You are on page 1of 8


cells have chromosomes that are abnormally

distributed and frequently present in excess
of 46 (REF. 1). Paradoxically, non-tumorous
DNA repair, genome stability and cells and plants with an asymmetrical or
unbalanced chromosome distribution
cancer: a historical perspective grew less vigorously than normal cells,
whereas cancer cells were characterized
by an enhanced growth capacity 1. Work in
Penny A. Jeggo, Laurence H. Pearl and Antony M. Carr Drosophila melanogaster had revealed that
chromosome aberrations correlated with
Abstract | The multistep process of cancer progresses over many years. The the formation of genetic variants and, in the
prevention of mutations by DNA repair pathways led to an early appreciation of a late 1920s, Hermann Muller demonstrated
role for repair in cancer avoidance. However, the broader role of the DNA damage that exposure of D. melanogaster to X‑rays
response (DDR) emerged more slowly. In this Timeline article, we reflect on how induced “transmutation” of a gene, causing
our understanding of the steps leading to cancer developed, focusing on the role both visible chromosome aberrations and
heritable phenotypic variations3,4.
of the DDR. We also consider how our current knowledge can be exploited for Intriguingly, in 1775, Percival Pott
cancer therapy. made the seminal observation that there
was a high incidence of scrotal cancer
Carcinogenesis has long been recognized carcinogenesis, leading to the much in boys who assisted chimney sweeps,
as a state of uncontrolled growth of our more recent but critical understanding and he linked this finding to exposure to
own cells. The earliest models proposed that oncogene-induced replicative soot 5–7. This represented the first evidence
the notion of a mutational event, even stress promotes genomic instability. We for a work-related and environmental
before James Watson and Francis Crick’s provide a perspective on how the notions cause of cancer. By 1955, shortly after
seminal discovery of the structure of of tumour initiation and progression the discovery of the structure of DNA,
DNA. By the 1980s, the initiation step emerged, describing the key concept that it was well appreciated that exposure to
of carcinogenesis was understood to tumour progression is inexorably linked chemical mutagens could increase DNA
necessitate the generation of mutations, with to disruption of the DDR (FIG. 1). Finally, mutation rates, with a correlation between
the concept of environmental mutagens and we consider the ironic conundrum that, mutagenesis and carcinogenesis being
failed DNA repair being central to many although targeting the DDR can provide hypothesized although certainly not
models. By contrast, an understanding of treatment strategies, it is the misregulation consolidated2. Remarkably, the suggestion
the basis underlying tumour progression or of the DDR that is often the route by which that there could be cancer susceptibility
outgrowth unfolded relatively slowly, and an tumour cells evade therapy. genes was also proposed2.
appreciation of the critical role of the DNA
damage response (DDR) took even longer Mutagenesis underlies carcinogenesis Linking mutagens, carcinogens and DNA
to solidify (see BOX 1 for useful definitions). In the early twentieth century, long With the definition of the structure of DNA
Indeed, even as the twenty-first century before the structure of DNA was defined, in 1953 (REFS 8,9) and the understanding
began, DNA repair remained a relatively Theodor Boveri proposed that a cancer cell that genetic mutation represented a
minor component of the broad field of was a changed normal cell and advanced the change in the chemical structure of the
cancer research. theory that tiny microscopic bodies called DNA molecule, the first clear connections
It is now appreciated that tumour chromosomes were abnormally distributed between the processes of mutation and
progression necessitates the downregulation in cancer 1. As the notion of hereditary units carcinogenesis were made by Phil Lawley,
of damage surveillance mechanisms and an or ‘determinants’, and later ‘genes’, emerged who worked at the Chester Beatty Research
increase in genetic and epigenetic instability alongside their relationship to chromosomes, Institute (United Kingdom; now known as
to achieve uncontrolled proliferation and the idea that permanent changes to these the Institute of Cancer Research). Working
the adaptability associated with aggressive ‘genes’ (defined as mutations) could with Peter Brooks, he showed that many
tumours. In this Timeline article, we underlie heritable biological phenotypes classic alkylating agents worked by reacting
describe early concepts of mutation and became conceptualized. From there, it did directly with DNA to form stable chemical
cancer that predate knowledge of the not require a great leap to appreciate that adducts that could disrupt the template
structure of DNA, and we summarize such mutations might underlie the origin function of the DNA molecule10,11. This led
how the links between mutagenesis and of the biological variation observed in directly to their seminal observation that
carcinogenesis were established. We go on to cancer (reviewed in REF. 2). In the 1930s, the carcinogenicity of polycyclic aromatic
discuss the early studies of viral oncogenes it was recognized that, whereas a normal hydrocarbons — the likely causative agents of
and the insights that they provided into human cell has 46 chromosomes, cancer scrotal cancer in Pott’s chimney sweeps and


