Global Transcriptional Analysis of Yeast Cell Death Induced by Mutation of Sister Chromatid Cohesin

Cohesin is a protein complex that regulates sister chromatid cohesin during cell division. Malfunction in chromatid cohesin results in chromosome missegregation and aneuploidy. Here, we report that mutations of MCD1 and PDS5, two major components of cohesin in budding yeast, cause apoptotic cell death, which is characterized by externalization of phosphatidylserine at cytoplasmic membrane, chromatin condensation and fragmentation, and ROS production. Microarray analysis suggests that the cell death caused by mutation of MCD1 or PDS5 is due to the internal stress response, contrasting to the environmental or external stress response induced by external stimuli, such as hydrogen peroxide. A common feature shared by the internal stress response and external stress response is the response to stimulus, including response to DNA damage, mitochondria functions, and oxidative stress, which play an important role in yeast apoptotic cell death.


Introduction
It has been increasingly evident that yeast as well as other unicellular organisms can undergo apoptotic cell death (for recent review, see [1]). The apoptotic cell death shares many morphological and biochemical features with mammalian apoptosis, including DNA fragmentation and chromatin condensation, exposure of phosphatidylserine on the outer membrane surface, leakage of Cytochrome C from mitochondria [2]. Apoptosis in yeast can be induced by external stresses, such as hydrogen peroxide [3], acetic acid [4], as well as mutations of certain essential genes, such as CDC48 [5], ORC2 [6], and PDS5 [7]. The yeast apoptotic cell death is promoted by proapoptotic genes, including the metacaspase YCA1/MCA1 [8,9] and apoptosis-inducing factor AIF1 [10].
Cohesin is a protein complex that regulates sister chromatid cohesion and is required for ordered chromosome segregation [11]. Cohesin physically links the sister chromatids during replication. This attachment makes it possible for the kinetochore of each sister chromatid to attach to microtubules from different poles. During anaphase, cohesin is released from chromosomes, and proper segregation occurs [12]. Cohesin contains at least four protein subunits: Smc1, Smc3, Mcd1/Scc1, and Scc3 [13,14]. In addition to the four subunits, several additional proteins are required to establish or maintain the chromatid cohesion but are not part of the cohesin complex,such as Pds5 [14][15][16][17], Scc2, and Scc4 [18]. Scc2 and Scc4, for example, are required for loading the cohesin complex onto chromosomes, but do not colocalize with the complex [18]. Pds5 is essential for both mitosis [15,16,19] and meiosis [20]. Pds5 localizes to chromosomes in a cell cycle-dependent manner and is required for sister chromatid cohesion, chromosome condensation, DNA repair, and proper chromosome segregation.
In addition to its function in chromatid cohesion, studies show that cohesin has a role in apoptosis [21,22]. Human RAD21/MCD1 is reported as a nuclear caspase target. Induction of apoptosis by diverse stimuli causes the cleavage of hRad21. The cleaved C-terminal product of hRAD21 is translocated from the nucleus to cytoplasm and acts as a nuclear signal for apoptosis. RAD21/MCD1 is also overexpressed in prostate [23] and breast cancer cells [24]. Suppression of RAD21/MCD1 reduces the ability for cell proliferation and increases the level of apoptosis [24]. In yeast (S. cerevisiae), our recent study showed that mutation of PDS5 causes cell death in early meiosis [7].
Here we report that mutation of MCD1 (mcd1-1) and PDS5 (pds5-1) showed an apoptotic phenotype, including exposure of phosphatidylserine to the outer membrane, accumulation of DNA breaks, condensation and fragmentation of chromatin, and production of ROS. Microarray analysis of the global transcript of mcd1-1 and pds5-1 mutants, in comparison with H 2 O 2 induced apoptosis, showed that the majority of the differentially expressed genes were upregulated. These genes were involved in a variety of functions, including response to stimulus, cell cycle regulation, mitochondrial functions, and cell wall organization. Our data suggest that the cell death caused by mutation of MCD1 or PDS5 is due to response to physiological or internal stress, compared to the H 2 O 2treated cells, which is considered response to environmental or external stress.

