Genome instability of ageing stem cells--Induction and defence mechanisms.

The mammalian organism is comprised of tissue types with varying degrees of self-renewal and regenerative capacity. In most organs self-renewing tissue-specific stem and progenitor cells contribute to organ maintenance, and it is vital to maintain a functional stem cell pool to preserve organ homeostasis. Various conditions like tissue injury, stress responses, and regeneration challenge the stem cell pool to re-establish homeostasis (Fig. 1). However, with increasing age the functionality of adult stem cells declines and genomic mutations accumulate. These defects affect different cellular response pathways and lead to impairments in regeneration, stress tolerance, and organ function as well as to an increased risk for the development of ageing associated diseases and cancer. Maintenance of the genome appears to be of utmost importance to preserve stem cell function and to reduce the risk of ageing associated dysfunctions and pathologies. In this review, we discuss the causal link between stem cell dysfunction and DNA damage accrual, different strategies how stem cells maintain genome integrity, and how these processes are affected during ageing.

DNA lesions and mutations accumulate with progressive ageing

There is substantial evidence for an ageing associated accumulation of DNA damage in various tissues despite the presence of different types of repair systems (Figs. 1 and 2; Dimri et al., 1995; Jackson and Bartek, 2009; Jiang et al., 2007). Markers of DNA damage accumulate in skin of ageing baboons (Herbig et al., 2006), in stem and progenitor cells of ageing murine intestine (Hewitt et al., 2012) as well as in ageing haematopoietic stem (HSC) and progenitor cells from mice and humans (Rossi et al., 2007a; Rübe et al., 2011). Although γH2AX foci in quiescent aged HSCs have recently been disputed to label sites of DNA damage (Flach et al., 2014), other studies demonstrate an accumulation of DNA breaks and mutations in ageing murine and human HSCs (Beerman et al., 2014; Busque et al., 2012; Corces-Zimmerman et al., 2014; Welch et al., 2012; Xie et al., 2014).

Accumulating mutations at stem and progenitor cell level appear to contribute to ageing associated defects in organ maintenance and an increase in cancer development (Visvader, 2011). Recently, whole genome sequencing approaches in the haematopoietic system revealed an age-associated increase of mutations in human HSCs and provided insight into the clonal evolution of mutations and development of leukaemia (Busque et al., 2012; Corces-Zimmerman et al., 2014; Welch et al., 2012; Xie et al., 2014). The data suggest that mutations are acquired in a lifelong random manner, and that most mutations behave as neutral ‘passengers’ at HSC level until an additional ‘driver’ mutation supports clonal amplification of a subset of passenger mutations (Welch et al., 2012). The initial driver mutation may only support increased self-renewal (Busque et al., 2012), but may still depend on additional ‘driver’ hits to transform the HSC into a leukaemic stem cell (Fig. 3; Welch et al., 2012). Similarly, it could be demonstrated that individual driver mutations in male germline stem cells contribute a selective advantage to the affected cells allowing the clonal amplification and inheritance of the mutations (Goriely et al., 2009; Goriely et al., 2003). It remains to be determined whether mutations in stem cells accumulate in a simple stochastic manner or whether ageing associated exponential dynamics of mutation accumulation occur. Given some evidence that checkpoint function may decline with age (Feng et al., 2007) and that older HSC reveal slower in vitro DNA damage repair (Rübe et al., 2011) it is possible to assume a progressive increase of mutation accumulation during ageing. The mechanisms that drive ageing associated increases in mutation accumulation in stem cells represents an emerging research field that could include cell intrinsic and extrinsic factors (DeGregori, 2013).

Fig. 3

A model how increased mutations could impair stem cell functionality. Young stem cells (SC) properly self-renew and generate functional progenitor cells (PC) to maintain tissue homeostasis. Aged, but still functional stem cells show accumulation of DNA lesions and passenger mutations. Driver mutations in critical genes render ageing stem cells dysfunctional. Clonal expansion of damaged stem and progenitor cells contributes to carcinogenesis, while diminished self-renewal capacities and generation of non-functional progenitor cells causes organ failure.

