Is a Lethal Yeast Prion
Total Page:16
File Type:pdf, Size:1020Kb
Suicidal [PSI+] is a lethal yeast prion Ryan P. McGlinchey, Dmitry Kryndushkin, and Reed B. Wickner1 Laboratory of Biochemistry and Genetics, National Institute of Diabetes Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD 20892-0830 Contributed by Reed B. Wickner, February 17, 2011 (sent for review January 28, 2011) [PSI+] is a prion of the essential translation termination factor codons in ade1-14 or ade2-1 is then sufficiently frequent that the Sup35p. Although mammalian prion infections are uniformly fatal, cells are Ade+ (grow without adenine and are white) instead of − commonly studied [PSI+] variants do not impair growth, leading to Ade (and red). A [PSI+] variant that efficiently inactivated all suggestions that [PSI+] may protect against stress conditions. We Sup35p (“killer [PSI+]”) would be lethal. As a permissive condi- report here that over half of [PSI+] variants are sick or lethal. These tion for killer [PSI+], we express a full-length chromosomal “killer [PSI+]s” are compatible with cell growth only when also SUP35 gene, and, on a plasmid, SUP35C encoding only the es- expressing minimal Sup35C, lacking the N-terminal prion domain. sential part but lacking the NM prion domain. This truncated The severe detriment of killer [PSI+] results in rapid selection of Sup35C cannot be converted to the [PSI+] prion form because it nonkiller [PSI+] variants or loss of the prion. We also report variants lacks the N domain that is essential for prion formation (18) and of [URE3], a prion of the nitrogen regulation protein Ure2p, that forms the core of the in-register parallel β-sheet amyloid structure grow much slower than ure2Δ cells. Our findings give a more re- that constitutes the prion form (19, 20). To detect killer [PSI+], alistic picture of the impact of the prion change than does focus on the plasmid-borne SUP35C must be expressed only enough to “mild” prion variants. keep cells alive but not enough for efficient termination, which − would make them Ade (Fig. S1). With a tetracycline-repressible rep lthough mammalian prions cause a devastating and uniformly promoter (Ptet ) (21), 10 μg/mL doxycycline sufficiently re- Afatal spongiform encephalopathy (reviewed in ref. 1), yeast pressed the plasmid-encoded Sup35C expression that cells lack- + and fungal prions, as described so far, are compatible with in- ing a chromosomal SUP35 were Ade (white) but could grow fi slowly (Fig. S2). Doxycycline at these levels does not detectably de nite growth, often at normal or near-normal rates. This ap- GENETICS rep parent contrast has led to proposals that the yeast and fungal slow growth of normal cells. The tet -SUP35C gene was on prions may benefit their hosts, even though prion formation may a URA3 CEN (centromere-containing) vector whose loss could be decrease the activity of the protein as a result of the aggregate/ selected by growth on plates with 5-fluorouracil (FOA; kills URA3 amyloid formation that is the prion change. The [Het-s] prion of cells but not ura3 mutants) (22). Podospora anserina is necessary for proper heterokaryon incom- Using strains 74-D694 or GT159 carrying PGAL1-SUP35NM rep + patibility of het-s with het-S strains (2), so it was proposed that this and URA3 Ptet -SUP35C plasmids, we induced [PSI ] forma- is a prion benefiting the host (3). However, [Het-s] is also part of tion by overexpression of Sup35NM on galactose media and a meiotic drive system, promoting inheritance of het-s by cheating selected Ade+ clones on plates with 10 μg/mL doxycycline. on meiosis (4), so the beneficial prion suggestion may be wrong. Artificially overproducing Sup35NM (the prion domain) can [PSI+] is an amyloid prion of the Saccharomyces cerevisiae Sup35p produce lethality because all Sup35p is drawn into the filaments (5–7), and was suggested to protect against stress (8) or to promote (23); thus, we repressed expression of Sup35NM by over 30 evolution by protecting against stress (9). All these tests of stress generations of growth on glucose to dilute out the Sup35NM resistance were done using the usual [PSI+] variants, which were before looking for colonies that grew poorly or not at all on FOA selected to be compatible with vigorous cell growth (10). medium (loss of Sup35C). Sup35p is an essential subunit, with Sup45p, of the translation We found that of 313 Ade+ guanidine-curable clones from termination factor. Like the mammalian prion “strains” (11, 12), 74-D694, 144 showed slow growth on FOA (“sick [PSI+]”) and [PSI+] has “variants,” several different stably propagated biolo- 24 did not grow at all (killer [PSI+]) (Fig. 1). Similar proportions gically and structurally different forms of the same prion protein of sick and killer [PSI+] were isolated from strain GT159 (Fig. (13, 14). In the usual [PSI+] variants, much