The mitochondrial ribosomal protein of the large subunit, Afo1p, determines cellular longevity through mitochondrial back-signaling via TOR1

A novel yeast mutant with reduced replicative aging

We compared the transcriptome of senescent yeast mother cells (fraction V) with young daughter cells (fraction II) after separating them by elutriation centrifugation [2]. Senescent cells were found to upregulate 39 genes and to down-regulate 53 transcripts. Deletion mutants [12] corresponding to these 92 genes were tested for their resistance or hypersensitivity to five different oxidants (hydrogen peroxide, tert-butyl hydroperoxide (t-BHP), diamide, cumene hydroperoxide, and menadione). Only two mutants were found to be consistently resistant against more than one oxidant (and not hypersensitive to any other oxidant). Among these two mutants only one, deleted for YGR076C/MRPL25 (later termed AFO1, see below) caused a mother cell-specific lifespan expansion on the standard media used by us (SC + 2% glucose) (Figure 1). This deletion mutation conferred resistance to diamide and t-BHP and a somewhat weaker resistance to hydrogen peroxide, as well as a 50% reduced ROS production (as compared to the BY4741 ρ⁰mutant). ROS production was measured by quantitation of fluorescence signals obtained after dihydroethidium (DHE) staining. The mutant displayed a 60% increase in the median and a 71% increase in the maximum lifespan (Figure 1). The mutant only grew on media containing fermentable carbon sources and hence is respiration deficient. We therefore asked if the respiratory deficiency caused the increased replicative life span. However, a bona fide BY4741 ρ⁰mutant did not show any extension in replicative life span (as compared to BY4741 WT cells), meaning that lack of respiration is not sufficient to confer longevity to mother cells (Figure 1). We also tested if afo1Δ cells displayed the retrograde response [3,13] by measuring CIT2 transcription and no effect of the afo1Δ mutation could be discerned (see Supplementary Figure 1). We conclude that the elongation of lifespan observed here is not caused by respiratory deficiency and is independent of the retrograde response as defined by Jazwinski [14] and Butow [15].

AFO1 codes for a mitochondrial ribosomal protein

AFO1 (YGR076C) codes for MrpL25p, identified by proteomic analysis as a component of the large subunit of the mitochondrial ribosome [16]. Because of its remarkable longevity phenotype, we re-named the gene AFO1(for aging factor one). A recombinant construct in which Afo1p was fused in its C-terminus with GFP (Afo1-GFP) was transfected into a heterozygous afo1Δstrain. Tetrad dissection revealed that the Afo1-GFP could replace endogenous Afo1p to enable growth on a non-fermentable carbon source. Confocal fluorescence microscopy confirmed that the protein is located in mitochondria irrespective of the cellular age and the genetic background (supplementary material, Supplementary Figure 2). The deletion mutant afo1Δ exhibited a ρ⁰petite phenotype, meaning that it failed to grow on glycerol media and lacked DAPI-detectable mitochondrial DNA. The mutant also showed negligible oxygen consumption when growing on glucose (data not shown). However, in contrast to a bona fide BY4741 ρ⁰petite mutant, which grew much more slowly than WT cells on standard media with 2% glucose as carbon source, the afo1Δ mutant grew as rapidly as WT cells. The growth properties of the mutant and its metabolic implications will be published in detail elsewhere. The average size of the afo1Δ mutant cells in exponential phase was equal to that of WT cells, while cells of the ρ⁰strain were about 20% larger. Importantly, disruption of the AFO1 gene in ρ⁰cells restored rapid growth, hence reversing the growth defect induced by the absence of mitochondrial DNA.