A reduced capacity to repair DNA leads to stem cell depletion

It is conceivable that the accumulation of lesions and mutations observed during ageing of HSCs may in part be caused by acquired defects in DNA repair pathways (Fig. 2). Germline mutations affecting DNA repair factors cause an increasing accumulation of DNA lesions and have the potential to cause progeria syndromes thus linking DNA damage accrual to progressive ageing. Classic examples are Werner syndrome, Hutchinson-Guilford disease or Cockayne syndrome (Burtner and Kennedy, 2010; Hoeijmakers, 2009; Chu and Hickson, 2009), while additional progeria susceptibility factors, such as SPRTN are being discovered (Lessel et al., 2014). Defective DNA repair can furthermore be directly linked to a premature exhaustion of the stem cell pool of certain tissues. A dysfunctional Fanconi anemia (FA) pathway, which repairs interstrand crosslinks (ICL) causes a premature failure of bone marrow haematopoiesis in humans. This is due to an accumulation of DNA lesions resulting in an overstimulation of DNA damage checkpoint responses in HSCs and their progenitors (Ceccaldi et al., 2012). Interestingly, lack of ICL repair recently was shown to sensitize murine HSCs to damage caused by endogenous aldehydes (Garaycoechea et al., 2012). In addition to the data on the requirement of ICL repair for stem cell maintenance, studies on mice deficient for nucleotide excision repair demonstrate a critical role also for this pathway in HSC maintenance and prevention of premature ageing (Fig. 2; Rossi et al., 2007a).

Fig. 2

Repair pathways revert distinct types of DNA lesions. On top different types of DNA lesions are highlighted, which are schematically represented in red. Indicated below is the pathway that repairs the specific subset of lesions.

HSC maintenance is furthermore affected by experimental manipulation targeting nonhomologous endjoining (NHEJ), which leads to defects in the haematopoietic reserve (demonstrated by mutation of Ligase 4, DNA dependent protein kinase catalytic subunit (DNA-PKcs), or loss of XRCC4-like factor (XLF)/Cernunnos; Avagyan et al., 2014; Nijnik et al., 2007; Zhang et al., 2011; Rossi et al., 2007a). However, also other organ compartments seem to rely on NHEJ, since reduced expression of Ku80 caused accelerated ageing of the skeletal muscle and muscle stem cells also known as satellite cells (Didier et al., 2012). NHEJ repairs double strand breaks (DSBs) by mediating the ligation of broken DNA ends after only minimal processing (Waters et al., 2014). In contrast to homologous recombination (HR), which also repairs DSBs and helps to restart stalled replication forks, NHEJ does not require a template and is thus considered error-prone. However, this independence of a template uncouples NHEJ from the cell cycle allowing execution of repair also during G0/G1. Since HSCs usually are quiescent and divide only rarely, they predominantly use NHEJ to repair DSBs (Mohrin et al., 2010). Although preferential use of the error-prone NHEJ allows quick repair of DSBs, this bears the risk of an elevated mutation rate. Recent evidence showed that DNA damage can remain unrepaired in quiescent HSCs possibly explaining an ageing associated DNA damage accrual. Furthermore, upon entry into the cell cycle, HR becomes available for repair (Beerman et al., 2014). In addition, HSCs harbouring non-repairable DNA damage, such as dysfunctional telomeres, are depleted by apoptosis upon cell cycle activation or blocked from cell cycle entry by induction of senescence (Wang et al., 2014). Both mechanisms (employment of HR as a repair pathway and blockage/removal of non-repairable HSCs) likely contribute to lower the risk of mutation accumulation in ageing HSCs. However, emerging data on the accumulation of genome mutation in ageing human HSCs indicate that DNA damage surveillance and protection mechanisms are not perfect and fail during ageing (Busque et al., 2012). In contrast to the studies on DNA repair deficient mouse models and progeria syndromes in humans (Hoeijmakers, 2009), relatively little is known about the evolution of DNA repair deficiencies during ‘physiological ageing’ and how this may affect the maintenance of genome integrity in ageing stem cells.

