|Home | About | Journals | Submit | Contact Us | Français|
Aneuploidy causes a proliferative disadvantage in all normal cells analyzed to date, yet this condition is associated with a disease characterized by unabated proliferative potential, cancer. The mechanisms that allow cancer cells to tolerate the adverse effects of aneuploidy are not known. To probe this question, we identified aneuploid yeast strains with improved proliferative abilities. Their molecular characterization revealed strain-specific genetic alterations as well as mutations shared between different aneuploid strains. Among the latter, a loss of function mutation in the gene encoding the deubiquitinating enzyme UBP6 improves growth rates in four different aneuploid yeast strains by attenuating the changes in intracellular protein composition caused by aneuploidy. Our results demonstrate the existence of aneuploidy-tolerating mutations that improve the fitness of multiple different aneuploidies and highlight the importance of ubiquitin-proteasomal degradation in suppressing the adverse effects of aneuploidy.
Aneuploidy, defined as any chromosome number that is not a multiple of the haploid complement, is associated with death and severe developmental abnormalities in all organisms analyzed to date (reviewed in (Torres et al., 2008; Williams and Amon, 2009)). Aneuploidy is the leading cause of mis-carriages and mental retardation in humans and found in 90 percent of human cancers (Hassold and Jacobs, 1984; Holland and Cleveland, 2009). Despite the high incidence of aneuploidy in tumors, its role in tumorigenesis remains uncertain (Holland and Cleveland, 2009; Schvartzman et al., 2010).
To shed light on the relationship between aneuploidy and tumorigenesis, we previously determined the effects of aneuploidy on normal cells. Twenty strains of budding yeast each bearing an extra copy of one or more of almost all of the yeast chromosomes (henceforth disomic yeast strains) display decreased fitness relative to wild type cells and share traits that are indicative of energy and proteotoxic stress: metabolic alterations, increased sensitivity to conditions that interfere with protein translation, folding and turnover (Torres et al., 2007), a cell proliferation defect (specifically a G1 delay), and a gene expression signature known as the environmental stress response (Gasch et al., 2000). These shared traits are due to the additional gene products produced from the additional chromosomes. Primary aneuploid mouse cells exhibit similar phenotypes (Williams et al., 2008). Based on these findings, we proposed that aneuploidy leads to an “aneuploidy stress response”. In this response, cells engage protein degradation and folding pathways in an attempt to correct protein stoichiometry imbalances caused by aneuploidy. This puts a significant burden on these protein quality control pathways resulting in increased sensitivity to compounds that interfere with protein degradation and folding. Synthesizing and neutralizing the proteins produced from the additional chromosomes also lead to an increased need for energy.
The increased sensitivity of many aneuploid yeast strains to cycloheximide and proteasome inhibitors suggests that ubiquitin-mediated protein degradation is one of the protein quality control pathways as being affected in aneuploid cells. During ubiquitin-mediated protein degradation multiple ubiquitin molecules are covalently linked to a substrate, which allows recognition by the 26S proteasome (Varshavsky, 2005). Upon recognition, ubiquitin chains are removed and substrates are fed into the catalytic cavity of the proteasome. Two deubiquitinating enzymes, Rpn11 and Ubp6, remove ubiquitin from substrates (Chernova et al., 2003; Hanna et al., 2003; Verma et al., 2002; Yao and Cohen, 2002). Both of these proteases are associated with the proteasome and are essential for ubiquitin recycling. In the absence of either protein, levels of free ubiquitin rapidly decline due to degradation of ubiquitin chains by the proteasome. In addition to a role in ubiquitin recycling, Ubp6 regulates proteasomal degradation. In its absence, proteasomal degradation of several substrates is accelerated (Hanna et al., 2006; Peth et al., 2009). The results described here indicate that Ubp6, through its role in protein degradation control, affects the proliferative abilities of several aneuploid yeast strains.
The consequences of system-wide aneuploidy of only a single chromosome are severe in all organisms analyzed to date (reviewed in (Torres et al., 2008)). In striking contrast, in most cancer cells aneuploidy is common, typically involving many chromosomes but proliferation potential in these cells is high (reviewed in (Albertson et al., 2003)). To resolve these contradictory observations, we hypothesized that genetic alterations must exist that allow cancer cells to tolerate the adverse effects of aneuploidy. To test this idea, we isolated aneuploid yeast strains with increased growth rates and characterized their genetic alterations. This analysis revealed strain-specific genetic changes and mutations shared between different aneuploid strains. We characterized further one of these shared genetic alterations, a loss of function allele in the gene encoding the deubiquitinating enzyme Ubp6. Our studies show that inactivation of UBP6 improves the proliferation rates of four different disomic yeast strains and suggest a mechanism for this suppression. Deletion of UBP6 attenuates the effects of aneuploidy on cellular protein composition. Our results demonstrate the existence of aneuploidy-tolerating mechanisms. Enhanced proteasomal degradation appears to be one of them.
