSI⁺] variant stability, i.e., the number of propagons per cell.^1,20 However, the results of our experiments did not match those that would be expected if curing were due to a toxicity-dependent selection for [psi⁻] cells, since curing would be expected to occur more often for the more cytotoxic strong variants than for weak variants. Because we observed exactly the opposite pattern of loss, we concluded that prion-loss was due to an inability of specific Sis1 constructs to support prion propagation.¹ Importantly, we were also able to confirm all results in 2 distinct yeast genetic backgrounds, allowing us to rule out the possibility that an uncharacterized polymorphism of a particular lab strain is responsible for a specific experimental outcome.¹ Figure 1.

Primary structure diagrams of Sis1 and Hdj1 (DNAJB1) expression constructs described in this manuscript. Protein regions are denoted using the following notation: J, J-domain; G/F, glycine/phenylalanine-rich region; G/M, glycine/methionine-rich region; CTD1/2, C-terminal peptide-binding domains I and II; DD, dimerization domain.^1,27 Dashed lines indicate where a region had been deleted.

One advantage of studying the prion-chaperone machinery in yeast is that specific functions that may be conserved in orthologous proteins can be easily tested using complementation assays. Of particular interest to us is the human homolog of Sis1, called Hdj1 or DNAJB1, which can not only rescue cell viability in a sis1-Δ strain but, incredibly, propagate the prion [RNQ⁺] in yeast.^27,28 As with Sis1, the minimum functional regions required for both these functions were narrowed down to the J-domain and glycine-rich regions.²⁸ Like [RNQ⁺], we found that Hdj1 can also propagate 4 different strong variants of [PSI⁺].¹ In sharp contrast, neither weak variant we examined was able to propagate, despite our reproduction of all experiments in 2 yeast genetic backgrounds, again illustrating that Hsp40 requirements are strongly affected by amyloid polymorphism.¹ This discovery that strong but not weak variants of [PSI⁺] can be propagated by Hdj1 was simultaneously reported by the True laboratory,²⁶ and yet these results conflict with an earlier report that Hdj1 was unable to propagate a specific variant of [PSI⁺].²⁰ Because the variant used in that study is also strong, it is likely that the discrepancy between our 3 reports is either an issue of protein expression levels or possibly an uncharacterized polymorphism in the genetic background used. Notably, both we and the True lab noted increased polymer sizes resolved on SDDAGE gels in cells expressing only Hdj1, indicative of a [PSI⁺] fragmentation defect in these cells.^1,26 These observations, together with the size increases observed with Sis1 truncation/deletion mutants described above, further support the notion that the primary role of Sis1/Hdj1 is in [PSI⁺] fragmentation as originally suggested by Higurashi et al. on the basis of kinetics data and SDDAGE analysis,¹⁰ and also indicate that Hdj1 is somehow deficient in that function, but not deficient enough to cause loss of strong variants of [PSI⁺].

