Stem Cell DNA Damage and Genome Mutation in the Context of Aging and Cancer Initiation

Abstract

Adult stem cells fuel tissue homeostasis and regeneration through their unique ability to self-renew and differentiate into specialized cells. Thus, their DNA provides instructions that impact the tissue as a whole. Since DNA is not an inert molecule, but rather dynamic, interacting with a myriad of chemical and physical factors, it encounters damage from both endogenous and exogenous sources. Damage to DNA introduces deviations from its normal intact structure and, if left unrepaired, may result in a genetic mutation. In turn, mutant genomes of stem and progenitor cells are inherited in cells of the lineage, thus eroding the genetic information that maintains homeostasis of the somatic cell population. Errors arising in stem and progenitor cells will have a substantially larger impact on the tissue in which they reside than errors occurring in postmitotic differentiated cells. Therefore, maintaining the integrity of genomic DNA within our stem cells is essential to protect the instructions necessary for rebuilding healthy tissues during homeostatic renewal. In this review, we will first discuss DNA damage arising in stem cells and cell- and tissue-intrinsic mechanisms that protect against harmful effects of this damage. Secondly, we will examine how erroneous DNA repair and persistent DNA damage in stem and progenitor cells impact stem cells and tissues in the context of cancer initiation and aging. Finally, we will discuss the use of invertebrate and vertebrate model systems to address unanswered questions on the role that DNA damage and mutation may play in aging and precancerous conditions.

STEM CELLS AND TISSUE DYNAMICS

Many adult metazoan tissues maintain long-term function through the ongoing elimination of terminally differentiated cells and the replacement of these cells by the newly divided progeny of cycling cells. An understanding of this process began almost 70 years ago through early lineage tracing studies of Charles Philippe Leblond using tritiated-thymidine injection in mice to reveal the turnover rates of labeled cells (Leblond and Walker 1956). These seminal studies initiated the stem cell theory of renewal and laid the foundation for modern labeling studies of cell turnover that confirmed age-associated mosaicism of adult tissues (Spalding et al. 2013; Arrojo e Drigo et al. 2019). Importantly, this work raised conceptual questions regarding how stem cells may endure the process of aging.

Aging is associated with an alteration in stem cell functionality and kinetics of tissue renewal in many tissues such as blood, skin, muscle, and the brain (Kuhn et al. 1996; Morrison et al. 1996; Conboy et al. 2003; Nishimura et al. 2005). An imbalance in tissue dynamics caused by deregulated self-renewal or cell turnover rates can compromise tissue function. Understanding how tissue dynamics are altered during aging or in pathological contexts is an important, yet highly complex question. At a molecular level, these changes may be induced by genetic, epigenetic, or metabolic alteration. Potential causes may include changes in stem-cell-intrinsic factors, alteration of niche properties, or modification of systemic signals. In this review, we will focus on stem-cell-intrinsic alteration through DNA damage and genetic mutation (for recent reviews on the impact of epigenetic and metabolic changes on aging stem cells, see Booth and Brunet 2016; Chandel et al. 2016; Brunet and Rando 2017).

DNA DAMAGE AND HOW IT LEADS TO MUTATION

All cells, including stem cells, are faced with the challenges of protecting their DNA from erosion. DNA damage is a deviation from the normal DNA structure with the introduction of damaged sites in the base-pairing or backbone structure. Multiple exogenous agents such as ultraviolet (UV) light, ionizing radiation (IR), and chemical mutagens, such as hydrocarbons present in tobacco smoke, can damage DNA. In addition, endogenous factors such as reactive oxygen species, telomere erosion, and replication errors can also be a source of damage. DNA replication, for example, is an opportune time for error, as the replication fork can slow down, or collapse, because of topological challenges including limiting nucleotides, repetitive sequences, non-B-form DNA, and collisions with transcription machinery.

It has been estimated that tens of thousands of lesions are experienced by a mammalian cell per day, with single-strand lesions making up the majority of this number (Lindahl and Nyberg 1972; Lindahl 1974; Lindahl and Barnes 2000). Damage involving a single strand can be accurately repaired using the other strand as a template. Small base lesions that do not significantly change the DNA helix structure are repaired by base-excision repair (Lindahl 1974). As for misincorporated bases, they are corrected by the mismatch repair pathway (Lahue et al. 1989). On the other hand, lesions involving bulky adducts and dimers are repaired by nucleotide-excision repair (Sancar and Tang 1993), whereas interstrand cross-links require the Fanconi anemia pathway (Zhang and Walter 2014). Additionally, DNA double-strand breaks (DSBs) also arise in the cell, often through replication fork collapse. This type of damage is particularly dangerous because more error-prone repair mechanisms are used that can lead to the loss of genetic information, contributing to genome instability. DNA DSB repair is primarily orchestrated by two pathways: if the cell has gone through S phase, duplicating its chromosomes, providing a template for the repair of the damaged chromosome, homologous recombination (HR) is used. If, on the other hand, the cell is in G₁, the more erroneous nonhomologous end-joining (NHEJ) pathway is used, which involves the ligation of the broken ends and the introduction of small deletions as a result.