To identify genetic alterations that suppress the adverse effects of specific aneuploidies or perhaps even multiple different aneuploidies, we sought variants of 13 different disomic yeast strains that proliferate well despite the presence of a disomic chromosome. To isolate variants of disomic yeast strains with decreased doubling time, we used continuous growth under conditions that select for the presence of the disomic chromosome rather than a traditional mutagenesis approach to keep the number of genetic alterations low (Experimental Procedures).
Environmental conditions, such as media composition greatly influence the outcome of evolution experiments (Gresham et al., 2008; Zeyl, 2006). Therefore, we initially chose two sets of disomic yeast strains, one that required growth in medium lacking uracil and histidine (−Ura−His medium) to select for the presence of the extra chromosome, and another that required growth in medium lacking histidine and containing the antibiotic G418 (−His+G418 medium). The doubling time of the disomic yeast strains was significantly longer in −His+G418 medium compared to −Ura−His medium (data not shown). We suspect that this is due to G418’s ability to cause frameshifts during translation (Davies and Davis, 1968; Davies et al., 1964). The increase in frameshifts further enhances the burden on the protein quality control pathways that help aneuploid cells deal with the proteins produced from the additional chromosomes. The greater difference in doubling time between wild-type and aneuploid cells in −His+G418 medium together with the finding that some disomic strains (e.g. disome V) appeared to lose large parts of the additional chromosome more readily in −Ura−His medium (data not shown) prompted us to perform the selection for disomic strains with increased proliferative rates in −His+G418 medium. Passaging cells in this medium initially led to an increase in doubling times in many strains (Figure 1A; Table S1). We do not yet understand the molecular basis for this transient slowing of cell proliferation but we note that it is reminiscent of the crisis period observed during serial passage of primary mammalian cells in culture (Todaro and Green, 1963). Populations with decreased doubling times emerged shortly thereafter (Table S1).
We isolated single colonies after 9 days (37 – 66 generations; Table S1) and 14 days (64 – 105 generations; Table S1). Doubling time measurements confirmed that 11 out of 13 disomic yeast strains had produced clones with significantly increased proliferation rates (Figure 1B) and changed the cell cycle distribution to be more similar to that of wild-type cells (i.e. Figure S1A). Because of the additive inhibitory effects of aneuploidy and G418 on cell proliferation in the disomic yeast strains, we predicted to obtain two types of suppressor mutations: mutations that improve the growth of disomic yeast strains only in −His+G418 medium and mutations that improve proliferation irrespective of which medium cells are cultured in. This appeared to be the case. All evolved isolates obtained from disomes IX, XI, XIII and XVI (the disomic strains whose proliferation is only minimally affected in YEPD medium to begin with) showed fitness gain only in −His+G418 medium but not in YEPD (Figure S1B). This phenomenon of genomic alterations being condition specific has been observed previously (i.e.(Dettman et al., 2007)). We conclude that aneuploidy-tolerating mutations exist that are growth condition specific and that improve proliferation in general.
To determine the basis for the decrease in doubling time in the evolved disomic strains, we first examined their karyotypes. Comparative Genome Hybridization (CGH) analysis revealed that the overall chromosomal composition was not altered in the majority of disomic strains (Table S2). Thus, the improved growth rates of these isolates must be caused by alterations that are undetectable by CGH analysis.
Descendants of strains disomic for chromosome IV experienced loss of the entire additional chromosome and most diplodized (Table S2). Isolates obtained from strains disomic for either chromosome XII, XIV or XV had lost large parts of one copy of the duplicated chromosome but also carried a duplication of a region of the left arm of chromosome XIII (TEL13L – YML046W; 183 kb, 345 genes; Table S2). It is highly likely that loss of all or part of the chromosome present in two copies is in large part responsible for the increase in proliferation rate seen in the evolved strains, but we speculate that genes exist in region TEL13L – YML046W, whose twofold increase in copy number improves proliferation of three different disomic yeast strains.
Truncations of the duplicated chromosome occurred in or next to Ty elements, retrotransposons that are scattered throughout the yeast genome. This correlation indicates that homologous recombination between these repeated elements was responsible for the loss of these regions. The ends of regions TEL13L – YML046W were also at or near Ty elements. Given that region TEL13L – YML046W does not carry a centromere but is nevertheless stably inherited, it is highly likely that the duplicated region TEL13L – YML046W represent a translocation caused by Ty element mediated recombination. Our results indicate that cells carrying an extra chromosome rapidly evolve and acquire genomic alterations. These include point mutations (see below), truncations, amplifications and whole genome duplications.
We showed previously that the majority of genes present on the disomic chromosome are expressed according to gene copy number exhibiting an average increase in gene expression of approximately 1.82-fold (Torres et al., 2007). Down-regulation of gene expression of the disomic chromosome, like loss of large parts of the additional chromosome could lead to increased proliferation rates. Gene expression analysis of the evolved strains that retained both copies of the disomic chromosome showed that gene expression of the chromosome present at two copies was not attenuated even though proliferation rates were increased (Figure 1C). Average expression of genes present on the disomic chromosome was increased an average of 1.84-fold compared to the rest of the genome. Thus, attenuation of gene expression of the disomic chromosome is not responsible for the improved proliferation rates.
