The relationship between visible intracellular aggregates that appear after overexpression of Sup35 and the yeast prion-like elements [PSI+] and [PIN+]


  • Ping Zhou,

    Search for more papers by this author
    • Present address: Department of Ophthalmology and Visual Sciences, Washington University, St Louis, MO 63110, USA.

  • Irina L. Derkatch,

    1. Laboratory for Molecular Biology, Department of Biological Sciences, University of Illinois at Chicago, Chicago, IL 60607, USA.
    Search for more papers by this author
  • Susan W. Liebman

    Corresponding author
    • *For correspondence. E-mail SUEL@uic. edu; Tel. (+1) 312 996 4662; Fax (+1) 312 413 2691. †Present address: Department of Ophthalmology and Visual Sciences, Washington University, St Louis, MO 63110, USA.

    Search for more papers by this author


Overproduced fusions of Sup35 or its prion domain with green fluorescent protein (GFP) have previously been shown to form frequent dots in [PSI+] cells. Rare foci seen in [psi] cells were hypothesized to indicate the de novo induction of [PSI+] caused by the overproduced prion domain. Here, we describe novel ring-type aggregates that also appear in [psi] cultures upon Sup35 overproduction and show directly that dot and ring aggregates only appear in cells that have become [PSI+]. The formation of either type of aggregate requires [PIN+], an element needed for the induction of [PSI+]. Although aggregates are visible predominantly in stationary-phase cultures, [PSI+] induction starts in exponential phase, suggesting that much smaller aggregates can also propagate [PSI+]. Such small aggregates are probably present in [PSI+] cells and, upon Sup35–GFP overproduction, facilitate the frequent formation of dot aggregates, but only the occasional appearance of ring aggregates. In contrast, rings are very frequent when [PSI+] cultures, including those lacking [PIN+], are grown in the presence of GuHCl or excess Hsp104 while overexpressing Sup35–GFP. Thus, intermediates formed during [PSI+] curing seem to facilitate ring formation. Surprisingly, GuHCl and excess Hsp104, which are known to promote loss of [PSI+], did not prevent the de novo induction of [PSI+] by excess Sup35 in [psi][PIN+] strains.


Prion diseases in mammals appear to result from the conversion of soluble PrP protein into a self-propagating abnormal structure (prion) (Caughey and Chesebro, 1997; Horwich and Weissman, 1997; Prusiner et al., 1998). The Saccharomyces cerevisiae non-Mendelian element, [PSI+] (Cox, 1965; Cox et al., 1988), was proposed to be a prion form of Sup35 (Wickner, 1994) on the basis of genetic evidence (Chernoff et al., 1993; Doel et al., 1994; Ter-Avanesyan et al., 1994), and this hypothesis has been supported by additional experiments (Chernoff et al., 1995; Derkatch et al., 1996; Patino et al., 1996; Paushkin et al., 1996; 1997; Glover et al., 1997; King et al., 1997; DePace et al., 1998).

Sup35, also called eRF3, is a subunit of the translational termination complex (Stansfield et al., 1995; Zhouravleva et al., 1995). Sup35 is composed of three domains. The C-terminal domain (Sup35C) is essential for viability (Ter-Avanesyan et al., 1993) and sufficient for translational termination. The middle region (Sup35M) has no known function. The N-terminal region of Sup35 (Sup35N) is the prion domain because Sup35N is required for [PSI+] propagation (Doel et al., 1994; Ter-Avanesyan et al., 1994), and overexpression of Sup35N alone is sufficient to induce the appearance of [PSI+] de novo (Derkatch et al., 1996). The induction of [PSI+] by the overexpression of Sup35 requires the presence of the prion-like factor [PIN+] (Derkatch et al., 1997). However, [PSI+] strains can be either [PIN+] or [pin] and, when [PSI+][pin] strains are cured of [PSI+], they cannot be reinduced to become [PSI+] by Sup35 overexpression (Derkatch et al., 2000).

In [psi] cells, Sup35 is in a functional soluble conformation but, in [PSI+] cells, most Sup35 takes on an alternative aggregated prion conformation (Patino et al., 1996; Paushkin et al., 1996), leading to the increased readthrough of stop codons detected as nonsense suppression. Remarkably, aggregated Sup35 can convert soluble Sup35 into aggregates in vitro (Paushkin et al., 1997). Based on these results and on the in vitro formation of amyloid-like fibres composed of Sup35 or its N-terminal prion domain, the molecular mechanism of the conversion has been suggested to be a seeded polymerization of soluble Sup35 into fibres (Glover et al., 1997; King et al., 1997; Paushkin et al., 1997; DePace et al., 1998).

Dot-like fluorescent foci were observed in most cells in [PSI+] cultures overproducing the green fluorescent protein (GFP) fused to Sup35 or its prion domain (Patino et al., 1996). Because in general such punctate foci were not observed in [psi] strains, these foci were hypothesized to be linked to [PSI+]. Furthermore, rare foci were seen in [psi] cells and were assumed to indicate the de novo induction of [PSI+] caused by the overexpression of the prion domain. Here, we examine dot and novel ring aggregates that appear in cells overexpressing constructs containing the [PSI+] prion domain and relate these aggregates to the appearance and/or presence of [PSI+].


