Extension of chronological life span by reduced TOR signaling requires down-regulation of Sch9p and involves increased mitochondrial OXPHOS complex density

Introduction

How and why we age has long been a fascination of humans.  In addition to being of intrinsic philosophical, evolutionary and biological interest, determining the molecular and cellular mechanisms underlying the aging process is relevant to understanding age-related pathology that ultimately limits human life and health span. Model organism studies have been instrumental in understanding aging, with many conserved pathways and factors having been identified in files, worms and yeast (and other organisms) that have physiological and pathological relevance in humans [1]. One general area that has been implicated strongly in aging and life span determination is nutrient availability/sensing. For example, dietary (i.e. caloric) restriction extends life span and ameliorates many of the age-associated declines in cellular function in virtually all organisms examined to date [2].

One major consequence of changing nutrient availability/sensing is alternation of cellular metabolism and mitochondrial respiration. Life span extension by caloric restriction, for instance, usually involves enhanced mitochondrial activity [2,3]. While best known for providing ATP via oxidative phosphorylation (OXPHOS), mitochondria are a major crossroads for anabolic and catabolic metabolism, as well as many other critical cellular functions such as apoptosis, signal transduction, and ion homeostasis [4]. Mitochondria also contain a DNA genome (mitochondrial DNA; mtDNA) that harbors a set of genes involved in OXPHOS and requires dedicated machinery for organellar DNA replication and gene expression that is encoded primarily by genes in the nucleus (e.g. mitochondrial DNA and RNA polymerase, ribosomes, transcription and translation factors, etc) [5,6]. Mitochondria also generate reactive oxygen species (ROS) as byproducts of the electron transport process, which is a major way they are thought to contribute to the aging process. For example, the "mitochondrial theory of aging", which builds on Harman's "free-radical" theory, posits that ROS from mitochondrial respiration damage cellular components, including mtDNA, and lead to declines in cell, tissue and organismal function over time [7,8]. As ROS are also signaling molecules, altered signal transduction is another potential contributor to aging phenotypes due to mitochondrial dysfunction [9]. While the mechanisms through which altered respiration affects life span are complex and have not been defined fully, differential ROS production is likely involved. For example, aberrant respiration due to defective RAS signaling [10], pharmacological inhibition [11], or imbalanced translation of mtDNA-encoded OXPHOS subunits [9] elevates cellular ROS and severely curtails yeast chronological life span (CLS). Conversely, mild uncoupling of mitochondrial respiration extends yeast CLS and decreases ROS [11].

Several kinase pathways serve as physiological switches in response to nutrient availability. For example, the conserved target of rapamycin (TOR) signaling pathway controls growth by positively regulating the processes of ribosome biogenesis and cytoplasmic translation when preferred nutrient supplies are available. In yeast, the TOR pathway also negatively regulates stress response genes, autophagy, and usage of alternate carbon and nitrogen sources [12]. Thus, when nutrients are limiting, TOR activity is reduced, energy is conserved (by shutting down expensive growth-promoting pathways) and diverted to provide stress resistance and access to alternate energy stores. The TOR kinase forms two multi-protein complexes, TORC1 and TORC2, with TORC1 functioning as the nutrient sensor [12]. In yeast, there are two TOR kinase genes TOR1 and TOR2. Both Tor1p and Tor2p can function in the TORC1 complex, but only Tor2p can function in the TORC2 complex. Thus, deletion of TOR1 results in reduced TORC1 signaling, but is not lethal. This is because Tor2p can partially cover the loss of Tor1p in TORC1, while still also functioning in TORC2. In contrast, deletion of TOR2 is lethal [13]. Reduced TORC1 signaling extends life span in a number of model organisms including yeast (S. cerevisiae), worms (C. elegans) and flies (D. melanogaster) [14-17]. We recently reported that a major mechanism underlying this phenotype in yeast is enhanced mitochondrial respiration driven, at least in part, by increased translation of mtDNA-encoded OXPHOS subunits [18]. In that study, we speculated that the extension of CLS by reduced TOR signaling involves an increase in the number of OXPHOS complexes per organelle that increases oxygen consumption, decreases ROS production in stationary phase, and thereby limits damage to cellular components. However, since mtDNA encodes only minority of the OXPHOS complex subunits (i.e. of the ~80 OXPHOS subunits only seven in yeast and thirteen in mammals are encoded by mtDNA) and mitochondria contain >1,000 proteins (encoded by nuclear genes and imported into the organelle), the possibility that TOR signaling regulates mitochondria in a more global fashion is likely. In fact, TOR-dependent changes in the mitochondrial proteome have been documented in human Jurkat T cells [19].

