Chromosomal duplication disrupts quiescence and life span

In the course of ongoing investigations, we discovered that haploid derivatives of wild oak-soil strain YPS1009 with different chromosome duplications showed abnormal arrest and regrowth after nutrient exhaustion from rich medium. To investigate systematically, we characterized each step along the progression to quiescence in a series of engineered YPS1009 aneuploid strains each with a full duplication of a different chromosome [57]. In this strain background, euploid cells shift to respiration as glucose is depleted from the media (day 1 of culturing), arrest as unbudded cells in G0 (day 1–2), silence their transcriptome (day 1–7), and become small and dense (day 4–7). All of these steps are important for healthy life span [58–60].

We found that all aneuploids tested had defects in these quiescence hallmarks, to varying degrees. One of the earliest steps is cell-cycle arrest upon nutrient exhaustion in a saturated culture. Whereas nearly 100% of euploid cells arrest as unbudded cells by 2 days of culturing, conservatively 1%–12% of cells, depending on the amplified chromosome, showed morphology indicative of budding (Fig 1A), even though the cultures had completely exhausted glucose (S1A Fig). The arrest defect is not specific to YPS1009, as two other strain backgrounds with extra chromosomes showed similar defects (Fig 1B). Supplementation with extra glucose, amino acids, or yeast extract did not correct the issue. In addition to the arrest defect, YPS1009 with an extra chromosome XIV (YPS1009_Chr14) showed an unusual morphology of bi-lobed cells with a single round nucleus in between, while YPS1009_Chr15 took on a multi-budded elongated state (Fig 1C). Importantly, YPS1009_Chr14 and _Chr15 aneuploids do not show these morphologies during log-phase growth (S1B Fig), indicating that these morphologies are specific to late-culture growth when cells normally arrest. Thus, chromosome duplication generally disrupts cell-cycle arrest upon nutrient exhaustion from rich medium, albeit with heterogeneity and some chromosome-specific effects.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Chromosome duplication disrupts multiple signatures of quiescence.

(A, B) Average and individual data points of percent budded cells is shown for visualization, at 2 days in (A) YPS1009 (n ≥ 3) or (B) oak strain YPS163 or vineyard strain DBVPG1373 (n = 2–3) with chromosome duplications as indicated. Asterisk, p < 0.05, Fisher’s exact test comparing counts of aneuploids vs. matched euploid (Eu) that were budded vs. unbudded. (C) Representative brightfield images from A. Blue, DAPI staining. Scale bar, 5 µm. (D) Left: Replicate-averaged log₂(fold change) expression of 963 genes (rows) repressed in euploid but significantly higher in abundance (FDR < 0.05) in all four aneuploids compared to euploid, in at least one time point (see Methods). Right: The replicate-averaged log₂(fold difference) in normalized transcript abundance for genes shown on the left, in each aneuploid vs. euploid at that time point. Purple indicates a repression defect. Data can be found in S2 Data. (E) Average of all genes shown in D for strains in the key to the left. Asterisk, p < 1 × 10⁻⁸, T test comparing the set of transcript abundances in each aneuploid vs. euploid at each time point. (F) Proportion of dense and light cells after 4 days (n = 3–5). Asterisk, p < 0.05, Chi-squared test. (G) Average fraction of colony-forming units relative to day 7 (YPS1009 strains, n = 3–6; YPS163 and DBVPG1373, n = 2). Asterisk, p < 0.05, replicate-paired T test comparing aneuploids to euploid at day 35. Euploid YPS163 and DBVPG1373 viability are indistinguishable from YPS1009. The data underlying this figure can be found in S2 Table.

https://doi.org/10.1371/journal.pbio.3003509.g001

Aneuploid cultures also displayed defects in subsequent steps of quiescence, including the stereotypical transcriptome silencing associated with quiescence [59,61,62]. Euploid cultures repressed thousands of transcripts, >256-fold below log-phase levels, beginning 1 day past exponential phase and dropping to stable levels by day 3 (Fig 1D and 1E). We measured bulk transcriptomes of YPS1009 with a duplication of Chr4, Chr12, Chr14, or Chr15, normalized by cell number using spike-in of Schizosaccharomyces pombe cells (see Methods). The transcriptional profiles of all aneuploids were dysregulated to varying extents: we identified 963 transcripts that were reduced in euploid cells but at statistically significantly higher abundance in all four aneuploids, in at least one time point (false discovery rate [FDR] < 0.05, see Methods). Two of the aneuploids (YPS1009_Chr4 and Chr12) eventually reached euploid repression after a delay. Transcriptome silencing is known to be important for quiescence, since defective silencing shortens life span [63]. In addition to the silencing defect, aneuploidy also disrupted cell densification that occurs during this time frame. Unlike euploid cells that densify after 4 days of culturing, all aneuploids tested showed statistically significant effects, with YPS1009_Chr4 and _Chr12 showing the largest defect (Figs 1F and S1C). Cells treated with salt (NaCl) stress as a control showed no difference, showing that the effect is not explained by generalized stress sensitivity of the aneuploid cells.

