The knowledge on the molecular aspects regulating ageing in eukaryotic organisms has benefitted greatly from studies using the budding yeast Saccharomyces cerevisiae. Indeed, many aspects involved in the control of lifespan appear to be well conserved among species. Of these, the lifespan-extending effects of calorie restriction (CR) and downregulation of nutrient signalling through the target of rapamycin (TOR) pathway are prime examples. Here, we present an overview on the molecular mechanisms by which these interventions mediate lifespan extension in yeast. Several models have been proposed in the literature, which should be seen as complementary, instead of contradictory. Results indicate that CR mediates a large amount of its effect by downregulating signalling through the TORC1–Sch9 branch. In addition, we note that Sch9 is more than solely a downstream effector of TORC1, and documented connections with sphingolipid metabolism may be particularly interesting for future research on ageing mechanisms. As Sch9 comprises the yeast orthologue of the mammalian PKB/Akt and S6K1 kinases, future studies in yeast may continue to serve as an attractive model to elucidate conserved mechanisms involved in ageing and age-related diseases in humans.
The budding yeast Saccharomyces cerevisiae has for several decades been used as a model system to unravel basic cellular mechanisms operating in eukaryotic cells. Indeed, numerous aspects of processes such as cell division and metabolism have been shown to be highly conserved between yeast and higher eukaryotes. As such, also the intriguing question on lifespan regulation is studied in yeast to identify genetic, cellular and environmental cues that are crucial in regulating the survival (and programmed cell death) of yeast cells. Two distinct ageing paradigms have been described in yeast: replicative and chronological ageing. When nutrients are readily available in the environment, yeast reproduce by an asymmetric cell division process known as budding, yielding a small daughter cell and a larger mother cell. The mother cells do not divide indefinitely, however, and the replicative lifespan (RLS) indicates the number of daughter cells a single mother cell can produce (Mortimer & Johnston, 1959; Steinkraus et al., 2008). When nutrients in the environment become limiting, cell growth and division ceases and yeast cells enter into the stationary (nongrowing) phase, wherein they remain viable for a period of time, which is annotated as the chronological lifespan (CLS; Muller et al., 1980; Fabrizio & Longo, 2003). Although it was long thought that upon nutrient starvation, yeast cells uniformly exit the cell cycle in the G1 phase to enter into a quiescent (‘G0’) state, it has been proven that these cells actually constitute a mixture of both quiescent and nonquiescent cells (Allen et al., 2006; Aragon et al., 2008; Davidson et al., 2011). These cells have significantly distinct properties, impacting on their survival in the stationary phase, as will be discussed further on (see section Improper G1 arrest induces DNA replication stress). The ability to study both RLS and CLS in yeast gives this organism the advantage of being a model for the ageing process for both dividing and nondividing cell types, exemplified by stem cells and terminally differentiated cells such as neurons, respectively, in mammals.
By studying yeast ageing, scientists hope to reveal principles of cellular ageing that might be conserved in higher organisms. Yeast studies revealed that such conserved mechanisms indeed exist. For example, one of the most universally effective interventions to extend lifespan across a wide range of eukaryotic species is the principle calorie restriction (CR, also called dietary restriction, DR), which encompasses a reduction in nutrient intake without malnutrition. The effects of CR appear to be largely mediated by two highly conserved nutrient signalling pathways, that is, the Ras/cAMP/PKA pathway and the target of rapamycin (TOR) pathway (Longo & Fabrizio, 2002; Fontana et al., 2010). In this review, we describe some of the molecular events linking the yeast TOR complex 1 (TORC1)–Sch9 nutrient signalling branch to downstream effectors mediating lifespan in yeast. Although TORC1 and Sch9 impact on both RLS and CLS in yeast, we focus primarily on lifespan regulation during chronological ageing. Thus, when we discuss the effects on ‘lifespan’ in this review, it typically means CLS, unless RLS is specifically stated.
TORC1 plays a central role in yeast nutrient signalling
In analogy with signalling by growth hormones and mitogens in mammalian cells, nutrient availability has a crucial impact on the physiology of yeast cells (Smets et al., 2010; De Virgilio, 2011). As such, when yeast cells are starved for an essential nutrient, such as carbon, nitrogen or phosphate, they arrest cell division and enter into the stationary phase. Cells entering the stationary phase are characterized by a global downregulation of transcription and translation, while inducing the expression of a specific set of genes that aid in their survival during nutrient absence, regulating aspects such as reprogramming of metabolism, reserve carbohydrate accumulation, induction of autophagy, cell wall remodelling and general stress resistance.
