OUP user menu

Mitochondria in ageing: there is metabolism beyond the ROS

Michael Breitenbach, Mark Rinnerthaler, Johannes Hartl, Anna Stincone, Jakob Vowinckel, Hannelore Breitenbach-Koller, Markus Ralser
DOI: http://dx.doi.org/10.1111/1567-1364.12134 198-212 First published online: 1 February 2014


Mitochondria are responsible for a series of metabolic functions. Superoxide leakage from the respiratory chain and the resulting cascade of reactive oxygen species-induced damage, as well as mitochondrial metabolism in programmed cell death, have been intensively studied during ageing in single-cellular and higher organisms. Changes in mitochondrial physiology and metabolism resulting in ROS are thus considered to be hallmarks of ageing. In this review, we address ‘other’ metabolic activities of mitochondria, carbon metabolism (the TCA cycle and related underground metabolism), the synthesis of Fe/S clusters and the metabolic consequences of mitophagy. These important mitochondrial activities are hitherto less well-studied in the context of cellular and organismic ageing. In budding yeast, they strongly influence replicative, chronological and hibernating lifespan, connecting the diverse ageing phenotypes studied in this single-cellular model organism. Moreover, there is evidence that similar processes equally contribute to ageing of higher organisms as well. In this scenario, increasing loss of metabolic integrity would be one driving force that contributes to the ageing process. Understanding mitochondrial metabolism may thus be required for achieving a unifying theory of eukaryotic ageing.

  • iron sulfur cluster
  • mitophagy
  • TCA cycle
  • chronological ageing
  • replicative ageing
  • hibernating ageing


It has been speculated since the 1950s that mitochondria execute a central role in the ageing process, yet only now is this field developing rapidly with many important recent discoveries. The ‘prehistory’ of ‘the mitochondrial theory of ageing’ (here called mTOA for shortness) has been pioneered by the ‘oxygen free radical TOA’ and the ‘somatic mutation TOA’ and therefore, indirectly, by Medawar (Medawar, 1952), Gerschmann (Gerschman et al., 1954), Harman (Harman, 1956), Szilard (Szilard, 1959), and Orgel (Orgel, 1963). However, mitochondria as a source of both reactive oxygen species (ROS; Harman, 1972) and antioxidant processes, exemplified by the biological role of superoxide dismutase, (McCord & Fridovich, 1969) were recognized much later. These canonical functions of mitochondria in ageing have been the subject of comprehensive timely reviews (Barros et al., 2010; Ugidos et al., 2010; Pan, 2011; Schleit et al., 2013) including two book chapters in a volume dedicated to the yeast ageing processes (Breitenbach et al., 2012; Longo & Fabrizio, 2012). Some of the seemingly disparate thoughts about the causes of ageing can now be integrated under a still preliminary but increasingly unifying hypothesis, in which mitochondrial physiology and the ageing-related metabolic changes are correlated with gene mutations that target proteins that are active in stress response and confer lifespan extension.

The last two decades have seen radical changes in the field of ageing and stress response. These include the elucidation of a large number of biochemical pathways for oxidative stress defence, which are in many cases specific for certain kinds of damage and particular subcellular compartments (Thorpe et al., 2004; Aung-Htut et al., 2012), and the role of ROS (superoxide and hydrogen peroxide) as signalling substances for growth and cell differentiation in situations where no particular oxidative stress occurred in the cells. Further, the role of mitochondria in programmed cell death and the ability of mitochondria to adjust to different physiological situations have been explored with respect to morphology, gene expression, protein traffic and cross talk with the nucleo/cytoplasmic system. Moreover, mitochondrial segregation during cell division, including rejuvenation, fission, fusion and mitophagy has been studied and found to change with ageing (Barros et al., 2010; Ugidos et al., 2010; Pan, 2011; Breitenbach et al., 2012; Mao & Klionsky, 2013).

However, mitochondria have, from a physiological perspective, major additional roles in cellular metabolism. They host the respiratory chain, the TCA cycle and the compartment of iron/sulphur cluster assembly in nonplant eukaryotes. Furthermore, mitochondria play a crucial role in the synthesis of lipids and membrane compounds such as ceramides (Aerts et al., 2008; Salminen et al., 2012). Ageing is a consequence of metabolic activities. During the last few years, a remarkable renaissance of metabolism research has been initiated, based on powerful new analytical and mathematical tools. It was demonstrated that metabolic adaptation – and this also applies to ageing – is faster and more dynamic at the enzyme and small metabolite level compared with the well-known transcriptional level, as rapid responses are necessary for survival when environmental situations change (Ralser et al., 2009; Gruning et al., 2010; Buescher et al., 2012). Furthermore, it has become clear that metabolic fluxes are to a large extent regulated on a post-translational and metabolic level, and hence, metabolic activities can only be extracted from transcriptome and proteome studies if combined with metabolite quantification and/or flux analysis (Bakker et al., 1999; Buescher et al., 2012).