Mutation of MCD1 and PDS5 Showed Major Markers of Apoptosis
Our previous report [7] showed that mutation of PDS5 caused apoptotic cell death in yeast during early meiosis.
Here, we demonstrated that mutation of PDS5 and MCD1, two cohesin related genes, caused apoptotic cell death during mitosis. Figure 1 shows some of the "hallmarks" of apoptosis-the translocation of phosphatidylserine from the inner leaflet to the outer leaflet of the plasma membrane ( Figure 1(a)), and nuclear fragmentation (Figure 1(b)). ROS accumulation was also observed in mcd1-1 and pds5-1 mutants, when grown at nonpermissive temperature (Figures 1(c) and 1(d)). Note that the ROS level was higher in mcd1-1 and pds5-1 mutants even at 21 • C (Figure 1(d)), compared to the wild type.Chromatin condensation was also observed by transmission electron microscopy ( Figure 2).

General Characteristics of Gene Expression
In attempt to understand the mechanism of the cohesininduced apoptosis in yeast,a global analysis of gene expression was performed in mcd1-1 and pds5-1 mutants undergoing apoptotic cell death. H 2 O 2 -treated cells were used as a positive control.The number of up-and downregulated genes were shown in Figure 3 (the original data is available at http://www.uwyo.edu/microscopy/cohesin.xls).The microarray data were further validated by reverse transcriptasepolymerase chain reaction (RT-PCR) ( Figure 4).5 different genes from each group were chosen for RT-PCR analysis. These genes represent genes that are up-and downregulated, and genes related to cohesin complex, which is of interest for further analysis. At both time points (see materials and methods for determination of the time points), more genes were upregulated than downregulated, suggesting a genetic program was being activated, rather than the general gene expression shutdown that might be expected in a rapid cell death. Only a small number of differentially expressed genes overlap among the three groups in both time points ( Figure 5). The total number of overlapped genes were 174 in time point 1 (167 upregulated, and 7 downregulated) and 349 in time point 2 (323 upregulated and 26 downregulated). This suggests that these three groups may undergo apoptotic cell death through different pathways. To better understand the pathways, the gene expression profiles were analyzed using Netaffx analysis software from Affymetrix (http://www.affymetrix.com/). Using Affymetrix's gene ontology (GO) mining tool, it was found that in H 2 O 2 -treated cells, the most prominent group of genes (the group with the lowest ρ-value) that were differentially expressed was "response to stimulus" (Table 2). In mcd1-1 and pds5-1 mutants, however, the most prominent   group is related either to physiological process (mcd1-1), or regulation of biological process (pds5-1). This may reflect the fact that the H 2 O 2 -treated cells mainly responded to the external stress, termed "environmental stress response" (ESR) [25], while in mcd1-1 and pds5-1 mutants, the gene expression responded mainly to biological or physiological changes due to the malfunction of MCD1 or PDS5.

Genes Involved in Response to Stimulus
When the 174 overlapped genes from the three groups in time point 1 were analyzed using the GO mining tool, it was found that the most prominent group of differentially expressed genes was related to "response to stimulus" ( Table 2), suggesting that, like the H 2 O 2 -treated cells, the mcd1-1 and pds5-1 mutants, when grown at nonpermissive temperature, were also under stress. An evidence of the stress was the production of ROS ( Figure 1). The changes of gene expression in mcd1-1 and pds5-1 mutants were referred to "internal stress response" (ISR), contrasting to the environmental, or external stress response (ESR    of the top three prominent groups that are differentially regulated (1st = regulation of biological process (Table 2), 2nd = development, 3rd = response to stimulus). Among the 58 differentially regulated genes response to stimulus, 31 (53%) were related to DNA damage ( Table 4), suggesting that more DNA damages were caused by the extended time of internal or external stress. It has been reported that both Pds5 and Mcd1 are involved in DNA repair [26,27]. The elevated number of differentially expressed genes response to DNA damage could be related in part, to the malfunction of mcd1-1 or pds5-1 in DNA damage repair.
Previous reports indicate that both Pds5 [27,28] and Mcd1 [26] have a role in DNA repair. To explore if the defect in DNA repair is involved in mcd1-1-or pds5-1-induced apoptosis, wild type, pds5-1-and mcd1-1 cells were exposed to UV radiation (150 J/m 2 ) using a UV cross-linker (Stratagene, Inc.), and then incubated for 20 hours at 37 • C before being assayed with DAPI. DAPI staining showed that in pds5-1, broken nuclei increased at least 18%, while in mcd1-1,   broken nuclei increased about 15%, compared to cells without UV radiation. At the same condition, UV radiation had almost no effect on wild type ( Figure 6(a)). ROS levels were much higher in mcd1-1 and pds5-1 mutants after UV radiation ( Figure 6(b)).These results suggest that (1) Mcd1 and Pds5 participate in DNA break repair, and (2) defect of mcd1-1 or pds5-1 in DNA break repair plays a role in apoptosis.