To obtain definite genetic proof that the AFO1 deletion caused the resistance against oxidative stress and the extension of the life span described above, we performed co-segregation tests in meiotic tetrads after out-crossing the afo1Δ strain in an isogenic cross. In 10 unselected tetrads, which all revealed a regular 2:2 segregation, we observed strict co-segregation of G418 resistance indicating the presence of the gene deletion, respiratory deficiency, and resistance against hydrogen peroxide stress (Figure 2A). We also tested the lifespan of all four haploid progeny of one tetrad and found consistent co-segregation of the deletion allele of afo1 with extended lifespan (Figure 2B). Furthermore, we also tested if the phenotype of the mutant might result from changes of the expression of the two neighboring genes of AFO1. No such effect was apparent (Figure 2C). We conclude that lack of AFO1 results in long lifespan and the oxidative stress resistance.

We next addressed the possibility that the deletion of other mitochondrial ribosomal genes might also lead to an increase in replicative life span. For this, we investigated the lifespan, growth properties and oxidative stress resistance of two additional deletion mutants in the genes MRP17 and PPE1, encoding mitochondrial ribosomal proteins of yeast. YKL003CΔ (mrp17Δ) was found to be resistant against diamide, t-BHP and juglone, but was hypersensitive to hydrogen peroxide and had a normal lifespan. YHR075CΔ (ppe1Δ) was resistant against diamide, yet had a normal lifespan (Figure 3). Therefore, the effect of the afo1Δ deletion mutant on lifespan is gene-specific.

Longevity mediated by the afo1 deletion is mediated by the TOR1 pathway

Two independent lines of evidence revealed that the afo1 deletion confers longevity and oxidative stress resistance through the TOR1 signaling pathway. First, we chromosomally integrated a C-terminally GFP-labeled version of the transcription factor, Sfp1p, at the SFP1 locus under the control of the native promotor in strains afo1Δ, BY4741 WT and BY4741 ρ⁰. Sfp1p is activated by the TOR1 and PKA pathways and is regulated by shuttling between the nucleus in its active form and the cytoplasm upon deactivation. Sfp1p is a major regulator of cytoplasmic ribosome synthesis and, consequently, of cellular growth [6]. As expected, addition of the Tor1p inhibitor rapamycin to WT cells induced the translocation of Sfp1p from the nucleus to the cytoplasm. In the bona fide BY4741 ρ⁰strain, Sfp1p was found constitutively in the cytoplasm, even in the absence of rapamycin. In stark contrast, in the afo1Δ mutant, Sfp1p was constitutively present in the nucleus, and rapamycin failed to induce the nucleo-cytoplasmic translocation of Sfp1p (Figure 4A). Similar results were obtained with an alternative Tor1p inhibitor, arsenite [17]. Arsenite induced the nucleocyto-plasmic translocation of Sfp1p in WT cells, while Sfp1p stayed in the cytoplasm of ρ⁰cells and in the nuclei of afo1 mutant cells, irrespective of the addition of arsenite (Figure 4B). Rapamycin failed to inhibit the growth of afo1 mutant cells [18]. Altogether, these data suggested that TOR1 signaling might govern the longevity of afo1 cells. The relation between TOR1 and AFO1 was further explored by epistasis experiments using double mutants (Figure 5). The lifespan of the double deletion strain (afo1Δ, tor1Δ) was similar to the lifespan of the tor1 deletion strain, i.e. about 15% longer than wild type (in good agreement with [11]). However, the double mutant afo1Δ,tor1Δ strain aged more rapidly than the single mutant afo1Δ strain (Figure 5). We conclude that a functional TOR1 gene is needed for exerting the lifespan-prolonging effect of afo1Δ.

We constructed single and double knockout afo1Δ, sfp1Δ mutant strains and tested their mother cell-specific lifespan and oxidative stress resistance. The median lifespan of sfp1Δ cells was shortened considerably as compared to WT cells, and the lifespan of the double afo1Δ, sfp1Δ mutant was longer than that of the sfp1Δ mutant, yet shorter than WT and afo1Δ (Figure 6A). Hence, the very short lifespan of sfp1Δ mutant cells is partially rescued by the afo1 mutation. Like the double afo1Δ,sfp1Δ mutant, sfp1Δ cells displayed a major growth defect. When the sfp1Δ strain was made ρ⁰with ethidium bromide, cell growth was not further inhibited (data not shown). Comparison of the strains on plates containing 1.6 mM t-BHP revealed that afo1Δ is moderately resistant, while single sfp1Δ and double afo1Δ,sfp1Δ mutants exhibited a similar degree of high resistance (Figure 6B). Taken together, these results show that the lifespan-extending effect of afo1Δis most likely independent of the presence of SFP1.

