Gain or loss of chromosomes resulting in aneuploidy can be important factors in cancer and adaptive evolution. Although chromosome gain is a frequent event in eukaryotes, there is limited information on its genetic control. Here we measured the rates of chromosome gain in wild-type yeast and sister chromatid cohesion (SCC) compromised strains. SCC tethers the newly replicated chromatids until anaphase via the cohesin complex. Chromosome gain was measured by selecting and characterizing copper-resistant colonies that emerged due to increased copies of the metallothionein gene CUP1. Although all defective SCC diploid strains exhibited increased rates of chromosome gain, there were 15-fold differences between them. Of all mutants examined, a hypomorphic mutation at the cohesin complex caused the highest rate of chromosome gain while disruption of WPL1, an important regulator of SCC and chromosome condensation, resulted in the smallest increase in chromosome gain. In addition to defects in SCC, yeast cell type contributed significantly to chromosome gain, with the greatest rates observed for homozygous mating-type diploids, followed by heterozygous mating type, and smallest in haploids. In fact, wpl1-deficient haploids did not show any difference in chromosome gain rates compared to wild-type haploids. Genomic analysis of copper-resistant colonies revealed that the “driver” chromosome for which selection was applied could be amplified to over five copies per diploid cell. In addition, an increase in the expected driver chromosome was often accompanied by a gain of a small number of other chromosomes. We suggest that while chromosome gain due to SCC malfunction can have negative effects through gene imbalance, it could also facilitate opportunities for adaptive changes. In multicellular organisms, both factors could lead to somatic diseases including cancer.
THERE is growing evidence that aneuploidy is an important factor in adaptive evolution. Although aneuploidy generally has adverse consequences (Torres et al. 2007, 2008; Williams et al. 2008; Oromendia et al. 2012), it may provide selective advantage under various stresses (Pavelka et al. 2010; Sheltzer and Amon 2011). The selective advantage of aneuploidy cells has medical implications as shown for tumors (Weaver et al. 2007; Chandhok and Pellman 2009) and drug resistance in pathogenic fungi (Selmecki et al. 2006, 2009; Semighini et al. 2011). Moreover, advantageous aneuploidy can be a common step in evolution. Recently it was suggested that in yeast, aneuploidy may be a “quick fix” to tolerate stress during adaptive evolution since an aneuploidy state was shown to be transient while more “refined” mutations take over the culture (Yona et al. 2012). Moreover, changes in aneuploidy can be an “on–off switch” for colony morphological changes (Tan et al. 2013).
For these reasons it is important to understand how different mutants and different physiological conditions affect aneuploidy. Most of the quantitative analysis in yeast of mutants that influence aneuploidy was done using chromosome loss assays. Relatively fewer studies address chromosome gain (Hartwell and Smith 1985; Spencer et al. 1990; Stirling et al. 2011, 2012). However, measuring chromosome loss does not provide a complete view of chromatid malsegregation or aneuploidy tolerance. For example, inability to repair double-strand breaks (DSBs) may cause loss of a chromosome that is not due to a defect in chromatid transmission as evident by the number of DNA repair strains that exhibit increased chromosome loss (Yuen et al. 2007) and especially proteins involved in recombination (Nakai et al. 2011; Song and Petes 2012). Unlike chromosome gain, loss of chromosomes cannot be measured in haploid cells that contain natural 1n complement of chromosomes, obscuring the ability to study ploidy-dependent effects on chromatid segregation. Ploidy may have an effect on chromosome transmission as evident by the diploid-dependent lethality of some temperature-sensitive spindle pole body mutants (Storchova et al. 2006). In addition, at least for the budding yeast Saccharomyces cerevisiae, most of the colonies that were selected for chromosome loss, based on loss of genetic markers following centromere deactivation, actually reduplicated the homologous chromosome and were not aneuploid (Reid et al. 2008). Therefore, measurements of stable aneuploidy cannot be made by these types of assays.