Despite the existence of these strategies to safeguard the integrity of DNA, glitches in the system arise frequently leading to sequence variants, structural variants, or aneuploidy. Sequence variants include indels and point mutations, arising, for example, through deamination of 5-methylcytosine in CpG nucleotides in vertebrates, resulting in a C-to-T substitution (Razin and Riggs 1980). Structural variants involve more large-scale changes to the DNA sequence and therefore are more likely to alter gene function. These include amplifications, deletions, and translocations, which can be caused by recombination and replication-based mechanisms, erroneous DNA DSB repair, or be a result of transposable element mobility (Carvalho and Lupski 2016; Bourque et al. 2018). Most genes are haplosufficient (Huang et al. 2010), and therefore inactivation of one copy may not impair cell function. However, problems may arise when one allele in the genome is already inactive in the germline and the second allele is inactivated somatically, leading to loss of heterozygosity (LOH). LOH can be driven by the aforementioned mutagenetic processes as well as recombination with the homologous chromosome, also known as “mitotic recombination,” which on cell division, leads to segregation of two mutant alleles into one daughter cell. Thus, DNA damage in stem or progenitor cells can alter the genome in numerous ways and potentially radically disrupt tissue function over the course of aging.

Some of the first evidence suggesting a potential causal link between cellular DNA damage and organismal aging came from the realization that inactivation of DNA repair genes such as in Fanconi anemia and Werner syndrome in humans lead to early aging or “progeroid” syndromes (Moskalev et al. 2013; Carrero et al. 2016). Due to reduced ability to repair DNA damage, DNA lesions persist, and somatic mutations accumulate. Patients with these syndromes show accelerated aging and present symptoms of loss of proper tissue renewal such as skin atrophy, loss and graying of hair, and higher susceptibility to cancer development. Although these studies suggest sufficiency of DNA damage to drive early aging phenotypes, they do not provide evidence that endogenous levels of DNA damage or mutation can impact aging. How DNA damage and genome mutation may impair stem cell function will be further discussed below. We will first examine mechanisms that can mitigate the effects of DNA damage in stem cells.

MECHANISMS PROTECTING THE STEM CELL AND TISSUE FROM THE EFFECTS OF DNA DAMAGE

Although coping with DNA damage is important for all cells, it is particularly vital for adult stem cells that renew tissues throughout adult life. What are the ways in which stem cells and tissues avoid the negative impact of DNA damage and mutation? Here, we will discuss a number of important protection mechanisms acting at the stem cell and tissue level.

Protecting the Stem Cell: DNA Damage Responses and Repair

Evidence suggests that at least some stem cells use distinct mechanisms from their downstream differentiated or more committed progenitor cells to prevent the accrual of genetic lesions, which can be detrimental to homeostasis of the tissue. DNA damage is managed via the DNA damage response (DDR), which is an evolutionarily conserved signaling pathway in which sensors, mediators, and effectors orchestrate DNA repair or by the elimination of the damaged cell by apoptosis or by exiting the cell cycle. Interestingly, adult mouse hematopoietic stem cells (HSCs) and hair follicle stem cells (HFSCs) of the bulge were found to have increased radioresistance with minimal apoptotic response and accelerated DNA repair compared with their more differentiated progeny (Mohrin et al. 2010; Sotiropoulou et al. 2010; Beerman et al. 2014). Aged HSCs are even more resistant than young HSCs to DNA-damage-induced apoptosis (Gutierrez-Martinez et al. 2018). This likely helps prevent depletion of the stem cell pool but could be at the cost of accumulating mutations.

Stem cells also differ in a tissue-dependent manner in terms of strategies used that help limit passing mutations to progeny, with some favoring robust repair (Fig. 1A) and others favoring apoptosis (Fig. 1B) or terminal differentiation (Fig. 1C). The small intestine for instance is sensitive to apoptosis driven by DNA damage, whereas stem cells of the colon are resistant to apoptosis (Merritt et al. 1995; Potten and Grant 1998). Intrinsic differences in cell-cycle properties could explain why stem cells differ widely in their DNA repair mechanisms between tissues. As shown in the hematopoietic system, when a DSB arises in a quiescent cell, DNA repair is mediated by the efficient NHEJ mechanism, which acts quickly and does not need the presence of homologous DNA for repair, but is error-prone (Mohrin et al. 2010). Proliferating HSCs on the other hand use high-fidelity HR to repair DSBs but have an increased likelihood of accumulating damage during S-G₂/M (Mohrin et al. 2010). Alternatively, another strategy of protecting the tissue from propagating a mutation is used by melanocyte stem cells that differentiate on DNA damage (Nishimura et al. 2005; Inomata et al. 2009). Similarly, during the aging process, it is thought that the HFSC are gradually lost owing to differentiation on repeated DNA damage acquisition during hair follicle cycles (Fig. 1C; Matsumura et al. 2016).

Figure 1.

