Chromosomes are arguably the most difficult structures a cell has to maintain over a lifetime. The DNA in each chromosome experiences thousands of chemical alterations and DNA breaks in a single day, and the information each encodes requires strict regulation to maintain cellular identity and function. To manage these tasks, eukaryotes have evolved a complex packaging system known as chromatin, in which DNA is wrapped around a protein core of four different histone dimers and forms a nucleosome, the basic building block of chromatin. Recent studies have indicated that chromatin is a highly dynamic form of nuclear organization that influences DNA stability and gene-expression patterns1, 2. The level of chromatin compaction can be modulated through the chemical modification of histones (Box 1) or of DNA. The more densely the nucleosomes are packed, the more protected is the DNA from chromosomal damage3, but the less accessible it is for transcription2. Highly compacted, transcriptionally silent chromatin is known as heterochromatin, whereas more accessible chromatin is known as euchromatin (Box 2).
Unfortunately, the eukaryotic system of DNA packaging is not immune to the ravages of time. All eukaryotes, including humans, experience changes in chromatin organization and gene-expression patterns as they age. In the late 1990s, a few researchers proposed that changes in chromatin organization underlie ageing-related changes in gene expression and the ageing process4, 5. Changes in gene expression were already known to contribute to cellular senescence6, a possible cause of ageing7, and may provide an explanation for the age-related decline in organ and tissue function in complex organisms.
Although chromatin reorganization was linked to ageing in budding yeast over 10 years ago8, 9, these ideas have remained untested. Recently, a growing appreciation for the importance of chromatin in regulating gene expression and maintaining genomic integrity in complex organisms has reinvigorated interest in the link between chromatin alterations and ageing. In the past 10 years, advances in nuclear imaging technologies have revealed a high level of chromatin organization that is known as the nuclear architecture. In fact, genes from different chromosomes are often in close physical proximity and form discrete foci dubbed transcription factories, which help to orchestrate their transcription and organize the genome in the three-dimensional nuclear space (reviewed in Ref. 10).
The long-term maintenance of the nuclear architecture is vital for the normal functioning of cells and tissues over a lifetime. The dramatic effect of a disturbed nuclear architecture is exemplified by Hutchinson–Gilford progeria syndrome (HGPS), in which a mutation that disrupts the nuclear architecture leads to a disease with symptoms that resemble aspects of normal human ageing, such as loss of hair, restricted joint mobility and atherosclerosis11. Even cells from normal individuals undergo significant nuclear architecture changes in response to stress12, and there are early hints that normal human ageing is associated with alterations in nuclear architecture13.
In this review, we discuss the causes and consequences of changes in nuclear architecture with age. We focus on the role of epigenetic gene regulation during the ageing process, with an emphasis on drawing parallels between observations in yeast and mammals. We propose that a conserved DNA-damage response induces cumulative changes in chromatin structure and nuclear architecture that are important driving forces behind the inexorable changes that occur in organisms over time. These changes include a decline in genomic integrity, alterations in gene transcription and a loss of vitality — the series of changes we commonly refer to as ageing.
Heterochromatin alterations in yeast
In the 1990s, a series of discoveries in the budding yeast Saccharomyces cerevisiae identified a mechanistic link between epigenetic silencing and ageing. The replicative age of a yeast cell is the number of offspring it produces before undergoing senescence
(
23–30). Like in all eukaryotes, heterochromatin in yeast serves two main purposes: it maintains certain genes (such as the yeast mating-type loci) in a silent state through cell division, and it stabilizes the highly repetitive parts of the genome (the telomeres and ribosomal DNA (rDNA)), preventing them from recombining, fusing and breaking. Accordingly, yeast cells that lack crucial heterochromatin factors are infertile because both types of mating-type genes are expressed concurrently. Furthermore, telomeres become extremely short and tend to fuse, which causes major problems during cell division (reviewed in Ref. 14).
Sir2 mediates heterochromatin formation. One of the key regulators of yeast heterochromatin is Sir2 (silent information regulator-2), an NAD+-dependent histone deacetylase that predominately removes acetyl groups from Lys16 of histone H4 (Ref. 15). At the mating-type genes and telomeres, Sir2 interacts with its structural partners Sir3 and Sir4 (Refs 16, 86), which regulate and direct its deacetylase activity. Binding of the Sir4–Sir2 heterodimer to DNA nucleates DNA silencing by recruiting Sir3 to form the Sir complex. Driven by Sir2-dependent histone deacetylation, the Sir complex promotes heterochromatin formation by spreading along chromatin through cycles of recruitment of other Sir complexes86.