Sch9p belongs to the AGC family of kinases and is a key downstream target of TORC1 signaling in yeast. For example, Sch9p is a functional ortholog of ribosomal protein S6 kinase, a key mediator of mTOR signaling in mammalian cells [20]. TORC1 directly phosphorylates Sch9p at multiple sites, which is important for modulating cytoplasmic translation and cell cycle progression. Sch9p is also a negative regulator of both chronological and replicative aging [14,21] and has recently been shown to similarly regulate mitochondrial respiration [22]. In fact, like deletion of TOR1, deletion of SCH9 extends yeast CLS in a respiration-dependent fashion, suggesting that Sch9p could be a downstream mediator of TOR-dependent mitochondrial OXPHOS regulation in this regard. In the current study, we have examined in greater mechanistic detail how the yeast TOR pathway influences mitochondrial gene expression, OXPHOS activity, and proteome composition, and the role of the Sch9p kinase as a downstream mediator of its effects on mitochondria.

Results

Discussion

This study provides significant new insight into the mechanism through which TOR signaling controls mitochondrial function to influence yeast CLS and elucidates which arm of the TORC1 pathway is involved. The primary conclusions we draw from the results obtained are that 1) reduced TORC1 signaling (via deletion of the TOR1 gene) increases respiration primarily through up-regulation of the number of OXPHOS complexes/organelle, not by increasing overall mitochondrial biogenesis, 2) the up-regulation of OXPHOS complexes involves both mtDNA-encoded and nucleus-encoded subunits and, in terms of mtDNA expression, occurs primarily via translational regulation, 3) in addition to its effects on OXPHOS complex abundance, TOR signaling controls other aspects of mitochondrial proteome dynamics, 4) TOR-dependent changes in mitochondrial function and CLS are mediated by the downstream Sch9p kinase, and 5) it is TOR-dependent alterations of mitochondrial function in the exponential and/or post-diauxic-early stationary growth phases that subsequently impact late stationary-phase survival and extend CLS, which suggests a role of "mitochondrial pre-conditioning" on yeast aging. The basis of these conclusions and additional interpretations are discussed below.

The increase in cellular mitochondrial oxygen consumption (i.e. respiration) in response to reduced TOR signaling reported herein (Figure 5A) and previously [18] could occur by one of several mechanisms that are not mutually exclusive. For example, it could be mediated by direct effects on the activity of existing OXPHOS complexes, by increasing overall mitochondrial biogenesis (resulting in more mitochondria/cell), or by increasing the number of OXPHOS complexes per organelle. Our results demonstrate that increasing organelle OXPHOS complex density is definitely one mechanism at play.

The basis for this conclusion is that, in mitochondrial extracts, we observe increased abundance of both nuclear and mitochondrial OXPHOS subunits, but not other mitochondrial markers (e.g. porin; Figure 2A). This result was substantiated by our initial 2D-DIGE proteomic analysis of highly purified mitochondria from wild-type and tor1Δ strains, in which we identified OXPHOS complex subunits (from three separate complexes) as proteins that are in significantly higher abundance in mitochondria from tor1Δ cells (Table 1). Finally, there was no increase in overall mitochondrial biogenesis as judged by mtDNA content (Figure 2C), labeling of mitochondria with a GFP marker and analyzing them by FACS (Figure 2B), and western blot comparisons of mitochondrial and cytoplasmic markers (Figure 5C). While increased OXPHOS complex density is clearly occurring, we have not eliminated the possibility that there are also TOR-dependent effects on the enzymatic activity of the complexes that contribute to the increase in oxygen consumption.

