Sequestration of essential proteins causes prion associated toxicity in yeast


*E-mail; Tel. (+1) 312 9964226; Fax (+1) 312 4132691.


Prions are infectious, aggregated proteins that cause diseases in mammals but are not normally toxic in fungi. Excess Sup35p, an essential yeast protein that can exist as the [PSI+] prion, inhibits growth of [PSI+] but not [psi-] cells. This toxicity is rescued by expressing the Sup35Cp domain of Sup35p, which is sufficient for cell viability but not prion propagation. We now show that rescue requires Sup35Cp levels to be proportional to Sup35p overexpression. Overexpression of Sup35p appeared to cause pre-existing [PSI+] aggregates to coalesce into larger aggregates, but these were not toxic per se because they formed even when Sup35Cp rescued growth. Overexpression of Sup45p, but not other tested essential Sup35p binding partners, caused rescue. Sup45–GFPp formed puncta that colocalized with large [PSI+] Sup35-RFPp aggregates in cells overexpressing Sup35p, and the frequency of the Sup45–GFPp puncta was reduced by rescuing levels of Sup35Cp. In contrast, [PSI+] toxicity caused by a high excess of the Sup35p prion domain (Sup35NMp) was rescued by a single copy of Sup35Cp, was not rescued by Sup45p overexpression and was not associated with the appearance of Sup45–GFPp puncta. This suggests [PSI+] toxicity caused by excess Sup35p verses Sup35NMp is, respectively, through sequestration/inactivation of Sup45p verses Sup35p.


Protein misfolding and the presence of protein aggregates are associated with numerous neurodegenerative diseases (Taylor et al., 2002; Ross and Poirier, 2004) that include aggregates of Aβ-42, mutant Huntingtin and PrPSc in Alzheimer's disease, Huntington's disease and prion diseases respectively (Prusiner, 1982; Prusiner et al., 1998; Nussbaum and Ellis, 2003; Gil and Rego, 2008). Since there is sometimes a poor correlation between the presence of aggregates and other markers of the disease, the role of the aggregate is now thought to be benign or even cytoprotective and toxicity has been attributed to a soluble oligomeric species (Arrasate et al., 2004; Haas and Selkoe, 2007). Although not mutually exclusive, an alternate hypothesis is that the toxic species (either soluble or aggregated) causes death by interacting abnormally with or sequestering key factors in various pathways (Caughey and Baron, 2006; Gil and Rego, 2008).

Among these protein-misfolding diseases, prion diseases are unique as they are infectious. These diseases are associated with the accumulation of the prion form of the PrP protein, PrPSc (Prusiner, 1982). Prions were first discovered in mammals as the causative agent of some fatal neurodegenerative diseases, such as Scrapie in sheep and Creuztfeldt-Jakob disease in humans (Prusiner, 1998). PrPSc is in a β-sheet rich, aggregated, amyloid form that can attract normal monomers mainly in an α-helical soluble form, to join the amyloid aggregate in an autocatalytic process (Pan et al., 1993). Interestingly, the same protein is able to cause different infectious forms (or strains) of the disease in inbred hosts with the same genotype (Bruce, 1996). These strains are distinguished by varying symptoms and are attributed to different structures of the amyloid form of PrPSc (Bessen et al., 1995; Caughey, 2003).

Although prions were first discovered in mammals, the prion paradigm has been extended to yeast (Wickner et al., 2004). The three best studied yeast prions [PSI+], [PIN+] and [URE3], are the prion forms of Sup35p, Rnq1p and Ure2p respectively (Wickner, 1994; Sondheimer and Lindquist, 2000; Derkatch et al., 2001). Although none of these proteins are homologous to the mammalian prion protein PrP, they all share the properties of being β-sheet rich, aggregated and infectious in their prion forms. In addition, yeast prions also exist as different strains, called variants, that appear to differ in their amyloid conformations, which leads to phenotypic and biochemical variations (Derkatch et al., 1996; Zhou et al., 1999; Schlumpberger et al., 2001; Uptain et al., 2001; Bradley et al., 2002; King and Diaz-Avalos, 2004; Tanaka et al., 2004; Krishnan and Lindquist, 2005; Toyama et al., 2007).

In this study we focus on the [PSI+] prion and its protein determinant Sup35p. Sup35p in its non-prion form associates with Sup45p to mediate translation termination (Stansfield et al., 1995). Sup35p and Sup45p, both encoded by essential genes, belong to the families of eukaryotic release factors, eRF3 and eRF1 respectively. Sup45p recognizes all three stop codons in the ribosome and facilitates the release of the nascent peptide (Bertram et al., 2000; Song et al., 2000), while Sup35p is the GTPase that promotes both these functions of Sup45p (Frolova et al., 1996; Salas-Marco and Bedwell, 2004; Alkalaeva et al., 2006). The interaction between these two proteins mediates the release of newly formed peptide chains.

Sup35p can be divided into three domains. The C-terminal globular domain associates with Sup45p, has a GTPase domain, and is necessary and sufficient for viability and the translation termination function (Ter-Avanesyan et al., 1993). The N-terminal (N) and middle (M) domains are dispensable for viability but are necessary and sufficient for the induction and faithful maintenance of the prion variant state (Derkatch et al., 1996; Patino et al., 1996; Liu et al., 2002; Bradley and Liebman, 2004). When Sup35p converts to its prion [PSI+] form, the activity of the C domain is compromised as most of the protein is in the aggregated form, leading to a decrease in translation termination efficiency (Cox, 1965; Patino et al., 1996; Paushkin et al., 1996). Variants of [PSI+] are distinguished by differences in their reduction in translation termination efficiency: strong [PSI+] variants have lower translation termination efficiency and more aggregated Sup35p compared with weak [PSI+] variants (Derkatch et al., 1996; Zhou et al., 1999; Uptain et al., 2001).

[PSI+] can be induced by the overproduction of Sup35p or Sup35NMp. Presumably, since there is more protein, there is a higher chance that some of the protein molecules will convert to the prion form, seeding the conversion of more monomers (Wickner, 1994; Derkatch et al., 1996). Induction occurs efficiently only in the presence of other prions, such as [PIN+] ([PSI+]Inducible) (Derkatch et al., 1997; 2001). This is explained by the cross-seeding model, which proposes that the [PIN+] prion behaves as an inefficient template to convert Sup35p into its prion form by a heterologous seeding mechanism (Derkatch et al., 2004). [PIN+], like [PSI+], also exists as variants, which are characterized by different levels of aggregated Rnq1p and varying degrees of induction of [PSI+] (Bradley et al., 2002).

