Human Bcl-2 Reverses Survival Defects in Yeast Lacking Superoxide Dismutase and Delays Death of Wild-Type Yeast

Strains of S. cerevisiae used in these studies are the following: EG103 (wild type) (DBY746; MATα leu2-3,112 his3Δ1 trp1-289 ura3-52 GAL⁺), EG118 (sod1Δ) (EG103 with sod1ΔA::URA3), and EG133 (sod1Δsod2Δ) (EG103 with sod1ΔA::URA3 sod2Δ::TRP1) (9, 23). In these strains the mutations in SOD1 and SOD2 are deletions of most of the coding region. In all the experiments reported in this paper, these strains were transformed with either plasmid pAD4, a multicopy yeast expression vector carrying the LEU2 selectable marker, or pAD4-bcl-2, the same vector carrying the human bcl-2 gene under the transcriptional control of the strong alcohol dehydrogenase (ADH1) promoter (19). Expression from this promoter is high during log phase growth and decreases at stationary phase. Transformations were carried out by the lithium acetate method (17). It should be pointed out that, as observed in our previous study (24), sod1 mutants transformed with plasmids lost viability somewhat less rapidly than untransformed mutants. This is a behavior that we have not yet explained, but which may be due to unavoidable selection for robustness during the transformation, or to an effect of the LEU2 marker, which was present in all the plasmids. To test the latter possibility, the experiments described herein were repeated with strains transformed with the plasmids pBMT116 (vector control) and pBMT116-bcl-2 (carrying human bcl-2 cDNA), which are similar to the pAD4 plasmids, except that they contain the TRP1 selectable marker. Results with these plasmids (data not shown) were similar to those reported in this paper using LEU2 plasmids.

The expression of Bcl-2 was verified by Western analysis. A 25-μg sample of total cellular protein was run on a 15% SDS–polyacrylamide gel. Purified human recombinant Bcl-2 was used as a control and Bcl-2 was detected by anti–Bcl-2 antibody (cat. # 1624 989; Boehringer Mannheim Biochemicals, Indianapolis, IN) using the Western-Light Plus Chemiluminescent Detection System (TROPIX, Inc., Bedford, MA). Bcl-2 antibody (primary antibody) was diluted 1:80 and anti–rabbit alkaline phosphatase– conjugated secondary antibody 1:10,000. In all cases where the bcl-2 gene was present (and in no cases where the gene was not present), protein that reacted with the anti–Bcl-2 antibody was detected.

Unless stated otherwise, all experiments in liquid media were performed in synthetic dextrose complete media (SDC) with 2% glucose, and supplemented with amino acids, adenine and uracil as described (18) as well as a fourfold excess of the supplements Trp, Leu, His, and uracil. Strains were streaked from frozen stocks onto YPD (1% yeast extract, 2% peptone, 2% dextrose, 2% agar) plates. After 3 d of incubation, cells from a single colony on the YPD plates were transferred to 3 ml of SDC overnight medium lacking the appropriate supplement (Leu, ura, and/or Trp) to select for plasmid-containing cells (LEU⁺) with the correct SOD deletion (URA⁺ for sod1Δ or TRP⁺ for sod2Δ). For normal oxygen (high aeration) conditions, overnight cultures grown in selective media were inoculated at an OD₆₀₀ of 0.01 into flasks with a flask volume/medium volume ratio of 5: 1 and grown at 30°C, shaking at 220 rpm. Growth was followed by monitoring the turbidity at 600 nm (OD₆₀₀). An OD₆₀₀ of 1 is equivalent to 10⁷ cells per ml. For long-term stationary phase cultures, after 72 h, the cells were washed twice, resuspended in sterile distilled water, and incubation with shaking was continued. Cell viability was measured by plating serial dilutions of the yeast cultures onto YPD plates. Colonies formed were counted after 2 d of incubation at 30°C. Plates from cell lines lacking CuZnSOD were incubated in a low oxygen atmosphere generated using CampyPaks (BBL Scientific, Cockeysville, MD). All experiments were repeated at least three times with duplicate samples.