Our previous analysis of the disomic strains revealed a transcription profile shared by different disomes (Torres et al., 2007). This aneuploidy signature was only seen under conditions that eliminated the differences in growth rate between aneuploid strains (cells were grown in the chemostat under phosphate-limiting conditions). Gene expression analysis of the evolved isolates grown under these conditions confirmed that global gene expression patterns were maintained, with each evolved strain clustering most closely with its parental disomic strain (Figure S2A). Interestingly, the gene expression patterns of the two evolved disomic strains that we analyzed were more similar to each other than to the parental disomic strain (Figure S2A). This result suggests that the genetic alterations in the different isolates affect the same pathways and lead to a similar transcriptional response in the evolved strains.
To determine whether the evolved strains share a transcriptional profile that is distinct from that shared by the parental strains, we subtracted the original disome expression values from that of the evolved strains. This analysis revealed a common expression pattern among the evolved strains (Figure S2, Table S3). Ion transport, especially iron, and a subset of ribosomal proteins were significantly enriched in the decreased expression cluster (Table S3). Genes with increased expression were enriched for genes involved in amino acid metabolism (p value = 9.69*10−20). This group includes many of the genes responsible for biosynthesis of aromatic amino acids, branched chain amino acids, and arginine (Table S3). The significance of this expression signature is at present unclear but we speculate that increased protein synthesis due to the presence of an additional chromosome (see below) may bring about the need for increasing production of amino acids. Strain-specific expression changes also occurred. For example, a small group of genes increased in expression in both isolates from disome IX (Figure S2B). However, these gene groupings were rarely enriched for particular classes of genes, although they may be more informative when combined with knowledge of the mutations carried by these strains. We conclude that descendants of disomic strains with improved growth share a gene expression signature.
Evolved aneuploid strains that proliferate faster yet have maintained both copies of the disomic chromosome likely harbor heritable alterations not detectable by CGH. We selected 14 strains in which to identify these genetic alterations because their proliferation rates were significantly improved compared to the parent strain (Figure 1B). Tiling arrays or deep sequencing identified 43 SNPs that led to non-synonymous changes (Table 1) and 4 SNPs that led to synonymous genetic alterations that were verified by Sanger sequencing (Table S4A). In two evolved isolates of disome XIII we could not detect any non-synonymous genetic changes. A one base pair deletion, 10 synonymous and 21 non-synonymous alterations were present in the parental disomic strains (Table S4B). We note that the mutations already present in the parental disomic strains were likely acquired during their construction and could also confer a growth advantage.
Each evolved strain contained between 2 and 7 SNPs and little overlap was detected among descendants from the same parent strain (Table 1), indicating that different alterations lead to improved proliferation in the different disomic strains. Identical point mutations were only isolated among different descendants of disomes XI and XIV indicating that a selective sweep had not occurred in the evolution experiments. Interestingly, all three evolved disome XVI strains contained unique mutations in the poorly characterized SVF1 gene (Table 1). The emergence of mutations in this gene in three independent isolates of disome XVI with improved growth properties suggests that inactivation or hyperactivation of this factor (we do not know how the identified point mutations affect SVF1 function) confers a selective advantage on strains disomic for chromosome XVI.
Mutations in two genes were identified in descendants of different disomes. Point mutations in the gene encoding the vacuolar-targeting factor VPS64 were identified in descendants of disome IX and XI (Table 1). Mutations (premature stop codons) in the gene encoding the deubiquitinating enzyme UBP6 were identified in descendants of disome V and IX. This finding raises the interesting possibility that mutations exist that improve growth rates of more than one disome.
Genes involved in chromatin remodeling, stress response and protein folding, as well as rRNA processing were among those mutated in the evolved disomic strains and could contribute to the improved proliferative abilities of the evolved disomic strains. Striking, however, was the fact that fast growing descendants of strains disomic for chromosomes V, VIII, IX, XI and XIV harbored mutations in genes encoding proteins involved in proteasomal degradation (UBP6, RPT1, RSP5, UBR1). These results suggest that changes in protein degradation lead to an improvement in fitness in multiple aneuploid yeast strains.
We decided to test whether a causal relationship exists between mutations in UBP6 and improved proliferation rates of the evolved strains, because sequence analysis identified premature stop codons in UBP6 in two different evolved disomic strains. Ubp6 contains an ubiquitin-like (UBL) domain in its N-terminus that mediates binding to the proteasome and a peptidase domain in the C-terminal half of the protein (Figure 2A). Strain Dis V-14.1 carries a nonsense mutation resulting in the conversion of glutamic acid 256 to a stop codon (ubp6E256X; Figure 2A). Strain Dis IX-14.1 harbors an UBP6 allele that carries a premature stop codon at position 404 (Figure 2A). Both mutations leave the UBL domain of the protein intact but cause enough of a truncation to inactivate Ubp6’s protease activity. To determine whether the expression of this truncated version of UBP6 was at least in part responsible for the decrease in generation time of strains Dis V-14.1 and Dis IX-14.1, we analyzed disome V cells carrying the ubp6E256X mutation.