Quiescence is fundamentally important for normal life span in yeast [31,32,64,65]—indeed, aneuploids have a substantially shorter chronological life span (Fig 1G). Euploids of multiple strain backgrounds exhibited near 100% colony-forming capacity over 5 weeks of culturing. Instead, all but one aneuploid tested exhibited a dramatically shorter life span, with few, if any, viable cells remaining by the end of the time course (Fig 1G). The lone exception, YPS1009_Chr13, whose viability was relatively stable over 5 weeks but started at a significantly lower levels (60%–70% of cells). The reduced life span was also seen in two other strain backgrounds, again indicating that the effect is not specific to YPS1009. Our results expand past studies showing that chromosome duplication in the aneuploidy-sensitized W303 lab strain shortens replicative life span [66]; however, interpretation of those results are confounded since the strain lacks functional Ssd1 that is linked to both life span and quiescence [20,60,67,68]. Our results show that even in the presence of SSD1, aneuploidy disrupts normal aging.

Collectively, our results show that chromosome amplification disrupts multiple signatures of quiescence and life span, independent of which chromosome is amplified, but with some chromosome-specific effects. This strongly suggests a generalizable consequence of chromosome duplication on aging and life span, overlayed with chromosome-specific impacts.

A genetic screen for life span extension implicates the RQC pathway

To understand the mechanisms of life span limitation, we conducted a genetic screen in YPS1009_Chr12 cells as a representative (Fig 2A). We transformed the euploid and aneuploid strains with a low-copy, barcoded plasmid library expressing each of ~5,000 yeast genes flanked with their native upstream and downstream sequences [69]. YPS1009_euploid or YPS1009_Chr12 cells transformed with the library were pooled separately, grown to saturation, and cultured for 28 days, in three biological replicates. At 1 and 28 days, an aliquot of culture was plated and viable cells that grew into colonies were collected, prepared, and barcode-sequenced. We identified plasmids whose relative barcode abundance was enriched at 28 days compared to the starting library in each strain (at a relaxed FDR cutoff of 0.1 given noise in the assay, see Methods). A positive enrichment suggests a fitness benefit to cells that caused them to rise in frequency over time.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Genetic perturbation of the RQC pathway modulates cell-cycle defects.

(A) Genetic screen approach, see text. (B) Representation of the density of all barcode counts in the library (gray) and an indicator for one gene enriched in the pool already at day 1 (orange) and a second gene enriched only at day 28 (pink). (C) Percent budded cells in euploid (Eu) and aneuploid cultures harboring empty vector (EV) or plasmids encoding RQC1, LTN1, or HEL2 or in a strain lacking RQC2 (rqc2Δ) after 2 days of culturing. Asterisk, p < 0.05 comparing strains with gene plasmids to EV or rqc2Δ to WT; n = 3, Fisher’s exact test applied to count data. (D) As in C but with plasmids expressing wild-type RQC2 or the rqc2 D98Y mutant. None of the D98Y mutants were statistically different from the paired rqc2∆ strain (p > 0.1, Fisher’s exact test). (E) Average and standard deviation of percent viability (n = 4 or as listed in S2 Table) over time in denoted strains. Listed p-values calculated using a two-way ANOVA comparing each strain across time to the EV control. Asterisks above each plot represent a one-tailed, replicate-paired T test assessing individual timepoints, colored according to the key. (F) Average optical density at day 4. * p < 0.05, + = 0.06, one-tailed T test compared to paired EV control. See S2 Table for count data and p-values. The data underlying this figure can be found in S2 Table.

https://doi.org/10.1371/journal.pbio.3003509.g002

We identified 62 genes that reproducibly increased in the pool over 28 days versus the starting library in YPS1009_Chr12 and that were at least 2-fold more enriched than in the euploid analyzed in the same way (see Methods). We reasoned that genes that impact the health of aneuploid cells early in the time course would be already enriched at 1 day of culturing (Fig 2B, orange gene count), whereas genes that specifically impact life span would be more enriched at 28 days (Fig 2B, pink gene count). Of the 62 genes identified, 22 were more enriched (>2-fold) in abundance at 28 days compared to 1 day, implicating a specific impact on life span (S1 Data). Nearly half of these genes encode proteins localized to the mitochondria, whose function is critical for quiescence and life span but known to be defective in aneuploid yeast and mammalian cells [70–76]. The other 40 genes were enriched over the starting library at 28 days but also at 1 day of culturing (see Methods), suggesting an early impact on culture growth and cell health. Among these are genes already linked to life span from other studies, including sirtuin Hst2 tenuously linked to life span [77–79], several genes involved in autophagy (ATG12, POR1), which is required for healthy aging [80,81], and others discussed more below (SGT1, SCP1) [82,83].

Among the most intriguing on the list of 40 genes was RQC1, a component of the RQC pathway that resolves stalled ribosomes [38,39,84,85]. This was interesting because aneuploid strains are sensitive to translational inhibitor nourseothricin (NTC) that binds the ribosome to disrupt translation [13,25,86–88]. Given that translation defects increase with cellular age [47–49], these results raised the possibility that chromosome duplication induces defects in the RQC pathway.

