Department of Molecular Biology and Biotechnology, University of Sheffield, Sheffield S10 2TN, UK
Dr Peter Piper, Department of Molecular Biology and Biotechnology, University of Sheffield, Sheffield S10 2TN, UK. Tel.: +44 (0)114 2222851; fax: +44 (0)114 2222800; e-mail: email@example.com
Dr Peter Piper, Department of Molecular Biology and Biotechnology, University of Sheffield, Sheffield S10 2TN, UK. Tel.: +44 (0)114 2222851; fax: +44 (0)114 2222800; e-mail: firstname.lastname@example.org
Yeast overexpressing SOD1, the gene for Cu,Zn-superoxide dismutase (Cu,Zn-Sod), was used to determine how Sod1p overexpression influences the chronological lifespan [the survival of non-dividing stationary (G0) phase cells over time], the replicative lifespan (the number of buds produced by actively dividing yeast cells) and stress resistance. Increasing the level of active Cu,Zn-Sod in yeast was found to require either growth in the presence of high copper, or the simultaneous overexpression of both SOD1 and CCS1 (the latter being the gene that encodes the chaperone dedicated to Cu2+-loading of Sod1p in vivo). Dual SOD1 + CCS1 overexpression elevated the levels of Cu,Zn-Sod activity six- to eight-fold in vegetative cultures. It also increased the optimized survival of stationary cells up to two-fold, showing this chronological lifespan is ultimately limited by oxidative stress. In contrast, several detrimental effects resulted when the SOD1 gene was overexpressed in the absence of either high copper or a simultaneous overexpression of CCS1. Both the chronological and the replicative lifespans were shortened; the cells displayed an abnormally high level of endogenous oxidative stress, resulting in a high rate of spontaneous mutation. Such harmful effects were all reversed through the overexpression of CCS1. It is apparent therefore that they relate to the incomplete Cu2+-loading of the overexpressed Sod1p, most probably accumulation of a Cu2+-deficient Sod1p to appreciable levels in vivo. The same events may generate the detrimental effects that are frequently, though not universally, observed when Cu,Zn-Sod overexpression is attempted in metazoans.
A number of studies indicate that aging might be slowed by an increased scavenging of reactive oxygen species (ROS). There are a number of reports of increased longevity in Drosophila as a result of overexpressing either the cytosolic copper,zinc superoxide dismutase (Cu,Zn-Sod) or the mitochondrial manganese-Sod (Mn-Sod) in adult flies (Sun & Tower, 1999; Tower, 2000; Sun et al., 2002); and also with the expression of human Cu,Zn-Sod in Drosophila motorneurons (Parkes et al., 1998; Spencer et al., 2003). Despite this, certain other recent studies have found that an increased Cu,Zn-Sod does not extend Drosophila lifespan (Orr et al., 2003); that the longevity effects of overexpressing human Cu,Zn-Sod in Drosophila motorneurons are markedly genotype- and sex-specific (Spencer et al., 2003); or that treatment with low-molecular-weight Sod mimetics does not retard aging in the nematode Caenorhabditis elegans (Keaney & Gems, 2003; Keaney et al., 2004).
In many organisms levels of stress resistance are probably not optimized for maximization of lifespan. Instead the levels of protective activities, as determined by natural selection, will be those that will optimize survival, as determined by rates of extrinsic mortality. There is essentially no benefit to diverting resources to the attainment of the higher stress-resistance levels that would maximize lifespans, if these lifespans cannot be attained in the wild. It is instead better to divert such resources to reproduction (Kirkwood, 2002). This is presumably the reason that the cells from long-lived mammalian species have a greater DNA repair capacity as compared with the cells from species that are short-lived due to high rates of predation (Grube & Burkle, 1992; Promislow, 1994; von Zglinicki et al., 2001). Long-lived Drosophila lines, obtained by the selective breeding of individuals with long lifespan, and long-lived C. elegans mutants obtained by direct selection, characteristically display elevated levels of stress resistance (Service et al., 1985; Lithgow, 1996). Parallel increases in stress resistance and lifespan are also seen when the p66shc gene is mutated in the mouse (Migliaccio et al., 1999).