Appearance of novel aggregates of Sup35 during the induction of [PSI+]

Overproduction of Sup35 in [psi] cells leads to the de novo appearance of [PSI+] (Chernoff et al., 1993; Derkatch et al., 1996), presumably because the excess Sup35 increases the possibility of the formation of prion seeds (Wickner, 1994). To visualize the formation of Sup35 aggregates in vivo during the de novo induction of [PSI+], fusions of GFP with Sup35 (Sup35–GFP), Sup35N (Sup35N–GFP) or Sup35NM (Sup35NM–GFP) driven by the copper-inducible CUP1 promoter were expressed in [psi][PIN+] strains 74-D694 and SL1010-1A for 48 h (see Experimental procedures). As described previously (Patino et al., 1996), single or multiple dot aggregates, which are believed to be indicative of the de novo induction of [PSI+], appeared in some cells (Fig. 1A). Interestingly, we also observed the appearance of novel types of aggregates that were in a ring (Fig. 1B–F and H–J), curve (Fig. 1B–E, G and I) or line shape (Fig. 1B), henceforth referred to as ring aggregates. These rings were usually smooth, but rings with a thicker spot or region (Fig. 1E) or branched rings (Fig. 1G) were occasionally observed. Ring-bearing cells usually had a single aggregate, but occasionally contained a few lines or rings (Fig. 1J). Some rings appeared to be twisted (Fig. 1C, E and F). The rings were usually nearly as large as the cells (Fig. 1B–F) but could be of different sizes (Fig. 1H and I). Occasionally, rings were seen to thicken (Fig. 2B), and small rings (some of which could have been dots) were seen to become big rings (data not shown). Aggregates were not seen when cells were transformed with control plasmids that overproduced only GFP, suggesting that the prion domain of Sup35 is responsible for the formation of the aggregates.

Figure 1.

Visualization of highly ordered aggregates of Sup35NM–GFP and Sup35 proteins.

A–J. Sup35NM–GFP was expressed in [psi][PIN+] 74-D694 for 48 h.

K–Q. Cells stained with anti-Sup35 antibody after [psi][PIN+] 74-D694 cells transformed with pEMBL-SUP35 were freshly grown on synthetic complete medium lacking uracil for 3 days.

A, E–G and I–Q. Fluorescence images.

B, C and H. Fluorescence images with differential interference contrast (DIC) to show the positions of ring-like aggregates within cells.

D. The same cells using a fluorescence image (left), fluorescence image with DIC optics (middle) and DIC image (right). Rings were often in different planes and required up and down focusing to view the complete structures but, in the cells shown, they were nearly in the same plane.

Figure 2.

Appearance of aggregates in the progeny of aggregate-bearing cells. Cells of [psi][PIN+] 74-D694 transformed with pSUP35NM-GFP were grown in SC–Ura + CuSO4 for 40 h to induce dots and rings and then transferred to slides containing a thin layer of SC–Ura with 5 µM CuSO4 or YPD (similar results, data not shown). Cells with a ring (A and B) or dot (C) were examined at the indicated time points under a fluorescence microscope.

The location of the aggregates was independent of the location of the nucleus as determined by DAPI staining. They were also not associated with actin. Indeed, when cells were stained with Texas red–X phalloidin, many tiny spots appeared throughout the cells, and these actin spots failed to co-localize with the Sup35NM–GFP aggregates. When cells were treated with 200 µM actin-depolymerizing agent latrunculin A, the vast majority of the actin spots were gone, but the Sup35NM–GFP aggregates remained undisturbed (data not shown).

Ring aggregates appear to be heritable, as cells containing them were often found in clusters (Fig. 1B–G). Some ring-like aggregates extended directly into daughter cells (Fig. 1C and D). Daughters with rings arose from mothers with rings, whereas daughters with dots arose from mothers with either dots or rings. Some daughters showed a delayed appearance of aggregates (Fig. 2), suggesting that some aggregates are too small to be visualized.

Sup35 ring-like aggregates are formed by overexpression of native Sup35

The appearance of ring-like aggregates is independent of the CUP1 promoter because rings, along with the previously reported (Patino et al., 1996) dots and [PSI+] induction, were also observed when Sup35NM–GFP driven by the glucocorticoid-inducible promoter (GRE) was used (data not shown). To test whether Sup35 without any tag is able to form rings, native Sup35 was overproduced in [psi][PIN+] 74-D694 by introducing the multicopy plasmid pEMBL-SUP35. Cells were fixed, spheroplasted and immunostained with Sup35 polyclonal antibody. Dots and rings, similar to those found in cells expressing Sup35NM–GFP (Fig. 1A–J), were observed in 4.7 ± 0.7% and 4.7 ± 1% cells, respectively (Fig. 1K–Q), whereas no aggregates were seen when control plasmids not overexpressing Sup35 were used. Similar results were also obtained in another strain, [psi][PIN+] SL1010-1A (data not shown).