The RQC pathway is directly involved in aneuploidy-dependent quiescence defects

We explored the impact of specific perturbations to the RQC pathway. Both Rqc1 and Ltn1 are stoichiometrically limiting in yeast and mammals [38,89,90]. Remarkably, the defect in cell-cycle arrest was significantly alleviated simply by duplicating RQC1 or LTN1 on a plasmid, in all aneuploids tested: significantly more cells in each culture arrested as unbudded cells by 2 days (Fig 2C). Conversely, deleting these genes made the arrest phenotypes worse (S2A Fig). We tested other players in the RQC pathway as well. Previous work showed that deletion of RQC2 can actually alleviate the stress of a dysfunctional RQC pathway. This led to the suggestion that CATylated intermediates may be toxic [45,54,91]. Indeed, deletion of RQC2 alleviated the arrest defect. Furthermore, expression of the rqc2 D98Y mutant unable to add CAT tails [42,43] complemented the defect to levels that are statistically indistinguishable from the full RQC2 deletion (p > 0.1, Fisher’s exact test, Fig 2D). We also tested the upstream sensor of stalled ribosomes, Hel2, which participates in multiple translational quality control pathways [92–94]; duplication of HEL2 also alleviated the arrest defect to varying degrees. We were unable to test the impact of Hel2 deletion, since the YPS1009_Chr4 HEL2 deletion strain was not viable and YPS1009_Chr12 hel2Δ cultures did not maintain aneuploidy, suggesting that these aneuploids are especially reliant on Hel2. Together, these results show that all of the tested players in the RQC pathway affect cell-cycle arrest in aneuploids entering quiescence. Augmenting RQC genes also affected other aneuploidy quiescence defects, improving morphology defects of YPS1009_Chr14 and _Chr15 cells (S2B Fig) and increasing final optical density and/or cell densification of YPS1009_Chr4 and _Chr12 aneuploids (S2C and S2D Fig).

An important question is if augmenting the RQC pathway improves aneuploid life span. We devised a more sensitive single-cell, live-dead assay to assess viability by day, in aneuploid cells over-expressing RQC1 or LTN1 as representatives. Increasing either gene improved viability in all three aneuploids scorable by this method, to comparable levels (Fig 2E, p < 1 × 10⁻⁵, two-way ANOVA comparing strains across timepoints to the isogenic control, S2 Table). Viability in aneuploids with the empty vector began declining as early as day 7 of culturing. In contrast, RQC1 or LTN1 over-expression increased viability compared to the empty vector out to 18 days (in the case of Chr4-RQC1, -LTN1) or 21 days (for Chr14-RQC1, Chr12-RQC1, LTN1, p < 0.05, one-tailed T test comparing each time point to the control, see S2 Table). Overproduction of the genes also increased the total cell number of YPS1009_Chr12 cells (Figs 2F and S2C), explaining why RQC1 produced a fitness benefit at day 1 of the genomic screen. The impact on viability is clearly partial, since none of the plasmids increased viability to euploid levels. Nonetheless, over-producing single RQC subunits improves aneuploid health and viability in all aneuploids tested.

Aneuploid cells accumulate stalled ribosomes and RQC defects

Results above point to a problem in RQC during quiescence in aneuploids. We therefore tested if aneuploid cells show signatures of ribosome stalling. We used a reporter with a stall-inducing stretch of 12 arginine codons between GFP and tdTomato coding sequences (Fig 3A). Healthy cells can readily dismantle and degrade ribosomes stalled on the reporter. We confirmed that euploids lacking RQC1 or LTN1 accumulate a smeared product consistent with CATylated GFP (Fig 3B). As expected, the absence of Rqc2 in euploids produced a crisp GFP band without the diagnostic smear of CATylation.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Aneuploids show signs of RQC defects.

(A) Ribosome stalling reporter, see text. (B) Representative anti-GFP Westerns of euploid (left) and aneuploid strains (right) carrying the ribosome stalling reporter, cultured for 4 days and normalized by cell number. (C) Average and individual data points (n = 4) of relative RQC signal normalized to Ponceau S (see Methods). Asterisk, p < 0.05, replicate-paired T test compared to euploid. (D) Representative GFP (top), tdTomato (middle), or bright field microscopy. Scale bar, 5 µm. (E) Ribosome traces (codon-level occupancy normalized to transcript-body occupancy, see Methods) were compared for each mRNA in euploid and aneuploid cells, and the correlation between traces was calculated as shown in the cartoon. The distribution of pairwise correlations comparing euploid and YPS1009_Chr12 cells is shown for cells in log phase (L) or entering quiescent (Q, day 4). (F) Average ribosome occupancy (as described in E) for peaks with higher occupancy (FDR < 0.05) in euploids (purple, top) or in YPS1009_Chr12 (orange, bottom). (G) Distribution of codon positions for peaks higher in euploids (purple) or in aneuploids (orange). Asterisk, p < 0.05, Wilcoxon test. Some outliers are omitted from display. Replicate distributions were not different from one another (p > 0.1 in all cases). (H) Amino acids (rows) at aneuploidy-increased ribosome occupancy sites that are enriched or depleted compared to the YPS1009 proteome, shown for statistically significant sites (FDR < 0.05, Fisher’s exact test). Asterisk, enrichments seen previously in yeast ribosome stall sites [47]. The data underlying this figure can be found in S2 Table.

https://doi.org/10.1371/journal.pbio.3003509.g003

In contrast to euploid cells, all of the wild-type aneuploids tested accumulated significantly more RQC intermediates by both Western analysis (Fig 3B and 3C) and microscopy (Figs 3D and S3). Interestingly, the more sensitive microscopy data indicate that these defects accrue over time: GFP signal indicative of stalled translation on the reporter was low in log-phase cells but accumulated as early as 1 day of culturing in all four aneuploids tested, preceding the budding defects seen at day 2 (S3 Fig). Interestingly, many cells accumulated both GFP and tdTomato signal by day 4 of culturing, reflecting translation through the stall site to produce full-length protein (Figs 3D and S3). These results suggest that aneuploid cells may suffer from different RQC defects at different chronological ages (see Discussion).