Yeast can provide a useful model for investigating the diverse influences of an altered Sod activity level over stress resistance, the levels of damaged cell components and aging. Moreover, it is possible to investigate the effects of this altered Sod level on two distinct lifespans: the replicative (budding, generational) lifespan [measured as the number of daughters produced by each actively dividing mother cell (Kennedy et al., 1994; Sinclair & Guarente, 1997; Lin et al., 2000; Lin et al., 2002)] and the chronological lifespan [measured as the ability of stationary (G0-arrested) cultures to maintain viability over time (Longo et al., 1996; Fabrizio et al., 2001, 2004; MacLean et al., 2001, 2003; Harris et al., 2003)]. The latter lifespan essentially provides a simple, tractable system for investigating several factors that ensure long-term survival and a low rate of spontaneous mutagenesis in non-dividing cells. To this end we recently determined both survival and spontaneous mutation in the stationary cells of an isogenic series of yeast DNA repair mutants, so as to determine the relative importance of different highly conserved DNA repair activities in ensuring efficient chronological maintenance of G0-arrested yeast (MacLean et al., 2003). In higher organisms, any attempt to establish the importance of these enzymes in long-term cell maintenance would be complicated by additional factors, such as poly (ADP-ribose) synthesis and telomere shortening [yeast does not catalyse poly (ADP-ribose) synthesis or inactivate its telomerase]. We have also constructed strains that overexpress different free-radical-scavenging enzymes (Harris et al., 2003), partly to investigate the effects of these overexpressions on the two yeast lifespans, but also to provide a yeast system for investigating the basis of the detrimental effects sometimes seen when Cu,Zn-Sod (Avraham et al., 1988; Reveillaud et al., 1991; Bar-Peled et al., 1996; Lee et al., 2001; Peter et al., 2001; Pias et al., 2003) and Mn-Sod (Li et al., 1998; Zhong et al., 1996) are overexpressed in metazoans. Recently we described the consequences of overexpressing mitochondrial Mn-Sod in yeast, showing this acts to extend the chronological lifespan but to shorten the replicative lifespan (Harris et al., 2003). Here we describe the diverse effects of overexpressing the gene for cytosolic Cu,Zn-Sod.
Overexpressing active Cu,Zn-Sod in yeast requires either the presence of high copper or for the Sod1p levels to be increased in parallel with an overexpression of Ccs1p
Enzymes that neutralize ROS before the latter can wreak damage to cellular components constitute the first line of antioxidant defence in aerobic cells (Halliwell & Gutteridge, 1998). In Saccharomyces cerevisiae two Sod enzymes are of paramount importance in this respect: a Cu,Zn-Sod (Sod1p) that is abundant in the cytosol and the intermembrane space of the mitochondrion and an Mn-Sod (Sod2p), present in the lumen of the mitochondrion (Westerbeek-Marres et al., 1988; Gralla & Kosman, 1992; Moradas-Ferreira & Costa, 2000; Sturtz et al., 2001). To determine the effects of increasing these Sod activities, as well as the activity of cytosolic catalase, strains were constructed with extra genes for these enzymes integrated into their genomes (Table 1). These were introduced on DNA cassettes for an ADH2 promoter-driven expression of the coding regions of SOD1 (encoding Cu,Zn-Sod), SOD2 (encoding Mn-Sod) and CTT1 (encoding cytoplasmic catalase) (Harris et al., 2003). To compensate for any insertional effects of the cassette integrations, all lifespan measurements were made on strains that possess pRS403 sequences inserted at the HIS3 locus and pRS406 sequences integrated at URA3 (Table 1).
Recently we investigated the conditions that maximize yeast survival in stationary phase. Optimizing such survival increases the possibility that stationary (G0) yeast might be a useful model for identifying factors that help to ensure the long-term survival of non-dividing eukaryotic cells in general. It is also necessary to optimize long-term survival if attempting to show whether such a factor influences chronological aging, as it must then be shown that this factor has the ability to increase the longest lifespans for the organism under study (Kirkwood, 2002). Yeast was found to exhibit the longest stationary survival when it was pregrown to stationary phase on a respiratory carbon source (MacLean et al., 2001, 2003). It was specifically for this reason that the enzyme overexpression cassettes used the promoter of the alcohol dehydrogenase isozyme II (ADH2) gene, a promoter strongly de-repressed in cells growing on respiratory carbon source. It has since been found that this adaptation to efficient respiratory maintenance enables the cells to retain their replicative potential during stationary maintenance (manuscript in preparation). This contrasts with yeast grown to stationary phase on glucose, which exhibits a progressive decrease in replicative lifespan with increased periods of stationary arrest, followed by re-entry to the cell cycle (Ashrafi et al., 1999).
As previously reported (Harris et al., 2003), the integrated ADH2–SOD2 and ADH2–CTT1 cassettes increased the activities of Mn-Sod and catalase seven- and three-fold, respectively, during respiratory growth. In contrast, the SOD1 overexpression cassette (ADH2–SOD1) generated only marginal increases in Cu,Zn-Sod activity in initial experiments. This was despite this cassette elevating the levels of SOD1 transcripts, as analysed by Northern blotting, as well as the levels of Sod1p protein, as measured by Western blotting (compare the data for the wild-type and SOD1 strains; Fig. 1b,d). It was subsequently found that a substantial increase in Cu,Zn-Sod activity could be obtained not just in this SOD1 strain containing the integrated ADH2–SOD1 cassette but also in the control cells of the FY wild-type strain, by the simple expedient of supplementing the growth medium with high copper (Fig. 1a,b). In the wild-type this Cu,Zn-Sod is derived from the expression of the chromosomal SOD1 gene, whereas in SOD1 cells it originates from the combined expression of this chromosomal gene and the integrated ADH2–SOD1 cassette. As shown in Fig. 1(a), such copper supplementation increased not just total Cu,Zn-Sod activity, but also the Sod1p level in both the wild-type and the SOD1 strain. This reflects the copper-activation of the Ace1p regulator of the native SOD1 locus (Labbe & Thiele, 1999) and the cells elevating their Cu,Zn-Sod levels as a protective response to the high, potentially toxic levels of intracellular copper (Culotta et al., 1995).