Although neither the presence of [PSI+] nor the overexpression of Sup35p is toxic independently, it has been known for some time that overexpression of Sup35p (or to a lesser extent the SUP35NM domain) is growth inhibiting to [PSI+] yeast (Dagkesamanskaya and Ter-Avanesyan, 1991; Derkatch et al., 1996). In cells lacking [PSI+], but that are [PIN+], increasing the level of Sup35p but not SUP35NM causes a growth defect (Derkatch et al., 1996; Bradley et al., 2002). However, SUP35NM fused to green fluorescent protein (GFP) overexpressed in [psi][PIN+] cells leads to the formation of large ring and worm-like structures that are a precursor to [PSI+] and dramatically decreases the viability of just the cells with the ring/worm aggregates (Zhou et al., 2001; Ganusova et al., 2006). Apparently, only cells primed to form the [PSI+] prion are incompatible with overexpression of Sup35NM–GFPp.

The source of this lethality has been attributed to either the accumulation of Sup35p isoforms that might be toxic to the cell, or the depletion of essential factors (Derkatch et al., 1996; 1998). A single copy of SUP35C or overexpression of SUP45 or SUP35C relieves this inhibition, which suggests that these two essential factors might be sequestered (Ter-Avanesyan et al., 1993; 1994; Derkatch et al., 1998; Ganusova et al., 2006). In this study, we systematically examine the lethality caused by various levels of overexpression of SUP35 and SUP35NM in different variants of [PSI+] and [PIN+]. We show that at high levels of overexpressed Sup35NMp, the mechanism of [PSI+] mediated toxicity is through the loss of Sup35p function; whereas at high levels of Sup35p, the essential protein Sup45p is sequestered due to its association with Sup35p, and Sup35p activity is not limiting.


Growth inhibition caused by overexpression of SUP35 or its prion domain in [PSI+] yeast is alleviated by a proportional increase of Sup35Cp

We tested whether growth inhibition associated with various levels of overexpression of SUP35 or its NM domains (SUP35NM) was due to sequestration of Sup35p itself. A wild-type strain or a ΔNM strain, containing only the chromosomally encoded C-terminal domain of SUP35, was transformed with plasmids capable of overexpressing SUP35 or SUP35NM at either a 10-fold (10×) or 100-fold (100×) level. When these two strains were mated to [psi], weak and strong [PSI+] yeast, the [psi-] and weak [PSI+] diploids homozygous for SUP35 and with 10×SUP35NM, showed no growth defects while strong [PSI+] diploids were very sick (Fig. 1A, Panel A). This growth inhibition was alleviated in the cross with ΔNM (which contains one copy of SUP35C) (Fig. 1A, Panel B). When the level of SUP35NM was raised to 100×, even weak [PSI+] diploids showed a severe growth defect that was also relieved by one copy of SUP35C (Fig. 1A, Panels C and D). With 10× full-length SUP35, strong [PSI+] diploids were inviable but could be rescued by one copy of SUP35C (Fig. 1B, Panels A and B). In contrast, the inviability caused by 100×SUP35 in the presence of weak and strong [PSI+] was not rescued by one copy of Sup35Cp (Fig. 1B, Panels C and D).

Figure 1.

SUP35C rescues growth inhibition caused by overexpression of SUP35 and SUP35NM in [PSI+] cells.
A. Sup35Cp rescues growth inhibition caused by overexpressed SUP35NM in [PSI+] strains. Diploids selected from different mating reactions with continued overexpression of SUP35NM: Wild-type (w.t.) or a SUP35C containing strain (ΔNM) overexpressing SUP35NM 10-fold (10×) (Panels A and B) or a 100-fold (100×) (Panels C and D) crossed to [psi], weak or strong [PSI+] variants. The colour variation on synthetic media is inconsistent and cannot be interpreted accurately.
B. Sup35Cp rescues growth inhibition caused by 10× but not 100×SUP35 in [PSI+] strains. Diploids selected from different mating reactions with continued overexpression of SUP35: Wild-type (w.t.) or a ΔNM strain overexpressing full-length SUP35 10× (Panels A and B) or 100× (Panels C and D) crossed to [psi], weak or strong [PSI+] variants.
C. Overexpression of Sup35Cp rescues growth inhibition caused by 100×SUP35 in [PSI+] strains. Diploids selected from different mating reactions with continued overexpression of 100×SUP35:ΔNM overexpressing full-length 100×SUP35 with either expression of a control vector (Panel A), a low level of SUP35C (Panel B) or a high level of SUP35C (Panel C) crossed to [psi], weak or strong [PSI+] variants. Diploids were selected to maintain overexpression of vector or SUP35C.
D. Overexpressed Sup35Cp does not dissolve [PSI+] aggregates. Lysates from [PSI+] or [psi]SUP35/ΔNM diploids, simultaneously overexpressing full-length SUP35 and SUP35C, subjected to high-speed centrifugation, resolved on an acrylamide gel and probed with an anti-SUP35C antibody. T, total lysate; S, supernatant after high-speed centrifugation; P, Pellet after high-speed centrifugation.

Since one copy of SUP35C did not relieve toxicity caused by highly overexpressed SUP35 in strains containing [PSI+], it suggests that at 100×SUP35 either a Sup35p binding protein and not Sup35p itself is being sequestered, or a toxic species is formed at this level of overexpression. If a Sup35p binding partner is being sequestered, toxicity should be relieved by the simultaneous overexpression of SUP35C, as it does not join the [PSI+] aggregate. Sup35Cp would therefore compete for the sequestered protein, drawing it out of the prion aggregate. To test this, GAL::SUP35C plasmids were transformed into the ΔNM 100×SUP35 strain and mated to [psi], weak and strong [PSI+] yeast. Indeed, when GAL::SUP35C was expressed on a high copy plasmid (high level) the growth inhibition was alleviated (Fig. 1C, Panel C). GAL::SUP35C expression on a centromeric plasmid (low level) was not enough to relieve the toxicity (Fig. 1C, Panel B).