At the rDNA locus, Sir2 is a crucial component of a network of protein–protein interactions that regulate both silencing and DNA stability. In all eukaryotes, rDNA is organized as one or more arrays that contain 100–10,000 repeating units that are sequestered in the nucleolus. It is the highly repetitive nature of the rDNA that renders it particularly susceptible to recombination. The control of rDNA stability in yeast is well understood. Yeast rDNA is silenced and stabilized by the regulator of nucleolar silencing and telophase exit (RENT) complex, which consists of Sir2, Net1 and Cdc14 in a 1:1:1 ratio. Net1 recruits Sir2 to the rDNA17, where it forms a complex with, and thereby sequesters, the Cdc14 phosphatase. Cdc14 is involved in cell-cycle regulation and is kept inactive while in the RENT complex18, 19. Whether it also has a role in rDNA silencing remains unclear.
The first clear genetic link between heterochromatic silencing and ageing came from a genetic screen that isolated a gain-of-function mutation in SIR4, known as SIR4-42 (Refs 8, 20). SIR4-42 generates a truncated Sir4 protein, which cannot bind to telomeres and mating-type genes, thus abolishing silencing at these loci. The mutant Sir4 protein targets a greater amount of Sir2 and Sir3 to the nucleolus, which, in turn, correlates with an increase in mean lifespan by 40%. Thus, recruitment of Sir4 to the nucleolus early in life seems to slow ageing and extend lifespan. Examination of wild-type yeast cells showed that the redistribution of Sir proteins to the nucleolus is not limited to mutant Sir4, but reflects a normal process during yeast ageing8, albeit one that occurs later in life than in a SIR4-42 mutant.
Around the same time as the characterization of SIR4-42, a study of the yeast WRN homologue also pointed to the nucleolus as an important site that influences ageing. In humans, loss-of-function mutations in the human WRN gene cause Werner syndrome (WS), a progeroid disease that mimics many aspects of normal ageing including atherosclerosis, diabetes and dramatically aged skin by age 40. WRN and its yeast homologue SGS1 encode RecQ DNA helicases that function in DNA repair and recombination21, 22. In the absence of RecQ helicases, genomes are highly unstable, especially at repetitive loci. Deletion of SGS1 results in hyper-recombination at the rDNA and premature ageing that is associated with the relocalization of Sir3 to the nucleolus, which becomes dramatically enlarged and fragmented23(Fig. 1a). Taken together, these findings suggest that the relocalization of chromatin-modifying proteins is a normal event during yeast ageing and can have a dramatic effect on genomic stability and lifespan.
Figure 1 | Comparison of age-related changes in nuclear architecture between yeast and mammalian cells.
a | Schematic of accelerated ageing (top) and normal ageing (bottom) in a replicating yeast nucleus. In young yeast cells, telomeres, mating-type loci (MATa and MAT
) and ribosomal DNA (rDNA; orange) are silenced by silent information regulator-2 (Sir2)-containing complexes (purple circles). In wild-type yeast, homologous recombination at the highly repetitive rDNA locus generates extrachromosomal rDNA circles (ERCs) during cell division. Lack of the DNA helicase Sgs1 causes genomic instability at the rDNA and leads to increased ERC formation, accelerated changes in nuclear architecture and premature ageing23 (top). Sites of DNA damage and both ERCs and rDNA recruit components of the Sir2-silencing complex, causing a loss of silencing at telomeres and mating-type loci (green represents areas of transcriptional derepression). b | Changes in nuclear architecture of human cells in a model of accelerated ageing (top) and normal ageing (bottom). Young cells show dense, transcriptionally inaccessible perinuclear heterochromatin surrounding less densely packed, transcriptionally active euchromatin. Grey circles represent sites of facultative heterochromatin and blue ovals depict constitutive, perinuclear heterochromatin (Box 2). In Hutchinson–Gilford progeria syndrome (HGPS), a defect in the nuclear lamina component lamin A leads to an accelerated loss of pericentromeric heterochromatin and concomitant changes in nuclear architecture that are accompanied by transcriptional deregulation13, 58, 59 (green areas). Similar changes have been observed during normal ageing. Cellular stress can cause the formation of repressed senescence-associated heterochromatin foci (SAHFs)12.