This copper rescue of the Cu,Zn-Sod activity indicated that our inability to increase Cu,Zn-Sod activity by ADH2–SOD1 cassette expression in preliminary experiments had been due to an inefficient copper loading of the overexpressed Sod1p. A similar rescue of activity by copper supplementation is seen with Escherichia coli engineered to express eukaryotic forms of Cu,Zn-Sod (Stenlund & Tibell, 1999). We had no wish to elevate Cu,Zn-Sod activity levels in yeast by a routine growth of cultures in the presence of high copper, especially because free Cu2+ engages in Fenton chemistry (Halliwell & Gutteridge, 1998). Instead we investigated whether the levels of active Cu,Zn-Sod might be increased through the overexpression of Ccs1p (Lys7p). Ccs1p is a cytosolic chaperone dedicated to the loading of the Sod1p apoprotein with Cu2+ and highly conserved from yeast to humans (Culotta et al., 1997; Rae et al., 2001). Cells of the ccs1(lys7) mutant, lacking in this chaperone function, are defective in SOD1 expression and highly sensitive to oxidative stress. Its defect is also largely rescued by copper supplementation (Gamonet & Lauquin, 1998; Rae et al., 1999).
A cassette was constructed for ADH2 promoter-driven overexpression of the CCS1 open reading frame (ADH2–CCS1; see Experimental procedures). This was then integrated into S. cerevisiae, both in the absence and in the presence of the ADH2–SOD1 cassette (thereby generating strains CCS1 and SOD1 + CCS1, respectively, Table 1). With the introduction of this ADH2–CCS1 cassette the levels of Cu,Zn-Sod activity were increased in respiratory cultures without any need for copper supplementation (compare the FY wild-type and CCS1; also the SOD1 and SOD1 + CCS1 strains; Fig. 1b,c). In the control FY wild-type and in CCS1 this Cu,Zn-Sod originates entirely from the expression of the native SOD1 gene, whereas in strains SOD1 and SOD1 + CCS1 it results from the combined expression of this native SOD1 and the introduced ADH2–SOD1 cassette. The increase in Cu,Zn-Sod activity with introduction of the ADH2–CCS1 cassette alone (strain CCS1) is an indication that Ccs1p chaperone levels are normally limiting for the expression of active Cu,Zn-Sod in respiratory yeast [Ccs1p is normally much less abundant than Sod1p (Culotta et al., 1997)].
As shown in Fig. 1(a), protein levels of Sod1p increase when the wild-type is exposed to high copper [consistent with the copper activation of SOD1 (Labbe & Thiele, 1999)]. Levels of this Sod1p are, however, abnormally low in the ccs1Δ mutant (Fig. 2b). Cu-deficient rats and CCS−/– mice also have severely depleted Sod1p levels and Cu,Zn-Sod activity (Prohaska et al., 2003). Sod1p that is present as the Cu2+-deficient, apo-form (Cu2+-free, apo-Sod1p) might therefore be significantly less stable in vivo as compared with Sod1p that is present as the Cu2+-complexed, enzymatically active Cu,Zn-Sod. However, despite the loss of Ccs1p in the ccs1Δ mutant leading to a depleted level of Sod1p, CCS1 overexpression did not increase the Sod1p protein level resulting from the expression of the native SOD1 gene, even though it increased the Cu,Zn-Sod activity resulting from this expression (compare the control wild-type and CCS1 cells; Fig. 1b,c). Interestingly, the level of CCS1 transcripts resulting from a combined expression of the ADH2–CCS1 cassette and the native CCS1 was much higher in the absence of any simultaneous SOD1 overexpression (compare strains CCS1 and SOD1 + CCS1; Fig. 1d). It would appear therefore that not only does CCS1 expression influence both Sod1p level and Cu,Zn-Sod activity, but that SOD1 expression might also influence the levels of CCS1 mRNA. However, the objective of this study was not to investigate the interrelationships between these two genes, but to construct a strain in which Cu,Zn-Sod activity was substantially elevated in the absence of copper supplementation. With the combined introduction of the ADH2–SOD1 and ADH2–CCS1 cassettes this was achieved, levels of active Cu,Zn-Sod being elevated six- to eight-fold in respiratory cultures of strain SOD1 + CCS1 relative to the FY wild-type (Fig. 1b,c).