Mitochondrial metabolism during yeast ageing

Two ageing models are intensively studied in the yeast, S. cerevisiae: chronological and replicative ageing. The study of both have led to the discovery of ageing factors (i.e. TOR, SCH9/AMPK,SIR2) that are now of prime importance in plants, insects and mammals (Blagosklonny, 2008; Steinberg & Kemp, 2009; Ralser et al., 2012). However, these ageing measures are physiologically different processes, especially with regards to metabolism. Chronological ageing is another term for survival of the stationary phase after nutrients, or space, become limiting and cells no longer divide (Fabrizio & Longo, 2003; Longo & Fabrizio, 2012). The metabolic requirements necessary for this ‘silent’ metabolism are very different from those of growing cells. Survival of the stationary phase has been called a model for the survival of postmitotic cells in the human body, like for instance, neurons in the central nervous system. However, at least a percentage of human postmitotic cells are metabolically very active and are in that respect not similar to stationary yeast cells. In addition, a stationary culture is not 100% silent. Occasionally, cells divide, and there are phases when subpopulations show proliferation, known as adaptive re-growth (Longo & Fabrizio, 2012).

A special case of chronological lifespan represents long-time survival under conditions with very low metabolic activity, as found at low temperature. Mutant colonies of the laboratory yeast strain BY4741 (an S288c descendant) that show a prolonged hibernating lifespan can survive several years at 4 °C. Again, the nature of the gene deletions identified indicates that their contribution to prolonged survival is facilitated by altered metabolism, including mitochondrial metabolism and respiration (Postma et al., 2009).

In contrast, eukaryotic and prokaryotic cells can even in the continuous presence of a nutrient supply undergo only a limited number of cell cycles. In yeast, this phenomenon is used as a measure for lifespan, termed mother cell-specific or replicative ageing (for review: Breitenbach et al., 2004). The old mother cells start to display irregular cell cycles after about 25 cell generations (median value for many laboratory strains). Then, they lose checkpoint control and start cell cycles before the previous cell cycle has been completed (Nestelbacher et al., 1999), leading to genomic instability (Veatch et al., 2009). Replicatively aged mother cells are very large (compared with their newborn daughters) and finally undergo apoptosis (Laun et al., 2001). Comparing about 500 yeast deletion mutations, it turns out that there is relatively little concordance between the replicative, chronological and hibernating lifespans (Laun et al., 2006; Postma et al., 2009), indicating that they represent distinct facets of the ageing process.

Evidence for a decline in mitochondrial integrity during ageing

In the search for a common feature of the different ageing processes, it was found that apoptosis, internal oxidative stress and characteristic changes of mitochondrial morphology are common to old mother cells and chronologically aged cells. In particular, the familiar mitochondrial network of growing yeast cells transforms to many small roundish mitochondria that are stained intensively with dihydrorhodamine or dihydroethidium (Breitenbach et al., 2012), indicating the presence of ROS. This phenotype resembles small (or fragmented), globose (roundish) mitochondria of cells challenged by H2O2. The typical mitochondrial network of a normal yeast cell, a cell exposed to hydrogen peroxide, and one representative aged cell is illustrated, picturing a mitochondria-localized GFP (Westermann & Neupert, 2000) by super-resolution microscopy (Fig. 1). The mitochondrial network of these cells is compared with the localisation of aconitase-GFP fusion protein (Aco1-GFP) within a representative cell of a young and a replicatively aged yeast population, as separated by elutriation centrifugation [reproduced from (Klinger et al., 2010) Fig. 1b]. The mitochondrial ageing phenotypes observed in yeast display a strong similarity to aged mitochondria of higher cells, where mitochondria change their physiology, separate into single globose units and show an overwhelming signature of oxidative stress (Shigenaga et al., 1994). Fragmentation of the mitochondrial network by treatment of yeast cells with 0.5 mM H2O2 is reversible. The yeast cells reform the normal network after the end of oxidative stress and continue growth. This is in contrast to the overall globular pattern of fragmented mitochondria in a terminally old cell, where the globular pattern is locked in, as these cells are unable to dynamically restore the mitochondrial network (Fig. 1a and b).


The mitochondrial network disassembles during ageing and oxidative stress. (a) Mitochondrial morphology in a young, aged and H2O2-exposed yeast cell as obtained by super-resolution fluorescence microscopy. Wild-type yeast (YSBN11) was transformed with an mtGFP expression plasmid (Westermann & Neupert, 2000). Cells were harvested, fixed with para-formaldehyde and mounted in Vectashield on glass cover slips. Images were acquired using an OMX microscope (General Electric), equipped with a 488 nm laser and a 60×1.4 NA oil objective in structured illumination mode with z-sections of 125 nm spacing. Images were processed using SoftWorx (General Electric) and 3D images were reconstructed using Volocity (Perkin Elmer). Upper panel: Exponentially growing (young) cells in G2 in synthetic complete media with 2% glucose as carbon source. Note the elaborate mitochondrial network. Middle panel: Exponentially growing (young) cells were treated with 0.5 mM H2O2 in water for 45 min. Mitochondria are now small and roundish. Lower panel: A representative aged cell from a culture grown to stationary phase is depicted. Note the fragmentation of the mitochondrial network. (b) Aconitase localization in a young and replicatively old yeast cell. Wild-type yeast was transformed with an inducible expression construct encoding aconitase (ACO1-GFP) and analysed by fluorescence microscopy (Zeiss Axioscope) and DIC (adapted from Klinger et al., 2010). Upper panel: Exponentially growing (young) yeast cells in G2 phase, upon growth in synthetic complete media with 2% glucose as carbon source. Note the elaborate mitochondrial network. Length bar: 5 μm, Lower panel: An old mother cell isolated as fraction V by elutriation centrifugation (Jarolim et al., 2004). Note that the old mother cell is larger than a young cell and contains fragmented roundish mitochondria. Length bar: 5 μm [pictures in Fig. 1b are reproduced with permission from the authors (Klinger et al., 2010)].