The yeast findings raised an intriguing question: how does the nucleolus affect ageing? In 1997, ageing of yeast was shown to stem from the inherent instability of rDNA9. rDNA is highly repetitive and therefore prone to homologous recombination. Recombination between rDNA sequences results in the excision of a single circular molecule of DNA (an extrachromosomal rDNA circle (ERC)) that is replicated during S phase. Roughly 12 divisions later, the nucleus becomes packed with >1,000 ERCs that cause cell death, presumably by titrating essential proteins from the rest of the genome9. In young yeast cells, these recombination events are controlled by Sir2, which directly binds to the rDNA and deacetylates the surrounding histones, resulting in chromatin compaction and rDNA stabilization.
This discovery provided a direct link between heterochromatin and ageing and led to a testable prediction. Decreasing heterochromatin at the rDNA locus should accelerate ageing, whereas increasing it should extend lifespan. This hypothesis was confirmed by manipulating SIR2: deletion of the SIR2 gene led to a loss of rDNA silencing, elevated rDNA recombination and accelerated ageing, whereas integration of an extra copy of the SIR2 gene increased rDNA silencing, suppressed rDNA recombination and extended lifespan by 30%17.
Heterochromatin reorganizes in response to DNA damage. Why does Sir protein redistribution occur during ageing? The answer may come from the growing appreciation of the importance of chromatin in maintaining genomic stability and facilitating DNA repair. During the DNA-repair process, chromatin needs to be unpacked and reassembled, which has an important influence on the rate and type of repair. One possibility is that the relocalization of Sir proteins is an active defence process that the cell initiates to stabilize its DNA.
In 1999, four studies showed that a single DNA break is sufficient to illicit a DNA-damage-checkpoint response that releases Sir proteins from mating-type loci and telomeres and relocalizes them to the DNA break, possibly to facilitate the repair process24, 25, 26, 27. Indeed, genomic DNA from SIR2 mutants was shown to be more susceptible to cutting by an endogenously expressed EcoRI endonuclease25. However, no defect in DNA end-joining was observed using a plasmid-based assay26, a discrepancy that may be explained by the lack of chromatin on plasmid DNA. This finding fits with the observation of Tyler and colleagues, who found that Sir2 and other histone deacetylases modify the chromatin that surrounds the break site in a temporally coordinated manner, which appears to be a prerequisite of efficient repair28. We refer to this process as the relocalization of chromatin-modifying factors (RCM) response (Fig. 1a).
The damage-mediated relocalization of Sir proteins appears to have little effect on long-term genomic silencing patterns in young yeast cells. However, the accumulation of DNA damage and increased rDNA instability in old yeast cells eventually leads to a chronic RCM response, alterations in silencing and irreversible genomic changes. A role for DNA damage in age-related genomic instability has also been reported for loci other than the rDNA. DNA-break-induced loss of heterozygosity at artificially generated heterozygous loci was found to increase dramatically with age and, depending on the locus, was aggravated in the absence of Sir2 (Ref. 29), again indicating a protective role for heterochromatin. Both increased susceptibility to DNA damage and an accumulation of defective DNA-repair enzymes might explain how ageing can promote this global genomic instability.
Nuclear changes in ageing mammals
How do these findings in yeast relate to what is known about the role of heterochromatin and epigenetic silencing in mammals? Although the human genome and human ageing exceed the yeast model in complexity, the yeast studies may help us to understand fundamental processes that govern conserved aspects of ageing. Indeed, changes in heterochromatin composition and structure with age have been reported in several species including humans.