For experiments in 100% oxygen, cells were grown to stationary phase and on day 3 viability was measured as described above, except that plates were incubated for 48 h in glass dessicators filled with 100% oxygen, followed by incubation in air for the remainder of the time. The dessicator containing the plates was filled by evacuating and refilling it with oxygen two times. Alternatively, oxygen was gently blown in to the bottom of the dessicator for 15 min, allowing gas to escape through a small hole in the top. In either case, the container was then sealed and incubated at 30°C. Oxygen was replenished after 24 h. Identical control platings were always made and incubated in low oxygen (CampyPaks).

Cells were grown as described above. At the desired time, 5 × 10⁸ cells were pelleted at 3,000 g and the supernatant was carefully removed. The tubes were held on ice. 100 μl of 3.5% sulfosalicyclic acid and 100 μl of glass beads (0.5 μm, acid washed and dried) were added. Cells were vortexed for 1 min and then placed on ice for 1 min. This vortexing/cooling process was repeated six times. The samples were then centrifuged at 15,000 g for 3 min. The acid soluble fraction was used to determine glutathione concentration. Total glutathione was measured according to the method of Tietze (36). Oxidized glutathione (GSSG) was measured using the method of Griffith (11) in which 2-vinylpyridine was used to derivatize the reduced form of glutathione. Standard curves used in this assay contained all components used to derivatize the reduced form of the glutathione.

Crude cellular protein extracts were prepared by glass bead lysis as follows. Cells were suspended in lysis buffer at 4°C (50 mM Tris 7.2, 150 mM NaCl, 5 mM EDTA, and 0.2 mM PMSF) with an equal volume of acid washed 0.5 mm glass beads, and vortexed for six to eight cycles of 30 s of vortexing followed by 30 s of cooling. The mixture was then microfuged for 2 min to remove the cellular debris and glass beads. The supernatant was frozen until assayed. The catalase activity was determined spectrophotometrically by monitoring the disappearance of hydrogen peroxide at 240 nm (25). SOD activity was determined by monitoring inhibition of the autoxidation of 6-hydroxydopamine (14, 15). Protein concentration was determined using the Bio Rad assay (Bio Rad Laboratories, Hercules, CA), with bovine serum albumin as protein standard.

Copper and manganese were determined by atomic absorption spectrophotometry, using a Varian SpectrAA (Varian Instr. Business, San Fernando, CA) equipped with graphite furnace. Whole cell samples were digested in 3% HNO₃ at 100°C overnight, and then diluted to 0.1% HNO₃ for analysis. Alternatively soluble protein extracts were made to 0.1% HNO₃ and incubated overnight before analysis.

Mutation frequency was estimated by determining the number of cells able to form colonies on SDC plates lacking arginine and containing 60 μg/ml canavanine. (Resistance to canavanine, an arginine analogue, arises when mutations in the arginine permease occur.) Approximately 100 million cells were removed from cultures after 72 h of incubation and plated on canavanine plates. Colonies were counted after 4 d. Mutation frequencies are reported as mutants per 10⁶ viable cells. Three separate experiments with duplicate platings of duplicate samples were performed.

To determine whether human Bcl-2 could overcome cell death associated with oxidative stress, we expressed Bcl-2 in strains lacking either or both SODs. In the paradigm used here (24), cells were grown to saturation in SDC medium. At 72 h, the cells were washed, resuspended in water, and the incubation continued. Under this treatment, strains lacking either SOD are known to lose viability rapidly (24), beginning at entrance into stationary phase, while wild-type cells survive up to 2 mo.