To confirm the stalling reporter results, we performed replicate ribosome footprinting on euploid and the YPS1009_Chr12 aneuploid as a representative, in log phase and after 4 days of culturing. Ribosome occupancy at each codon was calculated as codon-level read count normalized to gene-body read count as done previously [47]. In exponentially growing cells, the transcript-specific traces of ribosome occupancy were generally very similar between euploid and aneuploid cells (Fig 3E). In contrast, many traces of ribosome occupancy were reproducibly poorly correlated in euploid and aneuploids entering quiescence, indicating differences in ribosome pausing on the same transcripts. We identified codons with statistically significant differences in normalized ribosome occupancy (FDR < 0.05, Fisher’s exact test, see Methods). We found 1,600 sites in 568 transcripts that displayed higher ribosome occupancy in quiescent euploid samples, whereas 1,203 sites in 305 transcripts with higher occupancy in the quiescent aneuploid (Fig 3F). While there was a slight enrichment for Chr12 transcripts over those from other chromosomes (p = 0.03, hypergeometric test compared to measured RNAs), these comprised only 10% of affected RNAs. Many transcripts showed multiple peaks with opposing effects, such that one site on the transcript had higher occupancy in the aneuploid but other peaks on the same transcript were larger in the euploid. While investigating this, we noticed that aneuploid-enhanced peaks tended to occur nearer to the start of the transcript (Fig 3G, p < 0.05, Wilcoxon rank-sum test, S4 Fig), for reasons that will require further study to dissect. Sites with higher ribosome occupancy in the quiescent aneuploid were enriched for residues previously associated with ribosome stalling in yeast, including lysine and asparagine at the P site and glycine and glutamate at the A site (Fig 3H) [47]. Importantly, peaks that were called significantly different between log-phase strains were typically subtle in magnitude (S4B Fig) and showed no significant difference in associated sequences (FDR > 0.05 at all sites). Thus, aneuploid YPS1009_Chr12 cells show increased ribosome occupancy at sequences known to induce ribosome stalling, but only during quiescence, consistent with results using the ribosome stalling reporter.

Inducing ribosome stalling in euploids disrupts quiescence and life span

If problems managing ribosome stalling drive quiescence defects in aneuploids, then increasing the level of ribosome stalling in euploids should cause similar defects. Indeed, this was the case. Low doses of NTC induce ribosome stalling, evident by increased RQC intermediates from the stalling reporter in normalized cell counts (Fig 4A). This dose of NTC was enough to significantly increase the number of euploid cells that remained budding at day 3 (Fig 4B). Remarkably, simply over-expressing the RQC reporter had the same effect, whereas expressing a control protein without the stall-inducing sequence (“-R12”) did not. Expressing the reporter in conjunction with NTC treatment or in the absence of RQC1 or LTN1 significantly exacerbated the defects, whereas stressing cells with sodium chloride as a control had only a minor effect (Fig 4B). Together, these results indicate that increasing ribosome stalling in euploid cells induces the arrest defect. It also slightly but significantly disrupted cell densification compared to the euploid control (S5 Fig). In contrast, deleting RQC1 or LTN1 from the euploid had only a minor effect on budding, unless cells also carried the RQC reporter. Thus, exacerbated stalling is required to produce the arrest defect. Consistent with the model that increased stalling shortens life span, these treatments in euploid cells significantly shortened life span to much greater levels than a salt-stress control (Fig 4C). Together, these results show that it is the increase in or perhaps quality of ribosome stalling that affects quiescence and life span, rather than a defect managing basal levels of stalling seen in the euploid.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Induction of ribosome stalling disrupts quiescence and life span in euploid cells.

(A) Representative anti-GFP western blot of euploid (Eu) and YPS1009_Chr12 cells with the stalling reporter, as described in Fig 3B, cultured for 4 days with and without low-dose (1 µg/mL) NTC. (B) Average and individual data points of percent budded cells in wild-type, rcq1∆, or ltn1∆ euploid cells exposed to 1µg/mL NTC, carrying the ribosome stalling reporter (“R12”) or non-stalling control reporter (“-R12”), or exposed to 0.7M sodium chloride as a control. Asterisk, p < 0.05, +, p < 0.1, compared to WT or as indicated, n = 3–6; Fisher’s exact test applied to cell count data.(C) Average (n = 3) and standard deviation of colony forming capacity in euploid cells normalized to 7 days. Asterisk, p < 0.05 compared to untreated control at 35 days; T test. The data underlying this figure can be found in S2 Table.

https://doi.org/10.1371/journal.pbio.3003509.g004

Interestingly, we noticed that NTC treatment and/or the stalling reporter significantly exacerbated defects seen in aneuploid cells, including odd morphologies of starved cells entering quiescence. In fact, NTC treatment of YPS1009_Chr12 induced a small number of bi-lobed cells characteristic of YPS1009_Chr14 entering quiescence, and in a few instances produced cells in which nuclear division occurred perpendicular to the division plane (S6A Fig). These morphologies are reminiscent of those caused by defective Cdc34, the E2 ubiquitin conjugase of the SCF complex that marks cell-cycle regulators for timed degradation by the ubiquitin-proteasome system [95–97].