Overexpressing active Cu,Zn-Sod extends the chronological lifespan of yeast cells optimized for G0 maintenance
S. cerevisiae cells that are pregrown to stationary phase on a respiratory carbon source can be maintained for weeks at 30 °C with little loss of viability (MacLean et al., 2001, 2003). However, we measure their survival (chronological lifespan) at the more stressful temperature of 37 °C, so that we can conduct the survival measurements over a timescale of days rather than months. Of the strains in Table 1, it was those overexpressing active Mn-Sod or Cu,Zn-Sod that displayed an increase in this 37 °C chronological lifespan relative to the control wild-type cells [Fig. 2; strain SOD2 and CTT1 strain data from Harris et al. (2003) being included here solely for the purposes of comparison]. As shown in Fig. 2(b), of all of the strains that were constructed, it was the combined SOD1 + CCS1 overexpressor that exhibited the greatest extension to this lifespan (a doubling of mean survival relative to the wild-type at 37 °C). This increase in chronological lifespan due to SOD1 + CCS1 overexpression was appreciably greater than that generated through SOD2 overexpression alone (Fig. 2). This may reflect Mn-Sod (Sod2p) being only active in the mitochondrial lumen and proportionally a smaller fraction of the total cellular Sod activity, relative to Cu,Zn-Sod (Westerbeek-Marres et al., 1988; Gralla & Kosman, 1992; Moradas-Ferreira & Costa, 2000; Sturtz et al., 2001). Overexpression of CCS1 alone also generated a small survival increase (compare strain CCS1 and FY wild-type; Fig. 2) that can probably be attributed to its ability to increase the levels of active Cu,Zn-Sod resulting from the expression of the native SOD1 gene (Fig. 1b,c).
The substantial increase in longevity generated by the combined expression of the ADH2–SOD1 and ADH2–CCS1 cassettes contrasts with a shortening of chronological lifespan when the ADH2–SOD1 cassette is expressed alone (compare the survival of the SOD1 strain with the control FY wild-type cells; Fig. 2). Impaired G0 survival of these SOD1 cells appears not to be due to an excessive production of hydrogen peroxide as, rather than being rescued, it is compromised to an even greater extent with the simultaneous overexpression of catalase (compare strains SOD1 and SOD1 + CTT1: Fig. 2a). Instead this impaired survival is probably one consequence of the high endogenous oxidative stress generated when the ADH2–SOD1 cassette is expressed in the absence of either added copper or a simultaneous overexpression of CCS1 (an effect discussed in more detail below).
The lifespans in Fig. 2 are of stationary cells maintained at 37 °C. For those strains that displayed an increased G0 survival relative to the FY wild-type at this temperature (CCS1, SOD2 and SOD1 + CCS1) we also measured chronological lifespans at the less stressful temperature of 30 °C. Stationary survival was much longer at this lower temperature, yet SOD2 and SOD1 + CCS1 still exhibited longer survivals relative to the control wild-type cells (Fig. 3a).
Overexpressing active Cu,Zn-Sod delays the acquisition of large-scale protein oxidation damage
Cells and tissues are thought to become steadily more pro-oxidant during the natural events of aging (Sohal & Weindruch, 1996). This leads to an increase in the level of oxidatively damaged cell components, conveniently monitored by the measurement of protein carbonyls (Levine et al., 2000). A dramatic increase in total protein carbonyl content was observed between the 28- and 35-day timepoints of 30 °C stationary maintenance, as the cell viability declined below 10% (Fig. 3b). This increase in protein oxidation was delayed in the SOD1 + CCS1 overexpressor (Fig. 3b), an indication that the increased Cu,Zn-Sod activity of this strain is delaying the accumulation of extensive oxidative damage. This delay to damage acquisition is a plausible explanation for the improved G0 survival of such SOD1 + CCS1 cells (Figs 2b and 3a).
Increased Cu,Zn-Sod activity has only a small affect on the yeast replicative lifespan
We next analysed the extent to which Sod1p overexpression would affect the yeast replicative lifespan. This is a lifespan that tends to be shortened by mutations which sensitize cells to oxidative stress (Barker et al., 1999; Wawryn et al., 1999; Nestelbacher et al., 2000; Van Zandycke et al., 2002). Measuring the replicative lifespans of the strains in Table 1, SOD1 overexpression alone (strain SOD1) and CCS1 overexpression alone (strain CCS1) were found to be associated with moderate and dramatic shortenings of this lifespan, respectively (Fig. 4b). In contrast, the combined overexpression of these two genes (strain SOD1 + CCS1) generated a small but significant (P = 0.0176) extension to the mean replicative lifespan relative to the control FY wild-type cells (Fig. 4b). We did not observe in such cells, engineered to overexpress active Cu,Zn-Sod, the dramatic shortening of the replicative lifespan that is seen when the mitochondrial Mn-Sod is overexpressed in yeast [an effect apparently related to mitochondrial segregation defects in old mother cells (Harris et al., 2003)].