To determine if the overexpression of SUP35C rescues growth inhibition by dissolving [PSI+] aggregates, lysates from [PSI+] and [psi] diploids simultaneously overexpressing SUP35 and SUP35C were subjected to high-speed centrifugation and analysed on an SDS-PAGE. Since in [psi] yeast, Sup35p can be found in the supernatant whereas in [PSI+] yeast Sup35p is absent from the supernatant fraction (Patino et al., 1996; Paushkin et al., 1996), we asked if simultaneous overexpression of Sup35Cp and Sup35p would cause Sup35p to appear in the supernatant in [PSI+] cells. However, Sup35p was not retained in the supernatant, indicating that [PSI+] aggregates were not disrupted by Sup35Cp overexpression (Fig. 1D).

SUP35C also rescues lethality caused by overexpression of SUP35 in [PIN+] cells

Growth inhibition was not observed in 10×SUP35[PIN+] diploids (data not shown). However, as described previously (Bradley et al., 2002) 100×SUP35 caused inviability in high and very high [PIN+] diploids, although other [PIN+] variants and the [pin-] diploids were unaffected (Fig. 2, Panel A). Expression of Sup35Cp at low levels did not relieve the toxicity, but at high levels of Sup35Cp, viability of high and very high [PIN+] diploids was rescued (Fig. 2, Panels B and C). Lethality was observed only in cells overexpressing the full-length Sup35p, not Sup35NMp (Fig. S1). This suggests that lethality is not due simply to the appearance of [PSI+], as overexpression of the SUP35NM domain leads to more induction of [PSI+] than the overexpression of the full-length protein (Derkatch et al., 1996). Rather it suggests that toxicity is caused by the formation of [PSI+] aggregates by the overexpression of full-length Sup35p that in turn sequesters an essential protein that normally binds to the C domain of Sup35p.

Figure 2.

SUP35C rescues growth inhibition caused by overexpression of 100×SUP35 in cells with [PIN+] variants. Diploids selected from different mating reactions with continued overexpression of 100×SUP35. A ΔNM strain overexpressing full-length 100×SUP35 with expression of either a control vector (Panel A), a low level of SUP35C (Panel B) or a high level of SUP35C (Panel C) crossed to either a [pin], high, very high, low or medium [PIN+] variant.

Sup35Cp alleviates growth inhibition of cells with large ring-like aggregates formed during the induction of [PSI+]

During the induction of [PSI+] by overexpressed SUP35NM–GFP in [PIN+] cells, large ring and worm-like structures have been observed that are a hallmark of cells that grow into [PSI+] colonies (Fig. 3A). These colonies have a markedly lower viability in comparison to their diffuse SUP35NM–GFP siblings (Zhou et al., 2001; Ganusova et al., 2006). We tested whether this reduced viability is due to the lack of Sup35Cp or one of its binding partners, by overexpressing SUP35NM–GFP in a high [PIN+] variant in the presence or absence of Sup35Cp. A single copy of Sup35Cp did not abolish the formation of these ring and worm-like structures (Fig. 3A). Ring containing cells without Sup35Cp had low viability (20%), whereas the diffuse cells were much more viable (87%) (Fig. 3B). Upon the addition of Sup35Cp, the viability of ring containing cells rose dramatically (73%).

Figure 3.

SUP35C rescues the toxicity of cells with SUP35NM–GFP‘rings’ without inhibiting the formation of [PSI+].
A. SUP35C does not inhibit the formation of NM–GFP rings in a high [PIN+] variant. A high [PIN+] variant overexpressing SUP35NM–GFP under a copper-inducible promoter, showing a ring-like structure, in the absence (vector) or presence of SUP35C. SUP35C is expressed on a centromeric plasmid under its own promoter. Arrow points to cells with a ring.
B. SUP35C increases the viability of [PIN+] cells with NM–GFP rings without decreasing de novo formation of [PSI+]. Viability and frequency of formation of [PSI+] in cells with diffuse SUP35NM–GFP and cells with SUP35NM–GFP‘rings’ in the presence or absence (vector) of SUP35C. (total number of colonies tested per category: minimum 80, number of independent transformants: at least 4 per category). Presence of [PSI+] was determined by colony colour and SUP35NM–GFP fluorescence (see Experimental procedures). Numbers indicate average percentages and error bars show standard error.

Since cells with rings frequently give rise to [PSI+] colonies (Zhou et al., 2001; Ganusova et al., 2006), we determined whether Sup35Cp increased viability by decreasing the frequency of [PSI+] appearance. However, ring cells with Sup35Cp gave rise to a similar amount of [PSI+] colonies as ring cells without Sup35Cp (Fig. 3B).

Overexpressed Sup35p appears to cause pre-existing [PSI+] aggregates to coalesce even in the presence of Sup35Cp

To determine whether Sup35Cp changes the morphology of pre-existing [PSI+] aggregates during overexpression of Sup35p, we took advantage of [PSI+] yeast containing an integrated copy of functional SUP35–GFP driven by a promoter repressed upon mating (MFA::SUP35–GFP) (Satpute-Krishnan and Serio, 2005). [PSI+] is observed as numerous, tiny Sup35–GFPp foci in this strain and since Sup35–GFPp is not synthesized upon mating, it is possible to monitor pre-existing [PSI+] aggregates in zygotes (Satpute-Krishnan and Serio, 2005). The 10×SUP35 wild-type and ΔNM strains were mated to the MFA::SUP35–GFP[PSI+] yeast and zygotes were monitored. As previously reported (see fig. 2c in Satpute-Krishnan et al., 2007), wild-type zygotes without overexpressed Sup35p retained the multitude of tiny foci (Fig. 4A). In contrast, both wild-type and ΔNM zygotes overexpressing Sup35p had large amorphous Sup35–GFPp aggregates, suggesting that pre-existing [PSI+] aggregates coalesce upon Sup35p overexpression (Fig. 4A).

Figure 4.