Genes Involved Mitochondria Functions
Mitochondria play an important role in yeast apoptosis [29,30]. In our current study, microarray analysis revealed that many differentially expressed genes were localized in mitochondria. In time point 1, among the 174 differentially expressed genes in all three groups, 38 were found localized in mitochondria ( Table 5). 26 of the 38 genes were localized only in mitochondria. Interestingly, 12 of the 38 have unknown functions. The genes that have known functions were involved in response to stimulus, constituents of mitochondrial membrane, regulation of ATP syntheses, and other enzymes. Of particular interest is the upregulation of CYC7, which encodes the isoform 2 of Cytochrome C. Cytochrome C has been reported to be released from mitochondria to cytoplasm after induction of apoptosis [31]. The increase of expression level of CYC7 may also contribute to the release and the abundance of Cytochrome C in cytoplasm after induction of apoptosis. The upregulation of CYC7 only appeared in time point 1, not time point 2, suggesting that the release of Cytochrome C is an early event during apoptosis. At time point 2, 40 (12 localized only mitochondria; 8 with unknown functions) of the commonly up-or downregulated genes were found localized in mitochondria. Although the number of genes was similar to that of time point 1, the majority of genes were different and only 7 appeared in both time points (bold letters in Tables 5 and 6). Of the 7 genes, 3 (ELG1, NTG1, and NTH2) were response to stimulus/DNA repair; 3 (GCV1, GLT1, and AEP1) were related to amino acid/protein metabolism; and one (YDR332W) with unknown functions. In general, the mitochondrial genes were mainly involved in response to stimulus/DNA damage repair, metabolism of DNA, proteins, and amino acids. BIR1, the gene encodes an antiapoptotic protein, was found overexpressed in all three cases (Table 6). When checked at time point 1, BIR1 was also overexpressed in all three cases, but the fold change in mcd1-1 was less than 2, and therefore was filtered out. Because BIR1 is also involved in coordinating cell cycle events for proper chromosome segregation, it is unclear that the overexpression of BIR1 is due to its role of cell cycle control, or apoptosis.

Genes Involved in Cell Cycle Regulation
Deregulation of cell cycle has been reported in yeast cells undergoing apoptotic cell death [32,33]. The apoptotic aging cells, for example, were described as to attempt the second budding cycle without the completion of the first one. Irregular expression of cell cycle related genes was also reported in old cells and orc2-1 cells that underwent apoptosis [33]. In our current study, we found that 62 genes that were differentially expressed in all three groups at time point 2 were related to cell cycle regulation (Table 7), while only 12 in time point 1 were related to cell cycle, suggesting that this deregulation occurs at the late stage of apoptosis. Cell cycle interruption was also observed at time point 1.
In mcd-1-1 mutant, for example, SEM images showed the daughter cell was unable to bud off the mother cell even when the size of the daughter cell was similar to the size of the mother cell (Figure 7(a)). The 62 all upregulated genes  (Table 7) were involved in both mitotic and meiotic cell cycles. 6 genes were involved in mitotic spindle checkpoint (MAD2, MAD3, DMA2, SGO1, IBD2, and MPS1); 3 in DNA damage checkpoint (PIN4, MEC3, LCD1) and two with unknown functions. Of the two genes with unknown functions, YOR019W interacts with ribosomes, and SPR6 was involved in sporulation. Stress, especially external (environmental, nutritional) stress, is the major cause of meiosis or sporulation [34]. The fact that 23 out of the 62 genes (Table 7) are involved in meiotic cell cycle suggests that yeast cells under the external (H 2 O 2 ), or internal (mc1-1 and pds5-1) stress may attempt to initiate meiotic division. REC8, for example, is a meiotic specific gene that is only expressed during early meiosis. Its expression, however, was detected under all three conditions (Table 7). On the mitotic side, many of the differentially expressed genes are involved in checkpoints of G1/S or G2/M transitions, such as SWI4, a DNA binding component of the SBF complex (Swi4p-Swi6p) that regulates late G1-specific transcription of targets, including cyclins, and genes required for DNA synthesis and repair. Another example is SFI1, a Centrin (Cdc31p)binding protein, required for progression through G2/M transition.