In the present study, sod1Δ and sod1Δsod2Δ mutants transformed with either the multicopy plasmid pAD4–bcl-2 or the vector control pAD4, were tested for survival. Early in the culture, when growth was logarithmic (early log phase, <2 × 10⁷ cells/ml) Bcl-2 had little effect on the viability (Fig. 1). At late log phase (∼6 × 10⁷ cells/ml, generally reached after 10–15 h of growth), differences began to be apparent. By the time stationary phase was reached at 24 h (10–13 × 10⁷ cells/ml, no further growth), differences were quite obvious. At 24 h, sod1Δ and sod1Δsod2Δ strains expressing Bcl-2 maintained their viability, while the viability of strains carrying plasmid controls was markedly reduced (Fig. 1). Over a culture period of 26 d, viability for all strains continued to decrease and the strain expressing Bcl-2 continued to show a much higher survival rate than the control. Table I (first two columns) summarizes the survival of the sod1Δ and sod1Δsod2Δ strains with or without Bcl-2 expression at 26 d. Because of the large difference in viability already evident at day 1 (Fig. 1) we also show the data normalized to the viability on day 1 (Table I, last two columns). For the sod1Δ strain, the difference was statistically significant (P < 0.01 by the two-sided t test) in both cases (raw data as well as normalized). The difference for the sod1Δsod2Δ strain was not statistically significant, although the trend was always the same as for the single sod1Δ mutant. Nevertheless, the major effect of Bcl-2 expression on survival in these experiments clearly occurred at the time of entry into stationary phase, rather than in the later stages.

Somewhat surprisingly, wild-type yeast with or without Bcl-2 expression grew well in 100% oxygen, exhibiting 80– 100% viability relative to growth in air (data not shown). On the other hand, yeast lacking CuZnSOD and/or MnSOD were quite sensitive to 100% oxygen and were unable to form colonies under these conditions. To test the effect of Bcl-2 expression on the growth of sod1Δ, sod2Δ, and sod1Δsod2Δ mutants under this intense oxidative stress, we tested the ability of these strains to survive a 48 h exposure to 100% oxygen. As shown in Fig. 2, expression of Bcl-2 in SOD-deficient cells substantially improved their ability to form colonies after incubation in 100% oxygen. The effect was particularly pronounced for the sod1Δ single mutant, but was also evident for the sod2Δ single mutant and the double mutant.

Culotta and co-workers (21–23) have recently identified several mutations that partially “rescue” the sod1Δ sod2Δ strain, and they showed that each of these mutations altered copper or manganese metabolism in some way. One of these, the pmr1 mutation, exerted its effects by increasing cellular manganese accumulation. (High intracellular manganese was previously shown to rescue sod1Δ mutants, probably by its own dismutase activity [2].) A side effect of the pmr1 mutation is an increased sensitivity to abnormally high manganese levels in the medium (21). We reasoned that, if Bcl-2 exerted its effects by a mechanism similar to that of pmr1, we should see increased sensitivity to manganese and increased manganese accumulation in strains expressing Bcl-2. In fact, we saw the opposite, expression of Bcl-2 in strains lacking SOD led to decreased sensitivity to elevated manganese levels as measured by cell survival (Fig. 3,A) and decreased manganese accumulation as measured by atomic absorption (Fig. 3 B). In other words, Bcl-2 expression caused the sod1Δ and sod1Δsod2Δ strains to exhibit a more wild-type–like phenotype.

Culotta and co-workers also observed that a mutation in the yeast BSD2 gene bypassed the growth requirement for SOD by altering copper homeostasis in a manner seemingly analogous to the pmr1 mutation. In this case, cells became hypersensitive to elevated copper concentrations (22). We tested sod1Δ and sod1Δsod2Δ strains expressing Bcl-2 for copper sensitivity and saw a slight, although not statistically significant, decrease in sensitivity in the Bcl-2 expressors (data not shown). This decrease came instead of the marked increase we would have expected if Bcl-2 were working through a mechanism like that of bsd2. We conclude that, unlike other suppressors of the sod1Δ phenotype, Bcl-2 does not work via effects on copper or manganese accumulation.