To assess the effects of this mutation on fitness, we performed a competition assay. In this assay, strains disomic for chromosome V carrying a GFP-PGK1 fusion integrated at URA3 were co-cultured with disome V cells carrying the ubp6E256X mutation also marked with URA3. We then monitored the fraction of GFP positive cells in the cultures over time by flow-cytometry. Control experiments showed that, with the exception of strains disomic for chromosome XIV, the GFP-PGK1 fusion did not affect the proliferation rate of the different disomic strains (data not shown).
Disome V cells carrying the ubp6E256X mutation proliferated significantly better than disome V cells wild type for UBP6 (Figure 2B; Figure S3). A truncation mutation in UBP6 was also identified in disome IX strains with improved proliferative abilities. In this strain too, replacement of the UBP6 locus with the ubp6E256X allele led to an increase in fitness (Figure 2B; Figure S3). Remarkably, the same allele also led to an increase in proliferation rates in strains disomic for chromosome VIII and XI (Figure 2B). The ubp6E256X allele did not improve the proliferative abilities of wild-type cells or of five other disomes (disome I, XII, XIII, XV, XVI) that we analyzed (Figure S3) and had adverse effects only in disome II and disome XIV cells (Figure 2B; Figure S3). Deletion of UBP6 had similar effects on disomic strains as expression of the UBP6 truncation. An increase in fitness was observed in co-culturing assays and in doubling time measurements (Figure 2C, D, Figure S4; data not shown). Analysis of cell cycle progression of disome V and disome XI cells lacking UBP6 revealed that the deletion suppresses the G1 delay of these two disomic strains (Figure S1A). Finally, we found that inactivation of UBP6 led to an increase in fitness of strains disomic for chromosome XI, and V in YEPD medium but not of strains disomic for chromosome VIII or IX (Figure 2E). We conclude that inactivation of UBP6 improves the growth rates of four different disomic strains in the presence of the translation inhibitor and proteotoxic compound G418. In two disomic strains growth improvement was also seen in the absence of the drug.
Next, we wished to determine the degree to which loss of UBP6 function contributes to the increased fitness of evolved Dis V-14.1 cells. We compared the doubling times of evolved Dis V-14.1 cells with that of disome V cells deleted for UBP6. Deletion of UBP6 did not affect cell cycle progression or doubling time in wild-type cells (Figure S1A). However, it led to a significant decrease in doubling time in disome V cells (4.2 +/− 0.2 hrs compared to 5.8 +/− 0.8 hrs; Figure 2D) but doubling times were not as short as those of the evolved Dis V-14.1 strain (3.8 +/− 0.1; Figure 2D). Conversely, restoring UBP6 function to the evolved Disome V-14.1 isolate reduced the proliferative potential of these cells (Figure S5). We conclude that loss of UBP6 function contributes to the increased proliferative abilities of Dis V-14.1 cells but other genetic alterations found in this strain also contribute to the increased proliferation rates of this isolate.
Loss of Ubp6 function causes ubiquitin depletion. This leads to cycloheximide sensitivity that can be suppressed by overexpressing ubiquitin (Hanna et al., 2003). Ubiquitin depletion was also observed in disome V ubp6Δ cells (Figure 3A). To determine whether ubiquitin depletion was responsible for the increased growth rate of disome V ubp6Δ cells, we examined the consequences of increased ubiquitin expression. Disome V and XI cells were co-cultured with disome V ubp6Δ and disome XI ubp6Δ cells, respectively. All strains carried a multi-copy plasmid expressing the ubiquitin-encoding gene, UBI4, under the control of the copper inducible CUP1 promoter. Addition of 100 μM CuSO4 significantly increased the steady state levels of free ubiquitin in all strains (Figure 3B). As expected, deletion of UBP6 suppressed the subtly adverse effects of overexpressing ubiquitin in wild-type cells (Figure 3C, F). However, high levels of ubiquitin did not abolish the growth rate improvements of disomic strains brought about by the inactivation of UBP6 (Figure 3D, E; Figure S6A). Similar results were obtained in disome VIII or XI strains harboring the ubp6E256X truncation allele (Figure 3G, H; Figure S6B) and in competition experiments where only the UBP6 deleted strains overexpressed ubiquitin (Figure S6C). Our results indicate that low levels of ubiquitin are not responsible for the improved fitness of disomic strains lacking UBP6.