A central role in the nutrient-induced signalling network affecting a plethora of these features is played by the nutrient-regulated, rapamycin-sensitive kinase complex TORC1, composed of the regulatory subunits Lst8, Kog1 and Tco89, and either the Tor1 or Tor2 kinase (Loewith et al., 2002). As TORC1 activity was shown to decrease in cells experiencing acute starvation for carbon, nitrogen, phosphate or amino acids, it appears that TORC1 is responsive to both the quantity and the quality of multiple nutrients in the environment (Urban et al., 2008; Binda et al., 2009,Fig. 1). Still, only for amino acid stimulation of TORC1 activity have the molecular mechanisms been partly elucidated (a). Here, the involvement of an evolutionary conserved mechanism was demonstrated, in which TORC1 activation by amino acids is mediated by the EGO complex, a specific group of Ras-family GTPases (Kim et al., 2008; Sancak et al., 2008, 2010; Binda et al., 2009; Flinn & Backer, 2010; Kogan et al., 2010). In addition to deprivation for several nutrients, TORC1 activity was reported to also decrease upon the application of numerous noxious stressors, such as acetic acid, high salt, redox stress or high temperatures (Kuranda et al., 2006; Urban et al., 2008; Almeida et al., 2009). The Rho1 GTPase was recently shown to be essential for mediating stress signals to TORC1, by directly binding to the Kog1 subunit and subsequently inactivating TORC1 (Fig. 1b; Yan et al., 2012). Accordingly, TORC1 activity seems to reflect an optimal growth condition, in terms of both nutrient availability and the absence of additional stress factors.
Although the effects of TORC1 signalling on yeast physiology are ample, only a few direct targets/effectors of this kinase have been identified. Of these, two key branches appear to relay the bulk of TORC1-mediated effects (Loewith & Hall, 2011). In the first branch, TORC1 phosphorylates, and thereby inhibits, Tap42, a regulatory protein controlling the activity and/or substrate specificity of type 2A (Pph21, Pph22 and Pph3) and type 2A-like (Sit4, Ppg1) protein phosphatases (Di Como & Arndt, 1996; Jiang & Broach, 1999; Duvel et al., 2003; Zheng & Jiang, 2005; Yan et al., 2006, 2012). The second branch is comprised of the AGC protein kinase Sch9. Phosphorylation of Sch9 by TORC1 occurs on multiple serine and threonine residues in its C-terminal region and was shown to be required for Sch9 to be active (Urban et al., 2008). Under the optimal growth conditions, TORC1 is active, resulting in Tap42 inhibition and Sch9 activation (Fig. 1a). Upon nutrient deprivation or the presence of noxious stresses, the Rho1 GTPase outcompetes Tap42 for binding to the Kog1 subunit of the TORC1 complex, thereby releasing Tap42 from the TORC1 complex, resulting in its activation (Fig. 1b; Yan et al., 2012). This Rho1 binding to TORC1 was also required for the dephosphorylation of Sch9 under the same conditions, although none of the Tap42–phosphatases were responsible for this dephosphorylation. Therefore, Rho1 impacts on the two key branches of TORC1 signalling by inducing both Sch9 dephosphorylation and the release of Tap42–phosphatases. For a more detailed review on all known TORC1-dependent signalling events in yeast, we refer to several recent excellent reviews (Smets et al., 2010; De Virgilio, 2011; Loewith & Hall, 2011). Here, however, we will focus on the Sch9 signalling branch of TORC1, as it has been shown that the effects of CR or TOR1 deletion on yeast longevity are mimicked by a deletion of SCH9 (Wei et al., 2008), indicating that Sch9 constitutes a major effector by which TORC1 signalling impacts on yeast lifespan.
As stated above, under optimal growth conditions, TORC1 is maximally active and as such will promote rapid growth by stimulating protein and ribosome biosynthesis, as well as cell cycle progression (Loewith & Hall, 2011). Sch9 plays a key role in this regulation as inhibition of Sch9 has been shown to lead to increased phosphorylation of the translation initiation factor eIF2α (Urban et al., 2008), which impairs its function (Hinnebusch, 2005). In addition, activated Sch9 phosphorylates and thereby inhibits the action of several transcriptional repressors, including Maf1, Dot6, Tod6 and Stb3, resulting in maximal expression of components required for ribosome biogenesis (Huber et al., 2009, 2011; Lee et al., 2009). Finally, Sch9 has recently been shown to phosphorylate the ubiquitin-conjugating enzyme Cdc34, thereby stimulating cell cycle progression in the presence of nutrients (Cocklin & Goebl, 2011). In addition to the stimulation of growth, TORC1 activity simultaneously suppresses a variety of stress response programs, as their activation is incompatible with rapid growth. Several of these stress responses are regulated via Sch9 and will now be discussed, as they are documented to play a key role in determining yeast survival in the stationary phase.
Coordinated entry into the stationary phase: a crucial role for the Rim15 protein kinase
Carbon source availability has a huge impact on the metabolism of Saccharomyces cerevisiae. When all other essential nutrients are present, yeast will preferably utilize glucose through fermentation, producing ethanol and acetate in the process. When glucose becomes limiting, yeast cells will enter the diauxic shift, during which metabolism is shifted from fermentation to respiration, allowing the usage of ethanol and acetate, which have accumulated during the fermentative growth phase. Finally, when these carbon sources are exhausted during the postdiauxic growth phase, cells will enter the stationary phase due to carbon source starvation (Smets et al., 2010). Besides adapting its metabolism and growth potential, nutrient limitation also induces several phenotypes annotated as stress responses, including the accumulation of the reserve carbohydrates glycogen and trehalose, and the transcriptional induction of a variety of genes known as the environmental stress response (ESR; Gasch & Werner-Washburne, 2002; De Virgilio, 2011). Typical ESR genes encode for proteins involved in chaperoning functions and scavenging oxygen radicals, which will aid in the survival of yeast cells during stressful conditions such as nutrient starvation.