Genes linked to ubiquitin metabolism alter aneuploid arrest phenotypes

To further dissect how RQC defects could impact cell cycle arrest, we returned to hits from our screen in YPS1009_Chr12. Several hits were linked to SCF-dependent protein degradation, including Cdc34 regulator UBS1 [98,99], chaperone SGT1 that associates with SCF [100], and POG1 that has been implicated in G1/S regulation and can suppress defects in E3 ubiquitin ligase Rsp5 [101–103]. Duplication of UBS1 and POG1 partly alleviated the arrest defect in YPS1009_Chr4 and/or YPS1009_Chr12, whereas SGT1 had a mild effect but missed the significance cutoff (Fig 5A). This raised the possibility that a defect in SCF and/or protein degradation may disrupt turnover of important cell-cycle regulators. Nearly all budded cells lacked nuclear Whi5-GFP consistent with elevated Cln3 activity (S6B and S6C Fig). If cyclin degradation is disrupted in this strain, then over-expression of cyclins may exacerbate phenotypes. Indeed, duplicating CLN3 or downstream G1 and S phase cyclins CLN2 or CLB6 statistically significantly increased budding in both YPS1009_Chr4 and YPS1009_Chr12 aneuploids, with only a weak effect on euploid cells (Figs 5A and S6D). These results raise the possibility that cyclins including Cln3 are not properly degraded in aged aneuploids and show that increased cyclin gene copy exacerbates defects. Cyclins are notoriously challenging to detect by western blot; however, we were able to detect degradation products of HA-tagged Cln3, but only in quiescent YPS1009_Chr12 aneuploid and not quiescent euploid cells (S6E Fig). Validation of the model that Cln3 cyclins are not fully degraded will require further study.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Genetic perturbation influences aneuploidy phenotypes.

(A) log₂ change in % budded cells for Chr4 (blue) or Chr12 (magenta) aneuploids with denoted genes vs. empty vector (data for RQC subunits reprinted from Fig 2 for comparison). Other tested genes had no significant effect (AFG3, ATG12, GRR1, OTU1, PCL2, PRE1, RPN11, YGP1, LEE1). Not all genes were tested in YPS1009_Chr4. (B) Average and individual data points (n = 3) of proteasome activity measured in permeabilized cells (left) or lysate (right) in aneuploids relative to the paired euploid, for caspase (C), chymotrypsin (Ch), or trypsin (T), see Methods. Asterisk, p < 0.05 for samples with less activity than euploid. (C) Percent cells with Hsp104-GFP foci (n = 3–4) for aneuploids harboring indicated gene duplications. Asterisk, p < 0.05 compared to empty vector control, Fisher’s exact test applied to count data. The data underlying this figure can be found in S2 Table.

https://doi.org/10.1371/journal.pbio.3003509.g005

One possibility is that chromosome amplification disrupts inherent proteasome function, as previously proposed in the aneuploidy-sensitized laboratory strain [104,105]. However, this does not appear to be the case in the wild strain background: we measured three separate proteasome activities in live cells and cell lysates using luminescent reporters that do not require ubiquitination for degradation [106]. Overall, aneuploids tested did not show grossly lower proteasome activities compared to euploids, and several activities were substantially higher than euploid (Fig 5B). Aside of YPS1009_Chr14, which showed slightly lower levels of chymotrypsin and caspase activity, the other scorable aneuploids showed activity equal to or even greater than wild-type cells, with some variation depending on the assay. (Chr15 was not tested due to challenges normalizing cell numbers). Furthermore, none of the aneuploids was more sensitive than the euploid to proteasome inhibitor MG132 in this strain background, in log phase or based on final cell number in saturated cultures (S7A and S7B Fig). Thus, inherent defects in proteasome function or assembly are unlikely to explain the common quiescence defects seen across multiple aneuploid strains.

However, in testing possible models, we found that duplication of the stress-induced polyubiquitin gene UBI4 significantly alleviated the aneuploid arrest defect in Chr4 and Chr12 aneuploids, as did duplication of deubiquitinase (DUB) UBP6 that is important for ubiquitin recycling (Fig 5A). The alleviating effect of UBP6 duplication is the opposite effect reported for the sensitized W303 strain, where deletion of UBP6 provided a benefit, reportedly by relieving proteasome inhibition that occurs through a separable Ubp6 domain [104,105,107]. To distinguish between these functions, we tested catalytically inactive ubp6-C118A that can still inhibit proteasomal processivity but lacks ubiquitin recycling activity. This mutant did not mitigate aneuploid arrest defects, indicating that the DUB activity is required (Fig 5A). In contrast, duplicating ubiquitin ligase HUL5, which antagonizes ubiquitin recycling by extending ubiquitin chains and increasing proteasome processivity, exacerbated arrest defects (Fig 5A). Duplication of a different DUB, UBP14, which disassembles unanchored ubiquitin chains and on specific targets [108], increased defects. Together, these results suggest an issue with ubiquitin recycling, availability, and/or metabolism, rather than an inherent proteasome defect, that underlies aneuploid phenotypes. Notably, none of the gene duplications tested alleviated the arrest defect to the same level as RQC1 or LTN1.