When Sup35NM tagged with a c-Myc epitope (Sup35NM–Myc) was expressed in the [psi][PIN+] strain 74-D694 and cells were stained with anti-Myc antibody, rings (and dots) were also observed (data not shown).

[PIN+] is required for the formation of both dot- and ring-like aggregates during de novo induction of [PSI+]

The de novo induction of [PSI+] by overexpression of Sup35 or Sup35NM requires the non-Mendelian element [PIN+] (Derkatch et al., 1997; 2000). Thus, if the dot and ring aggregates were associated with the induction of [PSI+], they should fail to appear in [psi][pin] strains. Indeed, after 48 h of overexpression of Sup35NM–GFP or Sup35NM–Myc in [psi][pin] derivatives of 74-D694 or SL1010-1A, none of the cells showed rings, dots (Fig. 3) or [PSI+] induction, although amorphous aggregates (not shown) appeared in ≈ 1–2% of the cells expressing Sup35NM–GFP. Likewise, overexpression of native Sup35 protein in these [pin] strains failed to produce rings that could be detected by anti-Sup35 antibody. Similar results were obtained when HSP104 null derivatives of 74-D694 and SL1010-1A were examined. This was expected as the propagation of both [PSI+] (Chernoff et al., 1995) and [PIN+] (Derkatch et al., 1997) requires the presence of the Hsp104 chaperone.

Figure 3.

Formation of dot or ring aggregates in a [psi] culture requires [PIN+]. Fluorescence images of pSUP35NM-GFP transformants of derivatives of 74-D694 that were [psi][PIN+] (A) and [psi][pin] (B), grown in SC–Ura + 20 µM CuSO4 for 65 h.

The frequency and appearance of ring and dot aggregates depends upon the length of the Sup35 prion domain and its level of overexpression

Ring and dot aggregates appeared most frequently in cells expressing Sup35N–GFP, less frequently in cells expressing Sup35NM–GFP and least frequently in cells expressing Sup35–GFP (Table 1). Likewise, Sup35N–GFP is more efficient in induction of [PSI+] than Sup35NM–GFP and much more efficient than Sup35–GFP (Table 1). Therefore, the frequency of the appearance of both ring and dot aggregates paralleled the frequency of the appearance of [PSI+] induced by these fusion proteins.

Table 1. The frequency of cells with different aggregates during the de novo induction of [PSI+] in [psi][PIN+] strain 74-D694.

Percentage of cells with
DotsRings [PSI + ] induction
  1. Strains transformed with plasmids encoding the indicated proteins were incubated for 48 h in liquid SC–Ura medium supplemented with 50 µM CuSO4. The frequencies of cells with aggregates and the induction of [PSI+] were scored in 200–600 cells. Averages from four transformants are shown. Similar results (data not shown) were obtained in another [psi][PIN+] strain, SL1010-1A.

Sup35N–GFP15 ± 8.424.2 ± 9.419.1 ± 3.7
Sup35NM–GFP9.1 ± 1.914.3 ± 3.49.6 ± 2.9
Sup35–GFP2.4 ± 0.91.8 ± 1.11.3 ± 0.4

To modulate the level of overexpression of Sup35NM–GFP, transformants were grown for 73 h in SC-Ura media supplemented with 0, 2, 10, 20 and 50 µM Cu2+. In cultures lacking supplemental Cu2+ (but containing residual Cu2+), about 1–3% of the cells could be seen to contain many small faint aggregates, whereas no rings and very few large dot aggregates were seen. The addition of even 2 µM Cu2+ caused the aggregates to become much larger and brighter, reduced the number of aggregates per cell to approximately one and led to the appearance of occasional cells with ring aggregates. As the concentration of Cu2+ was increased, the fraction of cells with aggregates increased, the aggregates became larger and brighter and cells with rings became much more frequent than cells with dots.

Cells bearing dot- and ring-like aggregates are [PSI+]

To determine whether cells bearing aggregates are [PSI+], we overproduced Sup35NM–GFP in [psi] strain 74-D694 for 40 h to induce rings and dots, and then single or budded cells were separated by micromanipulation (see Experimental procedures). Colonies arose from 11 out of 26 single or budded cells with ring-like aggregates in each cell, from 11 out of 14 single or budded cells with dots in each cell and from 209 out of 285 cells with no visible aggregates. The viability of ring-bearing cells was thus statistically lower than that of cells with no aggregates (P < 0.005). Although only eight of the 209 colonies from cells with no visible aggregates gave rise to any [PSI+] progeny, all 10 colonies tested from cells with rings and nine out of 10 colonies tested from cells with dots contained 20–100% [PSI+] cells. Both strong and weak [PSI+] variants, distinguished by the efficiency of nonsense suppression (Derkatch et al., 1996), were obtained. Interestingly, each individual ring- or dot-bearing single or budded cell gave rise to only one type of stable [PSI+] variant. Because some cells with rings gave rise to strong [PSI+] variants, whereas others gave rise to weak [PSI+] variants, and the same was true of cells with dots, we conclude that the strength of the [PSI+] induced does not correlate with the type of aggregate seen. Re-expression of Sup35NM–GFP for 48 h in [PSI+] colonies derived from either ring-bearing or dot-bearing cells revealed that each of them had dots in most cells and rings in just a few cells as is typical for [PSI+] (see below).