Genes Involved in Cell Wall Organization
Scanning electron microscopy was used to investigate the cell morphology under the external and internal stresses. Consistent with the microarray data, the cell surface structure of the pds5-1-and the H 2 O 2 -treated cells was similar, in which cell wall sunk, forming a hole on the cell surface (Figure 7(a)). The mcd1-1 cells appeared to be more damaged than pds5-1-or H 2 O 2 -treated cells. More cells collapsed, and cell surface was not as smooth as the other two, suggesting a potential cell composition change. The mcd1-1 cells were generally larger than the wild type or the other two. Further examination of the gene expression related to cell wall revealed that only 7 genes in time point 1 and 5 genes in time point 2 were differentially expressed in all three cases ( Table 8), suggesting that the cytoskeleton, rather than the cell wall composition or organization, may contribute more to the morphological changes observed by SEM. To test the possibility of the involvement of cytoskeleton in cell-morphology change, rhodamine-conjugated phalloidin was used to stain actin. As shown in Figure 7(b), for the wild type and mutants grown at permissive temperature, actin appeared to be in both polymerized (filaments) and nonpolymerized (monomers) forms. The filamentous actin was usually oriented parallel to the longitudinal axis of the cell, as well as along the cell wall. In H 2 O 2 -treated cells, and mutants grown at nonpermissive temperature, actin was present mostly as monomers, indicating the damage of the cytoskeleton.

Apoptotic Cell Death at Nonpermissive Temperature
Temperature shift in several temperature-sensitive mutants, includes CDC48 [5], ORC2 [6], and PDS5 [7], showed apoptotic cell death. We speculate that all cells undergo apoptotic cell death at the nonpermissive temperature. To test this speculation, wild type cells were incubated overnight at 42 • C, a nonpermissive temperature for wild type cells.  Figure 6: UV radiation caused increase of apoptotic cell death in mcd1-1 and pds5-1 mutants. (a) wild type, pds5-1 cells, and mcd1-1 cells were exposed to UV radiation (150 J/m 2 ) then incubated for 20 hours at 37 • C. Cells were stained with DAPI to identify the fragmented nuclei. Data are from 3 independent experiments; at least 100 cells were counted for each experiment. (b) ROS was produced in pds5-1 and mcd1-1 mutants. Wild type and the mcd1-1 and pds5-1mutants were exposed to UV radiation (150 J/m 2 ) then incubated for 2 and 18 hours at 37 • C. Dihydrorhodamine 123 (Priault et al. pages 684-91) was added into cell culture and incubated for another 2 hours, ROS production was measured using a microplate reader.
Apoptotic cell death was revealed in majority of wild-type cells. More than 90% of wild-type cells showed positive Annexin V staining, nuclei fragmentation, and chromatin condensation. About 50% of cells showed positive TUNEL staining, while none showed positive TUNEL staining at permissive temperature (Figure 8). ROS level was much higher than cells grown at 30 • C. Wild type showed no significant difference from the mcd1-1 or pds5-1 mutants when both incubated at 42 • C (data not shown). This result suggests that high temperature stress causes apoptosis in yeast and can be used as an apoptosis inducer for studying the mechanism of apoptotic cell death in yeast.