Our 2D-DIGE results are not entirely consistent with a recent proteomic study of rapamycin-treated yeast cell [23], where fewer OXPHOS proteins were identified as up-regulated. We found that addition of rapamycin during the growth phase impacts mitochondrial oxygen consumption on a longer time scale and to a lesser degree compared to adding rapamycin from the beginning of the growth experiment (i.e. at inoculation; Supplementary Figure 1). The latter condition is in some ways more similar to the tor1Δ strains analyzed in this study in that, in this case, TOR signaling is reduced throughout all stages of growth. Thus, differences in the timing and/or degree of TOR inhibition may explain the different results obtained in the two studies.

To increase OXPHOS complex density as a means to increase mitochondrial oxygen consumption is to our knowledge a unique mechanism of mitochondrial regulation levied by the TOR pathway. We originally hypothesized that this would lead to greater mitochondrial membrane potential due to the increase in electron transport activity and perhaps also a higher cellular ATP. However, this was not the case; there was instead a decrease in membrane potential in tor1Δ strains (Figure 2D) and no change in cellular ATP (data not shown). Thus, in tor1Δ strains, there in an increase in electron transport activity (i.e. oxygen consumption) and a decrease in mitochondrial membrane potential, which equates to a mitochondrial network with overall lower energy capacity on average. One potential explanation for this result is that reduced TOR signaling is leading to an increase in uncoupled respiration. This would lead to increased oxygen consumption in an attempt to maintain the membrane potential in the face of the proton leak and an inability to simultaneously increase ATP production. Since mild uncoupling also increases CLS [11], this indeed may prove to be the mechanism through which TOR signaling influences aging in yeast. Testing this hypothesis is a logical area of future investigation, but certainly other explanations can be envisioned.

In order to affect an increase in OXPHOS complexes per mitochondrion, the cell needs to increase the production and/or stability of both mtDNA-encoded and nucleus-encoded OXPHOS subunits, while not inducing a full mitochondrial biogenesis response. How reduced TOR signaling accomplishes this remains to be determined, yet several insights are gleaned from our results. First, we observe an increase in both mtDNA-encoded and nucleus-encoded OXPHOS subunits (Figure 2A, Table 1), thus TOR signaling is affecting both mitochondrial and nuclear gene expression simultaneously. According to our results, this is occurring both at the mRNA level (Figure 3) and at the translational level (Figure 1) in mitochondria, but not at the level of protein stability to any obvious degree (Figure 5B, Supplementary Figure 2). Since transcript-tion and translation are coupled in mitochondria [24-26], these changes probably work together to mediate the increase in OXPHOS complex abundance in tor1Δ strains. Although, the translational control appears to contribute to a greater extent, given only modest changes in mitochondrial transcript levels are observed. However, the change in mitochondrial transcripts of tor1Δ strains might represent a reduction of glucose repression, which is known to induce mitochondrial transcription [27,28] and mimic the effects of tor1Δ on respiration and CLS based on our previous study [18]. Interesting in this regard is the key role of the Snf1p kinase in the glucose repression phenomenon [29]. Snf1p is the yeast ortholog of mammalian AMP kinase, which negatively regulates mTOR signaling in response to energy charge by activating Tsc2, an inhibitor of mTORC1 [30]. Though a Tsc2 ortholog appears to be absent in yeast, these correlations might suggest an evolutionarily conserved regulatory framework that links glucose metabolism, TOR signaling, mitochondrial gene expression and life span.