As shown in Fig. 5, addition of either 0.1 mm or 1 mm Cu2+ to the culture medium imposed a number of interesting effects upon these replicative lifespans. In essence, this Cu2+ addition exerted opposing influences on replicative potential in the wild-type and the SOD1 strain, shortening the replicative lifespan in the control wild-type strain (Fig. 5a) but lengthening it in the SOD1 overexpressor (Fig. 5b). Remarkably the shortened lifespan of the latter could be rescued completely by the addition of 1 mm Cu2+, becoming essentially identical to that of wild-type cells growing in the absence of any added Cu2+ (Fig. 5c). The copper supplementation is most probably allowing the overexpressed Sod1p of these SOD1 cells to become active (cf. Fig. 1a), thus providing protection both against the increase in copper level (Culotta et al., 1995) and against the harmful effects of expressing Sod1p in a Cu2+-deficient form (detrimental effects that are discussed in more detail below).
The marked shortening of the replicative lifespan in the CCS1 overexpressor (Fig. 4b) might be a metalloinsufficiency effect, with high levels of Ccs1p acting to co-ordinate Cu2+ so efficiently that Cu2+ insertion into critical targets (for example, the cytochrome a/a3 complex) is inhibited. However, as Fig. 5(d) shows, neither 0.1 mm nor 1 mm Cu2+ could restore full replicative potential to this CCS1 strain. It is therefore still paradoxical why this CCS1 strain should display a shortened generational lifespan. Nevertheless it is highly significant that an increased copper level (Fig. 5a), SOD1 overexpression (Fig. 5b) or CCS1 overexpression (Fig. 5d) all individually act to shorten the replicative lifespan, but that this lifespan is restored to that of the control wild-type cells when SOD1 gene overexpression is combined with either a Cu2+ supplementation or CCS1 overexpression (Fig. 5b,e). In combination, all three of these influences even generated a small lifespan extension, the SOD1 + CCS1 cells growing in the presence of 1 mm Cu2+ displaying a small increase in mean (P = 0.0084) and maximal replicative lifespan relative to the control wild-type cells grown without any added Cu2+ (Fig. 5f). This increase in lifespan is, however, small as compared with the dramatic effects of an altered Sir2p histone deacetylase activity or altered redox status on the replicative lifespan of yeast (Lin et al., 2000, 2002).
Other phenotypes associated with overexpressions of SOD1, CCS1, SOD2 and CTT1
We were intrigued by the detrimental effects of expressing the ADH2–SOD1 cassette in the absence of either copper supplementation or CCS1 overexpression. Both the chronological lifespan and the replicative lifespan are shortened by such expression (compare the data for the control wild-type and SOD1 strains; Figs 2, 4a and 5b). In the case of the replicative lifespan, this lifespan-shortening effect can be reversed by copper supplementation (Fig. 5b). Equally, the effects of the ADH2–SOD1 expression acting to shorten both lifespans are counteracted by the simultaneous overexpression of CCS1 (compare SOD1 and SOD1 + CCS1,Figs 2b and 4b). It is most probable therefore that these reversals of the damaging effects of ADH2–SOD1 cassette expression are operating by the increased copper, or the CCS1 overexpression, acting to elevate the levels of active Cu,Zn-Sod (cf. Fig. 1). By inference, these harmful effects almost certainly relate to Sod1p being overexpressed in a Cu2+-deficient form. They are particularly exacerbated in the respiratory cultures of strain SOD1 where it is probable that this Cu2+-deficient Sod1p is accumulating at an appreciable level as the result of the combined expression of the chromosomal SOD1 gene and the introduced ADH2–SOD1 cassette (cf. Fig. 1b,c).
As shown in Fig. 6(a), direct measurements of in vivo ROS production revealed the ADH2–SOD1 cassette expression was resulting in very high endogenous ROS production. ROS production in these SOD1 cells was almost as high as in the sod1Δ mutant, a strain totally deficient in Cu,Zn-Sod (Fig. 6a). By contrast, gene overexpressions that elevate the activity of either Mn-Sod or Cu,Zn-Sod (strains SOD2, CCS1 or SOD1 + CCS1) resulted in ROS levels being lower than in the control wild-type cells, fully consistent with Sod overexpression leading to higher levels of superoxide scavenging in these strains relative to the wild-type (Fig. 6a). Not only was the endogenous ROS production in such glycerol-grown SOD1 cells abnormally high, but these cells also displayed a high rate of spontaneous mutation, as measured by CAN1s to can1r mutagenesis (Fig. 6b) and a slightly reduced maximal temperature of growth (Fig. 7a). The high ROS production is almost certainly the cause of the latter two phenotypes [antioxidant functions are known to counteract spontaneous mutagenesis in yeast (Huang et al., 2003)]. This high ROS production is reversed by a simultaneous CCS1 overexpression (Fig. 6a), as indeed are all of the other detrimental effects of ADH2–SOD1 cassette expression (compare strains SOD1 and SOD1 + CCS1; Figs 2 and 4–7). It is apparent therefore that these detrimental effects relate to the incomplete in vivo Cu2+ loading of the overexpressed Sod1p, most probably this protein accumulating in the cells in Cu2+-deficient form.