© 2016 Macmillan Publishers Limited. All rights reserved


also the classic carcinogenic components any significant way. There was, however, an removal of helix-distorting UV‑induced
of tobacco smoke — was directly correlated emerging recognition of the role that DNA photodimers from DNA, which explains the
with their ability to form DNA adducts, repair mechanisms might have in mediating highly specific skin cancer predisposition
providing an unambiguous link between the resistance to alkylating agents used for of patients with XP.
initiation of cancer and chemical changes cancer chemotherapy 21. A second clear example of a defective
in DNA12. The importance of these early DNA repair pathway being responsible
findings and the functional link between Insights from DNA repair disorders for cancer initiation emerged in the early
mutagenesis and carcinogenesis are A notable exception to this picture was 1990s: patients with Lynch syndrome
demonstrated by the development and later a seminal observation made in 1969 by (also known as hereditary non-polyposis
adoption of tests classifying carcinogens on Jim Cleaver22, who was studying the rare colorectal cancer) — a familial pattern
the basis of this relationship13. autosomal-recessive cancer predisposition of colorectal cancer characterized by
Lawley and Brooks were also among the syndrome xeroderma pigmentosum (XP), microsatellite repeat instability — were found
first to document biological repair processes which affects 1 in every 250,000 individuals. to carry mutations in the human homologues
for chemical and environmental damage Patients with XP have a 1,000‑fold increased of the bacterial mismatch repair proteins
to DNA10, a theme very actively adopted chance of developing skin cancer but MutS and MutL24–27. Both of these inherited
by many laboratories14–19 (see REF. 20 for a display almost normal levels of tumour diseases reinforced a model of cancer
review). Research over the next 30 years presentation in other tissues23. Cleaver initiation in which random unrepaired point
progressively revealed a plethora of repair found that cells from patients with XP mutations eventually result in an alteration
pathways for chemical lesions of DNA, were defective in the ability to repair DNA of the coding sequence of a key oncogene
primarily in bacteria. Subsequently, human damage caused by ultraviolet (UV) light 22. or tumour suppressor, thus initiating the
homologues for many of these repair The DNA repair defects in most (but not first step in cell proliferation and enabling a
enzymes and pathways were identified. For all) XP cells were subsequently shown to subsequent cascade of mutagenic events in
the most part, these systems — or possible result from mutations in components of the these precancerous cells.
defects in them — were not associated with human nucleotide excision repair system22,23. A prediction that emerged from
the initiation or the progression of cancer in Crucially, this process is responsible for the these studies of patients with hereditary
predisposition to cancer was that mutations
in DNA repair genes might frequently arise in
Box 1 | DNA damage response processes of relevance for cancer
cancer cells. As we discuss below, this has
DNA repair certainly proved to be the case. However, the
There are multiple DNA repair pathways, with subpathways providing lesion specificity. Nucleotide early studies were carried out when there
excision repair removes bulky DNA lesions; DNA non-homologous end joining and homologous was not a comprehensive understanding
recombination repair DNA double-strand breaks (DSBs); mismatch repair corrects mismatched of DNA repair pathways or the DDR of
base pairs; and base excision repair repairs damaged bases and links to single-strand break (SSB) humans and when sequencing technology
repair. Mutations in these pathways increase cancer susceptibility.
was considerably less sophisticated than
DNA damage response signalling it is now. Thus, the attempts to address
There are two DNA damage response (DDR) signalling pathways: ataxia telangiectasia mutated this prediction were not very revealing.
(ATM)-dependent signalling is activated by DSBs; and ataxia telangiectasia and RAD3‑related
In those early studies, polymorphisms and
(ATR)-dependent signalling is activated by single-stranded regions of DNA. DDR signalling can
activate apoptosis and checkpoint arrest, and can influence DNA repair. Mutations in ATM
tumour-­associated mutations in DNA base
signalling components confer cancer susceptibility. However, ATR-deficient mice show reduced excision repair enzymes — for example,
capacity for tumour formation93. 8‑oxoguanine DNA glycosylase (OGG1)
Cell cycle checkpoints and apurinic–apyrimidinic endonuclease 1
DNA integrity is constantly monitored, with DNA damage triggering a checkpoint response that (APE1; also known as APEX1) — and
prevents cell cycle progression; arrest can be permanent or transient. Checkpoints prevent in components of the downstream
progression from G1 to S phase and from G2 to M phase, and an intra‑S phase checkpoint regulates XRCC1‑based part of base excision
fork progression or origin firing. Many tumours have inactivated checkpoint responses. repair were identified in some tumour
Apoptosis cells28,29. However, the penetrance of such
Apoptosis represents a programmed cell death pathway that functions in some tissues during polymorphisms is weak, and the clinical
normal development but also prevents proliferation of damaged cells. Apoptosis can be p53 relevance of these to the overall cancer burden
dependent or independent. p53 is commonly mutated in cancer. in the population was unclear 30. Subsequently,
Fidelity of replication several complex conditions in which cancer
Multiple processes function to maintain the accuracy of replication and enhance recovery from predisposition is a feature — such as Bloom
replication fork stalling or collapse. Homologous recombination has a key role, and genes involved syndrome, Werner syndrome and Fanconi
in this process are commonly mutated in cancers. anaemia — have been shown to arise from
DNA re‑replication genetic defects in DNA repair systems, as
Re‑replication can cause aneuploidy and subsequently genomic instability. Several mechanisms have subsets of familial breast, ovarian,
prevent DNA re‑replication. For example, increased cyclin-dependent kinase (CDK) activity is prostate and pancreatic cancers31–36.
required for origin firing but is inhibitory for origin licensing.
Telomere length Studies of radiation exposure
Shortened telomeres lead to senescence, and so cancer cells need to maintain telomere length to That X‑ray exposure confers an increased
survive. Activation of telomerase or an alternative pathway to maintain telomere length is common risk of malignant disease, including
in cancers.
leukaemia and skin cancers, became