To determine if Bcl-2 had a direct effect on the status of other antioxidant proteins in yeast, we measured catalase activity in both wild-type and sod1Δ mutant cells expressing Bcl-2 (Fig. 4). Yeast has two catalases (Catalase A and Catalase T), but only the cytoplasmic Catalase T is expressed under our conditions. Catalase A is peroxisomal and only expressed under special conditions (8). We observed an increase in catalase activity for wild-type cells expressing Bcl-2 relative to vector-transformed controls. The sod1Δ mutant had a lower level of catalase than wild type. Bcl-2 expression caused an increase in catalase activity, but the activity was still lower than that of the wild-type strain (Fig. 4).

We also monitored levels of reduced and oxidized glutathione (GSH and GSSG, respectively) in wild-type cells with or without expression of Bcl-2. Total glutathione (GSH + GSSG) levels were unchanged by Bcl-2 expression, but the GSH:GSSG ratio increased from 85:1 for the vector-transformed control to 105:1 for the Bcl-2 expressor, indicating a slightly more reducing condition in the cells expressing Bcl-2. This result was significant to the P < 0.1 level.

Steinman reported that expression of Bcl-2 in Escherichia coli led to increased catalase expression and a higher mutation rate, and he therefore proposed a pro-oxidant mechanism for Bcl-2 in this system (33). To determine whether a similar phenomenon occurred in yeast expressing Bcl-2, we measured mutation frequencies, using forward mutation to canavanine resistance. As expected based on earlier work (9), mutation frequencies were clearly higher in sod1Δ and sod1Δsod2Δ mutants than in the wild-type parental strain. However, in all three cell lines, Bcl-2 expression decreased mutation frequencies (Table II). This result is consistent with an antioxidant activity for Bcl-2, since oxidative damage is believed to be the cause of much spontaneous mutation and is almost certain to be the cause of the excess spontaneous mutation found in strains lacking CuZnSOD.

We recently observed that yeast cells left in expired synthetic glucose medium (SDC), but not in rich medium (YPD) or water, lost viability quickly (Longo, V.D., P. Morcos, T.E. Johnson, E.B. Gralla, and J.S. Valentine, manuscript submitted for publication). To determine whether Bcl-2 could affect this type of cell death, we studied survival in water and in expired SDC medium of wild-type yeast expressing Bcl-2. Wild-type yeast left in complete glucose medium (SDC) rapidly lost viability beginning at day 5. The viability loss was significantly delayed by expression of Bcl-2 such that there was a five- to sixfold difference in the number of survivors by day 10 (Fig. 5). To be sure that the viability loss was not caused by the nutritional auxotrophies of the EG103 wild-type strain, survival in spent medium was also measured for the prototrophic strains D27310b, which also lost viability prematurely (data not shown). By contrast, no effect of Bcl-2 was seen on survival of wild-type yeast under conditions optimal for stationary phase survival, i.e., switched to water at day 3, probably because survival was already very high (Fig. 5).

The yeast sod1Δ and sod1Δ sod2Δ strains provide a simple model system for studies of the properties of Bcl-2. The presence of Bcl-2 prevented short-term death of both yeast sod1Δ and sod1Δ sod2Δ mutants at the point of entry into stationary phase (Fig. 1). In addition, it improved, although less dramatically, their long-term survival (Table II). Furthermore, Bcl-2 enabled sod1Δ, but not sod1Δsod2Δ, mutants to grow in 100% oxygen (Fig. 2) and increased indicators of antioxidant defense, i.e., catalase (Fig. 4) and reduced glutathione. These results confirm an antioxidant property for Bcl-2 and suggest that it functions in this regard by acting on basic intracellular mechanisms present in all eukaryotes (in human cells as well as in yeast).

Mechanisms of oxidative stress and antioxidant defenses in vivo are intimately connected with metal metabolism. Most pathways that have been proposed for generation of significant fluxes of reactive oxygen species in vivo involve participation of redox active metal ion catalysts (usually copper or iron ions). Defensive mechanisms also often involve metal ions, e.g., superoxide dismutase and catalase are metalloenzymes. Studies on sod1Δ yeast have demonstrated the importance of metal ions in oxidative stress resistance. First, high levels of extracellular manganese ion rescued sod1Δ strains, apparently because excess manganese accumulated in the cytoplasm and was able to function as a weak dismutase (2). In addition, sod1Δ strains have been shown to be partially rescued by excess copper in the medium. This rescue depended upon intracellular binding of the metal by the yeast copper metallothionein protein (34), rather than on a “free” metal ion as is postulated for manganese, but is otherwise similar.

