Download - Cancer as a Dysregulated Epigenome
-
7/27/2019 Cancer as a Dysregulated Epigenome
1/14
Even before the discovery of epigenetic mod-ifications in cancer, classical tumour biologysuggested that generalized disruption of geneexpression might underlie the key proper-ties of unregulated tumour growth, invasionand metastasis. Perhaps the earliest personto recognize the importance of gene expres-sion in cancer was Sidney Weinhouse, whodescribed a generalized disruption of the
biochemistry of cancer cells that was focusedon isozymes that were primarily related tometabolism1. However, since the discovery ofoncogene mutation in human tumours2, theprincipal focus of cancer genetics has beenon mutations. We argue in this Opinion arti-cle that, although key mutational changes arenecessary for the initiation of what we cur-rently recognize as neoplastic growth and arelikely to be required for escape from a cellularniche, epigenetic modifications also have acrucial role: these modifications allow rapidcellular selection in a changing environment,
thus leading to a growth advantage for thetumour cells at the expense of the host. This
view does not contradict and indeed col-laborates with the genetic model, but it putsepigenetics at the very heart of cancer biol-ogy, from normal precursor cells at the siteswhere cancer arises, and through all stages oftumour progression, to advanced metastaticdisease.
The first experiments on DNA methy-lation in human cancer, which comparedsamples of human colorectal cancer withmatched normal mucosa isolated fromthe same patients, showed widespreadhypomethylation involving approximatelyone-third of single-copy genes3. In responseto the discovery of tumour suppressorgenes4, later studies focused on identifyingsilenced genes as surrogates for mutation,beginning with the observation of promoterhypermethylation ofRB1 by Horsthemkeand colleagues5,6. During the 2000s, the
maturation of microarrays and the adventof next-generation sequencing technologiesin combination with the rise of data-drivendiscovery in biology have led to importantnew insights. These include the discoveryof genome-wide loss of epigenetic stability,which is common across disparate tumourtypes. This seems to be the underlying mech-anism for both the hypomethylation and thehypermethylation of individual genes, whichwas the historical focus of this field7. In addi-tion, recently discovered mutations in theepigenetic apparatus probably contribute
to epigenetic disruption in cancer. We reviewthese recent discoveries and point to thepossibility that cancer is a state in which theepigenome is allowed to have greater plas-ticity than it is supposed to have in normalsomatic tissues. This increased epigeneticplasticity is a normal component of develop-ment or postnatal responses to injury, but itsconstitutive activation in cancer causes epige-netic heterogeneity that leads to most of theclassical cancer hallmarks. We discuss belowhow this perspective provides new researchavenues for diagnostics and treatment.
Large epigenetic structures
Just as the field of cancer epigenetics waspresaged by early studies of abnormal geneexpression, the role of large epigeneticstructures in cancer was indicated by theearliest studies of cancer epigenetics byTheodor Boveri, who described abnormalchromatin in cancer cells in photomicro-graphs in 1929 (REF. 8). Alterations in nuclearshape are often used for diagnosis and arepotentially symptomatic of the disorganiza-tion of this carefully regulated state9. In addi-tion, nuclear lamina proteins (which serve
to retain nuclear organization) show alteredgene expression in cancer10. We describebelow advances from whole-genome ana-lyses that begin to provide molecular detailto these altered structures in cancer.
Chromatin LOCKs and LADs.Euchromatinrefers to genes that are more open to tran-scription owing to post-translational modi-fications of histones and lower nucleosomedensity, whereas heterochromatin is the oppo-site: genes that are less open to transcriptionowing to greater nucleosome density and
E P I G E N E T I C S A N D G E N E T I C S O P I N I O N
Cancer as a dysregulated epigenomeallowing cellular growth advantageat the expense of the host
Winston Timp and Andrew P. Feinberg
Abstract | Although at the genetic level cancer is caused by diverse mutations,
epigenetic modifications are characteristic of all cancers, from apparently normalprecursor tissue to advanced metastatic disease, and these epigenetic
modifications drive tumour cell heterogeneity. We propose a unifying model of
cancer in which epigenetic dysregulation allows rapid selection for tumour cell
survival at the expense of the host. Mechanisms involve both genetic mutations
and epigenetic modifications that disrupt the function of genes that regulate the
epigenome itself. Several exciting recent discoveries also point to a genome-scale
disruption of the epigenome that involves large blocks of DNA hypomethylation,
mutations of epigenetic modifier genes and alterations of heterochromatin in
cancer (including large organized chromatin lysine modifications (LOCKs) and
lamin-associated domains (LADs)), all of which increase epigenetic and gene
expression plasticity. Our model suggests a new approach to cancer diagnosis and
therapy that focuses on epigenetic dysregulation and has great potential for risk
detection and chemoprevention.