Ubp6 deubiquitinates substrates at the proteasome. This activity serves two purposes: recycling of ubiquitin and rescue of proteasome substrates from degradation. UBP6 not only antagonizes the proteasome through its deubiquitinating activity but also through a non-catalytic mechanism (Hanna et al., 2006; Peth et al., 2009). To determine whether the catalytic or non-catalytic function of Ubp6 was involved in modulating the fitness of disomic yeast strains, we examined the consequences of replacing the catalytic cysteine 110 with alanine (ubp6CA). Expression of the ubp6CA allele did not affect the proliferative abilities of wild-type cells (Figure 4A, Figure S7). In contrast, co-culturing of disome VIII, IX and XI cells with disomic cells carrying the ubp6CA allele showed that strains harboring the catalytic dead version of the protein quickly outcompete disomes carrying the wild-type UBP6 allele (Figure 4B-D). Our results demonstrate that Ubp6’s protease activity antagonizes proliferation in several disomic yeast strains.
Inhibition of the catalytic activity of the mammalian homolog of Ubp6, Usp14, leads to accelerated degradation of a number of proteins (Lee et al., 2010). These findings lead us to hypothesize that increased proteasomal degradation of an unknown number of proteins improves the fitness of disomic yeast strains. A prediction of this hypothesis is that lowering proteasomal activity decreases the fitness of disomic yeast strains. This appears to be the case. We previously showed that several disomic strains exhibit increased sensitivity to the proteasome inhibitor MG132 (Torres et al., 2007). Furthermore, a conditional loss-of-function allele in the proteasome lid subunit Rpn6 encoding gene (Ben-Aroya et al., 2008), was synthetic lethal with disomy XII and disomy XIV (data not shown) and decreased the proliferative abilities of almost all disomic strains tested (Figure 4E). Finally, we found that the ubiquitin profile in strains disomic for chromosome V, VIII or XI resembles that of hypomorphic proteasome mutants: the levels of free ubiquitin are slightly reduced (Figure 3A, B). Our results indicate that proteasomal degradation is a rate-limiting pathway in most, or perhaps all, disomic yeast strains.
To test the idea that increased protein degradation leads to improved fitness of disomic strains, we examined the effects of deleting UBP6 on the proteome of a yeast strain whose fitness is improved by the deletion of UBP6 (disome V) and one that is not (disome XIII). To measure relative protein abundance in disomic and wild-type cells, we utilized stable isotope labeling with amino acids in cell culture (SILAC) based quantitative mass spectrometry (Supplemental Materials).
SILAC analysis of disome V and XIII relative to wild-type cells revealed quantitative information for 2,953 proteins (60.7% of all verified ORFs) and 3,421 proteins (70.3% of all verified ORFs), respectively (Figure 5C, E; Table S5). The analysis of the average abundance of proteins encoded by the genes located on chromosome V and XIII demonstrated that the average protein levels of chromosome V-located and chromosome XIII-located genes were increased by 1.8 fold and 1.9 fold compared to the non-chromosome V or XIII encoded proteins, respectively. This correlation is best seen when proteins are sorted with respect to the chromosomal position of their encoding genes (Figure 5C, E). To control for artifacts caused by growth in medium containing heavy lysine, we performed a reverse labeling experiment, growing disome V cells in light medium and wild type cells in heavy medium and compared the results of both analyses. We obtained quantitative information on 2,755 proteins, of which 2,433 were detected in both forward and reverse experiments (r2 = 0.59). Of these, 431 proteins show significant up- or downregulation in disome V with high reproducibility (0.49 < log2 ratio < −0.49; r2 = 0.78, n = 431; Supplemental Materials).
An interesting additional aspect of the quantitative assessment of the protein composition of the disomic strains is that we are able to determine whether there are proteins whose levels do not increase according to gene copy number. A comprehensive analysis of multiple disomic strains will be presented elsewhere but several general conclusions are summarized here. We previously analyzed the abundance of a small number of proteins in disomic yeast strains and found that the levels of several of these, especially subunits of macromolecular complexes such as ribosome subunits, did not exhibit a coordinate increase between gene copy number and protein levels (Torres et al., 2007). Consistent with these observations, we find that a considerable fraction of proteins located on chromosome V, 30 of a total 135 proteins detected in both disome V experiments, were not upregulated according to gene copy number. 90% of the proteins that exhibit this property are part of macromolecular complexes. Similar results were obtained with disome XIII cells. 21 percent of proteins (65 of 312) did not show co-regulation of protein levels with gene copy number. 68% of these proteins function in large macromolecular complexes. A discrepancy between gene copy number and protein levels was most evident for ribosomal subunits, but was also observed for subunits of ribonucleotide reductase and the vacuolar ATPase. The enrichment of protein complex subunits in the group of disome-encoded proteins that does not show a coordinate up-regulation with gene copy number is of high statistical significance, when compared to all proteins encoded by chromosome V or XIII that are part of protein complexes (p value = 1.1*10−10 for disome V; p value = 3.8*10−3 for disome XIII). Analysis of RNA and protein levels indicates that down-regulation of gene expression occurred either at the level of transcription (14 genes in disome V and 22 genes in disome XIII) or posttranscriptionally (16 genes in disome V and 43 genes in disome XIII). Characterizing the feedback mechanisms that ensure accurate stoichiometries of these proteins will be an important aspect of understanding the effects of aneuploidy on cell physiology.