36 | JANUARY 2016 | VOLUME 16

© 2016 Macmillan Publishers Limited. All rights reserved


accepted within a few decades after the damage response signalling and direct DNA nature of carcinogenesis coupled with
discovery of X‑rays in 1895. However, repair 42 (BOX 1). Notably, DDR signalling the evident chromosome changes led
radiation studies were disappointing when frequently has a greater impact on genomic to several models that, at their core,
it came to gaining mechanistic insight stability in response to DNA damage involved clonal evolution: that is, the
into the aetiology of cancer. Nonetheless, compared with DNA repair pathways, which progressive selection of rare mutated
reports by the International Commission more dramatically influence cell survival. cells50. Two extreme models were proposed.
on Radiological Protection (ICRP) and In the first, carcinogenesis required the
the United Nations Scientific Committee activation of multiple oncogenes and/or the
on the Effects of Atomic Radiation What emerged more slowly, inactivation of tumour suppressor genes
(UNSCEAR) provide an invaluable resource and each rearrangement contained an
however, was an appreciation
for rationalizing the emerging concepts amplification or loss of a specific gene. This
(for example, see REF. 37). In particular, that DDR mechanisms ... are was supported by the identification of p53
the analysis of atomic bomb survivors essential for cancer avoidance as a tumour suppressor, the loss of which
provided a wealth of epidemiological data enabled the evolution of rare mutated cells51.
and revealed, for example, that there can be Such hypotheses also suggested that elevated
a long induction period before the onset of Competing models and research fields genome instability per se (a mutator
cancer, and in the UNSCEAR 1958 report The concept that cancer involves at least phenotype; see below) would not necessarily
it was concluded that radiation-induced one genetic mutation was well accepted be required, as each acquired mutation
mutations are biologically relevant for by the 1970s. However, the notion that could theoretically increase growth
carcinogenesis37. However, the relationship oncogenesis is a multistage process potential independently. The alternative
between chromosome aberrations or was proposed by Isaac Berenblum and extreme model proposed that a mutator
rearrangements (which X‑rays avidly Phillipe Shubik as early as 1948, based phenotype characterized by an intrinsically
induce) and point mutational changes on studies showing that tumour cells high level of mutations occurred in the
(which X‑rays inefficiently induce) remained induced by carcinogen treatment could founding clone. This suggested that
puzzling. A further important concept remain in a latent stage until outgrowth the vast majority of rearrangements had
emerged from these early radiation studies: was promoted by subsequent treatment no phenotypic consequence but rather
it was demonstrated that the frequency with croton oil, which contains phorbol represented ‘passenger mutations’. Although
of cell transformation — which was used diester and activates growth signals43. controversial at the time, the mutator
as a surrogate for carcinogenicity — did From 1970 to the turn of the twenty-first phenotype model remains actively discussed
not correlate with the cell-lethal effects century, a range of studies — including as part of working models today 52–54.
when cells were exposed to different epidemiological analysis of atomic bomb Advances in single-cell sequencing
types of radiation37,38 (note that ionizing survivors, studies in mice and cell culture procedures are demonstrating the enormous
radiation of different types produces DNA models of transformation — provided sequence changes between cells within a
damage of different complexities). Thus, strong evidence that cancer is a multistage single tumour and have shown that the level
the carcinogenic impact of radiation cannot process44. It was understood that cancer of plasticity within a tumour correlates with
simply be attributed to its capacity to cause incidence increased dramatically with age aggressiveness55. However, these findings
cell death. and that exposure to ionizing radiation do not entirely allow the distinction to be
These early studies that evaluated the brought forward the age of onset of most drawn as to whether multiple mutations
response to radiation damage raised an cancers, but that it still involved a marked and a mutator phenotype cause malignancy
additional dilemma: although patients with lag period. Two contrasting models arose or are instead a consequence of it. It will
ataxia telangiectasia are cancer prone and to explain these observations: one model thus be crucial to deduce the stage at which
profoundly radiosensitive, DNA was not stated that cancer involved a mutagenic genomic instability arises.
hypermutated in patient cells that were initiation step, followed by a long period of In parallel with the emerging concepts
exposed to X‑rays and, despite their marked outgrowth45, and the other stated that cancer that carcinogenesis necessitated DNA
radiosensitivity, cells from these patients did was a multistage process, involving multiple sequence changes, thinking also focused
not display an obvious defect in the repair of mutational hits46. Slowly, the concept of a on the fact that cancer is predominantly
X‑ray-induced DNA damage39,40. The cause multistage process became the predominant a disorder of deregulated growth that
of this apparent paradox only became clear model, supported in part by the observation is likely to involve changed patterns of
after the discovery that the gene defective that the transformation of cultured cells differentiation or dedifferentiation. By
in patients with ataxia telangiectasia, ATM, occurred more rapidly, and at a higher this time, it was generally accepted that
encodes a protein kinase that triggers a frequency, if cells were transfected with differentiation during development was
signalling cascade that regulates cell cycle two oncogenes rather than one47 (see, for epigenetic56. This led to experiments in
arrest and cell death responses rather than example, a review written in 1993 (REF. 48)). which tumour cell nuclei with a normal
a DNA repair enzyme41 (nonetheless, The marked number of mutational and modal chromosome number were
ATM signalling can indirectly influence chromosome changes present in cancer transplanted into anucleated eggs to
DNA repair processes). This important cells, which were evident from the early generate adult animals57–59. Notably, such
distinction between signalling responses studies, and the fact that the median injections allowed for the development of
and direct DNA repair has proved crucial number of rearranged chromosome arms normal animals, potentially demonstrating
in the context of cancer avoidance. Indeed, correlated with cancer prognosis49 made a developmental totipotency that
the wider response to DNA damage (known an important contribution to the shaping suggested a non-mutational basis for
as the DDR) is now usefully subdivided into of models and thoughts. The multistep transformation to malignancy. In the


© 2016 Macmillan Publishers Limited. All rights reserved


Ames test established to

identify carcinogens via
Link reported (1928–1961) Established Xeroderma analysis of mutagens13
between that hereditary material (Circa 1958) pigmentosum
environmental (DNA) can be damaged Reported that identified as the Mutator phenotype Concept of
exposure and by endogenous and DNA damage can first DNA repair Apoptosis for cancer cells oncogenes
cancer6 exogenous agents4,10,86 be repaired20,94 disorder22 defined95 proposed52 introduced65

1775 1900 1928 1953 1955 1958 1960 1969 1972 1974 1975 1981

(1900–1950) Helical structure Alkylating agents shown to (1975–1985) Ataxia telangiectasia

Concept of of DNA reported8,9 react with and damage DNA10 Viruses reported to be a
hereditary proposed as radiosensitive
material (1955–1961) Link (1969–2015) Individuals with disorders a major disorder character-
introduced1–4 between mutagenesis resulting from defective DNA repair cause of ized by cancer
and carcinogenesis shown to be cancer prone; hence, failure cancer61,62 predisposition39
established2,10,11 to repair DNA contributes to cancer22,31

Figure 1 | Timeline showing the key concepts and findings relating to the role of the DNA damage response in the development of cancer. Key
concepts are shown in blue and key findings in grey. DDR, DNA damage response; MMR, mismatch repair; MSH2, MutS homologue 2; ROS, reactive
oxygen species.

context of current knowledge, such Any remaining controversy concerns with models postulating that a mutator
experiments were likely to be flawed, or at the magnitude of the viral aetiological phenotype underpinned cancer
least exceptional. However, developments contribution, which cannot easily explain, development 72. Although the interpretation
in the DNA methylation field provoked for example, diet- and smoking-related of these findings at the time was
research into methylation changes tumorigenesis. controversial, the emerging consensus
associated with cancer, leading to was that it was replication stress that,
proposals that methylation changes drive Oncogenes and oncogene-induced stress either directly or indirectly, caused the
expression changes and thus cancer Oncogenes were first identified by studying observed DNA damage. The link between
development 60. Indeed, we now know that retroviruses: the prototype oncogene, src, oncogene expression and DNA damage
epigenetic changes are commonly found in is a chicken Rous sarcoma virus gene that and/or its repair generated considerable
cancer cells and, like mutagenesis, provide was hijacked from the chicken genome65. interest: for example, MYC expression
a route to changed gene expression and It soon became clear that a defined, but in non-cancerous cells was shown to
thus function. significant, number of oncogenes were reduce DNA repair efficiency and induce
During the latter part of the twentieth involved in cancer initiation and that p53 and its ATM–ATR-dependent (but
century, the various models tended to oncogene expression caused neoplastic p14ARF-independent) phosphorylation73,74.
be considered as exclusive, generating transformation66. In the early 1990s, it was Concurrently, it was shown that p53 and
unfortunate conflicts that also influenced reported that genome instability occurred p21 prevented cell proliferation if cyclin E
the battle that emerged regarding a viral rapidly after HRAS oncogene expression, or cyclin-dependent kinase 2 (CDK2)
aetiology for cancer. The identification and subsequent reports demonstrated was overexpressed and that this operated
of numerous oncogenes from work that this was not an isolated phenomenon, through an ATM–ATR-dependent
on viruses, along with revelations that as it was observed after expression of and p14ARF-independent mechanism75.
many viruses encode proteins related to other oncogenes67–69. By the late 1990s, Cyclin E expression had been previously
human growth factors and that expression it had been shown that the product of an shown to cause chromosome instability 76,
of these could promote deregulated growth, alternative reading frame (ARF) within and it was later demonstrated to interfere
led to a widespread belief that the majority the cyclin-­dependent kinase inhibitor 2A with replication77. The scene was thus
of cancers were of viral origin61,62. The (CDKN2A) locus, p14ARF, which binds to set: oncogene-induced proliferation of
viral community, in part because of its MDM2 and upregulates p53, responded otherwise normal cells perturbed DNA
huge contributions to oncogene discovery to RAS and MYC expression by activating replication mechanisms, producing
(see below), gained considerable influence. the p53–p21 pathway to drive senescence measurable DNA damage and genome
In hindsight, the strength of the arguments or apoptosis70,71 (BOX 1). As it was known rearrangements and activating the
made by the viral community discouraged that DNA damage activated the p53–p21 p53–p21 pathway via ATM–ATR-dependent
full consideration of a genomic instability pathway to drive senescence or apoptosis, mechanisms. In 2005, two key papers78,79
model for cancer development. With our this led to the suggestion that oncogene clearly demonstrated both the activation
current knowledge, a model that proposed expression directly caused DNA damage72. of DDR signalling, including proteins
genomic instability as the origin of cancers It was initially proposed that deregulated required for cell cycle checkpoint arrest
would predict that viral infections could metabolism due to MYC-dependent (BOX 1), and increased expression of DNA
be carcinogenic, given the ability of transcription increased the production damage markers in precancerous tissue.
many viruses to subvert components of reactive oxygen species (ROS) and These papers proposed that this reflected
of the DDR (for example, see REFS 63,64). thus DNA damage, an idea consistent oncogene-­induced DNA damage arising