Sister chromatid cohesion (SCC) is a process that tethers the newly replicated chromatids until anaphase and provides fidelity of chromosome transmission (Guacci et al. 1997; Onn et al. 2008; Xiong and Gerton 2010). Defects in SCC are associated with several developmental defects (Bose and Gerton 2010) and cancer (for example, Solomon et al. 2011 and summarized in Pfau and Amon 2012). SCC is primarily accomplished by the four-subunit cohesin complex containing Smc1, Smc3, yMcd1/hRad21, and yScc3/hSA1 or hSA2. Cohesin is deposited across chromosomes by the SCC2/4 cohesin loader. Cohesin becomes cohesive during DNA replication through acetylation by Eco1 (Ivanov et al. 2002; Rolef Ben-Shahar et al. 2008; Unal et al. 2008; Zhang et al. 2008; Heidinger-Pauli et al. 2009). Activation of cohesin is linked to DNA replication via proteins like Ctf4 and Ctf8 (Lengronne et al. 2006; Skibbens 2009) that facilitate the acetylation of cohesin. Ctf4 contributes to SCC also in an Eco1-independent manner (Borges et al. 2013). Cohesin is specifically enriched around the centromeres (Glynn et al. 2004), which in yeast is due in part to the protein Mcm21 (Ortiz et al. 1999; Poddar et al. 1999; Eckert et al. 2007; Ng et al. 2009). The centromere enrichment of cohesin facilitates sister chromatid biorientation before mitosis (Ng et al. 2009; Stephens et al. 2013), assuring proper chromatid segregation and the prevention of aneuploidy. This function may be independent of SCC, occurring through intra-DNA molecule cohesion (Stephens et al. 2011). Aneuploidy due to defects in SCC can occur even if SCC is established properly. Failure to maintain SCC or failure to disrupt SCC before mitosis should lead to aneuploidy. Wpl1 (the yeast homolog of the oncoprotein hWAPL) (Oikawa et al. 2004) is considered to be an important regulator of the SCC process. Recently, Wpl1 was proposed to have a role in preventing establishment of SCC at G2 by counteracting acetylation of Smc3 (Guacci and Koshland 2012; Borges et al. 2013; Lopez-Serra et al. 2013). On the other hand Wpl1 participates in maintenance of SCC once it is properly established (Rolef Ben-Shahar et al. 2008; Rowland et al. 2009; Sutani et al. 2009) and it controls chromosome condensation (Lopez-Serra et al. 2013). The effect of deletion of WPL1 on genome stability is not fully understood although evidence suggest it leads to increased loss of heterozygosity (Yuen et al. 2007).
In addition to SCC, cohesin has a role in the proper function of the kinetochore and chromatid biorientation as mentioned above (Ng et al. 2009; Stephens et al. 2011, 2013). Cohesin is also important for gene expression and DNA repair (Sjogren and Nasmyth 2001; Kim et al. 2002a,b; Unal et al. 2004; Bauerschmidt et al. 2010; Wu et al. 2012).
As we and others have shown, cohesin facilitates DSB repair between sister chromatids and suppresses recombination between homologous chromosomes (Sjogren and Nasmyth 2001; Covo et al. 2010; Heidinger-Pauli et al. 2010). Cohesin is recruited to DSBs (Strom et al. 2004; Unal et al. 2004) and stalled replication forks (Tittel-Elmer et al. 2012). SCC is activated in response to DNA damage (Strom et al. 2007; Unal et al. 2007, 2008; Heidinger-Pauli et al. 2008, 2009). Defects in SCC-mediated recombination might lead to aneuploidy, since inefficient resolution of homologous recombination intermediates can cause whole chromosome gain (Acilan et al. 2007; Ho et al. 2010).
Here, we show that different mutations in the SCC pathway can result in highly different increases in the rate of chromosome gain. Yet, the focus of this work is the use of our chromosome gain assay to answer questions that were not addressed previously in classical chromosome loss assays. We were able to show increase in chromatid malsegregation in diploid vs. haploid strains. In addition, we showed a clear effect of DNA damage on chromosome gain. Finally, we were able to show, for the first time, that defects in cohesin can cause multiple whole chromosome gains, including chromosome amplification (fast acquisition of multiple copies of one chromosome) that allows cells to survive toxic exposure. Based on these findings, we propose that the genome plasticity of diploid cells defective in SCC may facilitate adaptive evolution of pathogenic fungi and provide a selective advantage to cancer cells.
Materials and Methods
Strains used in this study are provided in Table 1.