Enhanced DNA repair in young stem cells

Some types of stem cells protect their genome by enhanced rates of DNA repair compared to more differentiated somatic cells. Hair follicle bulge stem cells in the epidermis avoid apoptosis after ionizing radiation (IR)-induced DNA damage via increased activity of NHEJ to repair DSBs (Sotiropoulou et al., 2010). Quicker repair is accompanied by attenuated p53 activation and expression of the anti-apoptotic factor B-cell lymphoma 2 (Bcl-2), which maintains bulge stem cell number after DNA damage infliction. Recently, a similarly enhanced NHEJ capacity was reported for satellite cells, which repaired DSBs more efficiently than their progeny (Vahidi Ferdousi et al., 2014). Interestingly, repair was not only faster, but also more precise, since satellite cells mainly used classical NHEJ and avoided alternative NHEJ. The latter makes use of microhomologies in sequences near the break site to align broken ends while intervening sequences are degraded and lost (see later section on telomere shortening induced chromosomal fusions, Betermier et al., 2014). Thus, the preferential use of classical NHEJ appears to contribute to enhance the maintenance of stable genomes in satellite cells (Vahidi Ferdousi et al., 2014). The small intestine is a highly proliferative tissue, which therefore strongly depends on functional stem cells. Subpopulations of intestinal stem cells (ISCs) seem to repair high loads of DNA damage with faster kinetics as compared to their progeny (Hua et al., 2012). However, in contrast to stem cells of the hair bulge or satellite cells, they make use of HR and NHEJ (Hua et al., 2012).

Prevention of DNA damage

One way to protect the genome is to avoid the formation of DNA damage in the first place. HSCs follow this strategy by sustaining a hypoxic status (Nombela-Arrieta et al., 2013) and by generating ATP mostly through glycolysis rather than mitochondrial respiration, which HSCs use upon cell cycle entry (Takubo et al., 2013; Yu et al., 2013). This greatly reduces the generation of reactive oxygen species (ROS) and its associated risks of damaging DNA. Mutations that change the metabolic activity and thereby increase ROS production diminish the self-renewal capacity of HSCs (Fig. 4; Chen et al., 2008). Thus, preserving the quiescent state reduces the risk to suffer damage from toxic intracellular byproducts of metabolism. Interestingly, loss of the DNA damage checkpoint kinase Ataxia telangiectasia mutated (ATM) also leads to increases in ROS, which influence cell cycle activity of HSCs. ROS reduce self-renewal by activation of the cyclin dependent kinase (CDK) inhibitor p16Ink4a and the tumour suppressor Rb, and block quiescence by induction of the stress response kinase p38 (Fig. 4; Ito et al., 2004; Ito et al., 2006). Furthermore, deletions of Forkhead box O (Foxo) transcription factors, which aid in detoxification of ROS, also compromise HSC self-renewal and quiescence by stimulating Rb and p53 (Miyamoto et al., 2007; Tothova et al., 2007; Yalcin et al., 2008). However, treatment with the ROS scavenger N-acetylcysteine rescues HSC maintenance (Chen et al., 2008; Ito et al., 2004; Tothova et al., 2007; Yalcin et al., 2008). Collectively, the data suggests that Foxo, Rb, and p53 control the hazardous effects of moderate ROS generation during normal stem cell maintenance but that sustained activation of these pathways contributes to stem cell dysfunction.

Fig. 4

Checkpoint signalling influences stem cell maintenance. The tumour suppressors p53 and Rb are two central checkpoint factors influencing stem cell maintenance. Both factors interfere with cell cycle control, and p53 additionally engages the cell death machinery. Upstream DNA damage response kinases ATM and ATR sense the cell's damage load and relay this information towards p53 and Rb. It is the magnitude of damage that decides between permanent solutions like senescence or apoptosis, or a transient response like reversible quiescence or self-renewal suspension. In order to compensate for the loss through senescence or apoptosis stem cells need to induce self-renewal divisions, which are started through growth promoting factors. Re-establishment of quiescence is vital in this context to prevent stem cell exhaustion through continuous self-renewal divisions. Interestingly, both tumour suppressors p53 and Rb are able to protect the organism from the accumulation of stem cells with too much damage (senescence and apoptosis) and additionally preserve the quiescent stem cell pool. The dashed connections are likely to exist but experimental evidence at stem cell level is partially lacking.