Stationary growth enhances the appearance of rings and dots

The time course of the appearance of [PSI+] cells and cells with dot and ring aggregates was examined in [psi][PIN+] 74-D694 overexpressing Sup35NM–GFP (20 µM CuSO4). [PSI+] induction began during exponential growth and increased further upon entrance into stationary phase (Fig. 4). Ring aggregates were hardly detectable during exponential growth, and their number increased dramatically upon entrance into stationary phase (Fig. 4). Cells with dot-type aggregates appeared earlier than ring-containing cells. The earlier appearance of the dot aggregates relative to the ring aggregates may explain why the ring aggregates were not described in previous papers (Patino et al., 1996; DePace et al., 1998), in which experiments were performed mainly on logarithmic cells.

Figure 4.

Time course of the appearance of rings, dots and [PSI+] cells. Strain [psi][PIN+] 74-D694 transformed with pSUP35NM-GFP was grown in SC–Ura supplemented with 20 µM CuSO4. Samples were taken at the indicated time points to score for [PSI+] (open columns), ring-like aggregates (filled columns), dots (hatched columns) and cell density (curve). Data represent the averages of six cultures. Cultures were inoculated at an OD600 of 0.01 and remained at a low OD for 24 h. Note that the high level (2.8%) of [PSI+] cells found in the starting cultures results from residual Cu2+ in the standard SC–Ura medium and/or leakage of the CUP1 promoter. However, when the [PSI+] induction experiment was repeated using the pGAL::SUP35 plasmid with the non-leaky GAL1 promoter, the time course of [PSI+] induction was essentially the same. Also note that 20 µM rather than 50 µM CuSO4 was used in this experiment because it was found to be less inhibitory to the growth of [PSI+] cells.

In a set of cultures parallel to those described above, but which were diluted into prewarmed media to an OD600 of 0.1 every 12 h after the cultures reached an OD600 of 1.5–2.3, ring- and dot-containing cells, respectively, constituted only 1.1 ± 0.6% and 3.2 ± 0.4% of the cells after 75 h, even though [PSI+] was induced efficiently (31.8 ± 2%). As the level of Sup35NM–GFP in stationary and periodically diluted cells was similar (32 ± 1.1 and 34.3 ± 2.1 fluorescence units per 1 OD600 unit of cells respectively), the appearance of visible SUP35–GFP aggregates is not merely the result of accumulation of Sup35–GFP during prolonged growth, but requires entrance into stationary phase.

Effects of overproduction of Hsp104 on ring formation in [PSI+] cells

Overproduction of Sup35NM–GFP for 48 h in [PIN+] strains carrying a weak [PSI+] revealed that most cells contained dots and just a few contained rings (Table 2). However, rings were seen at high frequency in [PSI+] strains that overexpressed Sup35NM–GFP for 48 h in the presence of excess Hsp104 (Table 2), which was shown previously to eliminate [PSI+] (Chernoff et al., 1995). As expected, cells that overexpressed Hsp104 showed frequent loss of dot aggregates (80% loss) and [PSI+] (77% loss). The appearance of rings, however, increased to 11%. Excess Hsp104 caused a similar effect when a [PSI+][pin] derivative of 74-D694 (Derkatch et al., 2000) was used. When this strain is cured of [PSI+], the [psi] derivatives cannot be reinduced to become [PSI+] by overexpression of Sup35 (Derkatch et al., 2000). Likewise, overexpression of Sup35NM–GFP in these cured derivatives does not induce the formation of rings, dots or [PSI+]. Therefore, the rings that appeared in the [PSI+][pin] strain in the presence of excess of Hsp104 were not the result of reinduction of [PSI+] after the original [PSI+] was lost and must have arisen from the original [PSI+]-bearing cells before the aggregates were entirely lost. This result suggests the existence of intermediates during the curing process.

Table 2. Effects of overexpression of Hsp104 on aggregate formation in [PSI+] cells.

Plasmids co-
transformed with
Percentage of cells with
DotsRings [PSI + ] loss
  1. The GAL1::HSP104 vector, pES5, or the control vector, pRS413, were co-transformed with pSUP35NM-GFP into [PSI+] strains carrying the weak [PSI+21] (see Experimental procedures). Transformants were grown in synthetic medium lacking uracil and histidine with galactose as carbon source and 50 µM CuSO4 for 48 h to express Hsp104 and Sup35NM–GFP. The appearance of aggregates and the presence of [PSI+] was scored in 200–600 cells in each transformant. Averages from four transformants are shown.