Gene inactivation was done by knockout of specific open reading frames using the KanMX cassette from the S. cerevisiae deletion collection. The primers that were used for WPL1 knockout were as follows:
5′ ATGTTTACTTCAGCCCTTTTT 3′ and 5′ ACGCTAGAAGGCTCATCAAA 3′ for CTF4; 5′ GTCCAATTTGAGTGTAAAATCACAGG 3′ and 5′ GGTGTTACACTGTTTAATCAAAGCTC 3′ for MCM21; 5′ ACCTGGGCCGTCTTAAATTT 3′ and 5′ AGCTTGCCTTGCCATTGTTT 3′ for CIN2;
5′ TCCGGTTGAAGAGGTTCCA 3′ and 5′ TTTCGAAGGAGAGCCTGAAT 3′ for MAD1; 5′ GGACAGTGAGGGTACATTTCAAGA 3′ and 5′ CAGCAACATCCGCAGATTTT 3′. mcd1-1 strains were created by popin/popout of pVG257 (Guacci et al. 1997) and the mutation was verified by sequencing. For the chromosome gain assay, we used as a starting point strains that were previously developed (Narayanan et al. 2006). These strains were modified by replacing the inverted Alu repeats with the LYS2 gene. Heterozygous mating-type diploids were created by transforming MATa haploid cells with a vector containing HO under a native promoter (YEpHO). Nonmating isolates were selected and diploid status was confirmed by low UV mutability of CAN1. MATa/MATa derivatives were then created by transforming diploid strains with pGAL-HOT, (HO under Gal10 promoter). Transformants were incubated in galactose media for 6 hr and mating colonies were selected after assuring that the pGAL-HOT plasmid was cured. To create CUP1/cup1Δ diploid strains, we inserted KanMax cassette using plasmid pFA6 targeted to the CUP1 locus of our diploid strain background (CS2335 for example) using primers 5′GCAGCATGACTTCTTGGTTTCTTCAGACTTGTTACCGCAGGGGCATTTGTCGTCGCTGTTACACCCCCGTACGCTGCAGGTCGACGGATCCCC3′ and 5′ATGTTCAGCGAATTAATTAACTTCCAAAATGAAGGTCATGAGTGCCAATGCCAATGTGGTAGCTG-ATCGATGAATTCGAGCTCGTTTTCGA3′. For knockout verification primers, 5′ CATTTCCCAGAGCAGCATGAC 3′ and 5′ GTTCAGCGAATTAATTAACTTCC 3′ were used.
General conditions for rate determination of chromosome gain are described below. Experiments were started by patching at least six single colonies from each genotype to YPDA-rich medium followed by incubation overnight in 30°, including mcd1-1 temperature-sensitive strains (mcd1-1 strains were grown and maintained at 23° prior to the experiments). Overnight patches were then spread on selective media (CuSO4) and diluted samples were spread on synthetic complete media. Putative chromosome gain was indicated by resistance to copper. To restrict the effect of the mcd1-1 mutation to the growth phase and not to the selection phase, plates were incubated at 23°. Plates were incubated for 2–4 days.
Chromosome copy number analysis
Copy number was estimated using array comparative genome hybridization (CGH). Genomic DNA preparation, labeling, hybridization, and data analysis procedures were as described earlier (Zhang et al. 2013).
Determination of copper-resistant rate as a measure of chromosome gain
Undiluted cultures of yeast patches were spread in synthetic complete media containing 0.9 mM CuSO4. In parallel, diluted samples were spread in synthetic complete media to determine the amount of cells in each patch. After 3–4 days the number of copper-resistant colonies was determined. For each genotype in the first few experiments, the copper-resistant colonies were replica plated to another CuSO4-containing plate. The great majority of the resistant colonies were able to grow again on CuSO4 plates after replica plating.
A genetic system to study chromosome gain
Defects in SCC are expected to affect chromosome gain as well as loss. Chromosome gain has been addressed primarily using systems based on haploid cells. We chose to determine chromosome gain in diploid cells to address the role of homologous chromosome interactions, segregation defects, and aneuploidy tolerance. In yeast, the CUP1 gene codes for copper metallothionein, which protects against excessive copper (Ecker et al. 1986) and increased copies of CUP1 result in resistance to high copper levels (Fogel and Welch 1982; Resnick et al. 1990). We used a previously developed strain (Narayanan et al. 2006), where a single copy of CUP1 has been placed near the telomeric CAN1 gene on chromosome V and the natural copy in chromosome VIII is deleted (Figure 1). Resistance to copper provided selection for gain in CUP1 copy number and potentially gain of chromosome V. After modifying the strain (See Materials and Methods) the resulting cells were sensitive to 0.4 mM CuSO4 and allowed robust selection for chromosome V gain when yeast cells were plated to 0.9 mM CuSO4 (or higher).
The vast majority of resistant colonies in our experiments were due to gain of at least one copy of chromosome V. Among 33 copper-resistant colonies examined (haploid and diploid strains and various genetic backgrounds), all exhibited whole chromosome V gain as determined by CGH. There were no chromosome V gains in the absence of copper selection (supporting information, Table S1) among mcd1-1 diploid cells. No other aberrations on chromosome V, such as local increased copy number of the locus surrounding CUP1 were observed, although other chromosomes were also gained (discussed below). Importantly, we did not find any false positives when 0.9 mM CuSO4 was used for selection. Since we cannot exclude some killing of cells with chromosome V gain, the rates of chromosome gain can be considered as minimal estimates. The median chromosome gain rates for all genetic backgrounds including 95% confidence of intervals as derived from the rate of copper resistance are presented in Table S2.