Proteostasis stress is alleviated by duplication of RQC or ubiquitin

Trisomy 21 in humans and chromosome amplification in sensitized W303 yeast is associated with increased protein aggregation, although the source of proteostasis stress is not known. We previously showed that wild aneuploid yeast strains do not show signs of proteostasis dysfunction unless stressed by SSD1 deletion or treated with translational inhibitor NTC [13,25]. Here we found that aging also induces protein aggregation in wild aneuploids: at 7 days of culturing, 50%–70% of aneuploids with extra Chr4, Chr12, Chr14, or Chr15 harbored foci of protein disaggregase Hsp104 (Figs 5C and S8), compared to ~5% of euploid cells. The proportion of aneuploids with such aggregates decreased substantially upon duplication of RQC gene LTN1 or polyubiquitin UBI4 as representative RQC and ubiquitin-affecting genes (Fig 5C). We confirmed in YPS1009_Chr4 and _Chr12 aneuploids that deletion of RQC2 also significantly reduced Hsp104-GFP signal (S7D Fig). These results suggest that age-induced protein aggregates can be alleviated by duplication of limiting RQC subunit Ltn1 or polyubiquitin, or deleting RQC2.

We considered several models for how RQC defects could perturb ubiquitin homeostasis. One possibility is that protein aggregation triggered by RQC defects depletes free ubiquitin in aged aneuploids, as proposed in several neurological disorders associated with aggregates [109–111], especially since non-mitotic cells rely on ubiquitin recycling for proteostasis maintenance [112]. But western blots of bulk-culture lysates revealed that monoubiquitin is still visible in aged aneuploids (S7C Fig), indicating that at least some cells in the culture harbor free ubiquitin (see Discussion). We next asked if ubiquitin localization was different in aneuploid cells using fluorescence microscopy. Indeed, the ubiquitin profile was markedly different in aged aneuploids (Fig 6A). Whereas ubiquitin was evenly distributed across most of the aged euploid cells, the majority of aged aneuploid cells showed significantly brighter ubiquitin signal, often in discrete puncta (Fig 6B). Importantly, duplication of RQC1, LTN1 and HEL2, or deletion of RQC2, significantly reduced ubiquitin signal in all four aneuploids tested (Fig 6C and 6D). Thus, the impact of RQC augmentation on the ubiquitin localization phenotype parallels its impact on cell-cycle arrest, cell densification, Hsp104 aggregates, and viability in aging aneuploid cells. We were unfortunately unable to study co-localization of ubiquitin and RQC stalling products due to challenges with multi-antibody microscopy in fixed quiescent cells, thus leaving several possible causal models discussed below.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Aneuploids exhibit altered ubiquitin distributions.

(A) Representative brightfield and fluorescent images of euploid and aneuploid cells at log-phase and day 4 stained with fluorescent anti-ubiquitin antibody. Scale bar, 5 µm. (B) Distributions of maximum pixel intensity of ubiquitin signal per cell in euploid and aneuploid cells at log-phase and day 4, normalized to median signal intensity of paired euploid control. * p < 0.05, ** p < 1 × 10⁻¹⁰ Wilcoxon rank-sum test. (C) As in B but using split violin plots to show distributions of aneuploids harboring HEL2, LTN1, or RQC1 overexpression plasmid vs. paired empty vector (EV) controls done side-by-side. (D) As in C but for wild type (gray) or rqc2Δ cells, according to the color key (D). * p < 1 × 10⁻³, ** p < 1 × 10⁻¹⁰ Wilcoxon rank-sum test. Some outliers (for Chr4 in B and Chr14 LTN1 EV in C) are omitted from display, see S2 Table for counts. The data underlying this figure can be found in S2 Table.

https://doi.org/10.1371/journal.pbio.3003509.g006

Strains and plasmids

Strains used in this study are listed in S1 Table. YPS1009 aneuploids were generated in Rojas and colleagues [57]. Most chromosome duplications are stable for many generations; maintenance of aneuploidy was confirmed periodically by plating cultures from each experiment onto synthetic complete media with selection for marked chromosomes (SC −HIS +NTC). Gene deletions were generated by homologous recombination of the Hph-MX drug resistance cassette into the designated locus, followed by diagnostic PCR to confirm correct integration and absence of the target gene. Aneuploidy was confirmed and periodically checked through diagnostic qPCR of one or two genes on the affected chromosomes normalized to a single-copy gene elsewhere in the genome – normalized ratios close to 2 reflect gene duplication, and ratios between 1.2 and 1.8× indicated partial loss of aneuploidy in the cell population. HSP104-GFP was generated by integrating a GFP-ADH2 terminator::HIS3 cassette into the HSP104 locus [138] via homologous recombination. Aneuploid Hsp104-GFP strains were then generated through mating and dissection, crossing AGY1970 to relevant aneuploids. The GFP-{CGV}12-tdTomato stalling reporter was generated by PCR sewing and cloning the generated fragment into a KAN-marked CEN plasmid (pJH2). Unless otherwise noted, plasmids used in this work were from the MoBy 1.0 plasmid library [69]. The RQC2 plasmid was constructed by PCR amplification of the coding sequence and flanking regions (450 bp upstream and 200 bp downstream) of S. cerevisiae RQC2, then cloned into the MoBy 1.0 vector (KanMX marked). The RQC2_D98Y mutation was constructed by site-directed mutagenesis using Q5 Site-Directed Mutagenesis Kit (NEB).