[PSI + ][PIN + ] pRS41399.5 ± 0.40.5 ± 0.40.4 ± 0.5
pES520 ± 5.811.2 ± 1.976.5 ± 2.4
[PSI + ][pin ] pRS41398 ± 0.41.2 ± 1.5
pES538.9 ± 4.318.4 ± 2.658.7 ± 2.4

Remarkably, in many (≈ 30%) of the ring-like aggregates arising in cultures with overexpressed Hsp104, dots appeared to be linked into rings (Fig. 5B–D). This was only occasionally observed in cells with a normal level of Hsp104. This could reflect either integration of dots into rings or partial breakage of rings. In either case, it provides direct evidence that Hsp104 can alter the surface of Sup35 polymers as proposed previously (Kushnirov and Ter-Avanesyan, 1998).

Figure 5.

Ring-like aggregates formed during overexpression of Hsp104. [PSI+] cells co-transformed with pSUP35NM-GFP and pES5 (B–D) carrying GAL::HSP104 or control vector pRS413 (A) were grown in SC–Ura-His overnight and subsequently diluted in SC–Ura-His + 50 μM CuSO4 with galactose as carbon source at 1:12 dilution and incubated for 48 h.

Unexpectedly, although overexpression of Hsp104 causes loss of [PSI+] in the absence of Sup35 overproduction (Chernoff et al., 1995), [PSI+] and ring as well as dot aggregates were induced in the [psi][PIN+] 74-D694 strain overexpressing Sup35NM–GFP and Hsp104 simultaneously (data not shown). Likewise, overexpression of Hsp104 did not prevent an excess of complete Sup35 from inducing [PSI+], e.g. in pEMBL-SUP35 transformants (data not shown and Y. O. Chernoff, personal communication).

Ring-type aggregates form in [PSI+] cells treated with GuHCl

GuHCl can eliminate [PSI+] as efficiently as an excess of Hsp104 (Tuite et al., 1981). We asked whether GuHCl could convert dot aggregates visualized in [PSI+] cells by overexpression of Sup35NM–GFP into rings, just as overexpression of Hsp104 does. Two independent [pin] derivatives of 74-D694 containing weak [PSI+] were used ([PSI+21][pin] and [PSI+5][pin]). Similar to an excess of Hsp104, GuHCl treatment of [PSI+] strains induced loss of Sup35NM–GFP-associated dot aggregates and [PSI+], but increased the appearance of Sup35NM–GFP-associated rings (Table 3). As we verified that overexpression of Sup35NM–GFP in cured [psi] derivatives of these strains does not induce the formation of rings, dots or [PSI+], the rings that appeared in the presence of GuHCl were not a result of the reinduction of [PSI+] but of the conversion of Sup35NM–GFP-associated aggregates into rings.

Table 3. Effects of GuHCl on Sup35NM–GFP aggregation and induction of [PSI+].

Genotype of

Percentage of cells with
DotsRings [PSI + ] induction [PSI + ] loss 
  1. a . More than 1000 cells in each transformant were checked. Independent derivatives of strain 74-D694 with various [PSI] and [PIN] elements were transformed with pSUP35NM-GFP and grown in SC–Ura + 50 µM CuSO 4 liquid medium with (+) or without (–) 5 mM GuHCl. The appearance of aggregates and the frequency of [PSI+] induction were scored in 200–600 cells in each transformant. Averages from four transformants are shown.

[psi ][PIN + ] 8.7 ± 2.914.2 ± 4.37.5 ± 2 
+5.6 ± 0.713.7 ± 46.6 ± 1.4 
[psi ][pin ] a  
[PSI + 13 ][PIN + ] 88.5 ± 30.5 ± 0.8 1.3 ± 1.3
+53 ± 8.114.9 ± 0.8 35 ± 1
[PSI + 5 ][pin ] 86 ± 8.20.5 ± 0.5 6 ± 2.3
+74.8 ± 7.29.9 ± 1.4 24.1 ± 4.8
[PSI + 21 ][PIN + ] 87.2 ± 11.30.8 ± 0.4 1.9 ± 1.3
+48.8 ± 6.517.8 ± 6.2 30.1 ± 3.3
[PSI + 21 ][pin ] 68.5 ± 19.3 14.2 ± 11.2
+51.8 ± 13.913.2 ± 2.7 37.5 ± 18.1

Surprisingly, although GuHCl causes loss of [PSI+] (Tuite et al., 1981), [PSI+] and ring as well as dot aggregates were efficiently induced in the [psi][PIN+] 74-D694 strain in the presence of GuHCl (Table 3). Therefore, GuHCl does not affect the induction of [PSI+] and the formation of different aggregates during induction of [PSI+] in [psi] strains. Moreover, GuHCl is unable to induce the appearance of rings or dots in [psi][pin] 74-D694 cells (Table 3), suggesting that GuHCl cannot facilitate the formation of rings or dots in the absence of [PSI+].