of the Sup35p is tied S3). Cells with killer [PSI+] grew very poorly after several days up in amyloid filaments and is unavailable for translation ter- on ½YPD containing 10 μg/mL doxycycline, confirming that mination, but enough is free to keep the cells alive. The shortage failure to grow on FOA is attributable to lack of Sup35C (Fig. of Sup35p increases the misreading of nonsense codons, which 2A) and not, for example, a variant that never loses the URA3 provides the basis for the genetic assay of [PSI+]. plasmid. Most of the colonies that did grow were pink/red and Ure2p is a regulator of nitrogen catabolism that can convert to proved to have lost [PSI+]. On ½YPD lacking doxycycline, killer an amyloid prion called [URE3] (5, 15, 16). Variants of [URE3] [PSI+] cells were red and grew well because of an abundant have also been described (16, 17). supply of Sup35C (Fig. 2B). Here, we show that more than half of [PSI+] variants are lethal The instability of sick and killer [PSI+] is shown in Fig. 3. or highly pathogenic, unlike the variants usually studied. Simi- [PSI+] isolates were streaked on both FOA (Fig. 3A, Left) and larly, we describe abundant [URE3] variants that slow cell ½YPD without doxycycline (Fig. 3A, Center). After several days growth dramatically in a background in which deletion of the on ½YPD, sick and killer cells grew more normally when URE2 gene has no effect on growth. These results show that streaked to FOA (Fig. 3A, Right), showing that toxic [PSI+] acquisition of a yeast prion may be disastrous for yeast. variants are not stable, even in the presence of excess Sup35C. This suggests that toxicity is not limited to undersupply of active Results Sup35p has two domains, the N-terminal (NM) domain, which is necessary and sufficient for prion formation (and has physiolog- Author contributions: R.P.M., D.K., and R.B.W. designed research, performed research, ical function as well, as discussed below) but is dispensable for contributed new reagents/analytic tools, and wrote the paper. growth, and the C-terminal C domain, which is essential for The authors declare no conflict of interest. translation termination, and can perform this function without 1To whom correspondence should be addressed. E-mail: [email protected]. + fi the NM domain. In [PSI ] cells, most of the Sup35p is in la- This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. ments, and is inactive. Readthrough of premature termination 1073/pnas.1102762108/-/DCSupplemental. www.pnas.org/cgi/doi/10.1073/pnas.1102762108 PNAS Early Edition | 1of5 Downloaded by guest on September 23, 2021 A Single colony 5-FOA ½ YPD 5-FOA from ½ YPD Non-toxic PSI+ Sick PSI+ Sick PSI+ Killer PSI+ Fig. 1. Screening of Ade+ colonies for killer [PSI+]. Isolated Ade+ colonies Killer PSI+ were restreaked three times on −Ade −Ura + 10 μg/mL doxycycline before being stamped on FOA. Circled colonies have killer [PSI+]. The fractions of PSI+ ½ YPD from 5-FOA isolates with mild, sick, and killer [ ] are shown below. B C D Non-toxic Sick PSI+ Sick PSI+ PSI+ Sup35p and/or that the toxic [PSI+] variants are inherently un- stable. Streaking sick [PSI+] isolates on ½YPD after selection on − FOA showed a mixture of small white (Ade+ Ura ) and large − − pink/red (Ade Ura ) colonies (Fig. 3B, Upper). These small, white, poorly growing colonies, on restreaking on ½YPD, formed a mixture of small and larger white Ade+ colonies and pink/red − Ade colonies, with the latter having lost [PSI+] (Fig. 3B, Lower). Thus, selection for prion loss or alteration attributable to prion toxicity, and likely inherent prion instability, makes isola- + tion of stable killer [PSI ] apparently impossible. ½YPD ½YPD ½YPD Growth of sick [PSI+] isolates with 3 mM guanidine (which PSI+ A PSI+ cures [PSI+]) (24) cured the prion, as shown by cells becoming Fig. 3. Toxicity and instability of [ ] isolates. ( ) Sick and killer [ ] − − − μ candidates were streaked on ½YPD and on FOA plates. After 2 d of growth Ade on the Ade Ura + 10 g/mL doxycycline plates. Once B–D PSI+ + on ½YPD, strains were restreaked on FOA plates. ( ) Individual sick [ ] cured of killer [PSI ], these cells could grow well on FOA plates and nontoxic [PSI+] colonies were streaked on ½YPD from FOA. (Lower) or on ½YPD with doxycycline, similar to those shown in Fig. 3A. Single colonies restreaked on ½YPD are indicated. This shows that the toxicity is indeed attributable to a variant of [PSI+]. Transformation of presumed [PSI+] cells with a plasmid expressing Sup35NM-GFP confirmed the presence of prion color of yeast colonies (16). Strain BY241 was transformed with aggregates (Fig. 4). An apparent correlation between [PSI+] the LEU2 centromeric plasmid pVTG12 expressing Ure2N-GFP toxicity and the number of Sup35NM-GFP aggregates was noted (a fusion of the Ure2p prion domain with GFP) under control (Fig.