38 | JANUARY 2016 | VOLUME 16

© 2016 Macmillan Publishers Limited. All rights reserved


Oncogene expression shown to lead

to activation of the p53–p21 pathway, (2002–2003)
and thus to senescence or apoptosis70 Oncogene
p53 mutations identified expression shown to
Concept of in cancers. Role of p53- Concept of Oncogene cause deregulated Significance of
tumour dependent surveillance caretaker and expression shown metabolism leading replication stress and
suppressors pathways recognized as gatekeeper genes to cause replication to ROS production replication fork stability
introduced96 cancer suppressive51,80 introduced98 instability76 and DNA damage72,73 appreciated78,79

1984 1989 1990 1993 1997 1999 2000 2002 2005 2014

Cell cycle Microsatellite instability (1997–1999) BRCA1 Acquired DDR reported to be an Multiple
checkpoints identified in Lynch syndrome and BRCA2, which are characteristics anticancer barrier in mutations in
proposed97 tumours and shown to be mutated in hereditary of cancer cells early-stage tumorigenesis; DDR genes
due to MMR deficiency. breast cancer, shown to defined101 mutations in DDR genes identified in
MSH2 identified as the first function in homologous shown to occur in later- cancers87,88
Lynch syndrome locus24,25 recombination99,100 stage tumours78,79

Nature Reviews | Cancer

from replication stress, synthesizing cell p14ARF–MDM2–p53 pathway 82. Slightly the DDR occurs in precancerous lesions
culture data and demonstrating a direct distinct but related examples also followed, and that p53 mutations occur subsequently
relevance to clinically derived cancer tissue. such as the demonstration that, although in late-stage tumours, strongly support
tumours in Brca2+/− mice undergo loss an order of events in which the onset of
DDR downregulation in tumours of heterozygosity, the proliferation of replication stress represents an early event,
As discussed above, the notions that homozygous BRCA2‑deficient tumour promoting the subsequent mutations that
endogenous and exogenous DNA damage cells demands additional mutations in allow outgrowth.
cause mutations that lead to carcinogenesis, spindle checkpoint genes83. The full breadth However, a more recent study involving
and that cancer avoidance necessitates of the relevant pathways that require ultradeep sequencing of cancer genes in
active DNA repair mechanisms, emerged downregulation, and subsequently their sun-exposed normal skin biopsies revealed
as key concepts early in work on the importance in contributing to tumour a substantial accumulation of cancer
aetiology of cancer. What emerged more progression, was slowly revealed, as was driver mutations (with the characteristic
slowly, however, was an appreciation that the realization that downregulation of signature of UV‑induced mutations) that
DDR mechanisms in general, as distinct these pathways could create a mutator had undergone positive selection in the
from specific repair pathways per se, are phenotype that causes genomic instability, absence of evidence for cancer formation85.
essential for cancer avoidance. Initially, as postulated many years earlier. This suggests a different aetiology in which
based on studies of apoptosis, an important there is a strong initial selection pressure to
concept for our understanding of cancer Lessons from history upregulate growth-enhancing genes; cells
onset emerged: it was not necessarily Current models of cancer would argue that with such changes then await further steps
the DNA damage itself that killed an initiating event (or several initiating that lead to a genetically unstable state. The
(or prevented the growth of) a cell, but events), often caused by a mutation, enhanced cancer predisposition caused
rather the signalling pathways activated leads to oncogene activation and ensuing by mutations in genes that affect both the
by such damage. Cell cycle checkpoint replication stress84. However, this must be early (for example, mutations enhancing an
pathways, initially defined as systems coupled with subsequent downregulation initiation event such as in XP) and perceived
that monitor genome integrity, are now of DDR mechanisms — possibly by genetic late (for example, damage surveillance
understood to be pivotal in precluding alterations as a consequence of replication mechanisms such as in Li–Fraumeni
the continued proliferation of damaged stress — to allow continued proliferation and syndrome) steps of cancer would be
cells80. p53 has a key role in this, and the continued genome instability, a prerequisite consistent with there not being a defined
frequent loss of p53 function in tumours for a cancer cell to rapidly adapt to its order of events that lead to carcinogenesis.
contributed to the emerging notion that ever-changing microenvironment. Although Our historical reflection in this article
DDR pathways must be downregulated the historical reflection in this article considers the many models or factors
to allow uncontrolled proliferation81. has considered the timeline of emerging that have been proposed to contribute to
In 1997, Manuel Serrano et al.70 seminal concepts, this does not reflect the carcinogenesis, which include viruses,
proposed that expression of the HRAS order of events in the aetiology of cancer epigenetic changes, DNA-damaging agents,
oncogene leads to activation of the (FIG. 1). The initiating event (or events) replication stressors and oxidative stress.
p53–p21 pathway, which drives senescence causing oncogene activation most likely On the basis of our current knowledge, all of
or apoptosis. Thus, for proliferation to precedes a state of replication stress, but it these factors are indeed valid contributors,
occur, the p53–p21 pathway must become remains unclear whether downregulation and they all merge into a model that
downregulated in HRAS-expressing of the DDR is always directly caused by ultimately leads to the generation of a
cells70. Similar findings were observed errors that arise from replication stress or if genetically unstable state, which is in most
following expression of the MYC oncogene, it could precede replication stress. Findings cases essential for carcinogenesis (FIG. 2).
although in this case proliferation from the laboratories of Jiri Bartek78 and Strikingly, this pinpoints the enormous
necessitated downregulation of the Thanos Halazonetis79, that upregulation of significance of the DDR processes: their