We have described the in vivo formation of ring-like Sup35 aggregates that are caused by excess Sup35. We show that, in [psi] cells, the formation of these aggregates, as well as the previously described dot aggregates (Patino et al., 1996), is strictly dependent upon the prion-like factor [PIN+], which is required for the de novo induction of [PSI+]. Furthermore, we show directly that both these types of aggregates only form in cells that have become [PSI+].

Individual aggregate-bearing cells each gave rise to only one type of stable [PSI+], weak or strong. This suggests that different types of seeds give rise to the different [PSI+] variants and that, once a [PSI+] seed is established, it propagates throughout the cell. Interestingly, either type of [PSI+] could segregate from different dot- or ring-bearing cells. Thus, the distinction between dot and ring aggregates is not a manifestation of different heritable variants of [PSI+]. We propose that dot and ring aggregates are formed after the establishment of [PSI+] seeds and that, although ring or dot aggregates can only form in the presence of [PSI+] seeds, any type of [PSI+] seed can give rise to either type of aggregate.

The distinction between dot- and ring-like aggregates is reminiscent of different packing arrangements of crystals formed under different conditions. Indeed, fibres of purified Sup35 formed in different buffers look somewhat different (Glover et al., 1997). The co-existence of a separate ring and dot aggregate in the same cell was rare, suggesting that, once a particular quaternary structure is initiated, it propagates throughout the cell. Therefore, the dot-prone and ring-prone conformations could be stabilized in the presence of dots or rings respectively. Furthermore, the aggregates already present in [PSI+] cells seem to template the formation of dots and inhibit ring formation.

The existence of flexible intermediates, which can switch between dot-prone and ring-prone conformations, is suggested by the finding that cells with rings can give rise to daughters with rings as well as daughters with dots (Figs 1, 2A and B). As visible aggregates were not always passed directly to daughters, the intermediates in the daughter cells may be free to switch conformations. Furthermore, in rapidly growing exponential cultures, cells bearing dot-prone intermediates may have a selective advantage because the viability of ring-bearing cells is reduced. The fact that [PSI+][pin] cells overexpressing Sup35NM–GFP and grown in the presence of GuHCl or excess Hsp104 gave rise to frequent daughter cells with rings can also be explained by the existence of flexible intermediates. These rings cannot be the result of reinduction of [PSI+] after the original [PSI+] was lost because, owing to the absence of [PIN+] (Derkatch et al., 2000), overexpression of Sup35NM–GFP did not cause the formation of rings, dots or [PSI+] in [psi] derivatives of these strains. Therefore, the rings must have arisen from the original [PSI+]-bearing cells before the [PSI+] aggregates were entirely lost. This suggests that the partial unfolding of the [PSI+] aggregates to an intermediate conformation seeds ring formation. These intermediates may or may not be sufficient to maintain [PSI+] and could also form during regular [PSI+] induction in [PIN+] strains. In either case, it appears that the intermediates do not form into visible aggregates until the cells reach stationary phase.

Our results show that, although GuHCl or excess Hsp104 causes [PSI+] loss in the presence of excess Sup35, neither treatment prevents the induction of [PSI+] by excess Sup35. Thus, the mechanisms of [PSI+] curing by excess Hsp104 and [PSI+] induction by excess Sup35 do not rely exclusively upon the altered balance of Hsp104 and Sup35 levels. Possibly an imbalance between these proteins and other cellular factors or, in the case of [PSI+] induction, merely the elevated cellular concentration of Sup35 are essential. The results also suggest that the de novo appearance of [PSI+] is not just the reverse process of the curing of [PSI+]. This is consistent with the finding that GuHCl only affects the propagation [PSI+] in dividing cells, but does not directly dissolve [PSI+] seeds (Eaglestone et al., 2000). Our previous finding that the induction, but not propagation, of [PSI+] requires [PIN+] (Derkatch et al., 2000) also indicates that the de novo formation of [PSI+] seeds is an event distinct from the propagation of [PSI+].

One way to explain the finding that the de novo formation of rings and dots occurs primarily in stationary phase cells is to propose that aggregate formation is enhanced by the continuous production of Sup35NM–GFP in non-dividing cells. This seems unlikely, however, because the level of Sup35NM–GFP in stationary and periodically diluted cells was similar. Furthermore, the same conditions do not induce rings efficiently in cells that are already [PSI+]. Rather, we hypothesize that entrance into stationary phase per se enhances aggregate formation. Perhaps this results from the change in concentration of chaperones, e.g. Hsp104, associated with stationary phase.