Thus far, all mutations identified that reverse growth defects of sod1Δ and sod1Δ sod2Δ yeast appear to cause accumulation of manganese or copper in the cytosol. If Bcl-2 were involved in the accumulation of manganese or copper, as are the products of the PMR1 and BSD2 genes, respectively (21, 22), we would have expected a higher sensitivity to elevated concentrations of these metal ions in cells expressing the oncoprotein. We found instead that Bcl-2 improves the growth of sod1Δ and sod1Δ sod2Δ mutants in the presence of both metals but especially manganese (Fig. 3), indicating that it works in another fashion. We have not yet determined whether this improvement is due to a general decrease of the oxidative stress in the cells or to a direct involvement of Bcl-2 in manganese homeostasis or subcellular distribution.

Mammalian neural cells expressing Bcl-2 have recently been shown to contain increased concentrations of glutathione (5, 19). The expression of Bcl-2 in E. coli was shown to induce catalase expression without exposure to exogenous hydrogen peroxide. It also led to increased mutation rates, leading the author to conclude that this oncoprotein functions as a prooxidant (33). In yeast, we also observed an increase in catalase activity in strains expressing Bcl-2 (Fig. 4), but the spontaneous mutation frequency was reduced by Bcl-2 expression (Table II). In addition, wild-type yeast expressing Bcl-2 had slightly higher GSH/ GSSG ratios than wild-type controls, although no difference in total glutathione was observed. Taken together, these data argue against a prooxidant effect of this oncoprotein in yeast and support a role for Bcl-2 in promoting antioxidant defenses.

We observed that wild-type yeast kept in expired SDC medium (not switched to water) during stationary phase lost viability rapidly. While the death of sod1Δ and sod1Δ- sod2Δ strains is known to be oxidative (24) it is not clear to what to attribute the death of wild-type cells in expired medium. The death of wild-type cells in expired medium was delayed by the expression of human Bcl-2 (Fig. 5). The precipitous viability loss we observed for yeast in expired minimal medium was quite surprising, especially in view of the fact that yeast transferred to water show no such dramatic viability loss and instead survive for months. A possible explanation is based on the assumption that yeast are likely to encounter a nutrient-depleted environment quite often in the wild and therefore might be expected to recognize and respond to such a condition. It is thus possible that death could be an evolved response to such a situation. One's first notion is that in a unicellular organism such as yeast, each individual cell competes against all others. However, this approach may not be the most effective for maintenance of the species. Yeast grow in colonies that are generally isogenic, so survival of the genes of a single colony is guaranteed even if only a few organisms survive. The cell death we observed in expired medium could be explained by a specific cell suicide program, evolved to kill most members of a colony so that a few better-adapted individuals could survive and grow on the limited nutrients. This putative death program would increase the chance of survival of the species. The better adaptation of the lucky individual(s) that survived could be due either to mutations (leading to natural selection) or simply to random variation in the micro-environment. (Note that such a suicide response would not be expected to occur in pure water. Since there are no nutrients to be shared, the best response is for each cell to last as long as possible.)

Ras2 is one of two yeast Ras proteins that play roles in growth control and have structural and functional homology to the ras family of mammalian proto-oncogenes. In yeast, Ras is coupled to a signal cascade that involves intracellular cAMP and is sensitive to nutritional status (7, 35). In another study, we found that the viability loss of yeast in expired minimal medium was prevented by a null mutation in ras2, suggesting that Ras2 is involved in a cell death pathway (Longo, V.D., P. Morcos, T.E. Johnson, E.B. Gralla, and J.S. Valentine, manuscript submitted for publication). Interestingly, extracellular cAMP and starvation are involved in initiating the cell death program of stalk cells in the developmental cycle of the primitive eukaryote Dictyostelium (4). In human cells, Bcl-2 has been shown to co-precipitate with both R-Ras and p21-Ras (3, 6) both of which are highly homologous to yeast Ras2 (7). These observations raise the possibility that Bcl-2 may prevent cell death in yeast by acting through the Ras pathway.