© 2016 Macmillan Publishers Limited. All rights reserved


Virus Exogenous DNA- Downregulation of DDR

damaging agent processes that prevent
• Compromised proliferation of damaged cells
Endogenous DNA- fork fidelity
damaging agent • Compromised Proliferation DDR
replication Invasive tumour cell

mutation Oncogene activation • Increased DNA
and increased replicative damage and mutations
and oxidative stress • Genomic instability
• Persistent genomic instability
Healthy cell Precancerous cell • Further loss of DDR processes

Figure 2 | How the DNA damage response pathways influence steps one of these DNA repair processes is compromised. However, although
leading to cancer. The figure shows how changes in the DNA damage enhanced replication stress increases the level NatureofReviews
DNA breakage,
| Cancer
response (DDR) pathways promote critical steps in the aetiology of mutation or rearrangement, a range of responses — for example, the
carcinogenesis. A healthy cell has a plethora of DDR processes to ability to accurately recover replication, the activation of checkpoint
protect its DNA from exogenous and endogenously arising DNA- arrest or other p53‑dependent responses — can prevent the
damaging agents, and respond to viral infections. None­theless, the proliferation of damaged cells. Progression from this precancerous state
processes are not perfect, and an early step in the aetiology of cancer is to ongoing proliferation requires the downregulation of these DDR
the generation of one or more mutational changes. This may directly or processes, thereby facilitating persistent genomic instability. For clarity,
indirectly result in oncogene activation, which leads to replicative these steps have been depicted to arise in a linear fashion, which may
and/or oxidative stress. Genetic predisposition to cancer can arise when not be the case.

importance was evident in the earliest studies profound interconnectedness, has only been arising in BRCA1- or BRCA2‑deficient
but has emerged to be far more substantial appreciated more recently. Similarly, the patients89. The key insight came from the
than originally predicted. Although early advent of tumour genome profiling by deep realization that BRCA1 and BRCA2 function
studies demonstrated that cells can recover sequencing has only recently demonstrated during homologous recombination, a key
from exposure to external DNA-damaging that DDR genes are frequently mutated process that promotes replication fork
agents, revealing that they harbour DNA in all cancer types, with many individual stability, and that drug-targeted inhibition
repair mechanisms20, perceptive scientists pathways or genes found to be mutated in of a specific enzyme (poly(ADP-ribose)
also saw that the DNA structure revealed more than 50% of cancers of a specific type polymerase 1 (PARP1)) confers a reliance on
by Watson and Crick could accumulate (for example, more than 50% of ovarian homologous recombination and hence drug
endogenously arising DNA damage, cancers harbour mutations in genes involved sensitivity 89. Although such an approach
predicting that DNA repair pathways might in homologous recombination)87,88 (FIG. 3). might be anticipated to uniquely benefit
be essential even during normal growth and The steps during carcinogenesis can be BRCA1- or BRCA2‑deficient patients,
metabolism86. However, these early studies summarized as shown in FIG. 2. Although an current studies revealing that homologous
did not predict that the DDR mechanisms understanding of these steps is of significant recombination can be downregulated in
encompass not only DNA repair processes academic interest, it also provides a key tool approximately 50% of ovarian cancers
that seek to avoid the initiator mutations, but for informed, targeted cancer therapy 87. dramatically expand the scope for such
also DNA damage surveillance mechanisms The enhanced sensitivity of many cancer therapy 88. A plethora of related studies are
that preclude the proliferation of genetically cells to DNA-damaging agents, including in progress, which include strategies to
unstable cells. Furthermore, there is also radiation exposure, has been evident for promote synthetic lethality based on changes
a requirement for an efficient replication many years and indeed has been exploited such as the subversion of apoptosis, altered
machinery that restricts replication errors for therapeutic benefit. The rationale for such pathways of DNA double-strand break
and for processes that allow recovery from sensitivity was poorly understood and was repair and loss of checkpoint arrest in cancer
the inevitable difficulties encountered during often unsatisfactorily attributed to the more cells87,89. Conversely, a similar but distinct
replication in a manner that maintains rapid growth of tumour cells. Consequently, phenomenon called ‘synthetic viability’ can
genomic stability (BOX 1). We now know the approach relied on serendipity, coupled allow the rescue of a cell-growth defect that
that cancer cells not only downregulate with random trial and error. Our current is imposed either by a preceding mutation
these pathways but often subvert them to knowledge of how the DDR pathways are in the cancer cell or by drug treatment
fast-track evolution and gain adaptability, changed in cancers is providing routes for (or by a combination of both). It is well
which is the ultimate driver of carcinogenesis more specific and rationally targeted therapy. established that cancer cells often gain
and metastasis87. A significant concept in this regard is mutational changes that confer synthetic
that of synthetic lethality, in which the goal viability in the context of drug therapy,
The future is to identify a drug that will cause lethality including in response to PARP inhibitors90.
This Timeline article highlights the to a cancer cell that has inherent DDR It is likely that many cancer cells have
significance of the role of DDR processes defects without harming a normal cell87. equivalent synthetic viability mutations
in cancer aetiology. However, the plethora The foremost and very elegant example (for example, p53 mutations, as discussed
of DNA integrity surveillance, repair of exploiting a synthetically lethal genetic above) that compensate for preceding
and signalling pathways, alongside their relationship is the treatment of breast cancers genetic changes91,92.

40 | JANUARY 2016 | VOLUME 16

© 2016 Macmillan Publishers Limited. All rights reserved


In conclusion, to fully exploit the also exploit imaginative ideas to progress All authors are at the Genome Damage
and Stability Centre, School of Life Sciences,
genome instability that is now recognized cancer treatment. Ironically, however, our
University of Sussex, Brighton BN1 9RQ, UK.
as an inherent property of most — if not understanding of the stages of carcinogenesis Correspondence to P.A.J., L.H.P. and A.M.C.
all — cancers, it is critical that we enhance has also provided an explanation for why our; laurence.pearl@
our understanding of the DDR pathways and targeted therapies will frequently fail.;

Published online 15 Dec 2015
a Acute myeloid leukaemia Ovarian serous carcinoma Glioblastoma
OD p53 CPF OD p53 CPF OD p53 CPF 1. Boveri, T. Concerning the origin of malignant tumours
by Theodor Boveri. J. Cell Sci. 121 (Suppl. 1), 1–84
FA NER FA NER FA NER 2. Burdette, W. J. The significance of mutation in relation
to the origin of tumors: a review. Cancer Res. 15,

NHEJ MMR NHEJ MMR NHEJ MMR 201–226 (1955).