The overexpressions of SOD genes studied here have other notable effects on phenotype. Strains with an elevated activity of either Cu,Zn-Sod or Mn-Sod (SOD2, CCS1 and SOD1 + CCS1) display improved respiratory growth in the presence of either ethanol or sorbic acid (Fig. 7b), chemical stress agents that exert a strong pro-oxidant effect on cells by elevating superoxide radical production by the mitochondrial electron transport chain (Piper, 1999). This mitochondrially generated superoxide also appears to be a factor determining the upper temperature limits of growth, because a 1 °C increase in maximal growth temperature is seen with increased activity levels of Mn-Sod, though not of Cu,Zn-Sod (Fig. 7a). This is consistent with other evidence for a greater impact of Mn-Sod relative to Cu,Zn-Sod in protection against the superoxide generated by the mitochondrial respiratory chain (Moradas-Ferreira & Costa, 2000; Sturtz et al., 2001).
Several studies report that overexpression of oxidant scavenging enzymes (Cu,Zn-Sod, Mn-Sod or catalase) can generate longevity increases in Drosophila (Parkes et al., 1998; Sun & Tower, 1999; Tower, 2000; Sun et al., 2002; Spencer et al., 2003). These investigations have been cited widely as support for the free radical theory of aging, the hypothesis that cells and tissues slowly lose function as a result of the steady accumulation of oxidative damage over time. Increased ROS scavenging does not, however, always lead to an increase in lifespan, as a recent study has found that Cu,Zn-Sod overexpression does not decrease age-related mortality in a long-lived Drosophila line (Orr et al., 2003). Nevertheless it would appear there may be a correlation among species between those with an increased level of intrinsic stress resistance and those with a longer intrinsic chronological lifespan (see Introduction).
Aging of adult Drosophila and C. elegans is primarily the chronological aging of postmitotic cells and tissues of the adult organism. These aging processes might therefore more closely resemble the chronological lifespan of stationary-phase, non-dividing yeast, rather than the replicative lifespan of yeast cells in division. Nevertheless stationary (G0) phase yeast is (like the dauer larval state of C. elegans) essentially a starvation condition. Here cells are surviving on their intrinsic reserves through low-level respiratory metabolism, having previously entered a state of high stress resistance (Werner-Washburne et al., 1996; Herman, 2002). Until now most of the studies on yeast stationary survival have used cells pregrown on respirofermentative (generally glucose) media. In such cultures the loss of major oxidant scavenging enzymes decreases survival (Longo et al., 1996, 1999), whereas the overexpression of Sod1p or Sod2p generates survival increases (Sturtz et al., 2001; Fabrizio et al., 2004). We recently investigated the conditions that maximize the survival of stationary yeast, because aging studies are more firmly grounded when they address the longest lifespans for the organism under study. Our work revealed that an optimized G0 survival of yeast requires the cells to be adapted to efficient respiratory maintenance, conveniently achieved by pregrowth of the cultures to stationary phase on a respiratory carbon source (MacLean et al., 2001, 2003). This study has focused exclusively on such cells for this reason.
Increasing either of the two Sod activities of yeast generates an increased G0 survival of cells adapted to efficient respiratory maintenance (Harris et al., 2003) (Figs 2 and 3). The optimized yeast chronological lifespan is therefore ultimately limited by oxidative stress. Of the two yeast Sod enzymes, it would appear to be Cu,Zn-Sod rather than Mn-Sod that exerts the greatest influence over this lifespan (although formal proof of this might require comparison between strains overexpressing identical activity levels of Mn-Sod or Cu,Zn-Sod). It should be noted that the earlier demonstrations of an increased yeast chronological lifespan all relate to cells pregrown to stationary phase on respirofermentative media (Fabrizio et al., 2001, 2004; Sturtz et al., 2001), cultures that consistently display shorter stationary survival in our hands as compared with cultures adapted to efficient respiratory maintenance (MacLean et al., 2001, 2003).
Cells become steadily more pro-oxidant as they age (Sohal & Weindruch, 1996). In such pro-oxidant cells, the antioxidant defences that rely on the reduced forms of glutathione and thioredoxins will be severely compromised, with the result that the activities that scavenge superoxide and peroxides will become even more critical for oxidative damage prevention. During long-term G0 maintenance, respiration-adapted yeast shows a dramatic increase in protein oxidation (Fig. 3). Both this protein oxidation and the loss of viability are delayed with Cu,Zn-Sod overexpression (Fig. 3), consistent with the large-scale oxidative damage to cell components being a major contributor to cell death.