DNA replication itself comprises the potential to induce DNA damage due to replication errors or breakage of elongating forks (Zeman and Cimprich, 2014). Impeding the efficiency of replication by conditional ablation of the checkpoint kinase Ataxia telangiectasia and Rad3 related (ATR) or its downstream kinase Chk1 is thus problematic for maintenance of different types of stem cells and causes premature ageing phenotypes in skin, bones, small intestine, and the haematopoietic system (Fig. 4; Ruzankina et al., 2007; Greenow et al., 2009). Lack of ATR or Chk1 triggers the rapid accumulation of DNA damage, which provokes p53-dependent and -independent checkpoint responses including apoptosis and cell cycle arrest (Ruzankina et al., 2009; Greenow et al., 2009). Similarly to the findings on HSCs, quiescence is still protective for hair follicle bulge stem cells since resting cells survive ATR deletion whereas cycling cells suffer p53-mediated deletion (Ruzankina et al., 2009).

Similar to stem cells of the hair bulge, HSCs and satellite cells also generally remain quiescent and are only activated when needed (Cheung and Rando, 2013; Wilson et al., 2008). Consequently, loss of HSC quiescence causes accelerated depletion of stem cells (Chen et al., 2008; Gan et al., 2010). During ageing satellite cells and HSCs preserve a quiescent state. However, de-repression of p16INK4a in satellite cells induces a permanent senescence-like state that prevents stem cell reactivation and consequently muscle regeneration (Sousa-Victor et al., 2014). Ageing HSCs show only little change in overall cell cycle activity and a delayed cell cycle progression (Noda et al., 2009; Rossi et al., 2007b). Similarly as to aged satellite cells, this has been attributed to an upregulation of p16INK4a, although this has not been universally observed (Attema et al., 2009; Janzen et al., 2006; Wang et al. 2014). Delayed cell cycle progression in aged HSCs could involve reduced capacity of the DNA replication machinery itself (Flach et al., 2014). Aged HSCs show a reduced expression of MCM4 and 6, subunits of the MCM complex, the helicase needed to establish pre-replicative complexes as well as for replication elongation. Reduced availability of the MCM complex thus could reduce origin firing during S phase and slow down fork movement, both of which has been observed in aged HSCs and was caused by knock down of MCM4 or 6 in young HSCs (Flach et al., 2014). Thus, inefficient DNA replication may contribute to reduce stem cell functionality in aged organisms.

Is there an immortal DNA strand?

Another strategy to avoid mutations in stem cells is the concept of the so-called immortal strand. This idea – which is under debate – postulates that stem cells would specifically keep the template DNA strands in an asymmetric division. Thus, if replication-associated mutations would occur only the differentiated daughter cell would obtain mutated DNA, while the original DNA strands remain in the renewed stem cell. Label-retaining stem cells at the +4 position in crypts of the small intestine led to this hypothesis (Potten et al., 2002), although the mechanistic details remain to be unravelled. Some evidence supports the immortal strand concept in satellite cells (Conboy et al., 2007; Shinin et al., 2006). However, this mechanism of genome protection seems not to be relevant in ISCs and HSCs (Escobar et al., 2011; Kiel et al., 2007; Schepers et al., 2011; Steinhauser et al., 2012).