The protein kinase Rim15 has been demonstrated to play a crucial role in the proper entry into the stationary phase upon nutrient starvation conditions (Swinnen et al., 2008). Upon nutrient limitation, Rim15 enters the nucleus and stimulates the activity of the transcription factors Msn2, Msn4 and Gis1 (Cameroni et al., 2004; Fig. 2). While Msn2 and Msn4 activate transcription of genes containing STRE (stress-responsive) promoter elements (Boy-Marcotte et al., 1998; Moskvina et al., 1998; Garreau et al., 2000; Gasch et al., 2000), Gis1 drives the expression of PDS (postdiauxic shift) element-controlled genes (Pedruzzi et al., 2000; Roosen et al., 2005). How Rim15 might stimulate Msn2/4 function still remains elusive. For Gis1, however, recent evidence pointed out that Rim15 stimulates its activity by phosphorylating the paralogues endosulphine proteins Igo1 and Igo2 (Bontron et al., 2013). Phosphorylated Igo1/2 directly inhibits the Cdc55 protein phosphatase 2A (PP2ACdc55), which preserves Gis1 in a phosphorylated state, promoting its recruitment to and activation of transcription of its target genes (Fig. 3). Intriguingly, this function of Rim15 seems to be conserved among eukaryotes, as the Xenopus Rim15 orthologous Greatwall kinase (Gwl) phosphorylates endosulphines (Ensa and Arpp19) to inhibit the activity of the protein phosphatase PP2A-B55δ (Gharbi-Ayachi et al., 2010; Mochida et al., 2010). Furthermore, phosphorylated Igo1/2 plays an additional role in transcriptional activation, as they have been shown to associate with the mRNA decapping activator Dhh1, thereby protecting the mRNAs from degradation by the 5'-3' mRNA decay pathway (Talarek et al., 2008).
Although originally studied for its role in transcriptional programming during the diauxic shift, it is now known that Rim15 activity is modulated by multiple nutrient signalling pathways (Swinnen et al., 2008). Today, at least 4 nutrient signalling protein kinases have been implicated in regulating Rim15 function, by affecting either its intracellular location or its kinase activity (Fig. 2). As such, it has become clear that Rim15 constitutes a major integration point for nutrient signalling, ensuring proper entry into the stationary phase upon depletion of essential nutrients. Interestingly, Rim15 and its downstream effectors Msn2/4 have been implicated in the regulation of autophagy by Sch9, although the mechanisms behind this control are not fully understood (Yorimitsu et al., 2007). Induction of autophagy depends on the activation of the Atg1-Atg13 protein kinase complex, whose activity is inhibited through phosphorylation (Cebollero & Reggiori, 2009). TORC1 also controls autophagy induction, independent of Sch9, by directly phosphorylating the regulator Atg13 (Kamada et al., 2000; Kamada, 2010). Similar to Rim15, the transcription factor Msn2 is known to shuttle between the cytoplasm and the nucleus, and TORC1 has been shown to inhibit nuclear translocation of Msn2, independent of both Rim15 and Sch9 (Gorner et al., 1998; Beck & Hall, 1999; Roosen et al., 2005). Gis1, on the other hand, constitutively localizes to the nucleus, independent of nutrient conditions (Roosen et al., 2005). Although Gis1 function depends largely on Rim15, additional evidence suggests Rim15-independent regulation of Gis1 via Sch9 through yet-unidentified mechanisms (Roosen et al., 2005; Zhang & Oliver, 2010; Pan et al., 2011). Interestingly, the transcription factor Rph1 was recently implicated in the coordination of glycerol and acetate metabolism in glucose-depleted cells, together with Gis1 (Orzechowski Westholm et al., 2012). Expression of RPH1 is induced upon glucose depletion (Gasch et al., 2000), and Rph1 was shown to be phosphorylated upon rapamycin treatment (Huber et al., 2009), indicating that Rph1 function is modulated by nutrient conditions. Still, the exact molecular mechanisms, for example, a possible involvement of Rim15, are unknown at present. Finally, we note that the FOXO family transcription factor Hcm1 was demonstrated to be required for proper preparation for the diauxic shift, by promoting mitochondrial biogenesis and stress resistance in yeast (Rodriguez-Colman et al., 2010, 2013). Regulation of Hcm1 function depends on the Sch9 protein kinase, as phosphorylation of Hcm1 by Sch9 inhibits its nuclear import (Rodriguez-Colman et al., 2013). Intriguingly, transcriptional profiling indicated that Hcm1 upregulates transcription of RIM15, as well as MSN2/4 and GIS1. As Hcm1 activation already occurs well before the glucose is completely consumed, this transcription factor may be an early response necessary to prime yeast cells for the adverse situation that lies ahead. It is thus clear that TORC1–Sch9 signalling impinges on a multitude of transcription factors to coordinate a proper entry into the stationary phase (Fig. 3).