Lessons from human progeroid syndromes. WS, HGPS and ataxia telangiectasia (AT) are rare genetic premature ageing disorders that demonstrate both the dramatic consequences of defects in nuclear architecture and the diverse sets of genes that are involved in its maintenance. As mentioned above, WS is characterized by a genome that is highly unstable owing to the lack of functional RecQ helicase22. HGPS shows similarities to WS but proceeds more rapidly11. One known cause of HGPS is a single base change in the LMNA gene, which encodes lamin A, an essential structural component of the nuclear membrane30. The mutant LMNA gene generates a truncated splice variant that disturbs the structure of the nuclear membrane and causes large changes in the nuclear architecture. Nuclei from patients with HGPS are characterized by a dysmorphic shape and a loss of heterochromatin-related proteins that are associated with the nuclear membrane, such as heterochromatin protein-1 (HP1), as well as altered histone-modification patterns that reflect a general disturbance in silent heterochromatin (Fig. 1b). Interference with aberrant LMNA splicing can reverse the structural defects that are typical for HGPS in cell culture, which demonstrates a direct causal relationship between the LMNA gene and HGPS31. The truncated LMNA splice variant has also been found in naturally old humans, which implicates changes in lamin A in the normal ageing process13. A conserved role for lamin A in the ageing process is consistent with the recent finding that neuronal cells in the nematode Caenorhabditis elegans also show changes in the nuclear architecture in aged animals, and a loss-of-function mutation in the worm orthologue of lamin A causes a decrease in life expectancy32.
Like WS and HGPS, AT is characterized by a defective nuclear architecture, progressive neurological degeneration, growth retardation, genomic instability and premature ageing33. Patients with AT have an inherited defect in the AT mutated (ATM) gene, which encodes a protein kinase that initiates the DNA-repair cascade. DNA repair and ATM in particular seem to be required for telomere maintenance34, 35, and defective ATM can disturb the interactions between telomeres and the nuclear matrix34. The contribution of this effect on AT pathology remains unclear and is complicated by the range of defects that are observed in ATM-defective cells. The yeast ATM orthologue Tel1 has also been linked to telomere maintenance36, which further suggests that pathways that are involved in the maintenance of nuclear architecture are highly conserved.
Chromatin-structure changes and epigenetic silencing. Changes in nuclear architecture do not appear to be restricted to defects in the structural components of the nucleus. An age-related loss of epigenetic silencing at certain repetitive elements was reported almost 20 years ago. Specifically, the major satellite repeats that form heterochromatic chromatin structures around the centromeres of every chromosome were shown to be more transcriptionally active in aged cardiac tissue, which suggests a progressive loss of silencing of these elements37. Given the number of repetitive elements in mammalian genomes, a reduction in repeat-associated heterochromatin would be consistent with significant changes in nuclear architecture. Shen et al. recently reported a possible mechanistic link between mammalian ageing and changes in heterochromatin38. Older individuals show altered activity in their histone-modifying enzymes, which causes a loss of perinuclear heterochromatin and concomitant changes in gene expression. These observations are reminiscent of the chromatin changes that occur during yeast ageing and in HGPS, and raise the possibility that changes in perinuclear architecture contribute to normal ageing in mammals (Fig. 1b).
Numerous other epigenetic changes in nuclear architecture and gene expression have been associated with ageing. More than a decade ago, Imai and colleagues showed that collagenase, a gene associated with cellular ageing, is differentially regulated during cellular senescence — a phenomenon that is often referred to as cellular ageing39. This effect appears to be due to changes in the subnuclear localization of the collagenase gene as cells undergo senescence. In young cells, the collagenase gene is repressed by the transcription factor OCT1. A considerable proportion of OCT1 was found in the heterochromatic nuclear periphery, where it colocalized with lamin B, a component of the nuclear membrane. This interaction was abrogated in senescent cells and, concomitantly, collagenase repression was lost39. On the basis of these findings, the authors proposed a model of age-associated heterochromatin reorganization that would account for such transcriptional changes in a global manner5.
This idea gained support from recent studies of cellular senescence, most notably by Lowe and colleagues, who found that senescence is associated with an overall increase in non-pericentromeric, facultative heterochromatin domains, known as senescence-associated heterochromatin foci (SAHFs; Fig. 1b)12. SAHFs form repressive chromatin structures that can be found at, but are not limited to, promoter regions of certain cell-cycle regulators, in particular target promoters of the cell-cycle regulator E2F. This finding led to the hypothesis that SAHFs promote senescence through direct repression of growth-promoting genes. Although the repression of cell-cycle regulators is an important function of SAHFs during cellular senescence, the frequency and distribution of these foci suggests a much broader impact of SAHFs. This notion is further supported by the finding that the formation of SAHFs appears to rely on the recruitment of proteins from promyelocytic leukaemia nuclear bodies40, which have been implicated in numerous cellular processes including transcriptional regulation, apoptosis and cellular defence in response to stress (most notably to DNA damage41).