SSC defects and DNA damage facilitate chromosome gain
There was an increase in the rate of chromosome gain for SCC defective cells in comparison to wild type (WT). The increase differed considerably among the mutants. Interestingly, homozygote deletion of WPL1, which has several roles in regulation of sister chromatid cohesion, had the least impact on chromosome gain (17-fold over WT, Figure 2). Homozygous deletion of CTF4 that links SCC establishment to DNA replication and MCM21 that facilitates SCC around the centromeres increased chromosome gain 180- and 100-fold over WT, respectively. The greatest effect was found for a cohesin temperature-sensitive mutant mcd1-1 grown at the semipermissive temperature of 30° (265-fold increase in chromosome gain over WT) (Figure 2A).
Chromosome behavior can be affected by differences in physiology between various types of cells or different stresses. Particularly relevant is exposure of cells to DNA damage, which can activate dormant cohesin molecules (Strom et al. 2004, 2007; Unal et al. 2004, 2007, 2008). Since cohesin mutants show defects in homologous recombination (Covo et al. 2010; Sjogren and Strom 2010) and since defects in resolution of recombination intermediate can lead to chromosome gain (Ho et al. 2010; Rodrigue et al. 2012) the effects of DNA damage and the role of homologous recombination on chromosome gain in WT and SCC defective strains were studied. We examined chromosome gain following growth of diploid MATa/MATα cells on plates containing a low level of the recombinogen methyl methanesulfonate (MMS; 1 mM). As shown in Figure 2B and Table S2, the high levels of spontaneous chromosome gain in the MATa/MATα wpl1Δ/wpl1Δ and MATa/MATα mcd1-1/mcd1-1 mutants were greatly increased by MMS, based on a comparison of rates between treated and untreated cells. Importantly, the MMS exposure and SCC defects resulted in synergistic increases in rates of chromosome gain. These results are consistent with the view that inefficient sister chromatid recombination in SCC-defective strains might lead to chromosome gain. Such recombination-associated aneuploidy is expected to depend on the function of RAD51, a gene that is central to homologous recombination. However, the rate of spontaneous chromosome gain in MATa/MATα wpl1Δ/wpl1Δ rad51Δ/rad51Δ diploid cells was significantly higher than in the wpl1Δ/wpl1Δ single mutant (Figure 2B, Table S2). Homozygous deletion of RAD51 alone also increased chromosome gain, although to a lesser extent than in the rad51Δ/rad51Δ wpl1Δ/wpl1Δ double mutant (Figure 2, A and B and Table S2). Just as for MMS, the rad51Δ homozygous deletion synergized with deletion of WPL1. Attempts to measure chromosome gain in the rad51Δ/rad51Δ wpl1Δ/wpl1Δ double mutant failed due to severe DNA damage sensitivity. These results indicate that recombination intermediates such as joint molecules are not the major source of aneuploidy in SCC defective strains. However, it is possible that DSBs, nicks, gaps, or unresolved replication structures (Sofueva et al. 2011) might lead to chromosome gain, especially in cells defective in SCC.
Mating-type controls of chromosome gain
We also examined the effect of cell type on chromosome gain. In yeast, changes in mating type greatly affect the transcriptional program (Galitski et al. 1999). Also, mating type influences several aspects of chromosome biology. Nonhomologous end joining is active only in haploid or diploid cells carrying one MAT allele, MATa or MATα, while homologous recombination is more active in heterozygous MATa/MATα cells (Kegel et al. 2001; Valencia-Burton et al. 2006; Fung et al. 2009). Therefore, we measured chromosome gain in a MATa/MATa strain. Surprisingly, for WT and SCC defective strains, the rate of spontaneous chromosome gain was increased at least 10-fold in diploids and up to 100-fold in MATa/MATa compared to MATa/MATα strains, as described in Figure 2C and Table S2. The high levels in the MATa/MATa strains did not depend on the end-joining gene DNL4 (end joining is suppressed in MATa/MATα cells) (Kegel et al. 2001) or on homologous recombination as determined using rad52Δ/rad52Δ mutants (Figure 2D and Table S2).The rates in rad52Δ/rad52Δ mutants were slightly, but statistically significantly, higher than for RAD+ MATa/MATa cells, in agreement with the effect of the rad51 deletion (Figure 2, B and D and Table S2).
The ability to form colonies on the CuSO4-containing medium was clearly dependent on the copy number of CUP1 since the rate of formation of copper-resistant colonies was greatly reduced, from 130 × 10−7 to 0.5 × 10−7, when there was only one CUP1 gene (i.e., MATa/MATa CUP1/CUP1 vs. MATa/MATa CUP1/cup1Δ; Figure 2D). These results indicate that the system is highly responsive to an additional copy of the CUP1 gene, leading us to conclude that CUP1 copy number gain is the driver of copper resistance in the homozygous MAT diploid strains.