The fact that the visible aggregates appear primarily in stationary phase cells, whereas [PSI+] can appear in exponential cultures suggests that most heritable [PSI+] seeds are too small to be visualized by fluorescence microscopy. Indeed, attempts to visualize [PSI+] without overexpression of Sup35 resulted in no signal when immunodetection was used (data not shown). Also, when [PSI+] but not [psi][pin] cells containing Sup35NM–GFP were grown in medium containing only residual levels of Cu2+, many faint small aggregates were seen in individual cells (data not shown) as opposed to one or a few large aggregates per cell observed upon overexpression of Sup35 in [PSI+] (Patino et al., 1996; also see Fig. 1).

Highly ordered protein aggregates are associated with prion diseases as well as other human diseases, including Alzheimer's (Citron et al., 1992) and Huntington's diseases (Scherzinger et al., 1997). Our finding that Sup35 ring-bearing cells have reduced viability suggests that certain large Sup35 aggregates are toxic. The induction of large Sup35 aggregates in yeast provides an in vivo model system for studying the formation and effects of protein aggregates in general.

Experimental procedures

Strains and media

The strains used in this study were 74-D694 (MATa ade1-14 leu2-3,112 his3-Δ200 trp1-289 ura3-52 [psi][PIN+]) (Chernoff et al., 1995), SL1010-1A (MATα ade1-14 met8-1 leu2-1 his5–2 trp1-1 ura3-52 [psi][PIN+]) (Zhou et al., 1999) and [psi][pin], [PSI+][PIN+] and [PSI+][pin] derivatives of these two strains obtained by overexpression of the complete SUP35 gene or its prion domain (Derkatch et al., 1997; 2000) and/or by GuHCl treatment. Three [PSI+]s derived independently in 74-D694 were used, [PSI+13], [PSI+5] and [PSI+21]. All these [PSI+] elements are weak and are therefore compatible with overexpression of the Sup35 prion domain, which would be lethal in the presence of a strong [PSI+]. [PSI+13] and [PSI+21] were induced by overexpression of complete SUP35 in [psi][PIN+] 74-D694. [PSI+5] was induced by overexpression of the modified N-terminal fragment of Sup35 encoded by pEMBL-ΔBalext in [psi][pin] 74-D694 (Derkatch et al., 2000). The 74-D694-Δhsp104 (Chernoff et al., 1995; Derkatch et al., 1997) and SL1010-1A-Δhsp104 strains (Zhou et al., 1999) are HSP104 disruption derivatives of 74-D694 and SL1010-1A respectively.

Standard rich medium YPD and synthetic complete medium (using Difco yeast nitrogen base) lacking uracil (SC–Ura) or lacking both uracil and histidine (SC–Ura-His) were used (Sherman et al., 1986). To induce the expression of the CUP1 promoter, 50 or 20 µM CuSO4 was added to synthetic medium selective for plasmids. To induce the expression of the GAL1 promoter, galactose was used to replace glucose in synthetic medium. When GuHCl was used, it was added to a final concentration of 5 mM. Yeast strains were grown at 30°C.

Plasmids, DNA manipulations and transformation

DNA manipulations and yeast and bacterial transformations were performed according to standard protocols (Sambrook et al., 1989; Rose et al., 1990). The centromeric URA3 plasmid pCUP::GFP encodes GFP driven by the copper-inducible CUP1 promoter. Plasmid pSUP35-GFP differs from pCUP::GFP in having SUP35 fused to the 5′ end of GFP. These plasmids were gifts from S. Lindquist. To fuse Sup35NM (1–254 residues) to the 5′ end of GFP, the 5′ region of SUP35 coding for Sup35NM was amplified from pEMBL-SUP35 (Ter-Avanesyan et al., 1993) using primers GCGGGATCCACAATGTCGGATTCAAACCA and CCATCCGCGGCATATCGTTAACAACTTCGT and cloned as a BamHI–SacII fragment into pCUP::GFP, resulting in pSUP35NM-GFP. A similar method was used to generate pSUP35N-GFP, which encodes Sup35N (154 residues) fused to the N-terminus of GFP using primers GCGGGATC CACAATGTCGGATTCAAACCA and CCAACCGCGGCAACTTGATACCGGAACTGG.

To tag Sup35NM with a c-Myc epitope, a fragment bearing three repeats of the c-Myc epitope was amplified from the SmaI-digested pMPY-3xMYC plasmid (Schneider et al., 1995) using primers CCATCCGCGGTCTGAGCAAAAGCT CATT and ATCGAGCTCTTAGCTACTATTAAGATCCTCCT CGG, and this was used as a SacII–SacI fragment to replace GFP in SUP35NM–GFP in pSUP35NM–GFP, resulting in pSUP35NM-MYC. pEMBL-SUP35 is a 2 µURA3 plasmid carrying wild-type SUP35 under the control of its native promoter (Ter-Avanesyan et al., 1993). Plasmid p2UGR1-3SGFP, which encodes SUP35NM–GFP under the control of the glucocorticoid-inducible promoter GRE, and the control vector pG-N795 (both kindly provided by S. Lindquist) have been described previously (Patino et al., 1996). The centromeric HIS3 plasmid pES5 carrying HSP104 under the control of the GAL1 promoter was a gift from S. Lindquist. Galactose medium was used to express HSP104. Plasmid pRS413 (Sikorski and Hieter, 1989) was used as a control.