A connection between senescence-associated heterochromatin formation and mammalian ageing has recently been made using baboon skin fibroblasts42. Tissue from older individuals accumulates cells containing heterochromatic foci that are reminiscent of SAHFs in senescent cells. These foci were found in >15% of the total cell population in aged tissues, which suggests that a significant fraction of aged tissue may be expressing markers of senescence and undergoing large-scale heterochromatic changes. Importantly, the emergence of heterochromatin foci occurred simultaneously with telomere shortening, which points to a shift from stable, perinuclear heterochromatin to induced, or facultative, heterochromatin. These studies raise the intriguing possibility that the age-associated loss of genomic silencing detected in previous studies may be linked to, or caused by, the formation of SAHF-like heterochromatic foci, a phenomenon reminiscent of the RCM response that occurs in yeast in response to DNA damage and ageing. However, it is important to keep in mind that cellular senescence — although it is a likely contributor to cancer and organismal ageing — does not equal the complex physiological processes that, together, define what is called ageing. The relevance of the aforementioned findings to the functional decline of higher organisms remains to be elucidated.
Changes in epigenetic gene regulation
In yeast, the redistribution of chromatin-modifying enzymes to the rDNA destabilizes telomeres and exposes them to degradation, and desilences the mating-type loci, causing sterility. In mammals, ageing has been associated with large-scale changes in both nuclear architecture and chromatin structure. How might these changes contribute to the ageing process? Because numerous genes are either directly or indirectly regulated by (nearby) heterochromatic regions2, it is possible that changes in the epigenetic make-up of a cell might alter its gene-expression patterns, thereby changing its genomic identity. In this section, we discuss evidence for age-related changes in gene expression across species and a possible role for DNA damage as an evolutionarily conserved mechanism that could drive changes in nuclear architecture and gene expression over a lifetime.
Gene expression changes with age. With the emergence of genomic technologies, age-associated alterations in gene-expression patterns have now been documented in several species (Table 1). An impressive example of age-related alterations in gene expression comes from a study by Yankner and colleagues43. The authors examined gene-expression patterns in the human cortex, covering a broad age range, and observed progressive changes in gene-expression patterns with age. Comparisons between age-related gene-expression patterns across tissues and even species reveals that several functional gene groups are similarly affected; the increased expression of stress-response genes and inflammatory genes is one example (Table 1). Such changes may reflect a response to age-related stress and are thought to counteract age-related tissue damage.
Could some of the reported transcriptional changes be a cause rather than a common consequence of ageing? The majority of significantly altered transcripts cover a broad and seemingly random range of genes, some of which may interfere with proper cell function; examples include the deregulation of cell-cycle genes in post-mitotic neurons44 and neuronal factors in muscle tissues45, 46. Such changes may pose a problem for proper cell function and thereby directly contribute to organ decline and ageing. In addition, although there are groups of genes that are shared between species, most age-related transcriptional changes are not shared, even between closely-related species such as monkeys and humans47. Even within species, the majority of transcriptional changes appear to differ between tissues48. If age-related transcriptional changes were solely a consequence of the ageing process, one might expect similar changes between species and, certainly, organs of the same animal. The observed variation between species and tissues underlines the apparent randomness of these changes. A stochastic component that affects gene-expression changes is also supported at the level of individual cells, as transcriptional profiles can differ between adjacent cells from the same aged tissue of a mouse49. However, the combined transcriptional changes within a given tissue appear to be rather reproducible, which implies that the pre-existing nuclear environment of a cell or tissue may also determine which genes become preferentially deregulated with age. This hypothesis is further supported by work in flies, which demonstrates that some genes that are deregulated during ageing localize to the same chromosomal region; this finding indicates a role for global changes in nuclear architecture with ageing50. Clearly, more work is required to understand the underlying causes of age-related transcriptional changes and their contributory relationship, if any, to the ageing process.