During ageing in vivo and in vitro, physiological changes resulting from oxidative insults are found in the mitochondria of higher cells as well. Mitochondrial mass in ageing rat brain and liver is severely decreased, as is the activity of mitochondrial enzymes, including respiratory complexes (Navarro & Boveris, 2004). In addition, a deficiency of mitophagy has been described (Hubbard et al., 2012), contributing to overall defects in the decomposition of damaged macromolecules or organelles.

Correlated with these impairments of mitochondrial structure and function, mutations in the mitochondrial DNA clearly increase in old age. This has been studied extensively in vivo in a mouse model that expresses a mitochondrial DNA polymerase with no active proofreading function (Polg mutant mice; Trifunovic et al., 2004; Kujoth et al., 2005), with new results recently summarized by (Bratic & Larsson, 2013). Polg mutant mice show increased mutation frequency in ageing, and the homozygous animals display a very impressive spectrum of premature ageing phenotypes (Trifunovic et al., 2004; Kujoth et al., 2005). However, the nature of the main cause of these phenotypes is questioned by the analysis of heterozygous animals which display a mutation frequency that is orders of magnitude higher than the mitochondrial mutation frequency in aged wild-type mice, yet they do not show signs of premature ageing (Khrapko & Vijg, 2007). Logic thus dictates that the accumulation of mitochondrial mutations during life can be excluded as a main cause of the ageing process of this mammal. Moreover, the second assumption that gene loss caused by deletions in mitochondrial DNA are causative for ageing (Vermulst et al., 2008) remains doubtful as well, as the ‘deleter’ mouse expressing a human dominant mutant form of the mitochondrial DNA helicase twinkle acquires mitochondrial DNA deletions similar in number to the ones found in very old wild-type mice. The mutant mice show a progressive defect in mitochondrial respiration like the human patients from which the mutant form of twinkle was isolated. However, they show no features of a premature ageing syndrome (Tyynismaa et al., 2005; Park & Larsson, 2011). Although other, yet unproven interpretations have been put forward (Ahlqvist et al., 2012; Bratic & Larsson, 2013), this controversy is still unresolved.

The debate about the correct interpretation of this animal model exemplifies a common problem in contemporary ageing research. It is relatively simple to find correlations of certain biochemical markers with age and lifespan, but extremely difficult to obtain proof for causal relationships. One candidate for the causal relationships of mutations inducing mitochondrial dysfunction and ageing are uncharacterized enzymatic functions that in their normal form serve to buffer cellular stress.

Additional support for this ‘buffering’ hypothesis comes from studies of mitochondrial and nuclear mutations that occur in aged wild-type yeast cells. The main information available on this comes from Gottschling and colleagues (McMurray & Gottschling, 2003; Veatch et al., 2009; Lindstrom et al., 2011). Their results reveal that mothers throw off respiratory deficient (‘petite’) mutant daughters. The reason why this effect is seen only at an advanced age is the heteroplasmic (the presence of a mixture of more than one type of an organellar genome) nature of the mitochondrial mutations, meaning that one cell harbours both wild-type and mutant mitochondrial genomes. These must become homoplasmic (when all mitochondrial genomes in a cell contain the same allele) through mitotic segregation. In the homoplasmic cells, only one type of mitochondrial (in this case mutant) genome is present, leading to a phenotype characteristic for a recessive mitochondrial mutation. At the replicative age where this becomes relevant, one can assume that the quality control process of mitophagy is already compromised. Furthermore, mother cells possess an increased frequency of gene conversion at the MET15 locus which is heterozygous in the genetic system BY4743 used to test this hypothesis (McMurray & Gottschling, 2003).

Unspecific biochemical reactivity and metabolite repair

Besides damage to proteins and nucleic acids, a third cause of the loss in mitochondrial integrity could be the accumulation of toxic metabolic intermediates. Several metabolites are prone to oxidation, and other unwanted chemical entities emerge from the side reactions of ‘promiscuous’ enzymes or by spontaneous chemical reactions (Linster et al., 2013). For frequent metabolic errors, evolution invented specific repair strategies. For instance, the accumulation of d-2-hydroxyglutarate, a side product of isocitrate dehydrogenase in the reaction to α-ketoglutarate, is prevented by an enzyme termed d-2-hydroxyglutarate dehydrogenase (Struys et al., 2005). A second example is cis-aconitate, which spontaneously interconverts into the more stable trans-aconitate and inhibits aconitase in the mitochondrial TCA cycle. The inhibition of aconitase, however, is prevented by a trans-aconitate methyltransferase, Tmt1 that converts trans-aconitate to a monomethylester (Cai & Clarke, 1999; Cai et al., 2001). It is less understood how cells deal with rare chemical modifications (Linster et al., 2013; Van Schaftingen et al., 2013). Intuitively, these metabolites lacking a specific clearance mechanism would accumulate over time and interfere with normal metabolic reactions. The contribution of this underground metabolism to the ageing process is without question very interesting, but only barely understood at present.