Effects of tor1Δ and sch9Δ on longevity mimic calorie restriction
In yeast, CR, imposed by decreasing the initial glucose concentration in the medium, is known to increase survival during stationary phase (Smith et al., 2007). Typically, CR is obtained by limiting the glucose concentration in the media from the standard 2% to either 0.5% or even 0.05%. As an alternative, yeast cells that have entered the stationary phase can be transferred to sterile water, which is viewed as an extreme form of CR (Piper, 2012). Similarly, one of the well-known phenotypes of tor1Δ or sch9Δ strains is their increased CLS, compared with wild-type cells (Fabrizio et al., 2001; Powers et al., 2006). Decreasing signalling through the Ras/cAMP/PKA pathway also significantly extends CLS, as seen in strains containing a mutated adenylate cyclase [i.e. the enzyme producing cyclic AMP (cAMP) in yeast], or deletion of RAS2 (Longo, 1999; Fabrizio et al., 2001). On the other hand, overstimulating these nutrient signalling pathways, for example, by overexpressing SCH9 or expressing a constitutively active Ras2Val19 variant, leads to a rapid loss of viability in stationary phase (Wei et al., 2009). These observations have led to the idea that the effects of CR on CLS of yeast are mediated, at least in a large part, by downregulating nutrient signalling through the TORC1–Sch9 and Ras/cAMP/PKA pathways. Intriguingly, CR appears to be one of the most universally effective interventions to extend lifespan across a wide variety of eukaryotic species. Furthermore, the TORC1 and protein kinase A (PKA) pathways are also highly conserved, and studies on higher eukaryotes indicate that downregulating signalling through these pathways has similar lifespan-extending effects. For instance, studies in mice indicate that an increased lifespan is observed after treating with rapamycin (Harrison et al., 2009) and that mice lacking S6K1, the mammalian orthologue of Sch9 (Powers, 2007), are long-lived and protected against several age-dependent defects (Selman et al., 2009). Also, mice lacking adenylate cyclase 5 or with reduced activity of PKA are long-lived and display protection against age-dependent diseases or damage/loss of function (Yan et al., 2007; Enns et al., 2009).
Due to the highly conserved nature of lifespan extension by CR or modulating signalling through the TORC1 and Ras/cAMP/PKA pathways, focus is now being put in finding the molecular mechanism(s) responsible for this phenomenon. In yeast, several models have been proposed, explaining why CLS is increased after CR, or in strains deleted for TOR1 or SCH9. Below we describe the most common models described for CR and/or TORC1–Sch9 effects.
Molecular mechanisms involved in lifespan extension by calorie restriction or downregulation of TORC1–Sch9 signalling
Upregulation of the yeast stress response
As cells progress through the diauxic shift, metabolism shifts from fermentation to respiration, which implicates that energy metabolism will depend on mitochondrial function. Besides supplying the cell with ATP, mitochondrial activity is also known to generate hazardous reactive oxygen species (ROS), which have been implicated in the process of ageing for several decades now (Harman, 1956, 1972). During the diauxic shift, yeast cells also induce the ESR (Fig. 3), resulting in the generally increased resistance to stresses exerted by stationary phase yeast cells (Gasch & Werner-Washburne, 2002). As discussed above, induction of the ESR depends on the Rim15 signalling module, and as such, extended CLS of both tor1Δ and sch9Δ strains depends to a large extent on the presence of a functional Rim15 protein kinase, as well as on its downstream effectors Msn2/4 and Gis1 (Wei et al., 2008). As the Hcm1 transcription factor was shown to induce the components of the Rim15 module, it is likely that Hcm1 also contributes to lifespan extension (Rodriguez-Colman et al., 2013), although this remains to be confirmed. In line with increased ROS production as cells enter stationary phase, a part of the ESR includes genes encoding enzymes functioning to detoxify several ROS, including catalases (detoxifying hydrogen peroxide) and superoxide dismutases (eliminating superoxide anions). Especially, superoxide anions seem to be detrimental to survival, and CR was shown to decrease these ROS in stationary phase yeast cells (Weinberger et al., 2010). Accordingly, the superoxide dismutase enzymes were shown to be of vital importance for stationary phase survival (Longo et al., 1996, 1999), and it has been reported that CLS extension of sch9Δ cells requires the mitochondrial superoxide dismutase Sod2 (Fabrizio et al., 2003), which protects several crucial enzymatic functions, such as aconitase and succinate dehydrogenase activities (Longo, 1999). These results indicate that (superoxide-induced) oxidative stress may represent one of the main physiological stress factors operating in stationary phase yeast cells and that the increased CLS of tor1Δ and sch9Δ strains largely depends on its increased resistance to this oxidative stress on account of the Rim15 module (Fig. 4).
ROS, Mitochondrial respiration and survival – the hormesis theory
As a major source of cellular ROS, mitochondria have been implicated in ageing and lifespan regulation in eukaryotic organisms. Earlier theories focused mainly on the damaging effects of high ROS levels, as stated in the ‘mitochondrial theory of ageing’ [reviewed in (Breitenbach et al., 2012)]. Simply put, according to this theory, cell ageing and death are primarily caused by damage to cellular constituents, inflicted through ROS, which accumulate over time. When cellular damage surpasses a critical threshold, cells will die by executing a programmed cell death mechanism (Herker et al., 2004; Carmona-Gutierrez et al., 2010). Nowadays, however, ROS are being recognized as signalling entities also, and several studies in yeast have proven the existence of conditions, in which exposure to mild oxidative stress, especially during early phases of growth, induces an adaptive response that protects yeast cells against further damage during later stages. This principle is known as the hormesis theory and seems to apply not only to yeast, but also to higher eukaryotes (Gems & Partridge, 2008; Ristow & Zarse, 2010; Pan, 2011).