3. Muller, H. J. Artificial transmutation of the gene.
TLS BER TLS BER TLS BER Science 66, 84–87 (1927).
CR DR CR DR CR DR 4. Muller, H. J. The production of mutations by X‑rays.
Proc. Natl Acad. Sci. USA 14, 714–726 (1928).
CS AM TM CS AM TM CS AM TM 5. Hall, E. J. From chimney sweeps to oncogenes: the
quest for the causes of cancer. Radiology 179,
[n = 200] [n = 522] [n = 596]
297–306 (1991).
6. Pott, P. The first description of an occupational cancer
Breast invasive carcinoma Colon adenocarcinoma Lung squamous cell carcinoma in 1777 (scrotal cancer, cancer of chimney sweeps).
OD p53 CPF OD p53 CPF OD p53 CPF Bull. Soc. Liban. Hist. Med. 1993, 98–101 (in French)
HR UR HR UR HR UR 7. Waldron, H. A. A brief history of scrotal cancer.
FA NER FA NER FA NER Br. J. Ind. Med. 40, 390–401 (1983).
8. Franklin, R. E. & Gosling, R. G. Evidence for 2‑chain
NHEJ MMR NHEJ MMR NHEJ MMR helix in crystalline structure of sodium
deoxyribonucleate. Nature 172, 156–157 (1953).
TLS BER TLS BER TLS BER 9. Watson, J. D. & Crick, F. H. Molecular structure of
nucleic acids; a structure for deoxyribose nucleic acid.
Nature 171, 737–738 (1953).
CS AM TM CS AM TM CS AM TM 10. Brookes, P. & Lawley, P. D. The reaction of mustard
gas with nucleic acids in vitro and in vivo. Biochem. J.
[n = 1,033] [n = 398] [n = 454] 77, 478–484 (1960).
11. Brookes, P. & Lawley, P. D. The reaction of mono- and
b Breast invasive carcinoma Colon adenocarcinoma Lung squamous cell carcinoma di‑functional alkylating agents with nucleic acids.
Biochem. J. 80, 496–503 (1961).
OD p53 CPF OD p53 CPF OD p53 CPF 12. Brookes, P. & Lawley, P. D. Evidence for the binding of
HR UR HR UR HR UR polynuclear aromatic hydrocarbons to the nucleic
FA NER FA NER FA NER acids of mouse skin: relation between carcinogenic
power of hydrocarbons and their binding to

NHEJ MMR NHEJ MMR NHEJ MMR deoxyribonucleic acid. Nature 202, 781–784 (1964).
13. Ames, B. N. A combined bacterial and liver test
TLS BER TLS BER TLS BER system for detection and classification of carcinogens
as mutagens. Genetics 78, 91–95 (1974).
CR DR CR DR CR DR 14. Setlow, R. B. The photochemistry, photobiology, and
CS AM TM CS AM TM CS AM TM repair of polynucleotides. Prog. Nucleic Acid. Res.
Mol. Biol. 8, 257–295 (1968).
[n = 1,033] [n = 398] [n = 454] 15. Pettijohn, D. & Hanawalt, P. Evidence for repair-
replication of ultraviolet damaged DNA in bacteria.
c Breast invasive carcinoma Bladder urothelial carcinoma Lung squamous cell carcinoma J. Mol. Biol. 9, 395–410 (1964).
16. Friedberg, E. C. & Goldthwait, D. A. Endonuclease II
OD p53 CPF OD p53 CPF OD p53 CPF of E. coli. Cold Spring Harb. Symp. Quant. Biol. 33,
HR UR HR UR HR UR 271–275 (1968).
17. Laval, J. & Laval, F. in Molecular and Cellular Aspects
FA NER FA NER FA NER of Carcinogen Screening Tests (eds Montesano, H.,

NHEJ MMR NHEJ MMR NHEJ MMR Bartsch, L. & Tomatis, L.) 55–73 (IARC Press, 1980).
18. Lindahl, T. An N‑glycosidase from Escherichia coli that
TLS BER TLS BER TLS BER releases free uracil from DNA containing deaminated
cytosine residues. Proc. Natl Acad. Sci. USA 71,
CR DR CR DR CR DR 3649–3653 (1974).
CS 19. Ljungquist, S. & Lindahl, T. A mammalian
AM TM CS AM TM CS AM TM endonuclease specific for apurinic sites in double-
[n = 800] [n = 55] [n = 229] stranded deoxyribonucleic acid. I. Purification and
general properties. J. Biol. Chem. 249, 1530–1535
Figure 3 | Many proteins involved in the DNA damage response are mutated in cancer. Data (1974).
from The Cancer Genome Atlas have revealed the extent to which mutations in DNA damage response 20. Friedberg, E. C. A brief history of the DNA repair field.
Nature Reviews | Cancer Cell Res. 18, 3–7 (2008).
(DDR) proteins are observed in cancers. These changes differ for different tumour sites. The concentric 21. Laval, F. Repair of methylated bases in mammalian
circles indicate the percentages of patients affected. Part a shows radial plots in which the radius cells during adaptive response to alkylating agents.
Biochimie 67, 361–364 (1985).
length indicates the proportion of patients with protein-coding mutations that are predicted to dis- 22. Cleaver, J. E. Xeroderma pigmentosum: a human
rupt particular DDR pathways in each cancer type. All mutations included are present in at least two disease in which an initial stage of DNA repair is
distinct patient samples. Part b shows copy number variation in the different DDR pathways (red indi- defective. Proc. Natl Acad. Sci. USA 63, 428–435
cates loss of gene copies; blue indicates amplification of genes). Part c shows expression level variation 23. Lehmann, A. R., McGibbon, D. & Stefanini, M.
in DDR pathways (red indicates decreased expression; blue indicates increased expression). AM, alter- Xeroderma pigmentosum. Orphanet J. Rare Dis. 6, 70
native mechanism for telomere maintenance; BER, base excision repair; CPF, checkpoint factor; CR, (2011).
24. Leach, F. S. et al. Mutations of a mutS homolog in
chromatin remodelling; CS, chromosome segregation; DR, direct repair; FA, Fanconi anaemia path- hereditary nonpolyposis colorectal cancer. Cell 75,
way; HR, homologous recombination; MMR, mismatch repair; NER, nucleotide excision repair; NHEJ, 1215–1225 (1993).
non-homologous end joining; OD, other double-strand break repair; TLS, translesion synthesis; TM, 25. Fishel, R. et al. The human mutator gene homolog
MSH2 and its association with hereditary
telomere maintenance; UR, ubiquitylation response. Figure reproduced from REF. 87 , Nature nonpolyposis colon cancer. Cell 75, 1027–1038
Publishing Group. (1993).