PERSPECTIVES
NATURE REVIEWS |CANCER ADVANCE ONLINE PUBLICATION |1
Nature Reviews Cancer| AOP, published online 13 June 2013; doi:10.1038/nrc3486
2013 Macmillan Publishers Limited. All rights reserved
-
7/27/2019 Cancer as a Dysregulated Epigenome
2/14
certain post-translational histone modifica-tions. Typically, facultative heterochromatin that is, a region that can switch betweentranscriptionally repressive states and acti-
vated states is examined at a local genelevel. However, in addition to small-scalechanges, the genome is partitioned into largeeuchromatic and heterochromatic domains,which have been given different names formostly overlapping structures by laborato-ries that have approached this organizationusing varying methods. We recently reportedlarge organized chromatin lysine modifications(LOCKs), which are defined by genomicdomains enriched for heterochromatin post-translational modifications, such as histoneH3 lysine 9 dimethylation (H3K9me2)11.LOCKs expand during differentiation and arelost in cancer11(FIG. 1a,b). Heterochromaticregions can also be defined by their organiza-tion and position within the nucleus: DNA
sequences associated with proteins in thenuclear lamina are known as lamina-associateddomains (LADs)12. Heterochromatic regionsdefined by histone modifications (LOCKs)and those defined by nuclear location (LADs)have been shown to have 80% overlap in dif-ferent samples11,13,14, but a causal relationship,as in LADs controlling chromatin or chro-matin informing nuclear location, has not yetbeen proved.
LOCKs and LADs change during develop-ment, generally increasing in size. Genesin LADs are typically transcriptionallyrepressed15, but by artificially reorganizingthe nucleus to move genes to the nuclearperiphery, transcription profiles and his-tone modifications of chromatin contain-ing these genes are drastically altered15.Genes encoding proteins that are involvedin organizing the nuclear membrane alsohave altered expression in many differentcancer types16. Different laboratories haveobserved dynamic changes in chromatinstate by examining different histone sites for example, H3K9me2, H3K9 trimethyla-tion (H3K9me3) or H3K27me3 but stillnote that the prevalence of heterochromatic
regions is associated with the differentiationstate of the cell17,18. LOCKs are also alteredin cells undergoing epithelialmesenchymaltransition (EMT), an important behaviour incancer progression: during EMT, chromatinis reprogrammed in bulk, which results in adramatic loss of H3K9me2 and an increaseof H3K4me3 and H3K36me3 (REF. 19).Chromatin immunoprecipitation followed bymicroarray (ChIPchip) experiments carriedout on mouse chromosomes 414 showedloss of H3K9me2 in 96% of LOCKs but notin non-LOCK regions19.
The study of LOCKs and LADs in canceris very new, and even chromatin modifica-tions that are known targets of mutations,such as H3K27me3, have not yet beenanalysed systematically at a genome-scalesequencing level in cancer. A great deal ofdetail and mechanism needs to be fleshedout. For example, LOCKs and LADs maythemselves be nuanced with regard to com-binations of chromatin marks that definephysiologically distinct domains20. To date,other than pilot studies, there has beenno systematic analysis of primary humancancers and matched normal tissues withrespect to LOCKs and LADs. More detailedstudy has been carried out on blocks of DNAmethylation (discussed below), but theseneed to be related to LOCKs and LADs toform a complete picture of large-scale epi-genetic alterations in cancer. Euchromatinislands provide a clue to a possible connection
between LOCKs and LADs; these islands aresmall regions within the larger LOCKs andLADs that have reduced amounts of hetero-chromatin and are enriched for DNase hyper-sensitive sites and differentially methylatedregions in cancer21.
Hypomethylated blocks. We recently made asurprising discovery by using whole-genomebisulphite sequencing of human colorectalcancer samples, and this finding helps toexplain the earliest observation in cancerepigenetics: the widespread hypomethylationof genes in cancer3. By comparing three sam-ples of colorectal cancer to matched normalmucosa from the same patients, we identifiedlong blocks of hypomethylated DNA in can-cer with a median size of 28 kb and a maxi-mum size of 10 Mb (a range of 5 kb10 Mb)7(FIG. 1a,b). In blocks, normal samples exhib-ited methylation levels of ~80%, and thecancer samples ranged from 40% to 60%.One-third of transcriptional start sites arecontained within the large hypomethylatedblocks. Furthermore, these hypomethylatedblocks mostly corresponded to LOCKs andLADs, uncovering a surprising relationship
between large nuclear domains of both DNAand chromatin that are disrupted in cancer7.These findings were subsequently confirmedby others22.
There is a point of confusion in the lit-erature that we wish to clarify: in olderliterature based on Southern hybridization oflong interspersed nuclear element (LINE) or
Alu sequences, it seems that the DNA hypo-methylation in cancer is due to repetitivesequences and not to single-copy genes23.However, modern whole-genome bisulphitesequencing methods have demonstrated that
repetitive sequences, although somewhatenriched in hypomethylated blocks, arein fact no more hypomethylated thannon-repetitive sequences in the blocks7.