The first, clear indication of a ROS-mediated adaptive response mediating yeast lifespan extension came from a study on CR effects on CLS. CR was shown to specifically increase hydrogen peroxide (H2O2) levels in the early stationary phase, leading to an induction of superoxide dismutase activity and thereby resulting in a net increase in CLS under CR conditions (Mesquita et al., 2010; Weinberger et al., 2010).
In an independent series of studies, in search for the mechanism behind the extended CLS for tor1Δ and sch9Δ strains, it was shown that both strains were characterized by an increased respiration, due to increased translation of mtDNA-encoded OXPHOS complex subunits (Bonawitz et al., 2007; Pan & Shadel, 2009), as well as an increased transcriptional activity of the nuclear respiratory regulon, via a Hap4-dependent mechanism (Lavoie & Whiteway, 2008), and possibly also involving the Hcm1 transcription factor (Rodriguez-Colman et al., 2010; Fig. 3). Intriguingly, these effects result in a higher mitochondrial potential and superoxide accumulation in tor1Δ and sch9Δ strains during growth, which gradually decline below wild-type levels in stationary phase (Pan et al., 2011). As such, it was proposed that the elevated superoxide level during growth might elicit an adaptive response (addressed as mitohormesis, due to the involvement of mitochondria), resulting in decreased ROS production and enhanced survival during the subsequent stationary phase. This theory was supported by the fact that overexpression of the mitochondrial superoxide dismutase, Sod2, negates the CLS extension of tor1Δ strains. Furthermore, treating wild-type yeast cells with sublethal concentrations of menadione, which generates mitochondrial superoxide (Castro et al., 2008), or rapamycin, specifically during growth, resulted in lifespan extension accompanied by reduced amounts of superoxide during stationary phase (Pan et al., 2011). Importantly, this study showed that this treatment is only effective in a critical window, that is, during growth, as application of these compounds during stationary phase did not extend CLS.
Both types of studies described above indicate that especially superoxide levels during stationary phase are crucial in the determination of lifespan in yeast. Intriguingly, the described hormesis effects operating under CR, as well as in the tor1Δ and sch9Δ strains, were both shown to be independent of the Rim15 module (Weinberger et al., 2010; Pan et al., 2011). However, in contrast to CR, H2O2 levels in the tor1Δ and sch9Δ strains remained lower than those in the WT strain (Weinberger et al., 2010; Pan et al., 2011), pointing to some mechanistic differences by which CR and downregulation of TORC1–Sch9 generate hormesis effects to extend CLS (Fig. 4). As such, critical questions that remain are the nature and time frame in which different ROS influence hormesis, as well as the ROS-sensing mechanisms and downstream effectors driving this process.
Sch9 promotes superoxide-induced DNA damage
Yeast ageing is known to be associated with a reduction in genomic stability, which can be detected by an increased frequency of different types of DNA mutations, including base substitutions, small DNA insertions/deletions and gross chromosomal rearrangements (Fabrizio et al., 2004; Madia et al., 2007, 2008, 2009). Deletion of SCH9 has been shown to decrease all these types of DNA damages during chronological ageing, in part by the lower superoxide levels present in sch9Δ cells and the increased protection against age-dependent oxidative damage to DNA (Fabrizio et al., 2004; Madia et al., 2009; Fig. 4). Additionally, the error-prone Rev1-Polζ DNA polymerase complex, which is involved in DNA repair by translesion synthesis (TLS), was shown to be inactivated in the absence of Sch9 (Madia et al., 2009). Transcription of REV1 was shown to be decreased in sch9Δ cells, and this repression was dependent on the Gis1 transcription factor (Fig. 3). Alternatively, as REV1 expression was also decreased upon overexpressing either SOD1 or SOD2, superoxide itself may also impinge on REV1 transcription, independent of the Sch9–Rim15–Gis1 signalling module. In addition to its role in regulating TLS, Sch9 was shown to contribute to genomic instability by stimulating error-prone recombination events (Madia et al., 2008).