Having established the effects of disomy V on the yeast proteome, we next wished to test the hypothesis that loss of UBP6 function improves the fitness of aneuploid cells such as disome V cells by increasing the degradation of proteins that are in excess in this strain. If this was the case, the protein composition of disome V ubp6Δ cells should be more similar to wild-type cells than that of disome V cells is to wild-type cells. This appears to be the case.
We obtained quantitative information on 2,895 proteins for disome V ubp6Δ cells (Figure 5D, Table S5) and on 3,491 proteins for cell lacking UBP6 (Figure 5B; Table S5). For the analysis of the effects of UBP6 on protein composition, we only included proteins for which quantitative information was obtained in all four strains (2,352 proteins). To determine whether deletion of UBP6 attenuates the effects of disomy V on the intracellular protein composition, we rank-ordered all the proteins according to their relative protein abundance levels in the strain disomic for chromosome V and then asked how the expression of these proteins changes in disome V cells lacking UBP6. To quantify a potential attenuating effect, we created three bins: one that encompasses all the proteins whose levels fall within one standard deviation (SD) of the distribution (between −0.49 and 0.49, 1,947 proteins; Figure 6A), one bin that encompasses proteins whose relative abundance was low in disome V cells (log2 ratio < −0.49; 141 proteins Figure 6A) and one bin that encompasses proteins whose relative abundance was high in the disome V strain (log2 ratio > 0.49; 264 proteins; Figure 6A). We then calculated the mean of the protein abundance changes for each strain for all three categories and compared them with each other.
The mean of proteins whose levels fall within one SD of the distribution (−0.49 and 0.49) was similar between wild type, ubp6Δ, disome V and disome V ubp6Δ cells (disome V = −0.02; disome V ubp6Δ = 0.00; n = 1947; Figure 6A). In contrast, deletion of UBP6 led to the attenuation in expression levels of proteins whose relative abundances were low (log2 ratio < −0.49) in disome V cells (disome V = −0.81; disome V ubp6Δ = −0.44; p value = 3*10−19; n = 141; Figure 6A). The effects of deleting UBP6 were most dramatic among the proteins with the highest relative expression levels in disome V cells (ratio > 0.49). Whereas the mean of this bin was 0.96 for the disome V strain, it was 0.34 for disome V ubp6Δ cells (n = 264; p value = 3*10−35; Figure 6A).
The attenuating effects of deleting UBP6 were also observed for proteins encoded by genes located on chromosome V although the effects were not as dramatic, which is most likely due to the limited number of proteins that could be analyzed. The standard deviation we used for this analysis was that of the distribution of proteins located on chromosome V, which was 0.60. The average log2 expression level of chromosome V proteins was 0.84. The mean of proteins whose levels fall within one SD of the distribution (0.24 and 1.44) was the same between disome V and disome V ubp6Δ cells (disome V = 0.84; disome V ubp6Δ = 0.84; n = 105; Figure 6C). For proteins with low relative expression levels in disome V cells (log2 ratios below 0.24) some attenuation was seen as a consequence of UBP6 deletion (disome V = −0.25; disome V ubp6Δ = 0.16; n = 16; p value = 6*10−3; Figure 6C). The attenuation seen for chromosome V proteins with the relative highest levels (ratios above 1.44) was striking. Whereas the mean of this bin was 1.93 for disome V strain, it was 0.93 for disome V ubp6Δ cells (n = 15; p value = 4*10−5; Figure 6C).
To determine whether transcriptional or posttranscriptional mechanisms were responsible for the attenuating effects of deleting UBP6, we measured RNA levels in these strains. Microarray analysis showed that deletion of UBP6 caused an up-regulation of transcription of proteins with low relative expression levels in disome V cells (Figure 6B). This finding indicates that transcriptional effects are responsible for the attenuating effects of deleting UBP6 on proteins underrepresented in disome V cells. In contrast, decreased transcription was not responsible for the attenuating effects of the UBP6 deletion on proteins with high relative expression levels in disome V cells (Figure 6B, D). These data show that inactivating UBP6 attenuates the effects of disomy V on the proteome in at least two ways: (1) Inactivation of the ubiquitin protease promotes the down-regulation of proteins with high relative expression levels in disome V cells by a posttranscriptional mechanism. We presume that increased protein degradation is this mechanism. (2) Deletion of UBP6 promotes the up-regulation of proteins with low relative expression levels in disome V cells by increasing their transcription, most likely by affecting the abundance of proteins that regulate transcription of these genes.
Are the attenuating effects of deleting UBP6 specific to disome V cells? Deletion of UBP6 had a similar effect on the proteins with high relative expression levels in disome XIII cells, even though the proteins whose levels are increased in disome XIII cells relative to wild type are different than in disome V cells (Figure 6E, G; p value= 2*10−22). Transcriptional profiling indicated that this attenuating effect occurred at the posttranscriptional level (Figure 6F, H). In contrast to disome V cells, deleting UBP6 did not increase the abundance of proteins with low relative expression levels in disome XIII cells (Figure 6E).