© 2016 Macmillan Publishers Limited. All rights reserved


26. Bronner, C. E. et al. Mutation in the DNA mismatch 55. Fox, E. J. & Loeb, L. A. Cancer: one cell at a time. 81. Baker, S. J. et al. Chromosome 17 deletions and p53
repair gene homologue hMLH1 is associated with Nature 512, 143–144 (2014). gene mutations in colorectal carcinomas. Science 244,
hereditary non-polyposis colon cancer. Nature 368, 56. Gurdon, J. B. & Uehlinger, V. “Fertile” intestine nuclei. 217–221 (1989).
258–261 (1994). Nature 210, 1240–1241 (1966). 82. Eischen, C. M., Weber, J. D., Roussel, M. F.,
27. Papadopoulos, N. et al. Mutation of a mutL homolog 57. McKinnell, R. G., Deggins, B. A. & Labat, D. D. Sherr, C. J. & Cleveland, J. L. Disruption of the
in hereditary colon cancer. Science 263, 1625–1629 Transplantation of pluripotential nuclei from triploid ARF–Mdm2–p53 tumor suppressor pathway in
(1994). frog tumors. Science 165, 394–396 (1969). Myc-induced lymphomagenesis. Genes Dev. 13,
28. Cui, Z. et al. Association between polymorphisms in 58. Mintz, B. & Illmensee, K. Normal genetically mosaic 2658–2669 (1999).
XRCC1 gene and clinical outcomes of patients with mice produced from malignant teratocarcinoma cells. 83. Lee, H. et al. Mitotic checkpoint inactivation fosters
lung cancer: a meta-analysis. BMC Cancer 12, 71 Proc. Natl Acad. Sci. USA 72, 3585–3589 (1975). transformation in cells lacking the breast cancer
(2012). 59. Illmensee, K. & Mintz, B. Totipotency and normal susceptibility gene, Brca2. Mol. Cell 4, 1–10 (1999).
29. Yamtich, J., Nemec, A. A., Keh, A. & Sweasy, J. B. differentiation of single teratocarcinoma cells cloned 84. Halazonetis, T. D., Gorgoulis, V. G. & Bartek, J. An
A germline polymorphism of DNA polymerase beta by injection into blastocysts. Proc. Natl Acad. Sci. USA oncogene-induced DNA damage model for cancer
induces genomic instability and cellular 73, 549–553 (1976). development. Science 319, 1352–1355 (2008).
transformation. PLoS Genet. 8, e1003052 (2012). 60. Bird, A. P. The relationship of DNA methylation to 85. Martincorena, I. et al. Tumor evolution. High burden
30. Karahalil, B., Bohr, V. A. & Wilson, D. M. Impact of cancer. Cancer Surv. 28, 87–101 (1996). and pervasive positive selection of somatic mutations
DNA polymorphisms in key DNA base excision repair 61. Deuel, T. F., Huang, J. S., Huang, S. S., Stroobant, P. in normal human skin. Science 348, 880–886
proteins on cancer risk. Hum. Exp. Toxicol. 31, & Waterfield, M. D. Expression of a platelet-derived (2015).
981–1005 (2012). growth factor-like protein in simian sarcoma virus 86. Lindahl, T. The Croonian Lecture, 1996: endogenous
31. Polani, P. E. in Human Genetics: Possibilities and transformed cells. Science 221, 1348–1350 (1983). damage to DNA. Phil. Trans. R. Soc. Lond. B 351,
Realities (eds Porter, R. & O’Connor, M.) 81–131 62. Waterfield, M. D. et al. Platelet-derived growth factor 1529–1538 (1996).
(Excerpta Medica, 1979). is structurally related to the putative transforming 87. Pearl, L. H., Schierz, A. C., Ward, S. E., Al‑Lazikani, B.
32. Moldovan, G. L. & D’Andrea, A. D. How the Fanconi protein p28sis of simian sarcoma virus. Nature 304, & Pearl, F. M. Therapeutic opportunities within the
anemia pathway guards the genome. Annu. Rev. 35–39 (1983). DNA damage response. Nat. Rev. Cancer 15,
Genet. 43, 223–249 (2009). 63. Lees-Miller, S. P. et al. Attenuation of DNA-dependent 166–180 (2015).
33. Gray, M. D. et al. The Werner syndrome protein is a protein kinase activity and its catalytic subunit by the 88. Cancer Genome Atlas Research Network. Integrated
DNA helicase. Nat. Genet. 17, 100–103 (1997). herpes simplex virus type 1 transactivator ICP0. genomic analyses of ovarian carcinoma. Nature 474,
34. Hartman, A. R. et al. Prevalence of BRCA mutations in J. Virol. 70, 7471–7477 (1996). 609–615 (2011).
an unselected population of triple-negative breast 64. Weitzman, M. D. & Weitzman, J. B. What’s the 89. Lord, C. J., Tutt, A. N. & Ashworth, A. Synthetic
cancer. Cancer 118, 2787–2795 (2012). damage? The impact of pathogens on pathways that lethality and cancer therapy: lessons learned from the
35. Cavanagh, H. & Rogers, K. M. The role of BRCA1 and maintain host genome integrity. Cell Host Microbe 15, development of PARP inhibitors. Annu. Rev. Med. 66,
BRCA2 mutations in prostate, pancreatic and stomach 283–294 (2014). 455–470 (2015).
cancers. Hered. Cancer Clin. Pract. 13, 16 (2015). 65. Bishop, J. M. Enemies within: the genesis of retrovirus 90. Lord, C. J. & Ashworth, A. Mechanisms of resistance
36. Banerjee, T. & Brosh, R. M. Jr. RECQL: a new breast oncogenes. Cell 23, 5–6 (1981). to therapies targeting BRCA-mutant cancers. Nat.
cancer susceptibility gene. Cell Cycle 66. Cooper, G. M. Cellular transforming genes. Science Med. 19, 1381–1388 (2013).
10.1080/15384101.2015.1066539 (2015). 217, 801–806 (1982). 91. Callen, E. et al. 53BP1 mediates productive and
37. Lindell, B. & Sowby, D. The 1958 UNSCEAR report. 67. Denko, N. C., Giaccia, A. J., Stringer, J. R. mutagenic DNA repair through distinct
J. Radiol. Prot. 28, 277–282 (2008). & Stambrook, P. J. The human Ha‑ras oncogene phosphoprotein interactions. Cell 153, 1266–1280
38. Yang, T. C., Craise, L. M., Mei, M. T. & Tobias, C. A. induces genomic instability in murine fibroblasts (2013).
Neoplastic cell transformation by heavy charged within one cell cycle. Proc. Natl Acad. Sci. USA 91, 92. Bouwman, P. et al. 53BP1 loss rescues BRCA1
particles. Radiat. Res. Suppl. 8, S177–S187 (1985). 5124–5128 (1994). deficiency and is associated with triple-negative and
39. Taylor, A. M. et al. Ataxia telangiectasia: a human 68. Mai, S., Fluri, M., Siwarski, D. & Huppi, K. Genomic BRCA-mutated breast cancers. Nat. Struct. Mol. Biol.