Sis1 is also required for the curing of [PSI⁺] by overexpression of Hsp104,²⁰ though the mechanism by which curing occurs is debated.^9,29,30 Our recent findings raise the question of whether Hdj1 has maintained this functionality of Sis1 as well, a question that we were unable to address in our original investigation but will address here. We examined the effects of Hsp104 overexpression in W303 strains expressing Hdj1 in place of Sis1 (Fig. 2A). Strikingly, while the plasmid pRS426-GPD-HSP104 produces Hsp104 levels sufficiently high to efficiently cure all variants of [PSI⁺] that we have examined to date, when moved into these cells, all 3 strong [PSI⁺] variants we examined were completely protected from Hsp104-dependent curing (Fig. 2A). As with other Sis1/Hdj1 experiments, we sought to confirm that these results were not due to an uncharacterized polymorphism of the W303 genetic background and so we repeated these experiments with “isoprionic” strains of the 74D-694 genetic background bearing the same 3 strong [PSI⁺] variants. In this case, the results were highly surprising: while 2 variants, [PSI⁺]^STR and [PSI⁺]^VH, were fully resistant to curing, 74D-694 cells bearing [PSI⁺]^Sc4 were cured consistently across multiple repetitions of the experiment (overall 39 out of 40 transformants examined over 4 experimental replicates, Fig. 2B). To confirm that the observed color phenotypes really correspond to the maintenance or loss of the prion per se, and not merely an alteration of the phenotype, we subjected Hsp104 overexpressing cells from both backgrounds to analysis by SDDAGE. Indeed, only the [PSI⁺]^Sc4 variant from the 74D-694 genetic background shows loss of SDS-resistant aggregates (Fig. 2C). Finally, we ruled out both the accidental co-expression of Sis1, or a lack of Hsp104 overexpression, as possible confounding factors affecting our results; as expected, Sis1 expression was absent from all strains tested (Fig. 2D) and Hsp104 levels are highly elevated, relative to cells prior to plasmid transformation, to a level that efficiently cures all 3 [PSI⁺] variants in both backgrounds in the presence of Sis1 (Fig. 2E). Notably, the level of Hsp104 in the [PSI⁺]^Sc4 74D-694 strain is slightly, but reproducibly, higher than in the strains maintaining other [PSI⁺] variants and so we cannot presently rule out whether the curing of this variant, specifically, and only in this genetic background, is due to this slight difference in Hsp104 levels. Regardless, the completely different experimental outcome observed for this one combination of both prion variant and yeast genetic background underscores the importance of reproducing experiments in multiple genetic backgrounds whenever feasible. Figure 2.

Hdj1 is deficient in replacing Sis1 in the curing of [PSI⁺] by Hsp104 overexpression. (A & B) [PSI⁺] cells of the W303 (panel A) or 74D-694 (panel B) genetic backgrounds bearing strong [PSI⁺] variants ([PSI⁺]^STR, [PSI⁺]^Sc4, or [PSI⁺]^VH) and expressing Hdj1 in place of Sis1 (left columns) were transformed by a plasmid overexpressing Hsp104 ([pRS426-GPD-HSP104]) that normally cures all variants of [PSI⁺] (right columns). Color phenotype assays are shown for representative transformants (n ≥ 10 for each variant) following passage on media selecting for the Hsp104 overexpression plasmid. (C) Maintenance or loss of [PSI⁺] in cells shown in panel B was also confirmed by semi-denaturing detergent agarose gel electrophoresis (SDDAGE). Detergent resistant Sup35 aggregates indicative of the presence of [PSI⁺] were resolved by SDDAGE and visualized by immunoblot analysis using an antibody specific for Sup35. Control [psi⁻] cells were included for comparison. (D & E) Protein expression levels in cells from the 74D-694 genetic background were measured for Sis1 (panel D) and Hsp104 (panel E). Cell extracts from isolates in panel B were subjected to immunoblot analysis using antibody specific for Sis1 or Hsp104. A band cross-reacting with each antibody is shown as a loading control. Control [psi⁻] cells were included for comparison. Sis1 and Sup35 antibodies and methods for SDDAGE, SDSPAGE, and immunoblotting are the same as described in Harris et al.¹