Next, we addressed the question as to whether the longevity phenotype of the afo1 mutation might originate from suppressing the yeast apoptosis pathway. As shown previously [2], old mother cells of the wild type display all of the known markers of yeast apoptosis while these markers are absent from young cells. To tackle this problem, we isolated young (fraction II) and old cells (fraction V) from WT and afo1Δ cells by elutriation centrifugation and tested several markers of apoptosis such as externalization of phosphatidyl serine and DNA strand breaks (Figure 7). Our data clearly indicated that afo1Δ cells did not lose the ability to undergo apoptosis. In spite of a 60% longer median lifespan, senescent mother cells finally succumbed to apoptosis. We conclude that the components of the programmed cell death pathway that a yeast cell has at its disposal, do not cause replicative aging, but that vice versa replicative aging finally leads to cell death via apoptosis.

We also investigated whether the longevity phenotype of the afo1Δ mutant might be mechanistically related to the production of extrachromosomal rDNA minicircles (ERCs) [19,20]. FOB1 encodes a protein required for the unidirectional replication fork block in rDNA replication. We analyzed the influence of the fob1 mutation on longevity, growth, and ERC content of WT and afo1Δ cells. In our analysis the fob1Δ mutation in the BY4741 strain leads to an increase of the replicative lifespan by about 5 generations, in good agreement with previous reports [11,21].

However, we observed a similar median life span of the fob1Δ, afo1Δ double mutant and the afo1Δ mutant cells (Figure 8A). As an internal control, both the fob1Δ single mutant and the fob1Δ, afo1Δ double mutant exhibited the absence of ERCs even in fraction IV and V old cells, while a continuous age-dependent increase in ERCs was found, in particular in fraction IV and V senescent mother cells from WT and afo1Δ cells (Figure 8B). Thus, the lifespan-extension observed in the afo1Δ strain occurs in the presence of ERCs and is not further increased when ERCs are absent, consequently ERCs do not influence longevity in the afo1Δ strain.

AFO1, the retrograde response and mitochondrial back-signaling

The retrograde response (as defined by Jazwinski [14] and Butow [15]) of non-respiring cells is transmitted through the transcription factor Rtg1/Rtg3p and allows for the transactivation of genes involved in peroxisome synthesis that compensate for the deficient amino acid biosynthesis of cells that lack a complete citrate cycle. As an indicator of the retrograde response, expression of the peroxisomal Cit2p citrate synthase is usually measured [14,15]. Yeast strains displaying a strong retrograde response increase their replicative lifespan as ρ⁰strains over that of the corresponding ρ⁺strain. The retrograde response is generally suppressed in 2% glucose but strong on raffinose as sole carbon source [13]. We have measured CIT2 transcription under the conditions used in this study and found no increase in the transcript of this gene (Supplementary Figure 1), explaining why the ρ⁰strain in the BY4741 series shows the same lifespan as wild type. When raffinose was used as a carbon source, CIT2 transcription was increased, and as expected, an increase in the lifespan of the bona fide BY4741 ρ⁰strain was observed (unpublished observation). However, during growth on 2% glucose when the retrograde response is absent in our strain background, we do observe the increase in lifespan described in the present paper. We therefore conclude that the mechanism leading to this increase must be different from the retrograde response.

The lifespan elongation described for the afo1Δ mutant strain depends on a signal transmitted from mitochondria to the nucleo-cytoplasmic protein synthesis system and has a strong influence on replicative aging, vegetative growth, and oxidative stress resistance (see below). We propose to call this regulatory signaling interaction „mitochondrial back-signaling" to dis-tinguish it from the retrograde response described by Jazwinski [14] and Butow [15].