Biosynthesis, degradation and function of Fe/S proteins as determinants of ageing

In nonplant eukaryotes, mitochondria are the only compartment where iron/sulphur clusters are assembled. These clusters retain iron in its reduced form, and function as cofactors in enzymatic redox reactions. Iron/sulphur proteins are of central importance for many biological processes; most prominent are the redox reactions of the mitochondrial electron transport chain. Because of the toxicity of free ferrous ions as well as sulphide, the synthesis of Fe2-S2 as well as Fe4-S4 clusters is highly regulated. In bacteria, three different and partially overlapping biosynthetic pathways for Fe/S clusters have been found [reviewed in (Fontecave & Ollagnier-de-Choudens, 2008)]. The ‘nitrogen fixation pathway’ occurs in bacteria and Archaea; the ‘sulphur mobilization pathway’ in bacteria and the chloroplasts of green plants, and the third pathway, ‘ISC (iron sulphur cluster) pathway’ occurs in bacteria and mitochondria (Rawat & Stemmler, 2011), and is best described in S. cerevisiae.

Iron/sulphur clusters prevent the presence of soluble, reduced iron within the cell, necessitating a complex transport pathway. In a first step, Fe2+ is transported across the inner mitochondrial membrane, and then, the ferrous iron is bound and oxidized by the iron chaperone, frataxin [Yfh1; reviewed in (Lill, 2009; Philpott, 2012)]. Before the iron/sulphur cluster is assembled at the scaffold proteins Isu1/Isu2, these proteins have to be sulphurylated. The cysteine desulfurase Nfs1 and its activator Isd11 form a persulphide at Nfs1 by abstracting sulphur from the amino acid cysteine that is converted to an alanine residue and afterwards this –SSH group is transferred to Isu1. The Fe/S cluster assembly at Isu1 is strictly dependent on reduction of the persulphide sulphur to sulphide and the electron is transferred from either NADH or NADPH to Isu via the ferredoxin reductase Arh1 and the ferredoxin Yah1 (Lill, 2009; Philpott, 2012). The resulting previously so-called ‘inorganic’ sulphide ions of the clusters (four in the Fe4-S4 and two in the Fe2-S2 clusters) can be liberated as toxic H2S molecules when the clusters are destroyed. For instance, the Fe4-S4 cluster of aconitase is extremely sensitive and is easily destroyed by molecular oxygen (Gardner, 2002).

Replicative ageing is accompanied with a gradual decline of the inner mitochondrial membrane potential, finally leading to dysfunctional mitochondria (Lai et al., 2002). As mitochondrial iron uptake via Mrs3/Mrs4 is strictly dependent on the mitochondrial membrane potential (Muhlenhoff et al., 2003), Fe/S cluster synthesis will be deeply affected in old yeast cells. In fact, it was demonstrated that loss or damage of mitochondrial DNA is associated with a reduced membrane potential and a reduced Fe/S cluster biosynthesis. This in turn has a direct effect on the nuclear genome by promoting its instability and loss of heterozygosity (Veatch et al., 2009), due to a lack of iron sulphur cluster proteins, including DNA polymerase subunits, that are required for DNA replication. The increase in mutations such as base substitutions, deletions, insertions and chromosomal rearrangements observed in chronologically aged yeast cells is thus indirectly related to a decline in mitochondrial membrane potential (Wei et al., 2011).

This effect may be amplified by the Fe/S cluster dependency of several repair enzymes such as the DNA N-glycosylase (Ntg2) that is involved in the base excision repair (Meadows et al., 2003); Rad3, a Fe/S cluster containing 5′ to 3′ DNA helicase, that is participating in the nucleotide excision repair as well as in the repair of double-strand breaks, needs for the fulfilment of its function among others two Fe/S cluster containing proteins: a DNA primase Pri2 (Sung et al., 1987; Holmes & Haber, 1999) and Pol3, the catalytic subunit of the DNA polymerase delta (Chanet & Heude, 2003). Consistently, nonlethal single gene deletions of the aforementioned nuclear Fe/S proteins leads to dramatic decreases in chronological lifespan (Powers et al., 2006; Fabrizio et al., 2010). The Fe/S cluster containing helicase/nuclease Dna2 belongs to the family of RecQ helicases (Hoopes et al., 2002) that includes the yeast Sgs1 helicase (Gangloff et al., 1994). A well-known representative of this family is the human helicase WRN, responsible for the autosomal recessive premature ageing disease, Werner's syndrome (Gray et al., 1997). Similar to this observation, nonlethal dna2 mutations lead to enlarged and fragmented nucleoli and a dramatic reduction (by 86%) in replicative lifespan (Hoopes et al., 2002).

Fe/S clusters and reactive oxygen species

Besides their important biochemical role, Fe/S clusters are – in concert with the mitochondrial electron transport chain – a source of reactive oxygen species. Superoxide produced by the respiratory chain can, if accumulated above normal cellular levels, damage redox sensitive proteins, such as the TCA enzyme aconitase, Aco1, by oxidizing its Fe4-S4 cluster, with the consequence that Fe2+ is released (Gardner, 2002). This release of Fe2+ is an additional source of ROS by producing hydroxyl radicals via the Fenton reaction (Liochev & Fridovich, 1994; Longo et al., 1999; Cantu et al., 2009). Eukaryotic cells have developed a sophisticated but not fully understood mechanism to cope with this problem. Part of the solution appears to be through asymmetric segregation. For instance, the active enzyme Aco1 is primarily passed on to yeast daughter cells, whereas the damaged, inactive protein is retained in the mother cells (Klinger et al., 2010).