Growth conditions

Unless otherwise noted, all experiments were performed in rich YPD (Yeast extract, Peptone, Dextrose) medium. Quiescent cultures were generated by inoculating liquid YPD medium at an optical density (OD₆₀₀) of 0.05. Cultures were allowed to reach saturation and then maintained at 30 °C in a shaking incubator for the number of days indicated, with no nutrient supplementation. Maintenance of aneuploidy was verified by plating an aliquot of aneuploid cultures onto rich YPD plates, then replica plating to SC −HIS +NTC after 24 hours to determine maintenance of the two chromosome markers. Aneuploidy was also periodically verified through diagnostic qPCR as described above. Where indicated, cells without the NAT-MX resistance cassette were treated with 1 ug/mL NTC after cells reached mid-log phase.

Microscopy

Cell-density fractionation

Density gradient sedimentation and fractionation of stationary phase cultures was performed using Percoll (Sigma, P1644) as described previously [140]. Gradients were split evenly between two fractions. Fractions were collected using 18-gauge needle and 10 ml syringe, harvesting the heavy fraction first, then washed once in PBS, and resuspended in 1 mL of PBS. Fractions were quantified using a hemacytometer and optical density.

Chronological life span assay

Chronological life span in Fig 1G was determined by plating for colony-forming capacity over time. At various times over long-term growth, an aliquot of culture was harvested, OD₆₀₀ measured, and cells diluted serially to a 40,000× dilution, which was spread on YPD plates. Plates were incubated for 48 hours, and viable colonies were counted using ImageJ Colony Counter Plug-in (ImageJ) to quantify colony-forming units. Colony-forming capacity was calculated as colony-forming units divided by optical density measured at the indicated day.

RNA sequencing

RNA-seq was performed using total RNA isolated from log-phase and quiescent cultures. Cultures were started at OD₆₀₀ 0.05. Log-phase cultures were harvested after precisely three generations. Day 1, 3 and 7 cultures were harvested 24, 72, and 168 hours after log-phase cultures were harvested. OD-normalized samples were pelleted by centrifugation and flash frozen with liquid nitrogen and maintained at −80 °C until RNA extraction. Samples were mixed with a defined number of flash-frozen Sz. pombe cells before RNA extraction, to later serve as per-cell normalization. Total RNA was extracted by hot phenol lysis [141]. Mechanical disruption was required to efficiently lyse quiescent cells: 425–600 µM glass beads (Sigma, G8772) were added to samples in phenol-lysis buffer such that glass beads accounted for 1/3 of total sample volume. Greater than 50% of empty space was maintained in sample tubes to ensure efficient lysis. Samples were then vortexed for 1 min in 10 min intervals for 1 hour. rRNA depletion was performed using the Ribo-Zero (Yeast) rRNA Removal Kit (Illumina, San Diego, CA). Libraries were prepared with TruSeq Stranded Total RNA kit and purified using a Axygen AxyPrep MAG PCR Clean-Up Kit (Axygen). Illumina reads were mapped to the S288c genome substituted with SNPs from YPS1009 as called in Sardi and colleagues [142], using bwa-meme [143]. Read counts for each gene were calculated by HT-Seq [144]. Normalization was conducted by setting the slope of Sz. pombe reads across samples to 1.0. Statistical analysis of log₂(fold change) transcript abundances was done in edgeR [145] taking genes with a FDR < 0.05 as statistically significant. Genes shown in Fig 1E were defined as those significantly repressed (FDR < 0.05) in the euploid strain and statistically significantly higher (FDR < 0.05) in all four aneuploids, in at least one time point comparing that aneuploid to the euploid. Hierarchical clustering was performed using Cluster 3.0 [146] and visualized in Java Treeview [147]. Data represent the average of biological duplicate and are available in GEO Accession #GSE269236. Data for Fig 1D are available in S2 Data.

Genetic screen

Dual-marked YPS1009_euploid (AGY1611) and _Chr12 strains (AGY1612) were transformed with Moby 1.0 low-copy expression library [69] and viable colonies were scraped and frozen at −80 °C. Pooled cells were used to inoculate rich medium and then grown in biological triplicate for 28 days in YPD medium + G418 to maintain plasmids. A portion of each culture was harvested at 24 hours and 28 days. The harvested portions were plated on multiple plates of YPD + G418 to select for cells that were viable and able to form colonies, thus representing quiescent cells that re-entered the cell cycle. After 48 hours of growth, lawns were scraped, collected, and flash frozen. Plasmid DNA was collected from the starting pools, day 1, and day 28 samples using Zymoprep Yeast Plasmid kit (Zymo Research, D2004), with the following changes: 425–600 µM glass beads were with the lysis reagent, and samples were vortexed for 10″ three times during lysis. Samples were incubated on ice for 30 min after adding neutralization buffer. Barcodes were sequenced as previously described [148,149]. EdgeR was used to TMM normalize samples as previously described [145]. Genes with a significant positive log₂(fold change) in barcode abundance (FDR < 0.05 or FDR < 0.1) at day 28 versus starting pool were considered enriched, i.e., beneficial. We then selected genes with a 2-fold or greater linear difference in enrichment between average YPS1009_Chr12 sample versus average euploid sample, which resulted in 43 candidate genes at an FDR < 0.05 and 62 genes at FDR < 0.1. We focused on the relaxed stringency list to select candidates for downstream validation. Twenty-two of those genes were >2× more enriched at d28 versus d1 in the YPS1009_Chr12 strain, whereas the remaining genes were within 2-fold enrichment at both d1 and d28. Hierarchical clustering was performed on the log₂(fold change) abundance differences using Cluster 3.0 [146] and visualized using Java TreeView [147]. Data are available in GEO Accession #GSE269237.