Pre-existing [PSI+] aggregates appear to coalesce in the presence of overexpressed Sup35p.
A. Sup35Cp does not inhibit the formation of large Sup35–GFPp aggregates upon overexpression of Sup35p in [PSI+] cells. Zygotes from a strain containing [PSI+] and SUP35–GFP under the MFA promoter (MFA::SUP35–GFP, which is repressed upon mating) (MATa) crossed to a wild-type strain overexpressing a control vector or SUP35, and ΔNM overexpressing SUP35 (MATα). The size of Sup35–GFPp aggregates in wild-type zygotes overexpressing Sup35p is not reproducibly larger than in the ΔNM zygote overexpressing Sup35p.
B. Large Sup35–GFPp aggregates formed in [PSI+] zygotes overexpressing Sup35p are not passed on to daughter cells. Zygote from a mating between a ΔNM strain overexpressing 10×SUP35 and strain containing [PSI+] and MFA::SUP35–GFP observed at different time points (number of zygotes = 6). The growing zygote was placed on media that retained the plasmid overexpressing SUP35. The arrow points to a large aggregate in the original zygote. The arrowhead shows the formation of a new aggregate, larger than a normal [PSI+] aggregate. Due to decay of signal, images at 7 and 9 h were brightened using photoshop to observe aggregates at those time points.
C. Large aggregates of Sup35–GFPp are formed in [PSI+] diploids with overexpressed Sup35p. SUP35–GFP/ΔNM diploids overexpressing either a control vector or SUP35. Arrows point to enlarged [PSI+] aggregates.
D. Biochemical verification of formation of large Sup35–GFPp aggregates upon overexpression of Sup35p in [PSI+] cells, in the presence and absence of Sup35Cp. Non-functional SUP35–GFP was transiently (2 h) overexpressed or not overexpressed in a strain containing [PSI+] with a control vector or SUP35C, and lysates were subjected to sucrose gradient centrifugation, resolved on an acrylamide gel and probed with anti-SUP35C antibody. The arrowheads point to the fraction in which genomic Sup35p is most abundant.

To determine if large aggregates are passed to daughter cells, we monitored the fate of zygotes with large Sup35–GFPp aggregates. While 10×SUP35/SUP35 zygotes (with MFA::SUP35–GFP) failed to divide, 10×SUP35/ΔNM zygotes did give rise to daughter cells. The very large aggregates in these zygotes remained visible after several generations and were not passed on to daughter cells (Fig. 4B). Although some tiny aggregates of [PSI+] are passed onto daughter cells where they are capable of forming aggregates larger than normal [PSI+] aggregates, none were comparable to the size of the initial large aggregates (Fig. 4B).

We also monitored fluorescence in SUP35–GFP/ΔNM diploids using endogenously tagged SUP35–GFP driven by the SUP35 promoter (Satpute-Krishnan and Serio, 2005) to enable us to observe [PSI+] aggregates in equilibrium with newly synthesized Sup35–GFPp. Initial SUP35–GFP/ΔNM zygotes overexpressing Sup35p also had large amorphous aggregates similar to zygotes with an extra copy of MFA::SUP35–GFP (data not shown). Daughters from such zygotes did contain some aggregates larger than normal [PSI+], although none were as large as in the initial zygote (Fig. 4C), suggesting that very large aggregates are formed initially but are not inherited or formed in daughter cells.

To confirm the formation of larger [PSI+] aggregates biochemically, SUP35–GFP was transiently overexpressed in wild-type [PSI+] yeast with untagged SUP35, in the presence or absence of Sup35Cp, and aggregates were analysed by sucrose gradient centrifugation. In the absence of overexpression of SUP35–GFP, genomic Sup35p was primarily in the third fraction whereas upon the overexpression of SUP35–GFP, genomic Sup35p exhibited higher sedimentation with or without Sup35Cp, indicating that the average size of [PSI+] aggregates increases upon the overexpression of SUP35–GFP (Fig. 4D).

SUP45, but not other Sup35p associating factors, relieves growth inhibition caused by overexpression of Sup35p but not Sup35NMp in [PSI+] cells

To determine which Sup35p interacting protein might be sequestered, a strain containing [PSI+] capable of overexpressing SUP35 was transformed with plasmids containing genes encoding essential proteins that interact with Sup35p. These genes included NAB3, NRD1, PAB1, SPT15, ARP2, ACT1, SIS1 and SUP45 (Stansfield et al., 1995; Cosson et al., 2002; Gavin et al., 2002; Sanders et al., 2002; Ganusova et al., 2006; Bagriantsev et al., 2008). If Sup35p overexpression caused growth inhibition because Sup35p sequesters these essential proteins into the [PSI+] aggregate, then overexpression of these proteins should eliminate the growth inhibition. As previously reported (Derkatch et al., 1998), Sup45p expressed on a high copy plasmid was able to rescue this inhibition (Fig. 5A), whereas we now show that a lower expression of Sup45p on a centromeric vector is unable to rescue the growth inhibition (Fig. 5A). All other genes failed to complement growth (data not shown).

Figure 5.

Overexpressed SUP45 rescues growth inhibition caused by overexpressed SUP35 but not SUP35NM in [PSI+] cells.
A. Overexpressed SUP45 rescues growth inhibition caused by overexpressed SUP35 in [PSI+] cells. Serial dilutions are of a strain containing strong [PSI+] with an integrated GAL::SUP35 transformed with a control vector, or SUP45 on a low copy or a high copy plasmid spotted on plasmid selective media where SUP35 is not (SD-Trp) or is (SGal-Trp) overexpressed.
B. Overexpressed SUP45 does not rescue growth inhibition caused by overexpressed SUP35NM in [PSI+] cells. Diploids were selected from different mating reactions with continued overexpression of 100×SUP35NM and selection for maintenance of vectors and overexpression of SUP45: a wild-type strain overexpressing 100×SUP35NM with either expression of a control vector (Panel A), a low level of SUP45 (Panel B) or a high level of SUP45 (Panel C) crossed to [psi], weak or strong [PSI+] variants.

As Sup45p was capable of rescuing the growth inhibition, three Sup45p interacting proteins were overexpressed to determine if they would restore growth. TPA1, DBP5 and MLC1 (Valouev et al., 2004; Keeling et al., 2006; Gross et al., 2007) were chosen for their ability to interact with Sup45p and their involvement in translation termination or the ability to complement a SUP45 temperature sensitive mutation. However, none of these factors were able to rescue the growth inhibition (data not shown).

Since the growth inhibition was rescued by overexpression of Sup45p, we asked if the toxicity was caused by a reduction in the level of Sup45p in the cell. Thus, we compared Sup45p levels in a [PSI+] strain transiently overexpressing Sup35–GFPp to [PSI+] cells that did not overexpress Sup35–GFPp (Fig. S2). As the level of Sup45p is not reduced, toxicity is not caused by a decrease in Sup45p expression.