In contrast, an increase in mutation frequency, or mutation fixation, during chronological ageing cannot, by definition, be a dominant mechanism, as replicative DNA synthesis does not take place in nondividing cells. However, cells are not fully silent, possess basic metabolism and repair synthesis of DNA does take place (Longo & Fabrizio, 2012). Consequently, mutations have been studied in postmitotic nondividing cells. These include mutations conferring canavanine resistance, reversion of missense and frameshift mutations, gross chromosomal rearrangements, and mutagenesis induced by homologous recombination (Wei et al., 2011).

Mitochondria-centric metabolic processes important for the ageing process

Fe/S cluster biosynthesis is important, but not the only mitochondrial metabolic process associated with ageing. Another obvious mitochondrial metabolic connection to ageing is the TCA cycle, a central metabolic pathway located to this organelle. Evidence that TCA activity changes during ageing originated in the 1960s from plant studies (Laties, 1964). During yeast chronological ageing, succinate dehydrogenase (SDH), enzymes of the glyoxylate cycle (GCL) (isocitrate lyase (ICL) and malate synthase (MLS)), and enzymes of ethanol oxidation (alcohol dehydrogenase (ADH) and acetaldehyde dehydrogenase (ACDH)) are activated, while the classic TCA enzymes citrate synthase (CS), α-ketoglutarate dehydrogenase (KGDH) and malate dehydrogenase (MDH) decrease in activity (Samokhvalov et al., 2004), indicating dynamic changes of the TCA cycle during ageing. These observations are supported by the observation that metabolic enzymes are subject to oxidative modifications while cells age chronologically (Shenton & Grant, 2003; Klinger et al., 2010; Brandes et al., 2013).

A causal connection of mitochondrial carbon metabolism with ageing is indicated by experiments that demonstrate altered ageing phenotypes when TCA enzymes are deleted or overexpressed. To illustrate this connectivity, we obtained phenotypic information from the yeast genome database (SGD; Cherry et al., 1998) and created a mitochondria-centric interaction network, taking into account genetic and physical interactions as annotated in the BioGrid database. This network illustrates that despite replicative, chronological and hibernating lifespan being physiologically distinct phenotypes, the genes coding for mitochondrial proteins that are causative for these phenotypes, are tightly interconnected (Fig. 2). As a caveat, we want to mention that the high-throughput genome data used for the connectivity network in many cases still need to be confirmed by precisely aimed experiments.


Mitochondrial metabolic genes link yeast lifespan phenotypes. Computational interaction network among mitochondrial metabolic genes that influence chronological, replicative and hibernating lifespan. Genes with both metabolic function and altered replicative (gene symbol in pink), chronological (green) or hibernating (blue) lifespan phenotypes, were extracted according to their annotation in the Saccharomyces Genome Database (Cherry et al., 1998) as of August 2013. The genes were annotated according to their primary localization (‘Compartments’) using OrganelleDB (Wiwatwattana et al., 2007). The physical and genetic interaction network was obtained employing BioGRID(Stark et al., 2006), and visualized with Cytoscape (Smoot et al., 2011). References for the individual interactions are given in the Supporting Information Table S1.

Examples of these interactions (Fig. 2) involve several metabolic enzymes. The deletion of citrate synthase CIT1, for instance, was identified to prolong hibernating lifespan (Postma et al., 2009). This enzyme catalyses the condensation of acetyl coenzyme A and oxaloacetate to form citrate and is thus responsible for the most rate limiting step of the TCA cycle (Suissa et al., 1984). CIT1 in turn connects directly to three enzymes that are important for chronological lifespan, DIA4,LPD1 and ACH1. The first of these genes, DIA4, is not a metabolic enzyme in the classic sense but is a (yet putative) mitochondrial seryl-tRNA synthethase. Aminoacyl-tRNA synthethases attach amino acids specifically to cognate tRNAs and are therefore directly or indirectly implicated in the regulation of amino acid homoeostasis and translation (Laxman et al., 2013).

The second enzyme, LPD1, has a central function in connecting glycolysis and the TCA cycle in the biosynthesis of glycine. It encodes for dihydrolipoamide dehydrogenase, located in the mitochondrial matrix as a component of the glycine decarboxylase complex (GDC), 2-oxoglutarate dehydrogenase and pyruvate dehydrogenase (PDH) complexes. Mutations in LPD1 abolish activity of these complexes (Sinclair & Dawes, 1995; Pronk et al., 1996; Dickinson et al., 1997; Zaman et al., 1999). Evidence for the importance of LPD1 in ageing comes from studies of LAT1, coding for dihydrolipoamide acetyltransferase, the (E2) component of the PDH complex. Deleting the LAT1 gene abolished a chronological lifespan extension induced by caloric restriction, while overexpression of this protein prolonged chronological lifespan, which was nonadditive with caloric restriction (Easlon et al., 2007).

The third gene, ACH1 possesses a CoA-transferase activity and catalyses the CoASH transfer from succinyl-CoA to acetate generating acetyl-CoA (Lee et al., 1990). Cells lacking this gene accumulate high amounts of extracellular acetic acid, and during chronological ageing accumulate reactive oxygen species, obtain mitochondrial damage and exhibit an early onset of apoptosis (Orlandi et al., 2012). Interestingly, this gene is connected to LSC1, coding for the alpha subunit of succinyl-CoA ligase, the enzyme that catalyses the nucleotide-dependent conversion of succinyl-CoA to succinate. Although the chronological lifespan of the LSC1 deletion mutant is very short (Powers et al., 2006; Fabrizio et al., 2010), deletion of this gene extends the hibernating lifespan phenotype (Postma et al., 2009). This discrepancy may be explained by the physiological difference between chronological and hibernating lifespan. These examples illustrate the close, yet often opposing, connection between chronological and hibernating lifespan phenotypes.