Resistance to high chronic exposure of CuSO4 is much more frequent in cohesin-deficient cells than in WT
Increased copy number of CUP1 is expected to provide protection against higher levels of CuSO4. We, therefore, investigated the ability of several yeast strains to form colonies when grown on 1.5 mM of CuSO4.
The rate of colony formation of WT MATa/MATα cells on 1.5 mM CuSO4 was extremely low, <10−9 events per cell division, as described in Figure 2E and Table S2. Similar to the results with lower CuSO4 levels, the rate of colony formation was much higher when cells were homozygous for mating type. Remarkably, there was more than a 10,000-fold rate increase in MATa/MATα mcd1-1/mcd1-1 cells that were resistant to 1.5 mM CuSO4 compared to WT MATa/MATα MCD1/MCD1. The rate reached nearly 10−5/cell/generation. A similar rate was observed with the MATa/MATa mcd1-1/mcd1-1 strain, possibly indicating that there is a limit to the number of additional chromosomes that a cell can tolerate (Figure 2E).
We also addressed the ability of cells to tolerate even higher levels of CuSO4. Unlike for 1.5 mM CuSO4, no colony-forming units were found among 109 MATa/MATa mcd1-1/mcd1-1 diploid cells spread on 2 mM CuSO4 plates. However, highly resistant cells appeared several days after inoculation of 1.5 mM CuSO4-resistant colonies to liquid medium containing 2 mM CuSO4 (growth conditions are described in the legend to Figure 3).
Multiple aneuploidy and chromosome amplification in diploid cohesin mutants chronically exposed to a high level of CuSO4
Since increased copper resistance as well as the stepwise adaptation to high CuSO4 exposure was likely due to a change in chromosome number, we analyzed by array CGH all the chromosomes from diploid mcd1-1/mcd1-1 copper-resistant cells grown in 0.9 and 2 mM CuSO4, as described in Table S1. Typically, array-CGH analyses would compare the chromosome content of a resistant culture that was grown in CuSO4 liquid media against a copper-sensitive culture that grew in CuSO4-free medium at permissive temperature (the reference culture). The array CGH was also performed by comparing a reference strain to mcd1-1 cells grown at the semipermissive temperature (without CuSO4) and then propagated in liquid cultures without CuSO4. In the absence of CuSO4, there was almost no aneuploidy (Table S1), indicating that partial inactivation of cohesin does not lead to frequent unselected stable aneuploidy events.
In all cultures of copper-resistant colonies, there was gain of chromosome V, as described in Table S1. Interestingly, there was aneuploidy for other chromosomes as well. (Occasionally, the array-CGH signal indicates a copy number change <1 n, for example, 0.75, which is likely due to heterogeneity in the population). Among 19 cultures of diploid MATa/MATa mcd1-1/mcd1-1 cells, there were 14 isolates showing a gain of chromosome II in addition to chromosome V (Figure 3A). This coincidence of gain was also observed for 11 of 14 other copper-resistant haploid or diploid isolates of different genotypes (SCC proficient or defective). Altogether, chromosome II was gained along with chromosome V in 25 of 33 (75%) of all the resistant cultures examined. The gain of chromosome II in response to copper is statistically significant. The P-value of Fisher’s exact test corrected by Bonferroni considering each of 16 chromosomes as an independent hypothesis is 0.0016.
Frequently, chromosomal gain for chromosomes other than V and II was also detected, as described in Table S1 and Figure 3. Based on results for all genotypes and conditions, chromosomes VII and XI were often increased (13/33 and 14/33, respectively), yet due to the relatively small size of the sample the P-value of Fisher’s exact test is not significant when corrected to multiple hypotheses.
Chromosome loss events in the copper-resistant cells were not as frequent as chromosome gain events (22 vs. 83, excluding gains of chromosome V). However, if the frequently gained chromosomes II, V, VII, and XI are excluded, there is much less bias (22 vs. 31), suggesting little difference in the probability of gain or loss for infrequent random events (Figure 3B). There appears to be no correlation between the number of genes on each chromosome and occurrence of either chromosome loss or gain (Figure 3C). Therefore, we could not observe an effect of the untargeted gene dosage imbalance. There might be an exception for chromosome I, which is the smallest and lost more frequently, and chromosome III, which is gained more often.