Mechanisms of stem cell protection from DNA damage by repair, elimination or cell cycle exit. (A) Repair: depending on the cell cycle status of the cell, the cell undergoes repair by either nonhomologous end-joining (NHEJ) (top panel) or homologous recombination (HR) (bottom panel). NHEJ is the quick and efficient mechanism used by quiescent stem cells when they are faced with damage. It involves the ligation of the broken ends and often results in the introduction of small deletions, but can also lead to translocation and genome rearrangements. HR is used if the cell is cycling and goes through S phase, duplicating its chromosomes, providing a template for the repair of the damaged chromosome. This is usually more accurate repair than NHEJ, though erroneous choice of the homologous chromosome, rather than the sister, can lead to loss of heterozygosity (LOH). (B) Elimination: by apoptosis. Some cells undergo apoptosis rather than repair. If this mechanism is preferentially used in the stem cell, there is a higher chance of stem cell depletion. (C) Cell-cycle exit: by differentiation (top panel) on DNA damage, or remaining in a state of quiescence.

Caught between balancing the need to maintain tissue function and the need to block the propagation of mutations, stem cells have evolved diverse modes to cope with DNA damage and repair, often sacrificing immediate survival of a given stem cell for the expense of long-term maintenance of genome integrity in the tissue. Further studies are important to better understand the sensitivity and resistance of adult stem cells to damage, the repair mechanisms used, and age-related changes in this process.

Protecting the Stem Cell: A Quiescent State

One way to limit DNA damage is simply to avoid undergoing cell division, which would restrict replicative and chromosome segregation errors (Fig. 1C). Indeed, many populations of adult stem cells including hematopoietic, muscle, and neural stem cells are primarily in a nonproliferating quiescent state of G₀ (Cho et al. 2019; van Velthoven and Rando 2019). Evidence suggests that quiescence serves a protective role in these contexts as these populations of stem cells become depleted or “exhausted” when driven into the cell cycle on transplantation, caused by stress, or on genetic manipulation (Harrison 1978; Chen et al. 2000; Kamminga et al. 2006; Gan et al. 2010; Sacco et al. 2010; Schaniel et al. 2011; Staber et al. 2013; Cavallucci et al. 2016; Yue et al. 2016; Baumgartner et al. 2018; Singh et al. 2018). Stem cell exhaustion in these contexts may be a result of the loss of niche signals. Alternatively, these studies raise the possibility that increased DNA damage or an increased mutational burden on loss of quiescence may lead to stem cell functional decline during aging (Sharpless and DePinho 2007). Consistent with this notion, when mouse HSCs were forced repeatedly out of quiescence, they acquired DNA damage and became depleted (Walter et al. 2015). Nevertheless, the extent to which stem cell exhaustion is related to increased DNA damage or acquisition of mutations is not entirely clear and may differ depending on stem cell type. Additional potential links between DNA damage and stem cell senescence will be discussed later in the review.

Protecting the Tissue: Competition between Cells and Lineages

In addition to stem-cell-intrinsic mechanisms of protection mentioned above, tissue-level protection also helps to ensure the survival of the fittest lineage. This may be especially important for stem cells such as the Crypt Basal (Lgr5⁺) intestinal stem cells (ISCs) and those of the skin epidermis, that actively divide. One such mechanism is neutral competition in which ISCs undergo dynamic stem cell replacement shown, both in the mouse (Lopez-Garcia et al. 2010; Snippert et al. 2010) and the Drosophila intestine (De Navascués et al. 2012), which likely helps to prevent their loss.

Aside from neutral competition, biased cell competition also occurs between cells. Initially described in Drosophila, biased cell competition is a phenomenon whereby differences in cellular fitness allow selection of “winner” cells, while weeding out less-fit “loser” cells (Fig. 2A; Morata and Ripoll 1975). A large body of work in Drosophila has revealed that this process plays an important role in shaping adult tissues and has elucidated many molecular mechanisms underpinning this process (for a review on this topic, see Levayer 2019). Competition between stem cells also occurs and can result in greater niche occupancy of a given genotype with selective advantage or greater production of progeny (Zhang and Kalderon 2001; Nystul and Spradling 2007; Jin et al. 2008; Issigonis et al. 2009; Amoyel and Bach 2014; Kolahgar et al. 2015).

Figure 2.

Cell competition selects for “winner” cells and weeds out less fit “loser” cells. (A) Differential Myc levels drive cell competition in both Drosophila (progenitors in the disc) and mammalian cells (mouse developing epidermal cells and adult epidermis). This figure shows that Myc mutants are “losers” and are outcompeted by adjacent wild-type (WT) cells. Extra levels of Myc also renders cells “winners” compared with WT cells (not shown). (B) DNA damage creates differences in fitness and this can drive cell competition. Irradiation of mouse hematopoietic stem and progenitor cells creates DNA damage, which results in increased Trp53 levels. When these cells are transplanted into mice with nonirradiated cells (4 days after the irradiation), the irradiated cells are outcompeted by the nonirradiated cells that have lower Trp53 levels and a higher expression of more competitive signaling molecules. Thus, DNA damage via irradiation creates a long-lasting “loser” cell status by inducing p53-mediated apoptosis or cell-cycle arrest. (C) Stem cell lineages with higher levels of COL17A1 and Mycn become “winners” in mouse skin epidermis. Genomic stress leads to the proteolytic degradation of COL17A1 and thus results in differential levels of COL17A1 expressed in the epidermis. Cells with higher levels of COL17A1 outcompete the cells expressing lower levels via symmetric cell division and the elimination of the losers. The higher expression of COL17A1 maintains a healthy skin phenotype, whereas COL17A1 deficiency causes skin atrophy, fragility, dyspigmentation, and alopecia. Similarly, during epidermal stratification skin lineages with higher levels of Mycn outcompete the cells expressing lower levels of Mycn, but it remains unclear whether genomic stress is what drives differential Mycn expression.