This yeast study and the several studies of Sod overexpression in metazoans (see Introduction) reveal that overexpressing a gene that encodes a superoxide dismutase can potentially generate detrimental or beneficial effects. Overexpressing active Cu,Zn-Sod or Mn-Sod in yeast generates an increased chronological lifespan (Figs 2 and 3a) and resistance to endogenous oxidative stress (Fig. 7a). Mn-Sod overexpression even generates a slight increase in the maximum temperature of growth (Fig. 7b). However, ADH2–SOD1 cassette expression in the absence of either copper supplementation or CCS1 overexpression is quite clearly detrimental, causing shortening of both the chronological lifespan and the replicative lifespan (Figs 2, 4b and 5), high rates of endogenous ROS production and mutation (Fig. 6), and a reduction in the maximum temperature of growth (Fig. 7a). All of these detrimental effects are reversed by a simultaneous overexpression of CCS1 (compare strains SOD1 and SOD1 + CCS1; Figs 2 and 4–7). It is apparent therefore that they relate to incomplete in vivo Cu2+ loading of the overexpressed Sod1p. This, in turn, will cause Cu2+-deficient forms of Sod1p to accumulate in the cell (by itself, CCS1 overexpression elevates the level of Cu,Zn-Sod activity but not protein levels of Sod1p; Fig. 1b).
In considering the beneficial and detrimental effects of Sod1p overexpression, it is important to differentiate between the effects that arise from a Cu2+-deficient form of Sod1p and the effects that are due to overexpressing Cu,Zn-Sod in its fully active form. Increasing active Cu,Zn-Sod will lower superoxide levels, but at the same time increase the superoxide reductase activity of this enzyme, in which an electron donor other than superoxide reduces the active site Cu(2) to Cu(1). It has been suggested that the reported deleterious effects of Cu,Zn-Sod overexpression might be explained if this reductant were to be an essential molecule, or toxic in its oxidized form (Liochev & Fridovich, 2000). However, were this to be the case, it is most improbable that these deleterious effects would be rescued by either copper supplementation or Ccs1p overproduction. Instead our data indicate that the harmful effects result directly from an incomplete Cu2+ loading of the overexpressed Sod1p. They appear to be associated with the high ROS production caused by ADH2–SOD1 cassette expression in the absence of either high copper or CCS1 overexpression (Fig. 6a), a high ROS production that is almost certainly the cause of the high mutation rate in strain SOD1 (Fig. 6b).
It is not yet clear what causes this high ROS production. Sod1p is dimeric (Culotta et al., 1997; Rae et al., 2001). Therefore, one possibility is that ROS are produced when just one subunit of the Sod1p dimer contains a redox-active Cu2+. This incomplete Cu2+ occupancy might, in turn, alter the co-ordination geometry of this single bound Cu2+. The latter, in turn, might result in the bicarbonate () anion-mediated peroxidase activity shown for certain mutant forms of human Sod1p linked with the neurodegenerative disorder familial amylotrophic lateral sclerosis (FALS) (Roe et al., 2002). The symptoms of FALS in humans are believed to be the manifestation of long-term effects of increased oxidative stress (Corson et al., 1998). One of the consequences of a high ROS production is thought to be the inactivation of those enzymes that contain labile [4Fe−4S] clusters. This [4Fe−4S] oxidation in turn liberates free iron which, through Fenton chemistry, can lead to the production of the highly toxic hydroxyl radical (Srinivasan et al., 2000).
Detrimental effects of Cu,Zn-Sod overexpression have also been described in more complex systems, for example in Drosophila (Reveillaud et al., 1991) and in transgenic mice (Avraham et al., 1988; Reveillaud et al., 1991; Bar-Peled et al., 1996; Gahtan et al., 1998; Golenser et al., 1998; Peter et al., 2001). Cu,Zn-Sod overexpression can also increase lipid and protein oxidation (Lee et al., 2001) and apoptosis (Pias et al., 2003). There is a distinct possibility that these harmful effects are all due to the in vivo accumulation of Cu2+-deficient Sod1p because, as with our SOD1 yeast strain, the mice engineered for Cu,Zn-Sod overexpression have increased endogenous oxidative stress and mutation rates (Peter et al., 2001). This study appears to be the first to show that these diverse detrimental effects of Sod1p overexpression can be reversed by the simultaneous overexpression of CCS1, a gene whose sole function is to encode the chaperone that inserts Cu2+ into apo-Sod1p (Culotta et al., 1997). These harmful effects relate therefore to an inefficient Cu2+-loading of the overexpressed Sod1p, most probably the in vivo accumulation of a Cu2+-deficient form of Sod1p capable of generating high levels of endogenous oxidative stress.