Improper G1 arrest induces DNA replication stress
Initially, it was generally assumed that most of the yeast cells in stationary phase reside in a uniform quiescent, nondividing state. More recently, however, it was found that yeast cells grown to stationary phase on rich (‘YP’) medium could be separated by density into two major cell fractions, that is, quiescent and nonquiescent cells (Allen et al., 2006; Aragon et al., 2008; Davidson et al., 2011). The quiescent cells resemble G1-arrested cells, in that they are mainly unbudded, exhibit low ROS levels, are resistant to stress, are long-lived and are able to re-enter the cell cycle. The population of nonquiescent cells, on the other hand, is more heterogeneously, containing both budded and unbudded cells, which contain higher ROS levels, are more sensitive to stress and rapidly lose the ability to reproduce. The presence of these two cell population has been recapitulated on synthetic medium (Madia et al., 2009), although here the relative fraction of nonquiescent cells was higher, possibly due to the higher metabolic rate of cells on synthetic medium (Fabrizio et al., 2003) and effects of medium acidification, which is absent on rich medium (Burtner et al., 2009; as will be discussed below in the metabolic programming section). The latter findings are in agreement with observations that yeast cells display a longer CLS on YPD than on synthetic medium, and point to a connection between proper G1 arrest and CLS (Weinberger et al., 2007). In fact, the extension of CLS by either applying CR or deleting SCH9 correlates with a more efficient G1 arrest, as determined by the lower percentage of budded cells and increased G1 DNA content detected by flow cytometry (Weinberger et al., 2007, 2010, 2013). Furthermore, stimulating cell cycle progression by ectopic expression of CLN3, a G1 cyclin downregulated by nutrient starvation, or deleting SIC1, an inhibitor of entry into the S phase, both result in improper G1 arrest in stationary phase, accompanied by a decrease in lifespan (Mendenhall & Hodge, 1998; Weinberger et al., 2007, 2010). As with the hormesis theory discussed above, both Rim15-dependent and Rim15-independent effects are observed for proper G1 arrest in CR cells (Weinberger et al., 2010). As such, apart from its well-known role in inducing stress resistance, molecular connections seem to exist, which link the activity of Rim15 to proper cell cycle arrest upon nutrient starvation. In this respect, the endosulphine–PP2A-B55δ signalling branch links the Rim15 orthologous Greatwall kinase to regulation of mitotic progression in Xenopus (Gharbi-Ayachi et al., 2010; Mochida et al., 2010). Therefore, the Rim15–Igo1/2–PP2ACdc55 signalling branch in yeast might have yet-undiscovered connections to cell cycle regulation. Alternatively, Rim15 may induce cell cycle arrest by the induction of Xbp1, a transcriptional repressor of G1 cyclin (CLN) genes, possibly via the Msn2/4 or Gis1 transcription factors (Mai & Breeden, 1997; Weinberger et al., 2007; Fig. 3). Finally, increased G1 arrest of sch9Δ cells may be mediated independent of Rim15, by a reduced phosphorylation of Cdc34, a ubiquitin-conjugating enzyme involved in cell cycle progression (Cocklin & Goebl, 2011). In summary, proper G1 arrest is clearly beneficial for surviving nutrient depletion, and it has been proposed that superoxide anions play a crucial role in this process, as elevated superoxide levels seem to impair this G1 arrest (Weinberger et al., 2010; Fig. 4).
When cells fail to arrest growth in the G1 phase upon nutrient depletion, subsequent entry into the S phase is known to induce DNA replication stress, that is, inefficient DNA replication leading to DNA damage and cell death. To cope with this type of stress, yeast has evolved several response pathways, and it was shown that loss of components involved in the response to replication stress, such as MEC1 or RAD53, results in further impairment of proper cell cycle arrest and a decline of lifespan (Weinberger et al., 2007). DNA replication stress is thought to be attributable to multiple factors, including altered initiation complexes, reduced dNTP pools and/or reduced expression of DNA replication proteins in cells in stationary phase. Cells experiencing DNA replication stress are also more vulnerable to subsequent DNA damage and cell death, in line with the observation that budded (nonquiescent) cells die faster than unbudded (quiescent) cells in stationary phase (Allen et al., 2006; Weinberger et al., 2007; Burhans & Weinberger, 2009, 2012). As both DNA replication stress and DNA damage are important components of ageing and cancer development in higher eukaryotic organisms, including humans, studies in yeast may serve as a model to elucidate the molecular mechanisms involved in these processes in metazoans (Burhans & Weinberger, 2012).
As entry into stationary phase implicates a major shift in metabolism, one can imagine that this shift will have a major impact on survival of yeast cells. Several studies have focused on the effects of CR and carbon source substitution on CLS, often looking into the effects on mitochondrial properties and ROS produced (Smith et al., 2007; Mesquita et al., 2010; Pan et al., 2011). However, other aspects of metabolic shift largely remain elusive. Work done by Burtner et al. (2009) showed that when lifespan was assessed under standard conditions, that is, on synthetic medium with 2% glucose without subsequent transfer to water, yeast acidifies its environment due to the production of several organic acids. This acidification could be avoided by decreasing the initial concentration of glucose to 0.5% or lower, which limits the amounts of organic acids produced. Interestingly, buffering the external medium of cells grown in 2% glucose at pH 6 was sufficient to extend CLS to a similar extent as cells grown in unbuffered medium at lower glucose conditions. Finally, acetic acid was identified as a cell-extrinsic factor mediating cell death under standard conditions, and lifespan extension in strains such as sch9Δ was mainly attributed to its higher cellular resistance to acetic acid. Others, however, argue that the low pH–acetic acid effects operate by stimulating growth signalling pathways, which is detrimental for survival during stationary phase (Burhans & Weinberger, 2009; Fig. 4). Although the effect of extracellular pH is commonly recognized, other studies have shown that effects of CR or deletion of TOR1 or SCH9 are to a considerable extent independent of extracellular pH and thus represent genuine cell-intrinsic effects on CLS (Mesquita et al., 2010; Weinberger et al., 2010; Pan et al., 2011). Moreover, studies using the sch9Δ strain have indicated significant changes in its metabolic fingerprint, showing an enhanced ethanol catabolism as well as increased glycerol production compared with the wild-type strain (Fabrizio et al., 2005; Wei et al., 2009). Although ethanol is a carbon source that yeast can utilize by mitochondrial respiration, its accumulation in wild-type cells impairs CLS. In contrast, glycerol production and accumulation in the sch9Δ strain was shown to be a major contributing factor to lifespan extension. Transcriptional upregulation of glycerol metabolism was demonstrated not only in the sch9Δ strain, but also in the long-lived tor1Δ and ras2Δ strains, indicating the presence of common downstream effectors controlled by these pathways (Wei et al., 2009). In this respect, the Gis1 transcription factor seemed crucial. As this also implicates Rim15, this model postulates that the latter kinase is involved not only in mediating stress resistance of sch9Δ cells, but also in regulating their metabolism. A possible involvement of the Rph1 transcription factor (Orzechowski Westholm et al., 2012) in directing glycerol metabolism of sch9Δ cells remains to be determined (Fig. 3).