Destabilized telomeres, an intrinsic source of DNA damage

Eukaryotic chromosome ends are composed of highly repetitive DNA sequences, the telomeres. Progressive shortening of telomeres is observed in various ageing human organs including colonic mucosa and HSCs (Hastie et al., 1990; Jiang et al., 2007; Vaziri et al., 1994). Studies on telomere biology provided functional proof that the accumulation of DNA damage at chromosome ends can induce premature ageing in mice (Blasco et al., 1997; Rudolph et al., 1999) and fish (Henriques et al., 2013). Moreover, dysfunctional Werner syndrome helicase (WRN) causes premature ageing by accelerated telomere shortening (Chang et al., 2004; Crabbe et al., 2004; Du et al., 2004). During each round of S-phase the end replication problem and processing of chromosome ends lead to loss of 50–100 terminal base pairs of linear chromosomes (Chow et al., 2012; Lundblad, 2012). To avoid loss of genetic information during replication and to protect chromosome ends from fusion, the long telomeric repeats are additionally associated with protective shelterin nucleoproteins (de Lange, 2009). However, in human cells this buffer is limited to 7–12 kilobases and telomeres progressively shorten to critically short lengths that are then recognized as DNA damage. This evokes p53/p21 dependent checkpoint responses inducing a permanent cell cycle arrest (replicative senescence) (Fig. 4; Brown et al., 1997; d’Adda di Fagagna et al., 2003).

Stem cells are protected from extended telomere loss by expression of the enzyme telomerase (Chiu et al., 1996; Wright et al., 1996). However, the level of telomerase activity apparently is not sufficient to prevent telomere shortening in HSCs and ISCs during ageing (Schepers et al., 2011; Vaziri et al., 1994). Studies on telomerase deficient mice provided the first experimental evidence that telomere shortening leads to p53/p21 dependent defects in self-renewal and functional capacity of germline stem cells (Chin et al., 1999) and somatic stem cells (Allsopp et al., 2003; Choudhury et al., 2007; Rossi et al., 2007a; Sperka et al., 2012). The contribution of telomere shortening to the functional decline of stem cells and disease evolution during ‘physiological ageing’ remains to be determined. Of note, studies on wildlife bird revealed a predictive value on lifespan of both telomere length at birth and telomere shortening rates during adult life (Heidinger et al., 2012; Salomons et al., 2009). Similarly, in humans, telomere length was identified as predictive marker for ageing associated diseases, cancer, and lifespan (Cawthon et al., 2003).

Cell culture experiments suggested that p53 and retinoblastoma (Rb) checkpoint networks respond to telomere dysfunction (Chin et al., 1999; Hara et al., 1991). Indeed, deletion of p53 rescued depletion of testicular germ cells usually observed in the telomerase (mTerc) single knockout mice (Chin et al., 1999). However, loss of p53 led to an increased tumour load (Artandi et al., 2000) and accelerated the ageing of somatic tissues by enhancing survival of chromosomally instable stem cells with abnormal differentiation potential (Begus-Nahrmann et al., 2009). Triggered by critically short telomeres, p53 mediates two main checkpoint responses: permanent cell cycle arrest via activation of the cell cycle dependent kinase inhibitor p21 and apoptosis by induction of the BH3-only pro-apoptotic factor Puma. Interestingly, deletion of either p21 or Puma improves stem cell and tissue maintenance in the context of dysfunctional telomeres, which elongated the lifespan of telomere dysfunctional mice without increasing tumour formation (Choudhury et al., 2007; Sperka et al., 2012). These data suggested that it is possible to increase the functional stem cell reserve temporarily by blunting one branch of the p53 response without increasing the risk for genome instability and tumour formation as the other branch of the p53-pathway remains intact.

In addition to the induction of checkpoints it is conceivable that induction of DNA repair contributes to stem cell failure in response to telomere dysfunction. Along these lines it was demonstrated that exonuclease 1 (Exo1) dependent DNA end-resection of shortened telomeres contributes to the induction of DNA repair. This results in the formation of chromosomal fusions (likely induced by alternative NHEJ), which triggers subsequent checkpoint induction (Maser et al., 2007; Rai et al., 2010; Schaetzlein et al., 2007). Chromosomal fusions represent a major problem as they lead to amplification of DNA damage and chromosomal instability. Thus, it will be important to determine the relative contribution of checkpoints and aberrant repair to trigger stem cell and tissue ageing.