As overexpressed Sup45p-rescued [PSI+] growth inhibition due to overexpressed Sup35p, we checked whether Sup45p could also rescue growth inhibition due to overexpressed Sup35NMp in [PSI+] strains. Wild-type strains simultaneously overexpressing SUP35NM at a 100× level and SUP45 either on a low (low level) or high (high level) copy plasmid were crossed to [psi], weak and strong [PSI+] strains. Weak and strong [PSI+] diploids did not show enhanced growth even in the presence of SUP45 overexpression on a high copy plasmid (Fig. 5B).

Sup45–GFPp co-aggregates with [PSI+] aggregates in the presence of SUP35 but not SUP35NM overexpression

Since toxicity is not caused by reducing the absolute level of Sup45p in the cell, we asked if toxicity could be caused by inactivation of Sup45p via aggregation. Thus, we checked visually whether Sup45p exists as aggregates in the presence of [PSI+] and overexpressed SUP35 or SUP35NM. A strain containing endogenously GFP tagged SUP45 was transformed with plasmids overexpressing SUP35, SUP35NM or a control vector. These were mated to either [psi] or [PSI+] yeast and Sup45–GFPp fluorescence was monitored. Approximately half the [PSI+] zygotes overexpressing SUP35 showed bright Sup45–GFPp punctate dots whereas 98% of the [psi] zygotes showed diffuse Sup45–GFPp (Fig. 6A, Table 1, rows 1 and 2). [PSI+] diploids overexpressing SUP35NM, or not overexpressing either SUP35 or SUP35NM (vector), also showed diffuse Sup45–GFPp (Table 1, rows 3 and 4).

Figure 6.

Sup45–GFPp is sequestered into [PSI+] aggregates in cells overexpressing SUP35.
A. Sup45–GFPp forms detectable foci in [PSI+] cells overexpressing SUP35. A strain with a genomic copy of SUP45–GFP, overexpressing 10×SUP35 (MATa) was mated to either a strain containing [PSI+] or [psi] yeast (MATα) and fluorescence was monitored in zygotes. Two representative zygotes for the [PSI+] mating are shown.
B. Sup45–GFP aggregates colocalize with [PSI+] aggregates. A strain with a genomic copy of SUP45–GFP, overexpressing 10×SUP35 (MATα) was mated to a [PSI+] strain (MATa) with a genomic copy of SUP35 tagged with RFP and fluorescence was monitored in zygotes.

Table 1.  Percentage of zygotes with punctate Sup45–GFP dots.
Crosses% of zygotes with Sup45–GFP puncta
Parent 1XParent 2
  1. Yeast strains with the indicated genotypes were mated for ∼4 h and zygotes with Sup45–GFP puncta were counted using a fluorescence microscope. SUP35C was expressed from a high copy plasmid and overexpressed on SD-Leu (See Experimental procedures). At least 5 independent mating reactions were performed and 100–150 zygotes were counted for each reaction. Standard deviation is shown.

SUP45–GFP, 10×SUP35X[PSI+]55 ± 8.3
SUP45–GFP, 10×SUP35X[psi]2.2 ± 0.44
SUP45–GFP, 100×SUP35NMX[PSI+]< 1%
SUP45–GFP, 10× VectorX[PSI+]0
SUP45–GFP, 10×SUP35X[PSI+], ↑ Sup35Cp9.3 ± 1.47

Next we checked whether these Sup45–GFPp aggregates colocalized with coalesced [PSI+] aggregates. The SUP45–GFP strain overexpressing Sup35p was mated to a [PSI+] strain, in which the endogenous Sup35 was tagged with RFP (kindly supplied by T. Serio; see Experimental procedures) and fluorescence was monitored in the zygotes. Sup45–GFPp puncta were either completely colocalized (Fig. 6B, bottom), or partially overlapped with Sup35-RFPp puncta (Fig. 6B, top), suggesting that Sup45–GFPp is sequestered into Sup35-RFPp puncta.

As we hypothesized that Sup35Cp rescues [PSI+] mediated toxicity in the presence of high levels of Sup35p by drawing an essential protein out of the prion aggregate (Fig. 1C), we asked if overexpressed Sup35Cp kept Sup45p out of the prion aggregate. The SUP45–GFP strain overexpressing Sup35p was mated to a [PSI+] strain overexpressing Sup35Cp and fluorescence was monitored. Indeed, significantly fewer [PSI+] zygotes simultaneously overexpressing Sup35p and Sup35Cp had Sup45–GFPp puncta (∼10%) (Table 1, row 5) compared with zygotes overexpressing only Sup35p (∼55%).


The presence of the [PSI+] or [PIN+] prion by themselves is not detrimental to yeast cells. However, the overexpression of SUP35 in [PSI+] and [PIN+] cells leads to growth inhibition (Dagkesamanskaya and Ter-Avanesyan, 1991; Derkatch et al., 1996). Two hypotheses have been proposed to explain the growth inhibition associated with overexpressed SUP35 in [PSI+] and [PIN+] cells (Derkatch et al., 1996; 1998). One proposes that a toxic species of Sup35p is formed that is incompatible with the cell. It appears that this is the case in toxicity caused by overexpression of Rnq1p in [PIN+] cells, wherein a toxic oligomer is associated with cell death (Douglas et al., 2008). The alternate, not mutually exclusive, hypothesis proposes that aggregated Sup35p sequesters, and thereby inactivates, an essential protein. Since Sup35Cp that does not join the [PSI+] aggregate alleviates the growth inhibition, it seemed possible that Sup35p itself was the essential sequestered protein (Ter-Avanesyan et al., 1993). Supporting this hypothesis, other groups and we have observed that one copy of SUP35C was enough to mitigate the growth inhibition of 10×SUP35 in [PSI+] cells (Ter-Avanesyan et al., 1994). However, at higher levels of overexpressed Sup35p, the hypothesis was not supported by our finding that one copy of SUP35C is not enough to relieve the growth inhibition of 100×SUP35 in [PSI+] cells. This is because if Sup35p were the essential factor being sequestered, even at a 100× level of overexpression one copy of SUP35C should be able to rescue growth inhibition, as Sup35Cp does not join the prion aggregate.