Analysis of aggregates of the GFP fusion proteins in vivo

To express GFP, Sup35NM–GFP, Sup35N–GFP, Sup35–GFP and Sup35NM–Myc, driven by the CUP1 promoter, yeast transformed with the appropriate plasmids was inoculated at an OD600 of 0.02–0.1 into synthetic complete glucose medium lacking uracil and supplemented with 20 or 50 µM CuSO4 (SC–Ura + CuSO4) and grown at 30°C for 48 h unless otherwise specified. To express Sup35NM–GFP under the control of the glucocorticoid-inducible promoter GRE, cells were transformed with plasmids p2UGR1-3SGFP and pG-N795 and treated with 1 µM 11-deoxycorticosterone for 28 h. Cells were examined under a fluorescence microscope (Axioskop; Carl Zeiss) for GFP aggregates. In all experiments, three or four independent transformants and 200–600 fluorescent cells in each transformant were analysed.

Visualization of Sup35 aggregates by immunofluorescence

The Myc-tagged Sup35NM protein and the untagged complete Sup35 were overproduced in pSUP35NM-Myc- and pEMBL-SUP35-transformed yeast cells by growth in liquid SC–Ura + 50 µM CuSO4 medium for 48 h and SC–Ura for 3 days respectively. The immunofluorescence was performed essentially as described previously (Pringle et al., 1991). Mouse monoclonal antibody 9E10 (Santa Cruz Biotech), which recognizes the c-Myc epitope, was used at 1:250 dilution. Mouse polyclonal antibody against Sup35, M12, was kindly provided by V. Prapapanich and was used at 1:1000 dilution. The secondary antibody in these experiments was fluorescein-congugated goat anti-mouse antibody (Santa Cruz Biotech), which was used at 1:500 dilution.

Analysis of induction and loss of [PSI+]

Transformants carrying the plasmids indicated were grown in liquid media supplemented with 50 µM or 20 µM CuSO4 for 48 h or other specified time periods. The induction or loss of [PSI+] was tested by plating 200–600 cells on YPD medium, on which [psi] colonies are red, weak [PSI+] colonies are pink and strong [PSI+] colonies are white as a result of different levels of suppression of the ade1-14 nonsense mutation. Averages from three or four transformants are reported.

Analysis of the [PSI+] phenotype of aggregate-bearing cells

Transformants of [psi] 74-D694 with pSUP35NM-GFP were grown in SC–Ura + 50 µM CuSO4 for ≈ 40 h. Single or budded cells were placed into different positions on slabs of YPD medium by micromanipulation under a light microscope. The slabs were then transferred to slides and examined under a fluorescence microscope for aggregates in cells. After cells formed colonies, coverslips were removed, and each colony was purified on YPD. Colonies that were white or pink, indicative of suppression of the ade1-14 nonsense mutation, and that were cured of this suppression by growth on YPD containing 5 mM GuHCl were scored as [PSI+].

Measurement of the intensity of fluorescence in exponential and stationary cultures expressing Sup35NM–GFP

To measure the intensity of fluorescence in exponential cells (OD600 = ≈ 1.5) and stationary phase cells (OD600 = ≈ 3), cells were washed with PBS buffer, and the cell density was adjusted to an OD600 of ≈ 0.5. Excitation and emission wavelengths of 480 nm and 510 nm, respectively, were used to measure the intensity of fluorescence of the cells using the Perkin-Elmer 650-10S fluorescence spectrophotometer.

Actin staining and latrunculin A treatment

[PSI + 21 ][PIN + ] and [psi][PIN+] 74-D694 transformed with pSUP35NM-GFP were grown in liquid SC–Ura + 50 µM CuSO4 for 48 h. Some samples were treated with 200 µM latrunculin A (Molecular Probes) (Ayscough et al., 1997) for 10 min. All samples were then fixed with 3.7% formaldehyde for 20 min. Cells were washed with PBS and applied to polylysine-coated slides. After treatment with acetone for 6 min, cells were washed with PBS supplemented with 1% BSA and then stained with Texas red–X phalloidin (Molecular Probes) in the same buffer for 20 min. Cells were washed with PBS before the addition of mounting medium and the coverslip.


We thank M. Bradley for help throughout this work, V. Prapapanich for M12 antibody, R. Levin for help with the growth curve experiment, A. Ellicott for help with the immunofluorescence assay, Y. O. Chernoff for permission to quote his unpublished data, and S. Lindquist and J. Liu for plasmids. We also thank P. Okkema for use of his fluorescence microscope, and B. Nichols for use of his fluorescence spectrophotometer and his help with the fluorescence measurements. This study was supported by grants from the National Institute of Health (GM56350) and the Alzheimer's Association.


  1. Present address: Department of Ophthalmology and Visual Sciences, Washington University, St Louis, MO 63110, USA.