Evidence for involvement of the TOR1 pathway in longevity of the afo1 deletion strain

The nature of the signal created by Afo1p is unknown, especially since we found this ribosomal protein to be located in mitochondria in all physiological situations tested, including senescent yeast mother cells. Nonetheless, two independent lines of evidence support the notion that increased activity of TOR1 determines the longevity of the afo1 deletion mutant. First, in the double mutant deleted for both TOR1 and AFO1, a lifespan is observed that is only moderately longer than that of the wild type and is identical with the lifespan of the tor1Δ single mutant (Figure 5). Thus, paradoxically, the relatively small but significant elongation of the lifespan of ρ⁺respiring tor1Δ cells depends on inactivation of Tor1p, while the large increase in lifespan in the non-respiring afo1Δ cells depends on activity of the Tor1p. Second, rapamycin fails to abolish the nuclear location of the transcription factor, Sfp1p, an indicator of Tor1p activity, in afo1Δ cells (Figure 4A). Likewise, arsenite, another inhibitor of Tor1p [17], fails to abolish the nuclear location of the transcription factor, Sfp1p, in afo1Δ cells (Figure 4B). Sfp1p is well known to be one of the major metabolic regulators of growth and ribosome biosynthesis, which is limiting for growth [6]. The data presented here seem to indicate that Sfp1p activity in the nucleus could be crucial for longevity. The double mutant experiments shown in Figure 5 indicate that the sfp1Δ and the afo1Δ mutations exert their influence on longevity independently of each other. Moreover, we tested TORC1 kinase activity in WT, ρ⁰and in afo1Δ cells (data not shown) and found that the long-lived mutant, like the ρ⁰strain displayed only very weak TORC1 kinase activity. These results indicate that the Tor1p activity needed for longevity in the mutant might be feedback-regulated by Sfp1p and/or maybe independent of TORC1kinase activity.

A tentative scheme describing the genetic interactions of mitochondrial back-signaling that we are discussing here, is presented in Figure 9.

Has the function of AFO1 been conserved in evolution?

In this paper we are presenting evidence for a possibly indirect interaction of the mitochondrial ribosomal protein, Afo1p, with the TOR1 signaling system of yeast. This interaction is independent of the primary function of Afo1p in translation. A reduction of TOR1 signaling in yeast [7], rodent and human cells [22] suppresses cellular aging in cell culture [22], and increases longevity in mice [23], worms [24] and fruit flies [25,26]. These effects were shown to be non-additive with caloric restriction suggesting that the TOR pathway in these organisms is crucial for transmitting the caloric restriction signal. In metazoa, cellular life, but not organismic life is possible in the absence of mitochondrial respiration. It is therefore difficult to draw conclusions as to the generality of the afo1 mutant-based longevity described in the present paper.

The protein complement of mitochondrial ribosomes of both yeast and human cells has been studied [16 and the literature cited therein, 27-30] and the non-translational or extra-ribosomal functions (mostly in transcriptional regulation) of ribosomal proteins have been extensively studied [31,32]. The published extraribosomal functions mostly concern cytoplasmic, not mito-chondrial ribosomes. Possible extraribosomal functions have to date been found for three of the yeast mitochondrial ribosomal proteins only. Mrps17p and Mrpl37p may play a special role in yeast sporulation [33]. The mitochondrial ribosomal protein of the small subunit, yDAP-3 [34] is well conserved between yeast and human cells and besides its translational role has a distinct function in apoptosis. Its role in the aging process has not been studied yet.

Afo1p is a protein of S. cerevisae for which an obvious homolog is known in Neurospora crassa [16], but which has no easily apparent counterparts in other eukaryotes (or in E. coli) as judged by sequence similarity alone. It is therefore impossible presently to draw conclusions about possible functions of homologs of this protein in aging of higher eukaryotes. However, this may change when the three-dimensional structure of mitochondrial ribosomes will be determined and structural and functional homologs of Afo1p in higher eukaryotes may be found.