Although informative, the studies mentioned above only cover a limited area of the complete metabolic rearrangements that occur in yeast cells during stationary phase entry. Therefore, a metabolomic approach was employed by Goldberg et al. (2009) to study the effects of CR on yeast physiology during lifespan experiments. Important to note is that this study was performed using rich medium, instead of synthetic medium, which may obscure comparison with data from yeast lifespan assessments in synthetic medium. For instance, in YP medium, effects of acidification by acetic acid proved to be negligible. Intriguingly although, CR enhanced ethanol catabolism and reduced lipid body accumulation, similar to what was seen in sch9Δ cells (Goldberg et al., 2009; Wei et al., 2009). Unfortunately, glycerol levels were not determined, so whether a similar shunt towards glycerol metabolism occurs under CR conditions in rich medium remains to be determined. In general, the metabolomic study indicates that CR impacts on a vast amount of properties, programming the cell's metabolic capacity and organelle organization in a diet-specific fashion, prior to entry into the stationary phase. Much of these data await further elucidation, but it is clear that such metabolomic studies can greatly improve our understanding of the processes that determine yeast cell death.
Autophagy promotes lifespan extension
Autophagy in yeast is a vacuole-mediated degradation system that responds to reductions in available nutrients, especially amino acids (Cebollero & Reggiori, 2009). This system is constitutively active at low levels and induced by nutrient limitation, CR as well as a number of environmental and cellular stresses. Autophagy performs two key functions in eukaryotic cells. First of all, through recycling cellular constituents, it maintains intracellular pools of building blocks, such as amino acids, that are needed for de novo synthesis during periods of nutrient scarcity or stress. Secondly, autophagy also performs a detoxification function by removing molecules and even whole organelles that are functionally impaired by molecular damage, for example, oxidized by ROS, misfolded and/or aggregated. As discussed above, TORC1 and Sch9 independently keep autophagy in check during optimal growth conditions. As such, downregulation of TORC1–Sch9 activity is required for the proper induction of autophagy upon entry into the stationary phase. Intriguingly, TORC1 and Sch9, as well as the amino acid-dependent activator of TORC1, the EGO complex, were all found to be concentrated at the vacuolar membrane, placing this module in an ideal environment to regulate autophagy in response to cytosolic and intravacuolar nutrient levels (Binda et al., 2009; Fig. 1).
Loss of autophagic function has been shown to be detrimental for survival of yeast cells upon nutrient starvation in wild-type cells (Alvers et al., 2009a, 2009b), and autophagy is required for the lifespan-extending effects of both rapamycin treatment and CR (Alvers et al., 2009a, 2009b; Aris et al., 2013). Loss of viability in the absence of autophagy was shown to be accompanied by mitochondrial dysfunction, concomitant with a build-up of high superoxide levels (Suzuki et al., 2011; Aris et al., 2013; Fig. 4). Next to nonselective autophagy, yeast cells can execute several forms of specialized autophagic pathways, including mitophagy, the specific removal of mitochondria to the vacuole (Suzuki, 2013). Intriguingly, specific loss of mitophagy did not lead to a decline in lifespan, indicating that the observed mitochondrial dysfunction is caused by a loss of nonselective autophagy (Alvers et al., 2009a, 2009b; Suzuki et al., 2011; Aris et al., 2013). In addition to CLS reduction by a loss of autophagic function, the administration of spermidine, a natural polyamine that induces autophagy, was shown to increase lifespan in yeast, as well as in a range of other higher eukaryotic organisms (Eisenberg et al., 2009). Hence, the lifespan-extending effect of autophagy seems to operate via a highly conserved mechanism, strengthening the value of using yeast as a model system (Madeo et al., 2010). Intriguingly, lifespan extension by spermidine correlated with a decrease in superoxide production as well as a decrease in budding index, indicating a more efficient G1 arrest (Eisenberg et al., 2009).