Western blotting

Yeast strains were grown as described above, with the following additions: G418 was used to maintain the ribosome stalling reporter, and cells exposed to NTC were treated with 1 µg/mL NTC after cells reached mid-log phase. 2 OD units were harvested and flash frozen. Samples were lysed in 2× Laemmli buffer with glass beads (500 µM, Sigma). Proteins in Laemmli buffer were resolved in 4%–12% SDS-PAGE gels and transferred to nitrocellulose membrane (0.2 µm, LICOR), unless otherwise indicated. Ubiquitin blotting was performed using 12%–14% Bis-Tris PAGE gels and transferred onto PVDF membrane (0.2 µm, Amersham Hybond LFP). Western blots were developed using anti-GFP (Abcam, ab290) for samples containing the ribosome stalling reporter, anti-ubiquitin (LifeSensors, VU101), or anti-HA (Cell Signaling Technology, C29F4). Blots were developed on a Li-COR Odyssey instrument (Model 9120). Li-COR Odyssey software was used to quantify signal intensity of GFP and ubiquitin antibodies. Repeated attempts to blot against common loading controls were unsuccessful in quiescent cultures; therefore, Ponceau S signal was used to normalize protein loading levels, as performed by others [150]. Ponceau S signal was quantified using FIJI (ImageJ).

Proteasomal activity

Proteasome-glo Cell-Based Assays (Promega) was used to measure proteasomal activities in vivo and in vitro in lysates. Reagents were prepared according to manufacturer’s instructions. An equivalent number of yeast cells were flash frozen then lysed via vortexing with 500 µM glass beads on ice. Cell lysate and Proteasome-glo reagents were combined 1:1 in an opaque, white-walled 384-well plate (Corning). Luminescence was measured using a Tecan M1000 Pro.

Ribosome profiling and analysis

Cells were harvested using vacuum filtration with Whatman Nylon filters (Cytiva, 7410-004). Cells were immediately scraped from filters, transferred to eppendorf tubes, and immediately flash frozen. Collection time was <60 seconds. Ribosome profiling was performed by Ezra Biosciences as previously described [151]. Samples were sequenced on an Illumina Novaseq instrument and processed as described in Schuller and colleagues [152] as follows: reads were trimmed with CutAdapt (version 3.5) [153] with command j 8 -g ^GGG -a A{10} -n 2 -m 15 --max-n=0.1 --discard-casava. Reads with poor quality at the 5′ end base (quality score ≤ 10) were removed, reads were mapped to noncoding RNAs from Schuller and colleagues [152], and remaining unmapped reads were mapped to the YPS1009 genome [57] using bowtie2 (version 2.5.1) [154]. The 5′ position of each read was scored, and the P site taken to be at 12 nt into the read [151]. Reads matched well to the expected frame in all samples (see S4 Fig for examples). For each gene, read starts were summed for each position from −72 of the gene ATG and + 60 of each stop codon in the YPS1009 genome. Genes without an annotated ATG were omitted from analysis, as were genes with introns. Read counts were summed for each codon, incremented by 1 pseudocount, and then normalized to the mean read counts per codon (with appropriate pseudocounts) in each gene body, from 60 nt (20 codons) into the gene to 60 nt (20 codons) from the 3′ end as done previously [47]. Genes with at least 50 reads per gene body were retained for further analysis.

The correlation between transcript profiles shown in Fig 3E was taken as the uncentered Pearson correlation for each transcript as measured in euploid and aneuploid, paired by replicate (Fig 3E). Significant differences in ribosome peaks across replicate-paired aneuploid-euploid samples was calculated using Fisher’s Exact test with Benjamini-Hochberg multiple test correction [155], by comparing read count in each sample at a given codon to gene-body read counts for that transcript (# reads at that codon, # reads in the gene body, for euploid versus aneuploid in each replicate separately). Peaks more abundant in the aneuploid were taken as those with FDR < 0.05 and for which the normalized ratio of read counts at that codon was greater in aneuploids; vice versa for peaks more abundant in the euploid. Motif analysis in Fig 3H was performed as follows: codons whose normalized read count differed between euploid and aneuploid samples (FDR < 0.05) were combined across replicates, and peaks more abundant in aneuploids versus euploids were selected. Ten amino acids flanking each peak site were retrieved from the YPS1009 proteasome. The frequency of each amino acid (and stop codon) at each position in the matrix was calculated as the number of occurrences of that amino acid divided by the number of peaks scored. Count and total values were compared at each position in the matrix compared to amino acid frequency in the YPS1009 proteome, using Fisher’s exact test and Benjamini-Hochberg FDR correction. Enrichments shown in Fig 3H were taken as the log2(fold difference) in frequency and shown only for statistically significant positions (FDR < 0.05, Fisher’s exact test). Quiescent aneuploids showed significant differences in amino acid composition at peaks detected in quiescence (Fig 3H); there were no significant differences for a comparable analysis done for log-phase cells (FDR > 0.05 in all cases). Data are available in GEO Accession #GSE269238. Custom scripts used are available in https://doi.org/10.5281/zenodo.17281997.