Evidence is slowly accumulating that some lower eukaryotes, even unicellular ones, exhibit programmed cell death, which, however, may or may not share the morphological features that define apoptotic cell death (1). For example, a cell death program was demonstrated in E. coli, consisting of a plasmid-encoded DNA restriction/modification system that caused DNA cuts but not other typical apoptotic morphologies (28). Cell death during stalk formation in Dictyostelium followed a distinct program that included chromatin condensation, but not DNA fragmentation (4). Other, nonapoptotic forms of PCD may occur in higher eukaryotes as well. For example, none of the features characteristic of apoptosis were observed in the PCD of intersegmental muscle of the moth Manduca sexta at the end of metamorphosis (31).

The results presented here are consistent with the presence of a programmed cell death pathway in yeast, but they are equally consistent with an antioxidant role for Bcl-2 which could be acting on pathways or proteins that promote compensatory antioxidant defenses in sod null mutants as well as wild-type cells under certain circumstances. Our searches of the yeast whole genome database did not reveal the presence of homologues of pro-apoptotic proteins such as ICE or CED-3, suggesting that a programmed cell death in yeast, if it does indeed exist, may not involve the proteins and mechanisms that cause apoptosis in higher eukaryotes. No Bcl-2 homologues with high identity were found either, but we cannot exclude the possibility that functional rather than sequence homologues of Bcl-2 may be present in yeast, or that bcl-2 is a relatively new feature of apoptosis, added later in evolution.

There are only a few reports in the literature examining members of the bcl-2 gene family expressed in yeast (10, 13, 30). These papers are based on the finding that expression of Bax, a cell death–inducing Bcl-2 family member, is toxic to yeast. Using the two-hybrid system, Reed and co-workers (13, 30) showed that this Bax-induced cell death was inhibited by co-expression of Bcl-2 and some Bcl-2 homologues which are known to interact physically with Bax. However, since expression of foreign proteins fairly frequently causes cell death, it is not clear in these studies whether the death observed is a specific triggered cell death pathway or simply the result of a non-specific toxicity. If the latter is the case, then rescue by Bcl-2 could be explained simply as a case of Bcl-2/Bax dimerization masking the toxicity. Our results demonstrate an antideath activity for Bcl-2 in yeast in the absence of Bax, increasing the possibility that a real cell death program is being observed.

Recent work (10) using a similar system, demonstrated that Bax-related growth inhibition could be experimentally separated from cellular death in S. cerevisiae and that the presence of functional mitochondria was necessary for cell death to occur. These results make it appear more likely that Bax is acting on an endogenous system. In our work, both the oxidative stress-related death of sod1Δ and sod1Δ sod2Δ mutants in stationary phase and the death of wild-type cells in expired minimal medium occurred without introduction of any foreign genes, i.e., they are naturally occurring processes. That Bcl-2 is able to prevent both kinds of death is therefore strong evidence that it is acting on, or in concert with, an endogenous pathway that already exists in yeast.

In conclusion, our results demonstrate that Bcl-2 can provide antioxidant protection in yeast and that it can delay natural death in two separate paradigms. The fact that Bcl-2 exerts an antioxidant-type protective effect in yeast is evidence supporting the centrality of this role. However, it is important to stress that the mechanism by which the antioxidant protection occurs and the connection between the antideath and antioxidant actions of Bcl-2 in our system, as well as in mammalian cells, remains unknown.

This work was supported by National Institutes of Health grants DK46828 (to J.S. Valentine), AG12282, (to D.E. Bredesen), NS25554, (to D.E. Bredesen), and a pilot research grant from the University of California at Los Angeles Center on Aging based on a generous gift to the center by Harold and Libby Ziff (to J.S. Valentine and E.B. Gralla).