An important question, however, is what types of fitness differences are being sensed during cell competition? Could stem cells with DNA damage or with less fit mutant genomes be selected against? Interestingly, mouse HSCs that have been treated with low-dose IR are less competitive than nonirradiated HSCs in a manner that is dependent on Trp53 levels and lasts for weeks (Fig. 2B; Bondar and Medzhitov 2010). This implies that a memory of the irradiation stress was kept, which is proposed to be linked to a Trp53-dependent long-term mark acting as a cellular memory for DNA damage. A recent study of the mouse skin showed that stem cell lineages compete based on levels of the hemidesmosome component, COL17A1 (Fig. 2C; Liu et al. 2019). Interestingly, like HSCs, exposure of epidermal stem cells to IR triggers a long-lasting memory of genomic stress, resulting in the proteolytic degradation of COL17A1. How this memory is achieved and whether it also relies on p53, unrepaired DNA damage, or could be linked to genomic mutations or epigenetic mechanisms, is not clear. Another recent study of the mouse skin epidermis showed, as previously shown in Drosophila (de la Cova 2004; Moreno and Basler 2004), that during epidermal stratification cell lineages with higher levels of Mycn, a bHLH transcription factor, become winners (Fig. 2A,C). It is not currently clear whether, like COL17A1, Mycn might respond to altered genomic stress and how these two mechanisms might overlap. Interestingly, mechanisms that may be akin to cell competition can also expunge aberrant tissues with altered tissue architecture, such as those expressing oncogenic Hras GTPase, as shown in mouse hair follicle using live imaging (Brown et al. 2017). This is very reminiscent of early work in the fly showing elimination of tumorigenic cells via cell extrusion (Brumby and Richardson 2003; Vaughen and Igaki 2016). Thus, cell and lineage competition are mechanisms that can help to maintain integrity of adult tissues and are likely one means of eliminating cells with harmful DNA damage or mutant genotypes.

WHEN PROTECTION MECHANISMS FAIL: ACQUISITION OF MUTATION

Despite the numerous mechanisms in place to protect stem cells from harmful effects of DNA damage, studies over the past 10 years revealed the extent to which genomic mutations arise in adult stem cells. Here, we will present data showing that genetic changes occurring in stem or progenitor cells contribute to tissue mosaicism. We will also highlight some of the recent literature from humans that has shown that somatic genetic mosaicism is not a rare pathological event but a phenomenon present in many of our healthy adult tissues.

Evidence of Surprising Diversity in Somatic Genomes

Finding and studying somatic mutations in subsets of cells within a tissue is extremely challenging. Although recent advances in genomic sequencing are beginning to unveil the extent to which somatic variation arises, classic genetic studies using visible marker phenotypes provided the first evidence of genetic mosaicism. Studies by Curt Stern using Drosophila first showed spontaneous LOH during development owing to mitotic recombination between homologous chromosomes (Stern 1936). Mitotic recombination is an important mechanism of LOH in cancer and other genetic disorders (Jonkman et al. 1997; Choate et al. 2010), though not yet well understood in healthy tissues. Somatic variation caused by mobilization of transposable elements was later studied in maize by Barbara McClintock (1950). Evidence from reporter mice and DNA sequencing–based approaches suggest that LINE1 element mobility contributes to genetic mosaicism in the nervous system (Muotri et al. 2005; Coufal et al. 2009; Upton et al. 2015; Erwin et al. 2016) and estimate a de novo LINE1 element insertion frequency of 0.2 events per neuron in humans (Evrony et al. 2012; for a more extensive review of the literature, see Faulkner and Garcia-Perez 2017). How somatic mobilization of transposable elements impacts adult tissues is only beginning to be understood.