Calorie restriction counters gene-expression changes. Calorie restriction, a dietary regimen that extends the lifespan of numerous organisms, also delays the majority of age-related gene-expression changes in mice and, to a certain extent, in flies45, 50. It is currently unclear whether the effect of calorie restriction on gene expression underlies its beneficial effect on lifespan or is merely a consequence thereof. Findings in yeast suggest that there may be a causal link: Sir2 not only facilitates heterochromatin and promotes DNA stability, but is also a mediator of calorie restriction51, 52. Furthermore, in rodents and humans, the levels and activity of the Sir2 orthologue SIRT1 increase in response to calorie restriction53, which raises the possibility that the enzyme may also be involved in age-related changes in nuclear architecture and could be a mediator of caloric restriction in mammals. It will be interesting to explore to what extent SIRT1 directly regulates gene expression (so far, only a few examples are known54, 55, 56) and whether SIRT1 facilitates heterochromatin formation or promotes genomic stability in mammals. If so, perhaps some of the age-related changes in gene expression and genomic instability in mammals can be traced to the relocalization of SIRT1 during ageing, as is the case for the yeast orthologue Sir2.
DNA damage causes genome-wide transcriptional changes. Why gene-expression patterns change during ageing is not known; however, it has been speculated that a major underlying cause of these changes is DNA damage (Box 3). Yankner and colleagues first demonstrated a direct link between global age-related gene repression and oxidative DNA damage to the promoters of the repressed genes43. Oxidative DNA damage is caused by an accumulation of reactive oxygen species (ROS), which can be observed with age. ROS are highly unstable, reactive by-products of mitochondrial respiration and can damage several cellular components including lipids, proteins and DNA. Oxidative stress has, therefore, been proposed to be a significant contributor to cellular and organismal decline and its role during ageing has been extensively investigated (reviewed in Ref. 57). The idea that increased ROS generation may be responsible for gene-expression changes is further supported by comparisons of cerebellar and cortical gene-expression patterns of aged monkeys and humans. The tissue with higher respiratory activity and presumably higher propensity for DNA damage, in this case the cortex, showed greater alterations in gene regulation with age47. However, more work is needed before a causal relationship can be declared between respiration, DNA damage and gene-expression changes in the brain.
As previously mentioned, gene expression is not only altered during ageing in mice, but can vary between single cells in a homogeneous tissue. These changes can be accelerated by oxidative DNA damage in cell-culture experiments49. This effect on gene expression was long-lasting, persisting up to 9 days after stress — a finding that is reminiscent of long-lasting stress-induced changes in chromatin structure12. This study in particular supports the idea that randomly distributed sites of DNA damage can influence gene expression with age. It also implies that much of the variability in transcription among single cells cannot be detected by whole-tissue analyses. Given that most of the studies so far have examined whole tissues, the number of genes that are deregulated with age is likely to have been underestimated.
The fact that DNA repair is impaired in mouse models of HPGS suggests that DNA damage also has a role in this premature ageing syndrome58. cDNA microarray analysis of fibroblasts from patients with HGPS showed a range of gene-expression changes that cover gene-ontology groups as diverse as signal transduction, transcriptional regulation, cell-cycle regulation and development, consistent with a global deregulation of gene expression59. A recent report further highlighted the dramatic effect of DNA-repair defects on age-related gene expression changes60: Hoeijmaker’s group identified a novel mutation in a member of the well characterized xeroderma pigmentosa (XP) complementation group, XPF. XPF is part of an endonuclease complex that is involved in the repair of single-nucleotide lesions and DNA interstrand crosslinks. Humans with this particular mutation show dramatic progeroid symptoms and usually die in their teens. A mouse model for this disease shows gene-expression patterns that resemble those of normally aged mice, and metabolic gene-expression changes appear to be particularly conserved. The authors suggest that this may reflect a common stress response that provides protective tissue maintenance. Although this correlation was impressive, a significant number of other ageing-like transcriptional changes that were reported for this XPF mouse model do not fall into stress-response categories. Together, these observations suggest that, although some of the age-related transcriptional changes constitute a stress response, a significant proportion of the changes occurs in a seemingly random fashion.