Our results indicate that deletion of UBP6 causes attenuation of proteins with high relative expression levels in disomic cells by posttranscriptional mechanisms, most likely by increasing protein degradation. We propose that in disome V cells this effect on the protein composition increases growth rates, because proteins that inhibit proliferation of disome V cells are among the proteins whose levels are lowered by the deletion of UBP6. This is not the case in disome XIII cells. We further suggest that the attenuation of low expressed proteins, which occurs in disome V cells but not disome XIII cells, contributes to the differential effect of the UBP6 deletion on the two disomic strains.
This study is to our knowledge the first to describe genetic alterations that allow cells to tolerate the adverse effects of aneuploidy. Our analysis of 13 evolved disomic strains identified gross chromosomal rearrangements, chromosome loss, polyploidization and point mutations associated with increased proliferation rates. Their characterization revealed a surprising diversity in genetic alterations leading to improved growth rates. We suspect that this is, to some extent, due to the experimental design. The number of evolved strains that we analyzed was small and clones with improved growth properties were isolated soon after cultures experienced a decrease in doubling time. Nevertheless, it appears that many different types of genetic alterations can lead to improved growth in aneuploid yeast strains. Conversely, most strains appeared to share a common set of gene expression changes, perhaps indicating similar phenotypic consequences.
Although our analysis is far from comprehensive, it was nevertheless striking that different types of genetic alterations predominate in different aneuploid strains. This observation raises the possibility that different disomic yeast strains evolve by different pathways. What determines this difference is not yet clear but perhaps different forms of genomic instability exist among the disomes that lead to the favoring of one form of evolution over another.
The genetic alterations we identified as causing aneuploidy tolerance fall into two classes: (1) genetic changes unique to a specific isolate or a disomic strain and (2) alterations found in descendants of several disomic strains. Of special interest are genetic alterations that affect the proliferation of multiple aneuploidies. We identified three potential cases: a duplication of 183 kb on chromosome XIII and mutations in VPS64 and UBP6. The UBP6 mutations indeed led to increased proliferation in four different disomes. It will be interesting to determine whether and how the other genetic alterations affect multiple different disomes.
We have demonstrated that inactivation of UBP6 improves proliferation of strains disomic for chromosome V, VIII, IX and XI. This effect was especially striking in −His+G418 medium, where we believe the combination of frameshifts induced by G418 and disomy places an especially high burden on the proteasome. How does inactivation of UBP6 improve the fitness of some aneuploid strains? Our analysis of UBP6 mutants indicates that Ubp6’s proteasome-antagonizing function is responsible for the increase in fitness of the aneuploid strains. Quantitative proteomic approaches further indicate that deleting UBP6 reverts the overall protein composition of disome V and XIII cells to a state that is more similar to that of wild-type cells. This appears to be mediated by direct posttranscriptional effects on high abundance proteins in disome V and XIII cells and through indirect transcriptional effect on low-abundance proteins in disome V cells.
Inactivation of UBP6 attenuates protein levels in both disome V and XIII cells, so why does this improve fitness in disome V but not disome XIII cells? Attenuation of down-regulated proteins, which we observe in disome V cells but not disome XIII cells, could be responsible for the differential effects of the UBP6 deletion. Another not mutually exclusive possibility is that the proteins that antagonize proliferation in disome V cells are more efficiently degraded in the absence of UBP6 because they are proteasome substrates. In contrast, proteins responsible for decreasing the fitness of disome XIII cells are not. The transcription factor Gcn4 illustrates this point. GO search terminology revealed that genes encoding proteins involved in amino acid metabolism were significantly enriched among the genes most highly expressed in disome V cells and down-regulated when UBP6 was deleted in these cells (49 out of 175, p value = 3*10−33). 84 of the 175 attenuated genes contain binding sites for the Gcn4 transcription factor in their promoters (http://rsat.ulb.ac.be/rsat). The GCN4 gene is located on chromosome V and the levels of the protein are increased in disome V cells. We did not obtain quantitative information on Gcn4 protein levels from disome V ubp6Δ cells, but previous work showed that Gcn4 degradation is accelerated in the absence of UBP6 (Hanna et al., 2006). Deletion of GCN4 did not improve the fitness of disome V cells (E. T. unpublished observations) but scenarios such as the one described for Gcn4 could be the reason for why deleting UBP6 affects the growth properties of some aneuploids but not others.