Additional mutagenic processes also shape somatic mosaicism. Sequencing clonally expanded human adult stem cells using organoids has shown that around 40 de novo point mutations are acquired per year in liver, colon, and small intestine (Blokzijl et al. 2016); 13 de novo point mutations mutations per year in muscle stem cells (Franco et al. 2018); and about 200–400 total point mutations impact neural precursors (Lodato et al. 2015). One prominent mutational signature found in both human and mouse precursors is C-to-T transitions at CpG dinucleotides, thought to be caused by deamination of 5-methylcytosine to thymine (Behjati et al. 2014; Blokzijl et al. 2016; Lodato et al. 2018). In addition, larger-scale gene deletion and rearrangements were detected using single-nucleotide polymorphism (SNP) array methodology, with ∼14% of human colon crypts bearing a large-scale deletion or LOH event (Hsieh et al. 2013), which has been also documented in other tissues (O'Huallachain et al. 2012). Whole-genome sequencing of colon also recently confirms SNP and copy-number changes in healthy tissue (Lee-Six et al. 2019). Aneuploidy and copy-number variation in the brain and other tissues have similarly been reported, though frequencies vary depending on the detection technique (Rehen et al. 2001; O'Huallachain et al. 2012; McConnell et al. 2013; Cai et al. 2014; Knouse et al. 2014). Thus, it is now abundantly clear that human tissues have high degrees of genetic mosaicism. It is, therefore, critical to perform functional studies to understand the full impact of mosaicism on young, aged, healthy, and diseased adult tissues.

Clonal Expansion in Blood and Solid Tissues

Mosaic patches of adult tissue, or “clones,” can result from a long-lived stem or progenitor cell acquiring a mutation driving positive selection owing to increased fitness, or from neutral drift of an alteration with no impact on fitness (Snippert et al. 2010; Traulsen et al. 2013). Evidence for age-dependent clonal expansion of mutant stem cell lineages in the blood dates back to the 1990s in which probes for the inactive X-chromosome were used and detected its skewing during aging (Busque et al. 1990; Fey et al. 1994). More recently, the study of “healthy” control blood using sequencing-based approaches led to surprising evidence for clonal expansion of lineages having somatic mutation in the genes TET2, DNMT3a, and ASLX1 during adult aging (Busque et al. 2012; Jacobs et al. 2012; Laurie et al. 2012; Welch et al. 2012; Holstege et al. 2014). We will discuss further the physiological implications of blood clonality below (for an extensive review on clonal hematopoiesis, see Jaiswal and Ebert 2019).

Mounting evidence similarly indicates that solid tissues also have a high degree of genetic mosaicism with mutant progenitor cells giving rise to expanding mutant lineages under positive selection. In the 1990s, it was recognized with polymerase chain reaction (PCR) and through whole-mount tissue staining that sun-exposed normal human skin acquires clones of mutant TP53 (Nakazawa et al. 1994; Jonason et al. 1996). In recent years, these finding were greatly extended using targeted deep sequencing of 74 cancer driver genes on biopsies of normal sun-exposed eyelid epidermis and normal esophagus tissue. Frequent mutation of genes was found, including in NOTCH1 and TP53, that expand clonally and accumulate with age (Martincorena et al. 2015, 2018; Yokoyama et al. 2019). Additional recent evidence for large clonal expansions across numerous tissues including breast and lung has been shown with mutational analysis of RNA-seq data (Yizhak et al. 2019). Furthermore, other tissues show clear examples of somatic mutation-driven clonal expansion. In humans, megaencephaly syndromes leading to a clonal overgrowth of part of the brain arise through activating mutations of the AKT/PI3K pathway that can be caused by somatic mutations arising in neural precursor cells (Lee et al. 2012; Poduri et al. 2012; Rivière et al. 2012). Interestingly, somatic mutations activating PI3K have also been found to lead to Proteus syndrome, with patients having overgrowth of fibrous and adipose tissues (Lindhurst et al. 2012). Thus, positive selection of mutant lineages is prevalent in human tissues. The implications on cancer initiation of somatic mutations in driving early lineage expansion and selection will be further discussed below.

DNA DAMAGE AND SOMATIC MUTATION IN ADULT TISSUES: ROLES IN CANCER INITIATION AND AGING

What is the impact of these mutations on tissues? Clearly cancer initiation is one detrimental consequence, but not all mutations lead to cancer. Here, we will highlight the functional implications of somatic genetic mosaicism.

Somatic Mutations and Cancer Initiation

For over a hundred years, it has been recognized that cancer cells are distinct from normal ones because of the presence of aberrant genomes (Boveri 1914). Therefore, recent revelations that normal tissues harbor extensive mutations raise important questions about the relationship between apparently healthy tissue and cancer: Do mutations that provide positive selection in a tissue actually promote the eventual acquisition of additional genetic mutations leading to cancer as described in a classical multistep carcinogenesis model? Alternatively, in some instances, might these be two distinct selection processes with cancer requiring a divergent path from one that optimizes growth within an otherwise healthy tissue? As we previously discussed, multiple modes of cell and lineage competition actively shape the nature of selection within a tissue and, in theory, could respond differently to expanding mutant lineages versus precancerous clones.

Evidence from clonal hematopoiesis supports a multistep process in which a first mutation in healthy tissue precedes additional mutation (Fig. 3), increasing cancer risk. Indeed, longitudinal studies of patients with clonal hematopoiesis detected by SNP arrays support a strong increased risk of developing not only hematological cancer (Jacobs et al. 2012; Laurie et al. 2012; Welch et al. 2012; Genovese et al. 2014; Jaiswal et al. 2014; Coombs et al. 2017), but also lung and kidney cancers (Jacobs et al. 2012). Exome sequencing revealed that known tumor suppressor genes of myeloid cancers such as TET2, DNMT3A, and ASXL1, were mutated in apparently healthy blood (Busque et al. 2012; Genovese et al. 2014; Jaiswal et al. 2014; McKerrell et al. 2015; Coombs et al. 2017). Thus, the acquisition of these mutations in healthy blood is thought to represent the earlier phase in the development of leukemogenesis and suggests a period of latency that precedes it. Therefore, an understanding of how processes such as stem cell competition for niche occupancy may influence the switch from a premalignant state to a malignant one is important (Fig. 3B,D).