In sum, the three enzymes discussed in exemplary form here represent in paradigmatic form the importance of metabolic flux in the ageing process. They are either constituents of the TCA cycle, or feed into or withdraw from it, and each one is controlled by oxidative stress.

Regarding mitochondrial metabolism, replicative lifespan appears to be metabolically more distinct from the chronological and hibernating lifespans, yet connected through peroxiredoxin and thioredoxin systems, which operate under conditions of more continuous oxidative stress. ROS and free-radical theory of ageing are not the focus of this review; however, metabolic redox reactions are among the first processes affected by an excess of oxidizing over reducing molecules. Special attention has to be given to the mitochondrial peroxiredoxin Prx1, whose activity is strongly associated with control of replicative lifespan (Unlu & Koc, 2007). In a similar vein, it was found that a gain of function allele of the cytoplasmic peroxiredoxin, Tsa1, causes premature replicative ageing (Timmermann et al., 2010). On the other hand, the presence of Tsa1 has been shown to be required for the extension of replicative lifespan by caloric restriction (Molin et al., 2011).

In summary, there is convincing, yet mostly correlative, evidence that the TCA cycle plays an important role during ageing. Remarkably, the TCA cycle seems to connect between the three yeast ageing phenotypes, although chronological and hibernating lifespan appear more extensively integrated than replicative ageing. However, the TCA cycle has been less studied in ageing than other metabolic processes, and much work remains to be done to clarify its relationship to lifespan.

Involvement of mitophagy in the yeast ageing processes

Mitophagy, the specific degradation of parts of the mitochondrial network through the process of autophagy, serves a number of distinct physiological needs: (1) adaptation to nutrients and carbon source such as the shift from glycerol to glucose that results in the degradation of many mitochondria, (2) adaptation to starvation in stationary phase, which includes the degradation of mitochondria to gain an extra supply of nutrients for survival of the stationary phase (Tal et al., 2007; Bhatia-Kissova & Camougrand, 2010) and (3) finally, as a form of mitochondrial quality control resulting in the removal of toxic waste and damaged macromolecules.

Autophagy, in general, and mitophagy, in particular, are needed for the rejuvenation process of daughter cells in cell divisions of replicatively aged mother cells (discussed below). The same mechanism that guarantees rejuvenation of the daughter cells and therefore survival of the species leads to deposition of damaged molecules in the ageing mother cells. Surprisingly, little is presently known about the exact chemical nature of the damaged macromolecules and organelles that exist in aged cells. Protein carbonyls (Aguilaniu et al., 2003) are one form of oxidation products accessible for study, as an easy method for their detection is available (Goto et al., 1999; Beal, 2002; Levine, 2002). Other products of cellular decomposition, however, maybe of equal importance. This includes already discussed substances that originate as side reactions from metabolism(D'Ari & Casadesus, 1998; Linster et al., 2013).

The mechanism and physiological significance of mitophagy has been elucidated during recent years in considerable detail [reviewed in (Bhatia-Kissova & Camougrand, 2010; Hirota et al., 2012; Novak, 2012; Palikaras & Tavernarakis, 2012)]. The pathway of mitophagy consists of a mitochondrial specific part (to be described below) that leads to engulfment of mitochondria in the phagosome. This part of the pathway is dependent on only a few specific proteins, while the ‘unspecific’ part of the pathway is to a large degree identical with macroautophagy, or other known specialized forms of autophagy (Bhatia-Kissova & Camougrand, 2010).

Yeast has been a leading model organism in deciphering autophagy and mitophagy. Original nomenclature named the involved genes either aut- (Thumm et al., 1994), apg- (Tsukada & Ohsumi, 1993), or cvt-mutants (Harding et al., 1995), but the gene nomenclature has eventually been unified to ‘ATG’ and involves 33 genes so far (Klionsky et al., 2003).

Mitophagy commences with recruitment of a special surface marker, Atg32, whose translocation is most probably induced upon a critical loss of mitochondrial membrane potential (Priault et al., 2005; Kondo-Okamoto et al., 2012). A decline of this potential in ageing is observed, as well as damage of membrane compounds (Dmitriev & Titov, 2010; Paradies et al., 2011). In parallel, loss of mitochondrial metabolic capacity could also be triggered by opening of the mitochondrial permeability transition pore (Rodriguez-Enriquez et al., 2004), thus creating a link between mitophagy and apoptosis, which occurs in replicatively as well as in chronologically aged yeast cells and is their main cause of death [reviewed in (Breitenbach et al., 2012; Laun et al., 2012)]. It follows that mitophagy has pro-survival as well as a pro-death functions (Abeliovich, 2007; Tal et al., 2007) depending on the physiological situation. Here, we focus on the potential survival-promoting function of yeast mitophagy, which appears weakened in old mother cells.