Sch9, more than an effector of TORC1
As discussed so far, it is clear that sch9Δ cells mimic to a large extent the effects of CR with respect to lifespan extension. As similar effects are observed when working with the tor1Δ strain, Sch9 seems to regulate longevity mainly by acting as a downstream effector of nutrient-sensitive TORC1. Although this may be true, several studies indicate that Sch9 also acts independently of TORC1 (Yorimitsu et al., 2007; Smets et al., 2008). In addition, Sch9 was shown to be phosphorylated by the Pkh1/2 kinases at its PDK1 site in its activation loop, and this phosphorylation appears indispensable for its function (Roelants et al., 2004; Urban et al., 2008; Voordeckers et al., 2011). In line with the suggested activation of Pkh1/2 by the long-chain base phytosphingosine (PHS), an intermediate in sphingolipid biosynthesis, the phosphorylation and activation of Sch9 by Pkh1 is enhanced in vitro by the addition of PHS (Liu et al., 2005) and lowered in vivo by pharmacological or genetic reduction in sphingolipid synthesis (Huang et al., 2012). Hence, these results indicate that Sch9 is also an effector of sphingolipid signalling (Fig. 5), a process that has also been implicated in the regulation of cell death and survival in both yeast and higher eukaryotes (Dickson, 2010; Van Brocklyn & Williams, 2012). In line with this, downregulating sphingolipid synthesis in yeast results in lifespan extension, which largely depends on the reduction in Sch9 activity (Huang et al., 2012). As a result, Sch9 may function as an integration point of both nutrient- and sphingolipid-derived signals for proper regulation of longevity in yeast. Although these inputs may seem to be separate, recent evidence demonstrates that sphingolipid metabolism is modulated in response to nutrient presence (Klose et al., 2012; Lester et al., 2013; Shimobayashi et al., 2013). Moreover, our own unpublished results indicate different sphingolipid profiles in sch9Δ cells, compared with wild-type cells. Even though the molecular mechanisms linking nutrient status with the regulation of sphingolipid metabolism are for the most part unknown at present, Sch9 may play a key role in this connection.
Although we did not elaborate on replicative ageing in yeast, we note that recently a molecular mechanism has been elucidated by which replicative ageing modulates the activity of Sch9 (Lu et al., 2011). In this study, it was shown that Snf1, the yeast orthologue of the mammalian AMPK kinase, stimulates Sch9 activity by phosphorylating it on residues, different from those which are phosphorylated by TORC1 and Pkh1/2. During replicative ageing, acetylation of Sip2, a regulatory β-subunit of Snf1, was shown to increase. This acetylation enhances interaction of Sip2 with Snf1, which results in a downregulation of Snf1-mediated phosphorylation of Sch9, ultimately leading to a slower growth, but extended replicative lifespan.
As a result, at least three independent phosphorylation events are presently known that induce Sch9 activity. As these upstream kinases regulate Sch9 in response to different stimuli, it is clear that Sch9 comprises a major integration point for the regulation of the lifespan of yeast in response to diverse intrinsic and extrinsic signals (Fig. 5).
Conclusion and perspectives
In this review, we presented evidence for molecular mechanisms by which the TORC1–Sch9 signalling branch in yeast regulates longevity. As many of these mechanisms appear to be mimicked by CR, it is widely accepted that lowering dietary intake increases lifespan, at least in part, by downregulating this nutrient signalling pathway. Although several different models have been proposed by which CR and/or TORC1–Sch9 signalling affect lifespan extension, they should not be considered as mutually exclusive (Fig. 4). Indeed, several common themes seem to emerge from these studies. For example, the detrimental effects of superoxide are observed in multiple models for lifespan regulation. Generation of superoxide (especially in mitochondria) curtails lifespan by inhibiting proper G1 arrest, leading to DNA replication stress and DNA damage. This superoxide-induced stress can be counteracted by increasing stress responses, such as the induction of superoxide dismutases or increased autophagy. In addition, hormesis effects indicated that not all ROS are equivalent in promoting cell death, and different ROS may affect lifespan in a different way, exemplified by comparing the effects of superoxide and H2O2 levels under different conditions. Importantly, many aspects of these models describing the molecular events involved in regulating yeast lifespan seem to be conserved in higher eukaryotes. In our example, it has been established that the PKB/Akt–mTORC1–S6K1 signalling pathway affects lifespan in mammalian cells in similar ways, supporting the value of yeast as a model system to study ageing and disease aspects in humans.
Future work is needed to fill in the gaps of the existing models and reveal molecular connections linking different models together. Examples of aspects currently still unclear include the molecular way by which cells generate and sense different ROS and signal their presence to downstream effectors. In addition, although metabolomic studies have been initiated to explore the effect of CR and/or deletion of SCH9 on whole-cell metabolic rearrangements, follow-up studies are needed to pinpoint regulatory aspects of specific metabolites and/or enzymes involved in their production, which are crucial in lifespan regulation. In a final example, much work is still needed to elucidate the mechanisms linking nutrient signalling to sphingolipid metabolism, both of which are intimately linked to cell survival in yeast and higher eukaryotes. Therefore, much is still expected from yeast, but we are confident that this model will continue to live up to expectations and keep on revealing its secrets of ageing mechanisms to the scientific community.
Research was funded by grants of FWO-Vlaanderen and KULeuven to JW, as well as a Ph.D. grant of the Agency for Innovation by Science and Technology (IWT) to TW and a postdoc grant of FWO-Vlaanderen to ES.