mutation with abnormal radiation sensitivity. Nature instability in MycER-activated Rat1A‑MycER cells. 17, 688–695 (2010).
258, 427–429 (1975). Chromosome Res. 4, 365–371 (1996). 93. Schoppy, D. W. et al. Oncogenic stress sensitizes
40. Shiloh, Y., Tabor, E. & Becker, Y. Cellular 69. Felsher, D. W. & Bishop, J. M. Transient excess of MYC murine cancers to hypomorphic suppression of ATR.
hypersensitivity to neocarzinostatin in ataxia- activity can elicit genomic instability and J. Clin. Invest. 122, 241–252 (2012).
telangiectasia skin fibroblasts. Cancer Res. 42, tumorigenesis. Proc. Natl Acad. Sci. USA 96, 94. Hill, R. F. A radiation-sensitive mutant of Escherichia
2247–2249 (1982). 3940–3944 (1999). coli. Biochim. Biophys. Acta 30, 636–637 (1958).
41. Kastan, M. B. et al. A mammalian cell cycle checkpoint 70. Serrano, M., Lin, A. W., McCurrach, M. E., Beach, D. 95. Kerr, J. F., Wyllie, A. H. & Currie, A. R. Apoptosis:
pathway utilizing p53 and GADD45 is defective in & Lowe, S. W. Oncogenic ras provokes premature cell a basic biological phenomenon with wide-ranging
ataxia-telangiectasia. Cell 71, 587–597 (1992). senescence associated with accumulation of p53 and implications in tissue kinetics. Br. J. Cancer 26,
42. Boulton, S. J. et al. Combined functional genomic p16INK4a. Cell 88, 593–602 (1997). 239–257 (1972).
maps of the C. elegans DNA damage response. 71. Zindy, F. et al. Myc signaling via the ARF tumor 96. Murphree, A. L. & Benedict, W. F. Retinoblastoma:
Science 295, 127–131 (2002). suppressor regulates p53‑dependent apoptosis and clues to human oncogenesis. Science 223,
43. Berenblum, I. & Shubik, P. The role of croton oil immortalization. Genes Dev. 12, 2424–2433 (1998). 1028–1033 (1984).
applications, associated with a single painting of a 72. Vafa, O. et al. c‑Myc can induce DNA damage, increase 97. Hartwell, L. H. & Weinert, T. A. Checkpoints: controls
carcinogen, in tumour induction of the mouse’s skin. reactive oxygen species, and mitigate p53 function: that ensure the order of cell cycle events. Science 246,
Br. J. Cancer 1, 379–382 (1947). a mechanism for oncogene-induced genetic instability. 629–634 (1989).
44. Farber, E. The multistep nature of cancer Mol. Cell 9, 1031–1044 (2002). 98. Kinzler, K. W. & Vogelstein, B. Cancer-susceptibility
development. Cancer Res. 44, 4217–4223 (1984). 73. Karlsson, A. et al. Defective double-strand DNA break genes. Gatekeepers and caretakers. Nature 386,
45. Cairns, J. Mutation selection and the natural history of repair and chromosomal translocations by MYC 761–763 (1997).
cancer. Nature 255, 197–200 (1975). overexpression. Proc. Natl Acad. Sci. USA 100, 99. Moynahan, M. E., Chiu, J. W., Koller, B. H. & Jasin, M.
46. Farber, E. Carcinogenesis—cellular evolution as a 9974–9979 (2003). Brca1 controls homology-directed DNA repair.
unifying thread: presidential address. Cancer Res. 33, 74. Lindstrom, M. S. & Wiman, K. G. Myc and E2F1 Mol. Cell 4, 511–518 (1999).
2537–2550 (1973). induce p53 through p14ARF‑independent 100. Sharan, S. K. et al. Embryonic lethality and radiation
47. Weinberg, R. A. The action of oncogenes in the mechanisms in human fibroblasts. Oncogene 22, hypersensitivity mediated by Rad51 in mice lacking
cytoplasm and nucleus. Science 230, 770–776 4993–5005 (2003). Brca2. Nature 386, 804–810 (1997).
(1985). 75. Minella, A. C. et al. p53 and p21 form an inducible 101. Hanahan, D. & Weinberg, R. A. The hallmarks of
48. Vogelstein, B. & Kinzler, K. W. The multistep nature of barrier that protects cells against cyclin E-cdk2 cancer. Cell 100, 57–70 (2000).
cancer. Trends Genet. 9, 138–141 (1993). deregulation. Curr. Biol. 12, 1817–1827 (2002).
49. Fearon, E. R. & Vogelstein, B. A genetic model for 76. Spruck, C. H., Won, K. A. & Reed, S. I. Deregulated Acknowledgements
colorectal tumorigenesis. Cell 61, 759–767 (1990). cyclin E induces chromosome instability. Nature 401, A.M.C. is funded by the Medical Research Council (grant
50. Nowell, P. C. The clonal evolution of tumor cell 297–300 (1999). G1100074). L.H.P. is supported by the Cancer Research UK
populations. Science 194, 23–28 (1976). 77. Ekholm-Reed, S. et al. Deregulation of cyclin E in Programme grant C302/A14532, and P.A.J. is supported by
51. Chen, P. L., Chen, Y. M., Bookstein, R. & Lee, W. H. human cells interferes with prereplication complex the EU Seventh Framework Programme (FP7) RISK – IR pro‑
Genetic mechanisms of tumor suppression by the assembly. J. Cell Biol. 165, 789–800 (2004). ject under grant agreement no. 323267. The authors also
human p53 gene. Science 250, 1576–1580 (1990). 78. Bartkova, J. et al. DNA damage response as a thank F. Pearl for useful discussions.
52. Loeb, L. A., Springgate, C. F. & Battula, N. Errors in candidate anti-cancer barrier in early human
DNA replication as a basis of malignant changes. tumorigenesis. Nature 434, 864–870 (2005). Competing interests statement
Cancer Res. 34, 2311–2321 (1974). 79. Gorgoulis, V. G. et al. Activation of the DNA The authors declare no competing interests.
53. Loeb, L. A. Mutator phenotype may be required for damage checkpoint and genomic instability in human
multistage carcinogenesis. Cancer Res. 51, precancerous lesions. Nature 434, 907–913
3075–3079 (1991). (2005). FURTHER INFORMATION
54. Fox, E. J., Prindle, M. J. & Loeb, L. A. Do mutator 80. Wahl, G. M. & Carr, A. M. The evolution of diverse The Cancer Genome Atlas:
mutations fuel tumorigenesis? Cancer Metastasis Rev. biological responses to DNA damage: insights from ALL LINKS ARE ACTIVE IN THE ONLINE PDF
32, 353–361 (2013). yeast and p53. Nat. Cell Biol. 3, E277–E286 (2001).

42 | JANUARY 2016 | VOLUME 16

© 2016 Macmillan Publishers Limited. All rights reserved