Discussion
Sister chromatid cohesin plays a crucial role in accurate chromosomal segregation in the eukaryotic cell cycle. Recent studies indicate that cohesin proteins are also involved in apoptosis. Two studies showed that the Rad21 of mammalian cells was cleaved during apoptosis induced by diverse stimuli. The C terminal of Rad21 cleavage product was translocated to cytoplasm and caused amplification of the cell-death signal [21,22]. In human breast cancer cells, expression level of RAD21 is higher than normal cells. Inhibition of RAD21 lows the survival rate of breast cancer cells and induces apoptotic cell death [24]. In S. cerevisiae, mutation of PDS5 causes apoptotic-like cell death in early meiosis [7]. When treated with H 2 O 2 , which causes apoptosis in yeast [3], the Mcd1 level was decreased (Yang & Zhang,   Temperature shift in several temperature-sensitive mutants includes CDC48 [5], ORC2 [6], and PDS5 [7], and also shows apoptotic cell death. Our current study suggests that the cell death caused by nonpermissive temperature is due to the internal stress response (ISR), compared to external stress response (ESR). A common feature shared by the ISR and ESR is response to stimulus, including response to DNA damage, oxidative stress, and heat. Our data also suggest that ESR may become ISR that ultimately causes the cell death.
Since all the genes in the temperature sensitive mutants are essential genes, it would be interesting to see if cell death caused by the malfunction of all essential genes is apoptotic and due to ISR. Another commonality of the yeast cell death is the apoptotic nature of cell death at nonpermissive temperature, because all yeast strains, including the wild type die at their nonpermissive temperature, that is, above 37 • C. Our result shows that even the wild-type cells undergo apoptotic cell death at their nonpermissive temperature ( Figure 8).
Several studies have shown that DNA damage stimulates apoptosis in budding yeast. Mutation in CDC13, for example, triggers apoptotic cell death due to the accumulation of DNA damage [35]. UV radiation also induces some features of apoptosis [36]. These findings suggest that yeast may share some of the apoptotic pathways induced by DNA damage in mammals, especially those p53-independent pathways. Sister chromatid cohesin also appears to be involved in DNA repair. In fact, the cohesin subunit Rad21 was first identified in Schizosaccharomyces pombe by its sensitivity to UV or ionizing radiation, suggesting a role in DNA repair [37]. A recent study showed that downregulation of RAD21 impaired DNA double strand break repair and cells were more sensitive to etoposide and bleomycin, which induce double strand breaks [24]. In fission yeast, mutants of PDS5 lost the proliferative ability after arrest in G 2 and were hypersensitive to DNA damage [27]. Our current study further confirms that DNA damage plays an important role in yeast cell death due to the fact that (1) a large amount of genes related to DNA damage repair are differentially expressed during apoptosis by either H 2 O 2 treatment or by mutation of the cohesin protein Mcd1 or Pds5; and (2) UV radiation caused significant increase of cell death in mcd1-1 and pds5-1 mutants. We speculate that this defect in DNA repair may trigger the signaling pathway that ultimately leads to the cell death.

Transmission Electron Microscopy
TEM sample preparation was conducted as described previous [34]. Briefly, cells were fixed with 2.5% (v/v) glutaraldehyde in phosphate buffered saline (PBS) (pH = 7.2) for 40 minutes at room temperature. Cells were further fixed with 2% potassium permanganate in water for 1 hour at room temperature, then stained with 2% uranyl acetate for 90 minutes. Fixed cells were dehydrated with 30%, 50%, 75%, 85%, 95%, and 100% ethanol, then embedded in Spurr resin (Electron Microscopy Sciences, Pa). 60 nm ultrathin sections were cut with a diamond knife, stained with lead citrate for 30 minutes, and examined using a Hitachi H-7000 electron microscope, equipped with a high-resolution (4 K × 4 K)cooled CCD digital camera (Gatan, Inc., Pa, USA).