DNA damage alters chromatin
In yeast, DNA damage induces an RCM response that disrupts heterochromatin and alters gene expression. A wealth of literature has recently implicated chromatin-remodelling enzymes in the DNA-repair process in yeast and other more complex organisms (reviewed in Ref. 61). DNA double-strand break (DSB) repair involves the recruitment of histone modifiers to the repair site, together with other repair complex components such as Ku70/80 and DNA ligase IV. DSBs trigger the DNA-damage sensor kinases ATM, ATR or DNA-PK, which phosphorylate the surrounding histones H2A and H2AX in yeast and mammals, respectively62, 63, 64. The extent of this modification can reach into megabases, potentially affecting the epigenetic regulation of several genes65.
In yeast, chromatin-remodelling factors that are involved in DNA repair, such as histone acetyltransferases (HATs) and histone deacetylases (HDACs) as well as histone methyltransferases, are then recruited to the break site28, 66, 67. Although the precise role for these chromatin-remodelling complexes during DNA repair is not fully understood, it is presumed that they dictate the type of repair and facilitate the repair process by changing the chromatin composition around the site of damage. In yeast, HATs (such as Esa1) and HDACs (in particular Sin3, Rpd3 and the sirtuin Hst1) are part of chromatin-remodelling complexes that promote an ordered and dynamic progression of histone modifications over the time-course of DSB repair28. Importantly, these proteins are not specialized DNA-repair enzymes but, rather, chromatin-modifying enzymes that have several functions outside DNA repair, including gene silencing68. Recruitment of such factors to sites of damage may therefore be accompanied by a loss of function at their original sites, as is the case for the Sir complex25, 27.
Although a role for histone deacetylation during DNA repair has not been demonstrated in mammalian cells, there is accumulating evidence for chromatin remodelling besides H2AX phosphorylation. For example, histone methylation has been directly linked to the recruitment of DNA-repair factors in human cells. Specifically, methylation of Lys79 on histone H3 recruits the p53-binding protein-1 (53BP1). Recruitment of these factors is thought to tether the DNA-repair complex to the site of damage69, 70. Importantly, this mechanism appears to be evolutionarily conserved between yeast and mammals because the fission yeast 53BP1 homologue Crb2 is also recruited to DSBs, a process that requires methylation of Lys20 on H4 (Ref. 66). The fact that 53BP1 can also be found in the DNA-damage-associated heterochromatin foci of aged monkey fibroblasts42 further corroborates the idea that histone modifiers may have a crucial role during the DNA-damage response in mammals and points towards DNA damage as an inducer of global changes in chromatin architecture.
The epigenetic balance hypothesis
As mentioned earlier, the formation of transient heterochromatic foci around sites of DNA damage may explain how DNA damage might directly mediate gene repression12. Consistent with this notion, many of the gene-expression changes that are observed in aged individuals occur in a stochastic fashion, as does most DNA damage49, 71. It is also conceivable that certain genomic regions are more prone to damage than others, which could explain some of the predictable, co-regulated changes that are observed between aged individuals of the same species. Indeed, DNA breaks occur more commonly in certain euchromatic, active regions of the genome3. Furthermore, fragile sites on chromosomes that are prone to breakage are well documented in mammals72. It will be interesting to investigate whether these sites of preferential DNA damage correlate with loci that become deregulated with age.
Although it is conceivable how DNA damage might lead to gene repression, it is less obvious how age-related stress and DNA damage could account for gene activation. This question is particularly important to address because the fraction of genes that are significantly upregulated with age roughly equals or even exceeds the fraction that is downregulated (Table 1). Moreover, transcript levels overall appear to be increased in old animals73. One explanation may be that DNA damage interferes with the expression of transcriptional repressors, leading indirectly to an induction of sets of target genes. However, the diversity of genes that show increased expression with age indicates that other processes may also be at work.
Based on what we know about yeast ageing and the DNA-damage-induced RCM response, we propose the following model for how DNA damage might lead to global changes in gene expression to promote ageing in mammals. We refer to this model as the ‘epigenetic balance hypothesis’ (Fig. 2). In this model, age-related gene-expression changes are manifestations of the redistribution of chromatin modifiers from one genomic locus to another. The model also encompasses the idea that DNA damage mediates chromatin remodelling and changes in nuclear architecture that occur over a lifetime, which fits with evidence that oxidative stress and DNA damage can accelerate the ageing process. Based on the observations of Tyler and colleagues28, it is plausible that chromatin modifications during DNA repair are never fully restored to their pre-damaged state, resulting in progressive alterations in both chromatin-modification patterns and gene expression28.