Figure 3.

Somatic mosaicism with age. (A) Mutations arise in stem cells of young tissues. (B) Age-dependent clonal expansion of mutant stem cells via positive selection or neutral drift give rise to mosaic patches. These may have a premalignant capacity and their persistence can lead to cancer initiation. (C) Clonal expansion can lead to the age-related collapse in clonal diversity with very few stem cells contributing to the aging tissue. The functional impact of collapse of clonal diversity is still not fully understood but it can impact age-associated lineage skewing in some cases. (D) Clonal expansion of cancer driver genes can lead to cancer initiation.

Recent studies in the skin and esophagus support the idea of healthy tissue acquiring premalignant drivers, but also suggest the intriguing possibility that healthy tissues may have distinct selective pressures than those in cancer. Targeted deep sequencing of normal esophageal epithelium from young and old donors revealed that the number of detectable mutations and the sizes of mutant clones increased with donor age (Martincorena et al. 2018). NOTCH1 and TP53, canonical drivers of esophageal squamous cell carcinoma (ESCC), were found to be under selection in normal tissue (Martincorena et al. 2018; Yokoyama et al. 2019). Thus, the presence of clonal expansions in the normal epithelium suggests that these clones have a premalignant capacity and their persistence can lead to cancer initiation (Fig. 3B,D). These data strongly support the concept of “field cancerization” (Slaughter et al. 1953), previously proposed to predispose the esophagus to development of subsequent multiple tumors via initial precancerous drivers such as p53 (Tian et al. 1998). Nevertheless, an intriguing finding is that mutations in NOTCH1 and PPM1D are much more prevalent in normal skin than in cancer (Martincorena et al. 2018; Yokoyama et al. 2019). This suggests that different fitness of certain mutations exist in “normal” tissue versus cancer, complicating the notion of a linear multistep mutation accumulation process. Future studies will be necessary to understand these fitness differences and potentially capitalize on them for clinical benefit.

An Impact of Mutations and DNA Damage on Aging?

Aside from initiating and driving cancer evolution, what impact do somatic mutations have on aging? Here, we will discuss some of the potential detrimental consequences of mutation on tissues.

Studies from clonal hematopoiesis have shown a collapse of clonal diversity with very few stem cells contributing to the aging blood (Fig. 3C). This results from “winner” HSC clones expanding and, in an apparent zero-sum game, “loser” HSCs failing to contribute to blood. This was strikingly shown from sequencing the blood of a hematologically asymptomatic supercentenarian (aged 115 years old) revealing that ∼65% of her healthy blood compartment was dominated by the progeny of two HSC clones (Holstege et al. 2014). Extending on earlier work that we discussed above (Busque et al. 2012; Jacobs et al. 2012; Laurie et al. 2012; Welch et al. 2012), a study using whole-genome sequencing from the peripheral blood of ∼11,000 Icelanders of different ages found that a striking 50% of patients older than 85 had clonal hematopoiesis (Zink et al. 2017). Thus, abundant evidence indicates that mutations arise in HSCs (or in very upstream precursor cells) during aging and lead to selection of mutant lineages; however, the functional impact of collapse of clonal diversity is still not fully understood. One feature of the aging hematopoietic system in humans and mouse is a bias toward myeloid lineages (Sudo et al. 2000; Yamamoto et al. 2017; Ganuza et al. 2019). Although unlikely to explain all of the myeloid bias of HSCs that occurs during aging, TET2 deletion is sufficient in mouse to lead to a myeloid disorder (Li et al. 2011) and is strongly associated with myeloid dysplasia in humans (Buscarlet et al. 2018). Thus, a failure to maintain the repertoire of differentiated cell types present in youth can arise from a loss of clonal diversity. Interestingly, a reduction in the clonality of mouse muscle stem cells on repeated injury was found (Tierney et al. 2018). Although the role of mutation or DNA damage was not evoked in this study, it is feasible that increased replication stress might indeed drive some stem cell lineages to contribute less to the tissue, possibly explaining the observed collapse in clonality in the muscle.

Hypercompetitive lineages may render other lineages “losers,” but deleterious mutations may also create “loser” lineages cell-autonomously through suboptimal growth, stem cell functional decline, or loss from the tissue of the stem cell or lineage. Is there evidence for this? Quantifying deleterious mutations is a difficult task as these mutations will be either lost or only be present in a few cells. As a workaround, techniques from evolutionary biology have been applied to look at negative selection of point mutations within somatic tissues. By considering the normalized ratio of nonsynonymous to synonymous mutations, one can deduce the amount of detrimental mutations that have been lost. Strikingly, no evidence of negative selection was found in human tissues or in numerous types of cancer (Martincorena et al. 2017; Franco et al. 2018), arguing that the arising point mutations were not detrimental to the survival of the cell in which they arose. It is not yet clear how other types of mutational processes may create burdens on the cell or be selected against. For example, it is more likely that large-scale deletions or mitotic recombination-based LOH, both affecting hundreds to thousands of genes, would reduce cellular fitness. Similarly, de novo transposition events may also impair cellular function through transcriptional deregulation. The extent to which this occurs or might trigger cell death or cell selection mechanisms at the tissue level is not yet known.