We show that at a 100×SUP35 level, a proportional increase in Sup35Cp is able to rescue the growth defect. This suggests that at 100×SUP35, rather than Sup35p being sequestered, a Sup35p binding protein is sequestered into the [PSI+] aggregates. Because Sup35Cp competes with Sup35p for this binding partner, a proportional increase in Sup35Cp can alleviate the growth inhibition. Furthermore, weak [PSI+] variants, which have less aggregated Sup35p (Zhou et al., 1999; Uptain et al., 2001), are less growth inhibited, indicating that the severity of growth inhibition decreases with an increase in soluble Sup35p.

Additionally, as growth inhibition in [PIN+] cells caused by 100×SUP35 could not be relieved by one copy of SUP35C, but could be relieved by overexpression of Sup35Cp, it indicates that [PIN+] associated toxicity with overexpressed Sup35p also involves the loss of a Sup35p binding partner. Since growth inhibition occurs in very high [PIN+] but not in medium [PIN+], although very high [PIN+] has more soluble Rnq1p than medium [PIN+] (Bradley et al., 2002), it appears that it is not the amount of aggregated Rnq1p but rather the Sup35p aggregate seeding potential that leads to toxicity. Since overexpression of Sup35NMp did not cause growth inhibition in [PIN+] cells, although it leads to a higher frequency of [PSI+] induction than overexpression of full-length Sup35p (Derkatch et al., 1996), the formation of [PSI+] cannot be the sole cause of toxicity. Rather, a combination of the formation of [PSI+] and the ability of the aggregated, inducing fragment to directly sequester a Sup35Cp binding partner must lead to the growth defect.

To identify factors responsible for [PSI+] associated toxicity with overexpressed Sup35p, we overexpressed various essential genes whose products are known to associate with Sup35p. If overexpressed Sup35p were sequestering these factors into the [PSI+] aggregate, their overexpression should relieve the growth inhibition. Indeed, we have previously showed that excess Sup45p caused rescue (Derkatch et al., 1998). While it is possible that factors that were not tested complement the growth defect, an overexpression screen (Tank and True, 2009 reported while this paper was under review) to identify factors that ameliorate the toxicity of excess Sup35p in [PSI+] cells revealed only Hsp104p (which cures cells of [PSI+]; Chernoff et al., 1995) and Sup45p. Also, we found that when factors known to associate with Sup35p and associated with a variety of processes from the actin-cytoskeleton machinery (ACT1) (Ganusova et al., 2006) to RNA processing (PAB1) (Cosson et al., 2002) were overexpressed, only SUP45 was able to rescue the cells. Likewise, cells were not rescued by overexpression of a variety of proteins that associate with Sup45p, and are involved in translation termination (e.g. DBP5) (Gross et al., 2007) that might have allowed Sup45p to remain soluble. This may reflect the fact that the association between Sup35p and Sup45p is stronger than other interactions, which would explain why only Sup35Cp and Sup45p were able to relieve to growth inhibition caused by overexpressed SUP35. Growth inhibition caused by sequestration of Sup45p could be caused either by Sup45p monomers directly binding to the C-terminal domain of Sup35p in the aggregates or by the transient formation of Sup35p:Sup45p complexes that then join the Sup35p aggregate via the Sup35p prion domain in the complex. In either case, we propose that Sup35Cp rescues growth inhibition due to overexpressed Sup35p by retaining Sup45p in the cytosol and that Sup45p overexpression rescues because there are then enough Sup45p molecules to carry on the function of translation termination.

We used microscopy to visualize [PSI+] aggregates when Sup35p is overexpressed and to show the accumulation of Sup45p in these aggregates. Previous work showed that when [PSI+] yeast with chromosomally tagged SUP35–GFP expressed under a promoter that is repressed upon mating is crossed to an untagged [psi] yeast strain, the approximate size of the SUP35–GFP aggregates is not altered in the zygotes. Furthermore, this was true even when prion fragmentation (Hsp104p) was inhibited in the zygote (see fig. 2c in Satpute-Krishnan et al., 2007). In these zygotes, inhibition of fragmentation presumably causes the aggregates to enlarge, but this is invisible because the newly made Sup35p joining and enlarging the aggregates is untagged. Thus, we were surprised when our similar mating of a [PSI+] strain (with chromosomally tagged Sup35–GFPp expressed under a promoter that is repressed upon mating) to an untagged [psi-] strain that overexpressed untagged Sup35p, caused the immediate appearance of large fluorescent foci in the zygotes.

While it is expected that the large excess of aggregates that result from the Sup35p overexpression will overwhelm the fragmentation machinery, this by itself, is unlikely to explain the appearance of the large foci in the zygotes because, as described above, even inhibition of Hsp104p did not increase fluorescent foci size. Thus, we propose that when untagged Sup35p was overexpressed in GFP tagged [PSI+] zygotes, the overexpressed Sup35p ‘glues’ the pre-existing fluorescent aggregates together into large, ill-defined foci. These large Sup35p aggregates are not toxic per se, as they are present even in the presence of Sup35Cp. However, only zygotes with Sup35Cp are able to propagate, suggesting Sup45p is sequestered in the [PSI+] aggregates. Indeed, we show that the essential protein Sup45p co-aggregates with these Sup35p foci and that Sup35Cp keeps some of the Sup45p out of the large aggregates.

The ‘gluing’ of pre-existing [PSI+] aggregates might be due to the SUP35NM domain having a high glutamine/asparagine (Q/N) content, which has a high propensity to aggregate, a property shared by all known native yeast prions (DePace et al., 1998; Osherovich et al., 2004; Ross et al., 2005). We propose that, initially, when [PSI+] aggregates and overexpressed Sup35p come together in the zygote, pre-existing aggregates coalesce and the prion shearing mechanism is overwhelmed, which leads to the formation of these large aggregates. In diploid cells continuously expressing chromosomally encoded SUP35–GFP and overexpressing untagged Sup35p, aggregates larger than normal [PSI+] aggregates are observed, although they are not as large as those in the zygote experiment. This suggests that either the prion shearing mechanism partially recovers in the diploid cultures, or that smaller aggregates are preferentially transferred to, and therefore propagated in, daughter cells.