Atg32 possesses close homologues among fungi but is weakly conserved in other eukaryotes (Kondo-Okamoto et al., 2012). However, Atg32 is not the only mitochondrial protein involved in leading mitochondria into mitophagy. This process also involves Atg33, Aup1, Fzo1 and Por1 (a synonym of yeast Vdac1), and potentially Uth1 (Kissova et al., 2004). These proteins possess a partially redundant function. Deletion of any one of them still allows mitophagy to occur (Kissova et al., 2004; Bhatia-Kissova & Camougrand, 2010). The situation is complicated by the fact that these ‘entry’ proteins are needed to different degrees depending on the subtype of mitophagy.

The transition of mitochondria in yeast cells during stationary phase and mother cell-specific ageing into a large number of small roundish, separated, mitochondria (Klinger et al., 2010), suggests that these small mitochondria may be more accessible for the phagosome (Mao & Klionsky, 2013). Indeed, in other situations where mitophagy is needed, the mitochondrial network disassembles in a similar fashion, and the fragments of the network are subsequently enclosed in the phagosomes (Twig & Shirihai, 2011). The disassembly procedure depends on the mitochondrial fission machinery. Staining with redox-sensitive dyes and with mitochondria-specific fluorescent protein markers (ro-GFP) indicates that parts of the mitochondrial network are slightly depolarized without compromising the rest of the network. The depolarized regions are separated by the fission process and incorporated into phagosomes (Twig & Shirihai, 2011; Mao & Klionsky, 2013).

In contrast to yeast, mammalian cells create the mitophagic ‘eat me’ signal (Vernon & Tang, 2013) in a different way. In response to mitochondrial damage, the E3 ubiquitin ligase parkin is recruited to the mitochondrial surface and phosphorylated by the mitochondrial outer membrane protein PINK1. Activated Parkin then ubiquitinates mitochondrial outer membrane proteins such as the mitofusins Mfn1/2 and VDAC1 (Rodriguez-Enriquez et al., 2004; Novak, 2012). As a consequence, mitofusins are degraded and the isolated mitochondrion cannot fuse again with the mitochondrial network, entering a one way street to mitophagy. Here, the ubiquitinated hVDAC1 serves as an anchor for the p62 module which firmly links the mitochondrion to the phagosome via LC3 (the human homologue of Atg8; Rodriguez-Enriquez et al., 2004; Novak, 2012). Several other mitochondrial outer membrane proteins are also ubiquitinated at this stage (Tom70, 40 and 20; Miro 1 and 2; Yoshii et al., 2011). Remarkably, the involvement of an ubiquitinating enzyme in mitophagy provides testimony for the crosstalk between the proteasome and the autophagic pathway in the decomposition of cellular compounds.

In human cells, mitophagy can be induced by iron-chelation with drugs (deferiprone) by a mechanism independent of PINK1 and Parkin, again emphasizing the connection between iron metabolism and mitophagy (Allen et al., 2013). Mutations in both PINK1 and Parkin can lead to the devastating neurodegenerative Parkinson's disease, a typical disease of old age, demonstrating once more the close relationship of mitophagy with ageing (Hauser & Hastings, 2013).

Once mitochondria are channelled to the autophagosome the mitochondria-specific part of mitophagy concludes. The continuing ‘macro autophagy’ pathway ultimately leads to enclosure of the mitochondrion in the lytic compartment (the yeast vacuole) where not only proteins but also lipids and other biomolecules are degraded.

Mitophagy throughout the stationary phase was studied in detail and serves the purpose of supplying cells with the nutrients necessary for restructuring their architecture (for instance, the structure of the cell wall) and metabolically adapting to survival in times of starvation (Kissova et al., 2004; Tal et al., 2007). The ability to do so was crucial during evolution, as microbial cells in the wild are frequently confronted with a scarcity of nutrients. As nutrients are limited during chronological ageing, it is suspected that mutations compromising mitophagy in stationary phase should shorten the chronological lifespan. This prediction is supported by existing whole genome high-throughput data (Powers et al., 2006; Fabrizio et al., 2010). In both databases, deletions of all but one (UTH1) of the proteins that have been identified so far as being directly or indirectly specific for mitophagy (Atg32, Atg33, Uth1, Aup1, Fzo1, Por1), are short-lived in stationary phase. The chronological lifespan of the corresponding deletion mutants was significantly shorter than wild type, although these mutants are respiration-active. In addition, the two interactors Atg11 and Atg8 (Palikaras & Tavernarakis, 2012) of Atg32 in mitophagy, together with more than 30 ATG genes (Nakatogawa et al., 2009) are generally needed for macroautophagy. They are expected to show a strong negative effect on chronological lifespan when deleted, which is confirmed by published data (Powers et al., ; Fabrizio et al., 2010). The reason for this strong effect seems to be the necessity to also degrade nonmitochondrial cellular components by autophagy to survive starvation.

Interestingly, the known pathways of regulating mitophagy are closely interconnected with metabolic regulation. The increase in both the chronological and replicative lifespan which is observed when treating yeast cells with low doses of rapamycin (Bjedov & Partridge, 2011) is dependent on rapamycin-mediated removal of the inhibitory action of yeast Tor1 on autophagy and mitophagy (Rubinsztein et al., 2011). Likewise, activity of the RAS/PKA system of yeast inhibits mitophagy (Budovskaya et al., 2004). If the up-regulation of mitophagy observed in wild-type cells upon reaching diauxie and the stationary phase is prevented by genetically activating RAS2, or inactivating its interacting genes such as WHI2, survival of the stationary phase and chronological lifespan are compromised leading to cell death (Mendl et al., 2011). Hence, the metabolic activity and the ability to adapt to nutrient starvation seem to be a major link in connecting mitophagy with ageing.