Contributions of Persistent DNA Damage to Stem Cell Decline

A large body of literature, including work on HSCs, NSCs, and muscle stem cells, has explored the effects of persistent, induced DNA damage. We will not extensively review this literature here but refer the reader to some excellent reviews of the subject (see Williams and Schumacher 2017; Niedernhofer et al. 2018). Although much of this work is not exclusively on stem cells, collectively these studies show that unrepaired DNA damage can perturb general cellular function in a number of ways including (1) leading to cell-cycle arrest; (2) driving apoptosis or cellular senescence; (3) physically disrupting transcription (Garinis et al. 2009); (4) causing large transcriptomic changes including growth signaling and metabolic pathways (Edifizi et al. 2017); and (5) altering chromatin organization through relocalization of factors to DNA damage sites (Oberdoerffer et al. 2008).

This work raises the question of whether endogenous levels of DNA damage can impact aging and, if so, by which mechanisms. Several studies show a link between increased cellular senescence and stem cell functional decline during aging. An increase in the expression of the senescence-associated cyclin-dependent kinase inhibitor, p16^INK4a, was observed during aging in HSCs, NSCs, pancreatic islet cells, and muscle stem cells, accompanied by a decreased functionality of these stem cell populations during aging that was ameliorated in p16^INK4a–/– mice (Janzen et al. 2006; Krishnamurthy et al. 2006; Molofsky et al. 2006; Sousa-Victor et al. 2014). Although these data support the notion of p16^INK4a-dependent effects on stem cells, it should be noted that p16^INK4a need not be activated through endogenous DNA damage, but could be linked to one of the DNA damage–independent modes of p16 activation, such as changes to chromatin (Martin et al. 2014). Consistent with this, a loss of silencing via BMI1 repression of p16^INK4a was shown to underlie muscle stem cell senescence (Sousa-Victor et al. 2014).

Ongoing DNA damage may also result from alteration in replication kinetics. Interestingly, in mouse adult HSCs, diminished expression of MCM4 and MCM6 during aging resulted in delayed replication kinetics in aged HSCs causing replication stress. Induced replication stress in HSCs resulted in preferential killing of old HSCs, therefore providing a mechanism for functional decline of cycling HSCs during aging (Flach et al. 2014) and a likely explanation for previous observations, in human and mouse HSCs, of increased marks of DNA damage during aging (Rossi et al. 2007; Rübe et al. 2011).

Despite these findings suggesting that DNA damage may impair stem cell activity, the effects on adult stem cells of persistent DNA damage versus genome mutation or DNA damage signaling, must be further teased apart. In addition, determining how different types of endogenous DNA damage or mutagenic processes impact adult stem cells will be important. Finally, future studies are needed to define tissue-specific differences in endogenous DNA damage and their effects on stem cells and niche signals.

TOWARD AN UNDERSTANDING OF DNA DAMAGE AND MUTATION IN ADULT TISSUES

With the recent influx of DNA sequencing of healthy human tissues with age, our views regarding the genomes of somatic cells have been radically challenged. Although these studies provide a descriptive snapshot of evolving somatic genomes, the use of genetically amenable model systems will further improve our understanding of molecular causes and tissue-wide consequences of endogenous DNA damage and somatic mutations in adult stem cells.

Model Systems to Quantify and Study Spontaneous Mutation in Tissues

Early model system studies investigating spontaneous mutation accumulation with age in vivo did so exploiting a transgenic mouse and Drosophila lines with an integrated LacZ reporter gene allowing quantification of mutation at this locus (Dolle et al. 2000; Dolle 2002; Giese et al. 2002, 2010; Busuttil et al. 2007). An age-dependent increase in spontaneous mutation and an intriguing tissue bias of LacZ mutations was found; although mostly point mutations were found in the small intestine, large genome rearrangements were found in the heart (Fig. 4A; Dolle et al. 2000). Though these studies only focus on a single, artificial transgene, they provided an important foundation to begin to study spontaneous mutation in vivo.

Figure 4.