In a previous investigation, one of us (JKH) and co-workers found that that [PSI⁺] is cured much slower than [RNQ⁺] upon repression of Sis1 expression and that both a strong and weak variant of [PSI⁺] could be propagated by Sis1ΔG/F, which cures [RNQ⁺].¹⁰ Based on those observations, we previously postulated a ‘stringency’ model for Sis1 activity where these 2 prions require a presumably singular function of Sis1, but to different extents.¹⁰ A presumption in that model is that the Sis1ΔG/F construct, which does not support [RNQ⁺], must be fundamentally less active than Sis1–121, which supports [RNQ⁺]. The results from our most recent investigation negate this stringency model as we found that these 2 Sis1 constructs reciprocally supported a variant of [RNQ⁺] and weak variants of [PSI⁺].¹ To rule out any issues involving small experimental variations, or differences in specific cell isolates, we conducted shuffling experiments to further clarify this point by using cells carrying 2 prions simultaneously ([RNQ⁺] and [PSI⁺]^Sc37). We demonstrated that either prion could be selected for, at the expense of the other, depending upon which construct of Sis1 was shuffled in.¹ Notably, this experiment verified that the domain requirements for the 2 prions are mutually exclusive and, more importantly, demonstrated for the first time that Sis1 must accomplish at least 2 biochemically distinct functions in prion maintenance that are differentially required by distinct prion variants.¹ Stein and True recently found similar results when examining 5 variants of [RNQ⁺] as the domain requirements for Sis1 were highly variable, and in the case of 2 variants, fully distinct from one another.²⁶ However, despite multiple investigations into the Sis1 domain requirements of [PSI⁺], still no construct has been identified that separates cell viability from strong [PSI⁺] maintenance.^1,19,20,26 Together all data support the notion that strong [PSI⁺] variants require minimal functional contribution from Sis1 to fragment aggregates at a rate that allows stable propagation, while, as noted above, Sis1 domain requirements for weak [PSI⁺] variants and [RNQ⁺] variants vary widely and in some cases are completely distinct. The emerging picture is one where multiple domains of Sis1 can coordinate in alternate ways to maintain specific prions and prion variants. Though not withstanding additional possibilities, these divergent Sis1 functions are most likely either alternate means of regulating Hsp70 activity to accomplish prion fragmentation, or alternate modes of binding and recruiting Hsp70 to aggregates. Previous work, also from the True lab, is consistent with the latter hypothesis, suggesting that [RNQ⁺] variants present distinct surfaces for chaperone binding by combining different sets of non-adjacent amyloidogenic regions into the amyloid cores of different variants.^21,25 Our data,¹ and that of others,³³ also support a similar notion that [PSI⁺] variants may likewise present distinct chaperone binding surfaces which affect their ability to interact with, and therefore become remodeled by, the Hsp40/70/104 chaperone machinery in a Sis1-dependent manner. Given the collective data from our 2 laboratories, we favor the latter model where combinations of Sis1 domains mediate alternate modes of binding to amyloids and thereby recruiting and activating Hsp70 at amyloid surfaces, but certainly additional experimentation is needed to make any definitive determinations.

Sis1 is absolutely required for the propagation of at least 4 prions,^10,12,13 and yet these prions exhibit enormous differences in requirements for Sis1 activity,^{1,18–20,24–26} demonstrating a diversity of functionality that has only been revealed due to the exploration of chaperone requirements for multiple amyloid structures, i.e., multiple prions and prion variants. Importantly, of the 10 identified amyloid-forming prions in yeast, investigations into the role of Hsp70 and its co-chaperones in prion propagation have so far been limited to just 4: [PSI⁺], [RNQ⁺], [URE3], and [SWI⁺]. As such, the potential roles of J-proteins, nucleotide exchange factors, or the 4 isoforms of Ssa have not been systematically investigated for [MOT3⁺], [ISP⁺], [OCT⁺], [NSI⁺], [MOD⁺], and [NUP100⁺]. Given the apparent, profound diversity of prion structures and chaperone requirements revealed so far, it is likely that other modes of chaperone-dependent or perhaps chaperone-independent prion transmission may exist. Indeed, the amyloid forming prion [ISP⁺] has been reported to propagate independently of Hsp104 activity, indicating that its propagation may be either entirely chaperone independent or rely upon other chaperone systems which remain to be identified.³⁴ Critically, the full spectrum of amyloid diversity in vivo, and thus the gamut of chaperone functional complexity, will remain obscured until more prion•chaperone interactions can be investigated.

No potential conflicts of interest were disclosed.