Figure 2 | Redistribution of heterochromatin-associated factors as a cause of age-related changes in nuclear architecture and gene expression.
In the young, the nuclear architecture of each tissue is well defined; it comprises tightly packed perinuclear heterochromatin (blue) and patches of tissue-specific, developmentally controlled facultative heterochromatin islands (grey; represented by gene A) in the otherwise transcriptionally active euchromatin (represented by gene B). This generates cell-type-specific gene-expression patterns. Repetitive DNA is part of perinuclear heterochromatin and is transcriptionally repressed. Position-effect variegation (Box 2) can cause repression of nearby coding regions (gene C). Silencing complexes in constitutive and facultative heterochromatin are different (grey and blue ovals), but contain several identical chromatin-modifying enzymes (green ovals). Green tags represent transcriptionally permissive histone modifications, whereas red tags represent non-permissive histone modifications. The age-associated accumulation of DNA damage triggers global changes in nuclear architecture, including the formation of senescence-associated heterochromatin foci (SAHFs) in euchromatic DNA (grey) and a gradual loss of perinuclear heterochromatin. This loss may be a direct consequence of a redistribution of essential silencing factors, in particular histone-modifying enzymes, to sites of DNA damage or SAHFs (arrows). This process causes changes in nuclear architecture and tissue-specific gene expression patterns. In this example, gene A is activated owing to the loss of facultative heterochromatin, gene B is silenced in response to DNA damage and gene C is derepressed owing to changes in position-effect variegation. The corresponding changes in histone modifications are shown.
The consequences of the redistribution of chromatin-modifying enzymes would be twofold. Previously silent regions may become transcriptionally active, leading to ectopic gene expression and, possibly, destabilization of previously heterochromatic repetitive DNA. By contrast, genes may become repressed near sites of DNA damage through remodelling processes that are similar to the formation of SAHFs. According to the model, nuclear structure and organization is progressively and inexorably altered over time, resulting in the functional decline of cells and tissues. The model is consistent with both the stochastic changes in gene expression as described by Vijg and colleagues49 and the reproducible tissue-specific transcriptional changes that occur as organisms age, which will be dictated by the original architecture of a given tissue.
Perspective
In this review, we propose that a redistribution of chromatin modifiers is a natural, protective response to DNA damage, but may lead to epigenetic changes that affect genomic integrity and, thereby (at least in part), account for changes in gene expression that appear to be a hallmark of the ageing process. Although this epigenetic balance hypothesis presents an appealing explanation of what we currently know about age-related changes in nuclear architecture and gene expression, it is certainly not the only way to explain the observed effects of ageing. For example, it can be argued that changes in gene expression mediate changes in chromatin structure, which in turn enhances susceptibility to DNA damage. In this scenario, gene-expression changes would precede DNA damage. Despite convincing evidence for DNA damage as a trigger of transcriptional changes49, 60, it is conceivable that a change in the transcription status of a gene determines its susceptibility to DNA damage. A comprehensive (computational) analysis or the genome-wide mapping of sites of DNA damage and localization of chromatin-remodelling enzymes may shed light on the complex interplay between transcriptional activity and DNA damage.
Several findings suggest that DNA damage is a main trigger of nuclear ageing, supporting the free-radical theory of ageing74 (see the accompanying Opinion article by Pelicci and colleagues in this issue). However, it could also be argued that chromatin structure is directly affected by the ageing process through an as-yet-unknown mechanism that leads to increased DNA damage and a permanent damage response that alters gene-expression patterns in a similar way to the model proposed in this review.
Over the coming years, as researchers use mammalian models to map the global pattern of chromatin modifications during ageing, it should become clear whether changes in the epigenetic balance due to the RCM response underlie aspects of the ageing process. For now, perhaps we should pause for a moment to consider the remarkable ability of cells to maintain their chromatin and gene-expression patterns for as long as they do, overcoming daily chemical and physical damage, in some cases for many decades.
Previous Entries
Categories
Tag Cloud
Blog RSS
Comments RSS
Last 50 Posts
Back
Void 
Life 
Earth 
Wind 
Water 
Fire