The use of model systems to assay spontaneous mutations in adult tissues. (A) A LacZ reporter to assay spontaneous mutations (Dolle et al. 2000). In this assay, genomic DNA extracted from young/old hearts and small intestines of transgenic mice containing the LacZ reporter is assessed for spontaneous mutation. LacZ encoding plasmids are recovered in bacteria tested for intact LacZ activity. A significant age-related increase in point mutations was detected in small intestines, whereas a significant age-related increase in genomic rearrangements was detected in hearts. (B) A transgenic nucleotide repeat reporter assay to mark spontaneous mutations in mouse intestinal crypts (Kozar et al. 2013). In this system, spontaneous slippage of the cassette during replication allows for expression of the reporter. Wholly populated crypts indicative of the fixation of mutations were shown to increase in an age-dependent manner. (C) Spontaneously arising somatic Notch mutations in Drosophila intestinal stem cells (ISCs) can be detected as clonally expanded Notch mutants/neoplasia in aged flies (Siudeja et al. 2015). Inactivation of Notch leads to neoplasia with an accumulation of ISCs and enteroendocrine cells. Whole-genome sequencing of aged male neoplasia revealed that Notch is inactivated via deletions or structural rearrangements.

Other studies using in vivo lineage tracing of mutant stem cells in the mammalian intestine allowed for the better understanding of stem cell dynamics and the fixation of mutations with age. Kozar and colleagues used mice containing a dinucleotide repeat tract within a reporter gene to mark whether strand slippage happens during DNA replication that consequently results in an in-frame reporter gene, marking the cell. Intestinal crypts that are wholly populated by these marked mutations increased with age (Fig. 4B; Kozar et al. 2013). Similarly, in the human colonic epithelium, mutant stem cell dynamics were revealed by marking known spontaneous mutations that continuously label the ISCs. Interestingly, the investigators found much slower kinetics of crypt clonality likely caused by slower stem cell turnover in humans compared with mouse (Nicholson et al. 2018).

Our laboratory has recently developed a powerful model system to investigate spontaneously arising mutations in adult ISCs in Drosophila (Siudeja et al. 2015). We observed that during aging, spontaneously arising intestinal neoplasia develop in ∼12% of adult males over a rapid period of 6 weeks of adult life. Through application of whole-genome sequencing, we could show that these arise largely as the result of structural variants deleting regions of the Notch gene. As Notch is X-linked and present in a single copy in males, loss of one copy is sufficient to fully inactivate Notch and block proper stem cell differentiation, thereby resulting in the accumulation of large clonal masses of stem cells (Fig. 4C). In addition, we uncovered a second means of genome alteration through LOH likely through mitotic recombination (Siudeja et al. 2015; Siudeja and Bardin 2017). These data suggest that spontaneous mutation occurs frequently in Drosophila adult ISCs, making them a useful model to decipher underlying causes and consequences of stem cell somatic mutation on adult tissues. Important advantages of this model include the rapid acquisition of mutations over 6 weeks of aging, the application of whole-genome sequencing, abundant genetic tools, and the ability to alter environmental conditions.

Alternative Model Systems: Diverse Evolutionary Strategies of Somatic Genome Stability

In addition to fly and mouse models, other models are providing important advances in our understanding of effects of somatic DNA damage and mutation on adult stem cells and tissues. Active work in Caenorhabditis elegans has led to insight into systemic effects of somatic DNA damage (Mueller et al. 2015; Williams and Schumacher 2017). Further investigation in alternative invertebrate models such as planaria, hydra, and nontraditional vertebrate models such as the naked-mole rat may provide surprising solutions to how organisms cope with DNA damage or somatic mutations. Hydra and planaria, for example, have stem cells with an unlimited capacity for self-renewal and do not show signs of aging (Boehm et al. 2013) and the naked-mole rat is a long-lived vertebrate that is cancer resistant (MacRae et al. 2015; Petruseva et al. 2017). Probing into mechanisms in these models may yield unanticipated new insight into potential ways to mitigate the negative effects of mutation.

CONCLUDING REMARKS

In this review, we described how alterations to the DNA of stem cells can disrupt their efficient self-renewal and differentiation, consequently changing the status quo of different tissues, eventually impacting aging and cancer initiation. We highlighted some cell- and tissue-specific mechanisms by which stem cells protect themselves from damage and mutations. Despite these protection mechanisms, damage and acquisition of mutation occur. Ironically, somatic mutation and errors in the DNA repair process are capitalized on in the soma in some instances such as the generation of antibody diversity in vertebrates (reviewed in Li et al. 2004) or in programmed genome rearrangement occurring in lamprey, actively eliminating potentially harmful germline genes from the soma (Wang and Davis 2014; Smith et al. 2018).

Over recent years, advances in the technology to detect mutations and rare events in asymptomatic healthy tissues have revealed the sobering fact that our tissues are peppered with mutations. Healthy tissues are actually mosaics of cell lineages derived from mutant stem and precursor cells. Future studies will better define the forces of selection in healthy tissues and how these relate to cancer. An additional challenge will be to unveil the functional impact of accumulating mutations, linking genotype to diseases and aging phenotypes.

Fundamental questions remain regarding how DNA damage and somatic mutation of stem cells can be manipulated to slow down aging and delay or evade cancer initiation. How can genomic damage be prevented from accumulating in stem cells? Might mechanisms of cell competition be harnessed to replace potentially harmful mutant cells with therapeutic cells? Can tissue-extrinsic factors such as changes in the environment be manipulated to control clonal expansions? Indeed, these questions remain open today and will be active areas of research benefitting from studying diverse genetic model systems.