Recently, it was demonstrated that long telomeres also accumulate irreparable DNA damage (Fumagalli et al., 2012; Hewitt et al., 2012). Compared to telomere shortening, the accumulation of DNA damage at long telomeres may not be as detrimental to stem cells and tissues since telomere binding proteins suppress the induction of checkpoints and repair responses at intra-telomeric damage sites. In future research, it will be important to better define the detrimental effects that are induced by DNA damage responses at the chromosome termini in ageing cells and how these drive the evolution of stem cell and tissue dysfunction.

Removal of damaged stem cells

Expansion of damaged stem cells is deleterious for organ maintenance and may lead to cancer formation. To this end it may be important to clear damaged stem cells from the pool of self-renewing stem cells by induction of checkpoints such as apoptosis. Increasing levels of ageing associated apoptosis are detected under homeostasis in germ cells (Wang et al., 1999) and HSCs (Janzen et al., 2006) and upon external stress in small intestine crypts (Martin et al., 1998) involving p53-dependent and -independent processes. An alternative way to remove damaged cells from the active stem cell pool is the engagement of differentiation. IR-induced DNA damage limits the self-renewal capacity of melanocyte stem cells in the epidermal hair follicle by induction of differentiation into mature melanocytes. This leads to depletion of melanocyte stem cells and hair greying (Inomata et al., 2009). Interestingly, deletion of ATM elevates IR-induced differentiation of melanocyte stem cells (Inomata et al., 2009). Perhaps loss of ATM impairs self-renewal via elevated ROS in melanocyte stem cells similar to what was observed in HSCs (see above). In addition, it was recently reported that DNA damage in the haematopoietic system induces the G-CSF/STAT3 axis leading to induction of Basic leucine zipper transcription factor—ATF-like (BATF) in HSCs (Wang et al., 2012). This causes lymphoid-biased HSCs to differentiate, which contributes to HSC depletion and myeloid lineage bias in response to IR. Together, induction of differentiation in stem cells emerges as a novel checkpoint response preventing self-renewal of damaged tissue stem cells. Interestingly, it was recently shown that ATM is required for the prevention of differentiation and the sustained self-renewal of leukaemic stem cells in mixed lineage leukaemia (MLL)-AF9 induced leukaemia (Santos et al., 2014). It remains to be seen whether differentiation-inducing checkpoints contribute to the prevention of the accumulation of mutations in ageing stem cells and cancer initiation.

Outlook

There is an increasing knowledge on molecular mechanisms that are important for genome maintenance in adult tissue stem cells. However, recent data on the accumulation of genome-wide mutations and DNA damage in ageing stem cells indicate that genome maintenance mechanisms fail during ageing. Given the close interaction between the accumulation of molecular damages and gene mutations with the evolution of diseases and cancer in ageing tissues, it is one of the biggest challenges to delineate mechanisms that contribute to the failure of genome maintenance in ageing stem cells. Recent studies indicate that the role of self-renewing progenitor cells in tissue maintenance during homeostatic conditions may currently be underestimated (Sun et al., 2014). Therefore, it will also be important to consider differences in DNA repair in progenitor cells and stem cells and to define the role of progenitor cell DNA mutations in the evolution of ageing associated diseases. In addition, it appears to be of increasing importance to study mechanisms in the cell environment (the stem cell niche and in the circulatory environment) that may lead to an enhanced selection of mutant stem and progenitor cells during ageing. Along these lines it is interesting to note that a high percentage of ageing associated mutations in stem cells affect regulators of the epigenome (Corces-Zimmerman et al., 2014), which may drive adaptive responses and clonal selection of stem cells in ageing tissues. A detailed understanding of the molecular causes of genome instability in ageing stem and progenitor cells should ultimately help to develop molecular therapies aiming to improve tissue functionalities and to prevent the evolution of ageing associated diseases thus allowing a healthier ageing.