Whether the increase in nuclear OXPHOS gene expression is mediated at the transcriptional or post-transcriptional level remains to be determined, as does the identity of the putative TOR-regulated mitochondrial factors that meditate the increase in mitochondrial mRNA transcription/stability and translation of mtDNA-encoded OXPHOS subunits. Certainly, nuclear transcription factors that are known to be downstream of TORC1 [31], involved in nuclear-mitochondrial signaling [32], or in glucose repression of mitochondrial function [33] are obvious candidates to test with regard to the nuclear gene expression response. And, with regard to TOR-dependent factors that regulate mitochondrial gene expression directly, the mitochondrial transcription machinery, mitochondrial ribosomes, or the various general and specific translational activators [5] are likely candidates to consider in future studies. Furthermore, since our results clearly implicate Sch9p as the key mediator of the TORC1-mitochondria-CLS pathway (Figure 6B), searching for mitochondrial substrates of Sch9p as potential downstream targets that execute changes in mitochondrial gene expression and OXPHOS activity would likely be fruitful.

The fact that up-regulation of mitochondrial oxygen consumption [18] and mitochondrial translation (Figure 1) in tor1Δ strains occurs only in log-phase and early stationary phase cultures (and not later in stationary phase) strongly suggests that TOR-dependent mitochondrial changes that occur early are responsible for the life span extension later in stationary phase. The concept of early mitochondrial-related events effecting life span has been promoted by others in aging studies in C. elegans [34,35] and is also consistent with the observation of Piper and colleagues that previous conditioning of yeast to respiratory conditions extends CLS in subsequent cultures [36]. While, at this point, the molecular explanation for this "mitochondrial pre-conditioning" effect is not clear, we consider ROS signaling as one potential model. This idea is attractive because the rate of production of ROS from the mitochondrial electron transport chain is likely an accurate reflection of mitochondrial OXPHOS activity and/or redox status that could be used by cells as a retrograde signal to modulate nutrient-sensing pathways. Although we have not observed a significant change in the steady-state level of superoxide in log-phase tor1Δ cells (data not shown), it is possible that other ROS species may be relevant or that the steady-state measurements are not accurately predicting the rate of mitochondrial ROS production. Alternatively, we observed up-regulation of Yhb1, a nitric oxide detoxifying enzyme in tor1Δ mitochondria, but not Sod2p (data not shown; [18]). These results might suggest a role for NO and/or other reactive nitrogen species as relevant. Interesting in this regard, as is the case in tor1Δ cells (Table 1), Yhb1p localizes to mitochondria under anaerobic condition [37]. This, coupled to our observation that hypoxic conditions bypass the extension of CLS by TOR1 deletion [18] might suggest that reduced TOR signaling and anaerobic conditions share a common route to impact life span that may involve NO metabolism. Future studies along these and related lines, as well as further characterization of TOR-dependent changes in the mitochondrial proteome should be most revealing in terms of understanding how the TOR-mitochondria axis controls aging and deciphering the complex relationships between OXPHOS activity, ROS (and/or other reactive species), nutrient sensing, and life span. This, in turn, may provide new inroads into understanding and perhaps counteracting age-related pathology in humans.

Materials and Methods

Yeast strains. Unless otherwise stated, strains of the DBY2006 (MATa his3-Δ200 leu2-3,-112 ura3-52 trp1- Δ1 ade2-1) background were used exclusively. The GS122 and GS129 strains are derivatives of DBY2006 that have plasmid-borne RPO41 and rpo41-R129D alleles covering a chromosomal disruption of the endogenous RPO41 gene and have been described previously [24]. These strains were used for the experiments presented in Figures 1 and 4. The TOR1 gene was disrupted in these strains as described previously [18]. The SCH9 gene was disrupted using a standard HIS3 knockout cassette [38]. Briefly, the HIS3 in pRS313 was PCR amplified with primers ACCACCGCTATTAGTCAGGACTTATATGCAATGGGCACAACAGGAATAACAAGATTGTACTGAGAGTGCAC (SCH9_LeftDel) and CATCATTGATGTCC TCGTCCCCGTCATCATCGATGACATCTTCGTCTG GACTGTGCGGTATTTCACACCG (SCH9_RightDel).

Gel-purified amplicons were used to transform wild-type and tor1Δ DBY2006. His+ transformants were selected on his- plates and single colonies were picked and verified by PCR. The mitochondrial GFP expressing yeast strains were generated by transforming wild-type DBY2006 and tor1Δ with pYX142-SU9-GFP [39].