Finally, the copy number of chromosome V (the target of selection for copper resistance) was determined. Gain of more than one copy was frequently observed among the various WT, mcd1-1, and wpl1Δ isolates (20/33) resistant to 0.9 and 1.5 mM CuSO4 (Figure 3D, Table S1). Interestingly, in the diploid MATa/MATa mcd1-1/mcd1-1 cells those were adapted to grow in 2 mM CuSO4 liquid culture there was a gain of even more copies of chromosome V. Among 11 highly resistant isolates, there was only 1 that acquired just a single extra chromosome. In the remaining 10 isolates, half of them gained 2 extra chromosomes and half gained at least 3 extra chromosomes, as described in Figure 3C. The karyotype of such cells resistant to 2 mM CuSO4 dramatically deviates from that of the normal diploid cells (an example is presented in Figure S1). The multiple changes in karyotype are reminiscent of the variations in chromosome numbers that can be seen in cancer cells. The severe imbalanced genome was observed also in wpl1-deficient diploids exposed to CuSO4, as can be seen in Table S1 isolates Csra84–87.
Diploid cells are more prone to failure in controlling chromosome gain caused by genetic defects
Using the copper-resistance system, we found that the rate of chromosome gain in haploids (MATa) is lower than in diploid cells (MATa/MATα). While this trend is true for all strains, the effect is modest in WT strains. In SCC mutants, the differences are more striking. For example, the rate for mcd1-1 in haploids vs. diploids is 170 as compared to 2660 events/107 cell divisions (Figure 4A, Table S2). Hence, in comparison to the haploid cells, there was a further 4-, 5-, 11-, and 17-fold increase in the rates of gain, respectively, in the diploids as compared to the haploid for the mcd1-1, mcm21Δ, ctf4Δ, and wpl1Δ cells. The overall diploid effect for chromosome gain was nearly two orders of magnitude greater for MATa/MATa vs. MATa strains (compare rates in Figure 4 and Figure 2C and also Table S2).
As shown in Figure 4A, this diploid effect extends to other components of chromosome transmission. Loss of the microtubule protein gene CIN2 and the spindle assembly checkpoint protein gene MAD1 in haploid cells resulted in 15- and 16-fold increases, respectively, in the rates of chromosome gain. The rates (events/107 cell divisions) in diploids were increased another 44-fold (658/15) and 89-fold (1433/16). Thus, reduction in the fidelity of chromosome transmission results in chromosome gain that is greatly increased in diploid cells.
The apparent diploid-dependent chromosome gain may stem from differential tolerance of chromosome gain, especially since there would be a lower gene dosage effect for chromosome gain in diploids as compared to haploids. To test this idea, an indicator of chromosome gain tolerance (tolerance index) was estimated by determining the frequency of copper-resistant cells within a culture. The tolerance index for haploid and diploid wpl1Δ and WT (only data for wpl1Δ is presented) strains was evaluated under three scenarios. First it was determined directly in 3–6 copper-resistant colonies freshly harvested from CuSO4-containing media. The chromosome gain tolerance index was close to 1 (i.e., most of the cells in the colony were resistant) for haploid- and diploid-resistant colonies (Figure S2A). Second, the tolerance index was determined for cultures propagated from several CuSO4-resistant colonies in the absence of CuSO4. As seen in Figure S2A, no major difference was observed between haploid and diploid cells. Finally, copper-resistant colonies were diluted and spread to media lacking CuSO4. This was followed by suspending several (8–16) colonies that arose and spotting them to CuSO4 and CuSO4-free media (Figure S2B). No major difference between haploid and diploid CuSO4 tolerance was observed (similar amounts of cells grew on media with and without CuSO4). We conclude that chromosome gain is well tolerated both in diploids and haploids for at least 25 generations (the number of generations from a single cell to a colony).