One hypothesis explaining cell death in neurodegenerative diseases proposes that the aggregating proteins might interact aberrantly with several factors (Caughey and Baron, 2006; Gil and Rego, 2008). For example, numerous transcription factors have been found to interact with the disease associated protein of Huntington's disease. Overexpression of one such factor, Creb binding protein (CBP), showed relief of toxicity in cell culture, suggesting that depletion of CBP function contributed to toxicity (Nucifora et al., 2001; Jian et al., 2006). Although Sup45p associates with Sup35p to perform the function of translation termination, the presence of Sup45p in [PSI+] aggregates is still unclear as one study found Sup45p associated with [PSI+] aggregates whereas another did not (Patino et al., 1996; Paushkin et al., 1997). Although we do not see Sup45p in the [PSI+] pellet in wild-type strains (data not shown), we do see Sup45–GFPp puncta colocalizing with [PSI+] aggregates in cells overexpressing Sup35p. We propose that upon overexpression of SUP35, Sup45p is abnormally sequestered resulting in cell death. Thus the association of Sup45p and Sup35p, which is required for translation termination, leads to toxicity during the overexpression of Sup35p in strains containing [PSI+].

In contrast to the toxicity caused by overexpression of the full-length Sup35p, toxicity due to overexpression of Sup35NMp in [PSI+] was not rescued by high levels of Sup45p. Furthermore, Sup45–GFPp formed puncta only in [PSI+] cells overexpressing Sup35p but not Sup35NMp, suggesting overexpression of Sup35NMp does not sequester Sup45p. Rather, we propose, that overexpressed Sup35NMp forms an aggregate that captures a good deal of full-length Sup35p, whose depletion leads to growth inhibition. Similarly, during the de novo formation of [PSI+] by overexpression of SUP35NM–GFP in [PIN+] yeast, cells with large ring and worm-like SUP35NM–GFP aggregates are less viable (Zhou et al., 2001; Ganusova et al., 2006), but we find they can be rescued by Sup35Cp. Since these ring-like aggregates are a hallmark of newly appearing [PSI+], we propose that the large SUP35NM–GFP ring-like structures interact with and sequester Sup35p causing the observed toxicity.

Although Sup35p is in an aggregated form in [PSI+] cells, there must be enough functional Sup35p and Sup45p to carry on the essential function of translational termination. Thus, in normal [PSI+] cells there is a fine balance between the [PSI+] aggregates and functional translation termination complexes. Our results suggest that in [PSI+] cells highly overexpressing Sup35p or Sup35NMp, sequestration of Sup45p or Sup35p, respectively, into [PSI+] aggregates leads to growth inhibition.

Experimental procedures


SUP35 and SUP35NM overexpression plasmids (p743 and p749 respectively) are in a pEMBL vector under their own promoters (Ter-Avanesyan et al., 1993). These 2 μ plasmids contain both the URA3 and LEU2-d markers. While these plasmids are normally maintained at 10–20 copies per cell in ura3 leu2 hosts, on media lacking uracil (SD-Ura), they can be amplified to around 100 copies by selecting on SD-Leu since the LEU2-d promoter is defective (Derkatch et al., 1996).

GAL::SUP35 in pRS403 (Sikorski and Hieter, 1989) is an integrative vector with a HIS3 marker (p1104). This was created by cutting the galactose-inducible promoter (using XhoI and BamHI) from a pRS316 (Sikorski and Hieter, 1989) vector containing the promoter and ligating it into pRS403 (also cut with XhoI and BamHI) (Sikorski and Hieter, 1989). SUP35 was cut from pVK71 (Derkatch et al., 1996) (using Xba1 and Sma1) and inserted into pRS403 containing the galactose-inducible promoter.

p1160 is a centromeric plasmid with a URA3 marker with SUP35C under its own promoter and p750 is a pEMBL vector with SUP35C under its own promoter (Ter-Avanesyan et al., 1993). p1594 is a centromeric pGAL::SUP35C vector that was created by amplifying SUP35C by PCR and ligating it into a pRS414 (TRP1 marker) (Sikorski and Hieter, 1989) vector containing a galactose-inducible promoter. Expression from this plasmid is defined as a low level of Sup35Cp overexpression. GAL::SUP35C in pRS424 (p1598) is a 2 μ vector with a TRP1 marker and was created by cutting p1594 with XhoI and NotI (removing the galactose promoter and SUP35C) and ligated at those sites in pRS424. Expression from this plasmid is defined as high level of Sup35Cp overexpression.

p1182 is a centromeric plasmid with a LEU2 marker that contains Sup35NM–GFPp under the copper-inducible promoter (Zhou et al., 2001). To induce Sup35NM–GFPp, transformants were grown overnight in SD-Leu and reinoculated into copper containing media for 24–48 h.

To create pGAL::SUP35–GFP, the SUP35–GFP fusion was cut from pCup1::SUP35–GFP (p1083) (Zhou et al., 2001) using BamHI and SacI and ligated into a pRS413 (Sikorski and Hieter, 1989) centromeric vector (HIS3 marker), containing a galactose-inducible promoter, cut with the same restriction sites.

Plasmids containing SUP35 and SUP45 interacting genes were kind gifts from various people: PAB1 (pKF142, URA3 2 μ) from Curt Wittenberg; ARP3 (pDW8, URA3 CEN) from Rong Li; SPT15 (M4480, LEU2 2 μ) from David Stillman; NAB3 (425-NAB3, LEU2 2 μ) from Karen Arndt; NRD1 (316-NRD1, URA3 CEN) from David Brow; ACT1 (pRB1454, URA3 CEN) from Dan Burke; TPA1 (pDB1005, TRP1 CEN) from David Bedwell; DBP5 (pCS833, LEU2 2 μ) from Charles Cole; MLC1 (pRS290, TRP1, 2 μ) from Trisha Davis; pRS316Gal-SSB1 from Yury Chernoff.

SUP45 with its promoter was cut from Yep13-SUP45 (p757) (Derkatch et al., 1998) using XhoI, BamHI, blunt ended and ligated into either pRS314 or pRS324 (cut with BamHI and blunt ended) to create SUP45 with its promoter in a centromeric vector (p1672) or a 2 μ vector (p1673) with a TRP1 marker. Expression from p1672 and p1673 are defined as low and high level respectively.