Strains and media
S. cerevisiae strains used in this study are listed in Table 1. Cells were grown aerobically at 30 °C in liquid YP medium (1%, w/v, Difco Yeast Extract, 2% Bacto peptone, 20 mg L−1 adenine) containing as carbon source either 2% glucose (YPD) or 3% glycerol (YPGlycerol).
Construction of CCS1-overexpressing strains pRS403(ADH2) and pRS406(ADH2) are derivatives of pRS403 and pRS406 (Sikorski & Hieter, 1989) that contain an ADH2 promoter fragment (Harris et al., 2003). The coding region of CCS1 was inserted into these vectors as a PCR product with BamH1 and Not1 termini, thus yielding pRS403(ADH2–CCS1) and pRS406(ADH2–CCS1), respectively. These vectors were cleaved at their unique Pst1 site, prior to integrative transformation into yeast, selecting for histidine prototrophy (pRS403-derived plasmids) or uracil prototrophy (pRS406-derived plasmids).
Analysis of stationary phase survival (chronological lifespan)
Chronological lifespan determinations were of cells grown to early stationary phase on YPGlycerol, then transferred to water, as described earlier (MacLean et al., 2001, 2003; Harris et al., 2003). At the indicated times during subsequent 37 °C or 30 °C aerobic maintenance three serial dilutions were prepared, these being plated on YPD agar. After 5 days at 30 °C, viable cells were determined by colony counting. At least 900 viable cells from each dilution were used for each determination, the figures showing the mean and the standard deviation for the data from the three separate dilutions.
Analysis of replicative lifespan
To determine replicative lifespans, exponentially growing yeast was spread at low density onto YPGlycerol agar and incubated for 3 h at 30 °C to allow bud emergence. Typical experiments used 60 virgin buds, these being moved to fresh areas of the plate using a Singer MSM Micromanipulator (Singer Instruments, Taunton, UK). Lifespans of these virgin cells were then determined by counting and removing each of the buds produced, until they no longer divided, as previously described (Kennedy et al., 1995). For determining the replicative potential of stationary phase cells, these were plated at low density on YPGlycerol agar until bud emergence was observed, whereupon each budding cell was micromanipulated to a designated position on the plate for measurement of its replicative lifespan, as above. Differences in survival were compared using the non-parametric Log-Rank test, where the test statistic is chi-squared (χ2) and P ≤ 0.05 indicates a significant difference in lifespan.
Measurements of mutation frequencies and free radical production
Mutation to canavanine resistance was determined as previously described (MacLean et al., 2003). Free radical production in vivo was determined as the mean of dihydrorhodamine fluorescence measurements from 225 individual cells from each strain (Dugan et al., 1997).
Protein extract preparation, Sod and catalase assays
Yeast extracts were prepared in 50 mm potassium phosphate buffer (pH 7.0) containing a cocktail of protease inhibitors. Protein concentrations were determined using the Bio-Rad protein determination kit and bovine serum albumin as standard. Catalase activity was determined as in Aebi (1984). Sod activity was measured either from its ability to inhibit the xanthine oxidase reaction form or its ability to inhibit reduction of nitro blue tetrazolium to formazan in gels (Flohe & Otting, 1984). Distinction of Mn-Sod from Cu,Zn-Sod was based, in the former assay, on the selective capability of 2 mm cyanide to inhibit Cu,Zn-Sod (Flohe & Otting, 1984) and, in the latter assay, on the different gel migrations of Mn-Sod and Cu,Zn-Sod.
Analysis of expression levels; protein carbonyls
Levels of Sod1p and Sba1p (a loading control) were analysed by Western blotting using polyclonal rabbit anti-Cu/Zn-Sod (Stressgen SOD-101) or anti-Sba1p antibodies (the latter raised in our laboratory) at 1 : 5000 dilution and the Amersham ECL detection kit. To analyse oxidized proteins, samples of total cell protein were derivatized by mixing an aliquot with one volume of 12% (w/v) SDS and two volumes of 20 mm 2,4-dinitrophenylhydrazine in 10% (v/v) trifluoroacetic acid [a blank control was treated with two volumes of 10% (v/v) trifluoroacetic acid alone] (Levine et al., 1995). After incubation for 30 min at room temperature in the dark, and neutralization, the proteins (10 µg per lane) were loaded on a 12.5% SDS-PAGE. After electrophoresis, proteins were electroblotted onto a nitrocellulose membrane, and the membrane was probed with rabbit IgG anti-DNP (Dako) (1 : 5000 dilution) and goat anti-rabbit IgG linked to horseradish peroxidase (Sigma) (also 1 : 5000 dilution) by standard techniques. Detection utilized the RPN 2109 kit (Amersham Biosciences).
This project was supported by BBSRC grants 31/G15970 and G17849.