Microarray Analysis
For pds5-1 cells and mcd1-1 mutants, cells were inoculated into fresh YPDA medium and cultured at 21 • C overnight to a density of 1.0 × 10 7 cells/ml, then shifted to 37 • C for 2 or 12 hours and harvested for RNA isolation. The 2-and 12-hour time points were chosen because these time points represented cells with an apoptosis rate of 1-5% and 50-60%, respectively. These two time points were referred as time point 1 (2 hours) and time point 2 (12 hours). For control, wild-type cells were treated as the same as above and RNA was isolated at the two same time points. H 2 O 2 -treated wild-type cells were used as a positive control. Wild-type cells were inoculated into fresh medium and incubated at 37 • C until the cultures reached a density of 5 × 10 6 cells/mL. 5 mM H 2 O 2 was added to the culture and the cells were continually cultured at 37 • C for 4 or 10 hours before harvested for total RNA isolation. The 4-and 10-hour treatment of H 2 O 2 represents an apoptosis rate of 3-5% and 50-60%, respectively, and as referred to time point 1 and time point 2, corresponding to mcd1-1 and pds5-1. The control sample was treated as same as above, except with the omission of H 2 O 2 .
Total RNA was extracted using the enzymatic lysis protocol of Qiagen RNeasy minikits (Qiagen; Valencia, Calif) with the following modification. Buffer Y1 was replaced with sorbitol buffer (0.8 M sorbitol, 2% potassium acetate, pH 7.0). 0.5 mg/mL (final concentration) of Zymolyase-100 T (MP Biomedicals, Aurora, Ohio) was used for cellwall removal and the enzyme digestion time was 5 minutes. RNA labeling and hybridization procedures were performed according to protocols from the Microarray Core at the University of Colorado Health Science Center.
At each time point, total RNA was extracted from 5-10 × 10 7 cells using 2 Qiagen columns. The quality of the purified total RNA was evaluated by Agilent bioanalyzer and NanoDrop Spectrophotometer. 2.5 μg of high-quality total RNA was used as template for cDNA synthesis by reverse transcription. Affymetrix one-cycle cDNA Synthesis Kit, cDNA Sample Cleanup Module, and IVT Labeling Kit were used to produce biotinylated cRNA target. The Agilent bioanalyzer and NanoDrop Spectrophotometer were used to determine the cRNA yield, purity, and size distribution. Approximately, 20 μg of cRNA were used for fragmentation with the buffer from Affymetrix Hybridization Control Kit. The fragmented biotin-labeled cRNA was quantitated, and hybridized to the Affymetrix GeneChip Yeast Genome 2.0 Array, which includes approximately 5 744 probe sets for 5 841 of the 5 845 genes present in S. cerevisiae.
Microarray data was first analyzed using Affymetrix's GeneChip Operating Software. The expression level of the cohesin mutants and H 2 O 2 -treated cells at different time points was compared with that of the corresponding control cells, using the software's default parameters (Alpha1 = 0.05, Alpha2 = 0.065, Tau = 0.015, Gamma1H = 0.002, Gamma1L = 0.002, Gamma2H = 0.002667, Gamma2L = 0.002667, Perturbation = 1.1). A call to each gene indicating the change in transcript level between the baseline (control) and the experiment array was assigned with one of the Genes with a value of NC, MI, or MD were all regarded as no change of transcription level, and only the genes with a value I or D were selected. Fold change of the differentially expressed genes was calculated based on the signal-log ratio. Only the genes with a fold change equal to or greater than 2 were selected to ensure the gene expression was truly differential. Microarray data were derived from two independent experiments. Values were average of the twoindependent experiments

Rhodamine-Phalloidin Staining of Actin
Cells at time point 1 were fixed in PBS buffer with 4% formaldehyde for 1 hour. After washing with PBS buffer three times, fixed cells were resuspended in 1 mL PBS containing 1% Triton X-100 and incubated at room temperature for 3 minutes for cell permeabilization. Permeabilized cells were washed with PBS buffer, resuspended in 20 μl rhodaminephalloidin (Invitrogen) (final concentration = 160 nM) for 1 hour. Cells were mounted on a glass slide and viewed with a rhodamine filter.

Conclusion
Our data suggests that the cell death caused by mutation of MCD1 or PDS5 is due to the internal stress response,contrasting to the environmental or external stress response induced by external stimuli, such as hydrogen peroxide. A common feature shared by the internal stress response and external stress response is the response to stimulus, including response to DNA damage, mitochondria functions, and oxidative stress, which play an important role in yeast apoptotic cell death.