Role of UTH1 in the two ageing processes

The uth1Δ deletion mutant was shown to be chronologically long-lived (Kissova et al., 2004). The tentative explanation given is that Uth1, a mitochondrial outer membrane protein, has no influence on autophagy and only a partial influence on mitophagy.

UTH1 was studied in detail with respect to replicative ageing as well. Like the chronological lifespan (Kissova et al., 2004), the replicative lifespan of uth1Δ yeast is increased significantly (Kennedy et al., 1995). The role of Uth1 in the ageing processes is still not fully understood and somewhat controversial. The reason for this may be the unusually large number of biochemical functions that has been attributed to this SUN-family mitochondrial protein (Camougrand et al., 2004).

Role of polyamine metabolites spermine and spermidine

Another, presumably critical, but yet to be fully elucidated role in the process of autophagy is attributed to the polyamine metabolites spermine and spermidine. These metabolites are considered to be free-radical scavengers and are highly concentrated within cells (Lovaas & Carlin, 1991). During ageing, they decline in concentration as observed in various organisms (Minois et al., 2011). Vice versa, spermine/spermidine supplementation extends lifespan in laboratory models, including yeast replicative ageing (Eisenberg et al., 2009; Morselli et al., 2009), and restores age-induced memory impairment in drosophila (Gupta et al., 2013). Both, lifespan extension and memory improvement correlate with the activation of autophagy (Eisenberg et al., 2009; Gupta et al., 2013), and as spermine and spermidine are time-keepers of the stress response (Kruger et al., 2013), their function appears centre stage for age-related autophagy.

Apparently, the multiple (but closely related) forms of autophagy, including mitophagy, cannot completely remove the unwanted and detrimental components of aged yeast mother cells. Therefore, another ingenious invention serves the purpose of quality control: asymmetric segregation of damaged material, including mitochondria (Aguilaniu et al., 2003; Klinger et al., 2010). Accordingly, the aged mother cell retains damaged substances upon cell division, which protects the young daughter cell from age-dependent damage. This asymmetric segregation is in our view a general mechanism of living cells, allowing a cellular ‘rejuvenation’ process occurring at least once per life cycle, that is essential for survival of a species.


Mitochondria are intensively studied in regard to ageing, but most research has focused on mitochondrial ATP production, ROS leakage from the respiratory chain, and age-dependent induction of apoptosis. Here, we discussed three central functions of mitochondrial metabolism that may be equally important: (1) The biosynthesis of Fe/S clusters which is an essential task of mitochondria in eukaryotes, (2) mitochondrial carbon metabolism (TCA cycle), a central source of reducing equivalents and biosynthethic intermediates for the cell and (3) the process of mitophagy, which disassembles the mitochondrial network as result of stress and ageing and helps to maintain the energy state during starvation and to decompose damaged molecules in old mother cells or stationary cell cultures (Fig. 3).


Schematic overview of carbon metabolism, Fe/S cluster biosynthesis and mitophagy in ageing. A sufficient Δψ across the inner mitochondrial membrane is necessary for protein import into mitochondria and for the biosynthesis of Fe/S clusters (ISC) and is essential for survival of the cell. Only one of the TCA cycle enzymes, aconitase, contains a structural (non-redox active) Fe/S cluster rendering it sensitive to oxidants. ATG32 is located on the surface of damaged mitochondria and interacts with ATG8 on the preautophagosome. ATG8 and ATG32, yeast autophagy genes; ISC, iron sulfur cluster; TCA, tricarboxylic acid cycle; Δψ, electrochemical potential gradient across the inner mitochondrial membrane.

Importantly, all three processes are highly dynamic metabolic processes and their interconnectivity may buffer adverse cellular conditions, either caused by mutation in one of their components or caused by intra- or extracellular toxic challenge. Evidence suggests that toxic waste load increases with time, and by interfering with normal cellular activities, it inhibits function of metabolic pathways. Once the dynamics of complete metabolic pathways are compromised, the ageing process may be irreversible.

Importantly, mitochondrial metabolic processes influence different facets of ageing and form connections between replicative, chronological, and hibernating lifespan in yeast as well as to ageing in mammals. Partially overlapping mechanisms that arise from their biochemical function, and the regulation of basic cellular metabolic activity, potentially explain these ageing phenotypes. A unifying concept to understand the role of metabolism in ageing thus requires considering the metabolic integrity of mitochondria, adding additional complexity to the immensely important role of oxidative stress and programmed cell death in the ageing process.

Supporting Information

Table S1. References and interaction details for the molecular network of mitochondria-specific ageing factors in yeast.


We are grateful for the support of N. Lawrence of the Gurdon Institute Imaging Facility (University of Cambridge) in super-resolution microscopy, and to Benedikt Westermann, University of Bayreuth, for providing the mtGFP expression plasmid. We are grateful to FWF (Austria) for grant S9302-B05 (to MB) and to the EC (Brussels, Europe) for project MIMAGE (contract 512020, to MB), the Wellcome Trust (RG 093735/Z/10/Z) (to MR) and the ERC (Starting grant 260809) (to MR). M.R. is a Wellcome Trust Research Career Development and Wellcome-Beit prize fellow.


  • Editor: Austen Ganley


View Abstract