Interestingly, the rates of chromosome gain in wpl1Δ and WT haploids were comparable (Figure 4B). The lack of impact by the wpl1Δ mutation is surprising since wpl1Δ cells have higher rates of chromosome gain and loss in diploid cells than WT cells as described above (Figure 3 and S. Covo, D. A. Gordenin and M. A. Resnick, unpublished results)
Cohesin cohesion and chromosome gain
In this study, we have examined the role of various genes involved with SCC, ploidy, and mating type in preventing aneuploidy due to chromosome gain. Specifically, we focused on the broader effect of defects in SCC on the karyotype of cells under stress. The chromosome gain rates obtained here for the different mutants in SCC are not statistically different from the rates we calculated for chromosome loss with the same mutants (S. Covo, unpublished results). Defects in the establishment of sister chromatid cohesion lead to chromosome gain as shown in ctf4∆ cells (Figure 2 and Figure 4). Defects in cohesion per se result in premature chromatid separation before the bipolar attachment is established, which may lead to a random segregation of the two chromatids, increasing the chance that both chromatids will migrate to the same daughter cell. Thus, one daughter cell gains a chromosome and the other loses one (Figure 1). mcm21∆ strains exhibit proficient SCC across the chromosome but show very similar rates of chromosome gain, suggesting that the cohesin activity around the centromeres is at least as important in preventing chromosome gain as chromosome-wide SCC. This is in agreement with the important role of cohesin and MCM21 in imposing steric constraints on kinetochore orientation to ensure biorientation. In mcm21-deficient cells, the chromatids do not migrate randomly, rather the kinetochore has high probability to be attached to the wrong pole. We suggest that defects in cohesin itself cause much higher rates of chromosome gain because both the centromeric and the SCC functions are compromised (Figure 2A and Figure 4), and, therefore, the probability of both premature separation and aberrant attachment of the kinetochore is raised. In addition, as shown in Figure 2, deficiency of RAD51 or RAD52 increases the risk of chromosome gain and, therefore, the defects of mcd1-1 in homologous recombination may also contribute to the high rate of chromosome gain. Yet, the contribution of homologous recombination function in a SCC mutant is hard to tease apart, until a mutant in cohesin that is only defective in DNA repair is isolated, because as seen in Figure 2, defects in recombination and defects in SCC synergize. Surprisingly, deletion of WPL1 affects chromosome gain to a much lesser extent in diploid cells and has no effect in haploid cells (Figure 2 and Figure 4).
Defects in sister chromatid cohesion are measured at the cellular level by the separation of florescent markers attached to the chromosomes (i.e., sister chromatid separation). ctf4Δ cells exhibit increased sister chromatid separation (for example, see Borges et al. 2013), which is translated to a high rate of chromosome gain (Figure 2 and Figure 4). Deletion of WPL1 has been shown to increase somewhat sister chromatid separation as well as decrease cohesin attachment to chromatin (Rowland et al. 2009; Sutani et al. 2009; Maradeo and Skibbens 2010).When compared directly, sister chromatid separation was reduced in wpl1Δ relative to ctf4Δ (Borges et al. 2013); based on that fact alone, the effect on chromosome transmission should be less profound in wpl1Δ than ctf4Δ . Yet, WPL1 is supposed to contribute to the fidelity of chromatid transmission also by preventing SCC establishment at G2. The fact that wpl1Δ/wpl1Δ cells show only modest increase in chromosome gain rates suggests that deactivation of the antiestablishment function of Wpl1p has relatively low or no impact on the fidelity of chromosome transmission.
DNA damage synergistically interacts with SCC defects to increase chromosome gain but not through unresolved recombination intermediates
Recently it was shown that unresolved recombination intermediates can cause chromosome gain, specifically due to inability to resolve joint molecules that were generated by Rad51p activity (Acilan et al. 2007; Ho et al. 2010). Defects in SCC may increase the prevalence of such intermediates because of the role of SCC in recombination. We hypothesized that the increased recombination between homologous chromosomes in SCC mutants may result in delayed recombination intermediates. Indeed, we found that the recombinogenic agent MMS greatly increased chromosome gain in SCC mutants (Figure 2B). However, Rad51p-dependent recombination intermediates are not required for chromosome gain as demonstrated for wpl1Δ/wpl1Δ rad51Δ/rad51Δ diploids and MATa/MATa rad52Δ/rad52Δ mutants (Figure 2). While the role of SCC in homologous recombination is not clear yet, we suggest based on these observations that SCC is not an important determinant in the efficiency of resolution of recombination intermediates. The nature of the MMS-derived lesions that exacerbate chromosome gain in SCC defective strains remains to be determined. One possibility is that DNA damage-induced separase, which removes established cohesin molecules (McAleenan et al. 2013), may cause premature chromatid separation especially in SCC defective strains. Alternatively, while unrepaired chromosomes activate the checkpoint response, this activation is transient and is eventually shut down. The segregation of unrepaired chromosomes (gaps, breaks) may be less accurate.
Mating type and ploidy along with defects in chromosome transmission increase the risk of chromosome gain
Interestingly, the rate of chromosome gain is greatly affected by mating type (Figure 2 C). The reason for the difference remains to be established, although many genes are under mating-type control in yeast (Galitski et al. 1999). Several genes associated with mitosis, such as components of the aurora kinase and the Ndc80 complex are underexpressed in MATa/MATa cells (Galitski et al. 1999). Whatever the reasons, the mating-type effects on chromosome stability have important implications. For example, in scenarios where diploid pathogenic yeast become homozygous for mating type, there might be opportunities for greater resistance to drug challenges. It was previously shown that aneuploidy in pathogenic Candida albicans can lead to drug resistance (Selmecki et al. 2006).