The identification of mutations that accelerate protein degradation as conferring aneuploidy tolerance and the observation that several disomic cells harbored mutations in components of the ubiquitin-proteasome system highlight the importance of ubiquitin-mediated protein degradation in the survival of aneuploid cells. Based on the observations that yeast strains carrying additional yeast chromosomes show synthetic interactions with mutations that affect proteasome function and exhibit an increased sensitivity to conditions that interfere with protein turnover and folding (and strains harboring non-yeast DNA do not), we previously proposed that aneuploid cells are more dependent on these pathways for survival than wild type cells (Torres et al., 2007). Excess proteins produced by the additional chromosomes place an increased burden on the cell’s protein quality control systems. The results presented here support this idea. The quantitative assessment of the cellular protein composition of disome V and XIII cells revealed that the additional chromosomes are indeed producing proteins. Although the proteins that engage the protein degradation and folding machineries will be different for each additional chromosome, the necessity to degrade and fold excess proteins, compromises the cell’s ability to fold and degrade proteins whose excess presence in the cell interferes with essential cellular processes. Well-known examples of such proteins are α– and β– tubulin (Anders et al., 2009; Katz et al., 1990) and histones (Gunjan and Verreault, 2003; Meeks-Wagner and Hartwell, 1986). We propose that in the absence of UBP6, clearance of excess proteins is increased. This improves the fitness of strains, in which the proteasome neutralizes the excess proteins that impair growth. It is important to note that the increased reliance on protein folding and degradation for survival and enhancement of these pathways to improve fitness will not apply to the condition of polyploidy. In polyploid cells, the entire genome is duplicated and protein stoichiometries are not affected.
In humans, more than 90% of all solid tumors are aneuploid. Whether and how aneuploidy promotes tumor formation remains controversial (Holland and Cleveland, 2009; Schvartzman et al., 2010). Irrespective of aneuploidy’s role in tumorigenesis, it is clear from our studies that for tumor cells to acquire high proliferative potential and to become malignant, they must overcome the anti-proliferative effects associated with aneuploidy. Obtaining a comprehensive list of genes that modulate the fitness of specific aneuploidies or the aneuploid state overall, could provide key insights into how cancer cells evolve to escape the adverse effects of aneuploidy. Interestingly, 12 of the 29 genes found mutated in the evolved yeast strains have human homologs, some of which have been found to be up-regulated in tumors.
Finally, our results raise the possibility that aneuploid cancers are under profound proteotoxic stress. This increased reliance of aneuploid tumor cells on the ubiquitin-proteasome pathway could provide the framework for the development of new cancer therapeutics with a broad application spectrum and provide the rational for the use of already approved proteasome inhibitors such as Velcade in the treatment of aneuploid tumors in general.
All strains are derivatives of W303 (A2587) and are listed in Table S6. The UBP6 deletion, UBP6 truncation alleles and the PGK1- yEGFP-CaURA3 were created using the PCR-based method described in (Longtine et al., 1998). The ubp6C110A allele was provided by D. Finley. The temperature sensitive rpn6-ts allele is described in (Ben-Aroya et al., 2008). Disomy of all strains was confirmed by CGH analysis (Torres et al., 2007) and is available at puma.princeton.edu and in the Gene Expression Omnibus under accession number GSE20464.
Batch cultures of wild-type and disomic strains were kept in exponential phase by manual dilutions twice a day into fresh selective medium (−His+G418) for 14 days at room temperature. Optical densities varied between OD600nm of ~0.1 and ~1.0. Doubling times were calculated daily.
Approximately equal amounts of cells with and without PGK1-GFP were mixed in selective medium at OD600nm = 0.2 and maintained in exponential growth phase. Relative cell populations in the cultures were measured by flow-cytometry as cells containing PGK1-GFP exhibit 3 orders of magnitude higher green fluorescence than the non-GFP cells.
DNA libraries were generated using the Illumina DNA preparation kit. A summary of the number of reads, total number of bases sequenced, and coverage are presented in Table S7. We used the assembled genome of S288C (http://downloads.yeastgenome.org) and aligned our wild-type strain (W303, A2587) sequences using the Maq software package (http://maq.sourceforge.net). We found 1,396 single nucleotide polymorphisms (SNP) in W303 compared to S288C. Using the assembled S288C genome and taking into account the SNPs found in W303, we created a reference genome. The methods of SNP identification are described in detail in the supplemental information.
Other Techniques are described in the Supplemental Materials.
Table S1. Disomic Strains During Continuous Growth Experiments
Table S2. Summary of Karyotypes of Evolved Strains
Table S3. GO Enrichment of Down and Upregulated Cluster of Genes from Chemostat Experiments (Excel file)
Table S4. Synonymous Genetic Changes in the Evolved Disomic Strains and Mutations in the Parental Disomic Strains
Table S5. Mass Spectrometry Data (Excel File)
Table S6. Strains Utilized in this Study
Table S7. Sequencing Coverage (Excel File)
We are grateful to Daniel Finley, Philip Hieter, and Juergen Dohmen for reagents and Daniel Finley, John Hanna, Frank Solomon and members of the Amon lab for suggestions and their critical reading of this manuscript. This work was supported by a grant from the National Institutes of Health grant GM56800 and a Charles King Trust postdoctoral Fellowship to ET. M.J.D. and C.M.T. were supported in part by National Institutes of Health grant P50 GM071508. A.A. is also an Investigator of the Howard Hughes Medical Institute.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.