Mitochondria purification. Mitochondria were isolated from yeast (from cultures grown to OD₆₀₀=1.0 in selective media) by differential centrifugation followed by sucrose-gradient fractionation as described [40]. For 2D-DIGE, the purity of mitochondrial preparations was checked by western blot analysis with anti-actin (Chemicon, 1:1000), anti-alkaline phosphatase (Molecular Probe 1:1000), anti-Dol-P-Man synthase (Molecular Probes, 1:1000) antibodies to control for contamination of cytoplasm, vacuolar membrane, and ER membrane respectively. Only residual ER contamination was present in the purified mitochondrial preparations (data not shown).

Mitochondrial translation assay. Unless otherwise stated, mitochondrial translation assays were performed as described [25], except the following: all reactions were carried out at 30⁰C, gradient gels (6-20%, Figures 1 and 4; 15-22.5%, Figure 5) were used to resolve translation products and electrophoresis was conducted at a constant current of 30 mA.

Chronological Life Span Assay. Chronological life span was assayed as described previously [9,18] Unless otherwise stated, viability was determined by staining with 0.4% trypan blue.

Measurement of mtDNA Copy Number. The mtDNA copy number was determined using a quantitative real-time PCR procedure as described previously [41,42].

Northern Analysis. Northern blots were performed as described previously [24,42]. Briefly, 5 μg of total RNA extracted from yeast (cultured to an OD₆₀₀=1) was separated on a 1.5% agarose-formaldehyde gel and then transferred to a nylon membrane by capillary action. Radiolabeled probes were synthesized by PCR with ³²P-dCTP an added to the membranes in rapid-hyb buffer (GE Healthcare) and incubated overnight at 42 ⁰C. The membrane was washed at room temperature with increasing stringency before visualization by auto-radiography as described in the references cited above.

Western Blot Analysis. Western blots of mitochondria and total cell extracts (from cultures at (OD₆₀₀=1) was performed as described previously [18,25]. Proteins were separated on a 10% SDS-PAGE, transferred to a PVDF membrane, and incubated with the indicated primary and HRP-conjugated anti-mouse secondary (Molecular Probes) antibodies as described previously [18, 25]. Anti-Cox4p (MitoSciences) antibody (not used previously) was diluted 1:1000 for incubation with the blocked membrane.

2D-DIGE. Mitochondrial extraction followed steps mentioned in the "mitochondrial extraction" section. 2D-DIGE was conducted by the W.M. Keck Facility of Yale University (http://keck.med.yale.edu/dige/). Briefly, protein samples were prepared by TCA-precipitation of mitochondrial extracts from DBY2006 and tor1Δ. The samples were further cleaned with 2-D Clean-Up Kit (GE Healthcare) and labeled with CyDye DIGE fluors (Amersham). 50 μg of the labeled samples were resolved on a 2D gel (Ettan DIGE system from Amersham). A representative 2D gel and the distribution of up-regulated and down-regulated proteins is shown in Supplementary Figure 3. In this study, 26 proteins that were up-regulated by 2-fold or more were selected for MALDI-MS/MS analysis. We were able to unambiguously identify 11 of these based on multiple peptide matches.

Flow Cytometry. All analysis was performed on a Beckton-Dickenson FACSCalibur. Analysis of yeast ROS using DHE was performed as described previously [9]. For measurement of mitochondrial potential, cells from a growing culture were pelleted by centrifugation and washed with phosphate-buffered saline (PBS). DiOC₆ (Molecular Probes) was diluted to a final concentration of 200 nM in PBS and used to resuspend the cells. The cell suspension was then incubated for 30 minutes at 30 ⁰C, washed twice with PBS, and analyzed by flow cytometry using the FL3 channel without compensation. For measurement of mitochondrial mass, cultured GFP expressing yeast cells were pelleted, washed once and then re-suspended in PBS, and subject to flow cytometry analysis using the FL1 channel without compensation.

Supplementary Materials

Supplementary Figure 1

Supplementary Figure 2

Supplementary Figure 3A

Supplementary Figure 3B