Yeast strains and media

The following yeast strains are derivatives of 74-D694 (MATa ade1-14 leu2-3,112 his3-Δ200 trp1-289 ura3-52) (Chernoff et al., 1993): [psi] (L1751); weak [PSI+] (L1759); strong [PSI+] (L1763) (Derkatch et al., 1997); low [PIN+] (L1943); medium [PIN+] (L1945); high [PIN+] (L1749), very high [PIN+] (L1953) (Derkatch et al., 2000; Bradley et al., 2002). Wild-type (L3059) and ΔNM (L3058) strains are derivatives of SL1010-1A (MATαade1-14 met8-1 leu2-1 his5-2 trp1-1 ura3-52) (Zhou et al., 1999). [PSI+] and [psi] yeast with endogenous SUP35 tagged with GFP (MATα) and strains containing an additional copy of GFP tagged SUP35 under the MFA promoter (MATa) were kind gifts from Tricia Serio (Satpute-Krishnan and Serio, 2005). Unlike previous C-terminal fusions of GFP to SUP35 that are not functional, SUP35 in these strains is tagged with GFP between the N and M domains and remains functional. MATa versions of the MATα strains were obtained by mating to an isogenic 74-D694 strain. [PSI+] yeast with endogenous SUP35 similarly tagged with RFP (MATa) was also a kind gift from Tricia Serio (SY831). YBR143C (MATa) is a strain containing an endogenously tagged SUP45–GFP from Invitrogen. To create zygotes with this strain, MATα versions of [psi] (L1754: MATα ade1-14 leu2-3112 ura3-52 trp1-289 lys9-A21) (Derkatch et al., 1997) and strong [PSI+] (L2717: made by cytoducing [PSI+] into L1754) were used. Mating type of YBR143C was switched to MATα by overexpressing the HO gene.

To make [psi] (L3014) and strong [PSI+] (L3013) yeast with an integrated GAL::SUP35, an integrative plasmid GAL::SUP35 in pRS403 (p1104, HIS3) was cut at the unique StuI site in SUP35, transformed into L1751 ([psi]) and L1763 ([PSI+]) and selected on SD-His.

Standard yeast media and cultivation procedures were used (Sherman et al., 1986). Transformants were grown in media selective for the plasmids, e.g. synthetic dextrose lacking tryptophan (SD-Trp). Diploids of 74-D694 and SL1010-1A were selected on SD-His with the appropriate selection for the various plasmids. For 10× overexpression of SUP35 or SUP35NM, diploids were selected on SD-Ura-His and replica plated to SD-Ura-His-Leu to increase overexpression to 100×. Complex glucose media (YPD) was used to distinguish colour of yeast strains.

To express SUP35C under the galactose-inducible promoter, the media contained 2% raffinose along with 2% galactose.

To monitor fluorescence in zygotes, strains were mated on YPD for 2.5–4 h and fluorescence was observed using a Zeiss AxioScope2 with a 63× lens.

Analysis of viability and [PSI+] phenotype of ring bearing cells

To determine the viability of cells with the Sup35p rings, a high [PIN+] variant transformed with a copper-inducible Sup35NM–GFPp plasmid (p1182), with or without SUP35C, was grown in plasmid selective media overnight and induced with 50 μM copper for 24–48 h. Individual cells with rings or diffuse Sup35NM–GFPp were micromanipulated onto agar patches while observing them under a fluorescence microscope at 20×. Patches were transferred to a YPD plate and colonies were grown and assayed for the [PSI+] phenotype.

To monitor the [PSI+] or [psi] state of 74-D694 strains, the level of readthrough of the premature stop codon in the ade1-14 allele was used (Chernoff et al., 1995). [PSI+] cells readthrough the premature stop codon, as there is less functional Sup35p. Thus, they are able to grow on SD-Ade and are white (strong [PSI+]) or pink (weak [PSI+]) on YPD. [psi] cells on the other hand, terminate efficiently at the stop-codon, and are thus unable to grow on SD-Ade and are red on YPD. White and pink colonies that grew on YPD from the micromanipulated cells were confirmed to be [PSI+] by mating to a [psi][pin] tester strain (L2174) bearing the copper-inducible Sup35NM–GFPp plasmid. [PSI+] cells have Sup35NM–GFPp foci whereas [psi] cells show diffuse Sup35NM–GFPp fluorescence.

Preparation of yeast cell lysates and protein analysis

Crude cell extracts were prepared using glass beads (Biospec, 0.5 mm) in 750 μl of lysis buffer containing 50 mM Tris-HCl, pH 7.5, 50 mM KCl, 10 mM MgCl2 and 5% (w/v) glycerol supplemented with a protease inhibitor cocktail (P8215, Sigma) and 5 mM PMSF. Cells were lysed by vortexing (Vortex-Genie 2) at high speed, three times for 2 min with cooling on ice for 1 min between each vortexing at 4°C. Crude lysates were pre-cleared by centrifuging two times at 600 g for 2 min at 4°C to remove unlysed cells and cell debris.

For high-speed centrifugation analysis, 500–750 μg of crude lysate in 250 μl was spun at 100 000 g for 30 min at 4°C in a Sorvall TLA100.1 rotor. The supernatant and pellet were run on a 10% Tris-HCl gel. Sup35p was detected using the monoclonal βe4 Sup35p antibody.

Sucrose gradient centrifugation was performed essentially as described previously (Bagriantsev and Liebman, 2004). Briefly, lysates were spun in a continuous 20–60% gradient in a swinging bucket rotor at 10 600 g for 40 min at 4°C. Fractions were diluted 1:2 in lysis buffer, resolved on a 10% acrylimide gel and transferred onto an immunoblot polyvinylidene difluoride membrane (Bio-Rad). This was then probed using the Sup35Cp βe-4 monoclonal antibody. Signal was detected with a Tropix kit (Applied Biosystems) using the manufacturer's specifications.


We thank Tricia Serio (Brown University) for sharing published and unpublished strains with endogenous SUP35 tagged with GFP and RFP, Viravan Prapapanich for preparation of anti-Sup35C antibody, Joo Hong for creating plasmid p1104, Vidhu Mathur for help with experiments, the people who kindly gave us the many plasmids mentioned in Experimental procedures and Anita Manogaran, Nava Segev and David Stone for comments about the manuscript. This work was supported by a grant (GM-56350) from the National Institutes of Health to S.W.L. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.