SCC malfunction increases genome plasticity and allows adaption to stress
To our knowledge, this study provides the first description of how defects in SCC can be beneficial via genome instability to the toxic effects of an environmental chemical. Compared to WT, the SCC showed increased survival rates to CuSO4 due to genomic changes. Previously, copper resistance was attributed to various modes of copy number increase in the CUP1 gene, including CUP1 amplification, but in these cases the amplification of CUP1 was derived by cis-DNA elements such as inverted repeats (Fogel and Welch 1982; Narayanan et al. 2006). In contrast, here the only mode of CUP1 copy increase was through whole chromosome gain without any adjacent DNA sequence that destabilizes the CUP1 locus (Table S1). This is despite the collateral effects of unbalanced expression of many genes (Torres et al. 2007, 2008). Importantly, chromosomes could undergo considerable amplification, increasing by up to four chromosomes in a diploid cell (i.e., a total of six copies).
In agreement with the reduction in chromosome transmission fidelity, other chromosomes were gained as well, the most frequent being chromosome II (75% of the chromosome V gain events). We consider several possible overlapping hypotheses that provide likely explanations for the frequent gain of chromosome II. In the first, chromosome II is gained because it harbors genes that facilitate tolerance of copper (such as BSD2, SCO1, and SCO2). In the second, chromosome II is gained to counterbalance a specific imbalance created by gain of chromosome V either at the proteomic level or in the geometry of the spindle pole body. We obtained preliminary results using a system similar to that presented here but in which selection for gain of chromosome copy number is based on selection in only 0.15 mM CuSO4 in combination with formaldehyde (J. L. Argueso, unpublished results). In this system co-gain of chromosome II and V is far less frequent, indicating that copper exposure drives co-gain of chromosome II and V. An alternative mechanism to how copper may shape the pattern of aneuploidy is by reducing the fidelity of specific chromosomes (in this case, chromosome II).
Ploidy effect on chromosome gain
Finally, we found that diploid cells defective in chromosome transmission are able to form copper-resistant colonies at a much higher rate than their haploid counterparts. However, the natural diploid yeast (MATa/MATα) shows only a modest diploid-dependent increase in chromosome gain. The diploid-dependent chromosome gain appears to be a general phenomenon that can be revealed under stress to chromosome transmission since it is observed in mutants defective in SCC, spindle body checkpoint (mad1Δ), or tubulin filament formation (cin2Δ) (Figure 4A). There are several possibilities to explain the diploid-dependent chromosome gain. Diploids might tolerate aneuploidy better than haploids because of milder gene dosage effects, although we could not find support for this notion. There was no diploid-dependent tolerance of aneuploidy based on our observation that both haploid and diploid cultures contained comparable fractions of copper-resistant cells even when the cultures were grown on media lacking CuSO4 (Figure S2). Alternatively, there are diploid-dependent effects relating to the fidelity of chromosome transmission itself rather than aneuploidy tolerance. As a support to this notion, cells deficient in kinetochore functions, such as dam1-10 and spc110-1, show temperature sensitivity that is diploid dependent (Storchova et al. 2006). Scaling up the ploidy may cause defects in chromosome transmission as shown for tetraploid vs. diploids (Mayer and Aguilera 1990; Storchova et al. 2006). Another parallel between haploid vs. diploids and a further scale up to tetraploids is the combined effect of ploidy change and mutation in SCC (Storchova et al. 2006). Most ascomycota fungi are found as haploids in nature although they can go through a diploid cycle. S. cerevisiae, which is also an ascomycete, is actually stable as a diploid (Gerstein et al. 2006; Nishant et al. 2010). Our results suggest that heterozygous mating type counteracts the effect of ploidy increase, probably through changes in gene expression.
Overall, we observed that even mild defects in SCC increase significantly the chance of beneficial gain of multiple chromosomes under selective pressure. The stability of aberrant karyotypes, which might be a “quick and temporary fix” (Yona et al. 2012) needs to be determined. Regardless, given the impact of SCC on chromosome gain, we propose that hypomorphic or even temporary defects in the SCC pathway may have a significant role in adaptive evolution and disease, such as cancer.
We thank Kerry Bloom for discussion of the results and useful advice. We greatly appreciate the critical evaluation of the manuscript by Jessica Williams and Thuy-Ai Nguyen. This work was supported by the Intramural Research Program of the National Institute of Environmental Sciences (National Institutes of Health, Department of Health and Human Services) under project 1Z01ES065073 (to M.A.R.), American Cancer Society grant ACS IRG no. 57-001-53, and a Webb-Waring Biomedical Research Award from the Boettcher Foundation (to J.L.A.).
Communicating editor: J. Sekelsky
- Received September 9, 2013.
- Accepted November 11, 2013.
- Copyright © 2014 by the Genetics Society of America