What is the potential role of hypomethy-lated blocks in cancer? An intriguing sug-gestion comes from an analysis of geneexpression. Although the overall level ofgene expression in hypomethylated blocksremains low in cancer, the hypomethylatedblocks contain the most variably expressedgenes in tumours compared with normalcontrols7. Furthermore, the DNA methy-lation levels in these regions were not onlyreduced, they were also extremely variablein the quantitative levels of DNA methy-lation7. Thus, although mean changes in geneexpression and DNA methylation in cancerare important, their heterogeneity may beequally or even more important in tumourheterogeneity and cancer progression and
may underlie tumour cell heterogeneity.Furthermore, similar structures known
as partially methylated domains (PMDs) have been found to be relevant in dif-ferentiation and reprogramming. PMDs arelarge regions that are differentially methy-lated between embryonic stem cells andfibroblasts24, as well as between inducedpluripotent stem (iPS) cells and fibroblasts25.These areas generally overlap with the blocksfound in cancer7 and are hypomethylated inmore differentiated cells. This reinforcesthe idea that there is a strong link between theepigenetic loci dysregulated in cancer andthe loci that show controlled alteration indifferentiation.
In addition, the hypomethylated blocksmay contribute to mutation. Hypomethylatedloci in cancer often coordinate with DNA-break hotspots, and may therefore contributeto copy number changes26. As these primaryobservations are so new, it is likely that addi-tional mechanisms linking these considerableregional DNA methylation changes to cancerwill be uncovered over time.
5-hydroxymethylcytosine is a pro-posed intermediate in the demethylation
of cytosine, but it is indistinguishablefrom 5-methylcytosine by bisulphite- orrestriction enzyme-based techniques27.Affinity-based methods (for example,5-hydroxymethylcytosine DNA immuno-precipitation (hMeDIP) and methyl-CpGbinding domain (MBD)-binding assays28)and chemical methods (for example, liq-uid chromatographymass spectrometry(LC-MS)29) can distinguish hydroxymethylfrom methyl modifications. Although theabsolute level of hydroxymethylation inmost normal tissue is low, recent work has
P E R S P E C T I V E S
2 | ADVANCE ONLINE PUBLICATION www.nature.com/reviews/cancer
2013 Macmillan Publishers Limited. All rights reserved
-
7/27/2019 Cancer as a Dysregulated Epigenome
3/14
Normal cell Cancercell
a b
c
RNA
RNA
Euchromatin(on)
Heterochromatin(off)
OnOff
Off
Poised
Selection in responseto varying environment
Antisense RNA Other?
DOT1L p300
TET2
MLL
Normal
Hypomethylated shore
Hypermethylated island
CpG island
LOCKs and LADs
shown a relative reduction in the levelsof hydroxymethylation in melanoma30,liver cancer29 and colorectal cancer31. Newsingle-base-resolution methods that aresensitive to hydroxymethylation are emerg-ing to aid in distinguishing differences inhydroxymethylation from total methylation
in cancer32,33.
Small epigenetic structures
The role of DNA methylation in smallerregions of DNA, such as CpG islands (CGIs),is part of the classical cancer epigeneticsliterature, but here too our perspective hasbeen greatly changed by the advent of newergenomic technologies. For example, theexistence ofCpG island shores (CGI shores)and of asymmetric division of nucleosomesduring DNA replication were unknown untilrecently.
CGIs and shores. In 1982, Wolf and Migeon34discovered highly CpG-enriched sequencesthat, when methylated on the inactiveX chromosome, are associated with silencingof housekeeping genes. Bird and colleagues35later identified what they termed islandsof CpG-rich sequences enriched at genes
throughout the genome.The observation of the hypomethylation
of genes in human colorectal cancer3 wasextended shortly thereafter to a larger seriesof tumours, including pre-malignant adeno-mas, with hypomethylation as an apparentlyubiquitous feature of cancer36(FIG. 1a,b). Theoverall global reduction of 5-methylcytosinein tumours was confirmed by quantitativehigh-performance liquid chromatography(HPLC)37. Many laboratories identifiedgenes activated by hypomethylation, includ-ing oncogenes, such as HRAS38, and the
families of genes expressed normally intestis and aberrantly activated in tumours,such as the melanoma-associated antigen(MAGE) family in melanoma39. Additionalhigh-throughput array-based methodshave identified hundreds of genes that areepigenetically activated in various cancers,
including lung, gastric, colon, pancreatic,liver and cervical cancers4048.
Arguing that epigenetic gene silencingmight involve tumour suppressor genes,in 1991 Horsthemke5 and Dryja6 indepen-dently identified hypermethylation of a CGIupstream of the RB1 tumour suppressorgene5,6. Many tumour suppressor genes havesince been associated with hypermethy-lated CGIs49(FIG. 1a,b). However, there areseveral conundrums in this work. One issueis that much of this research was dedicatedto the analysis of stable tumour cell lines and
Figure 1 | Alterations in the cancer epigenome that can cause epi-
genetic dysregulation. a | Large organized chromatin lysine modifica-
tions (LOCKs) and lamin-associated domains (LADs) (shown here in large
scale) are associated with the nuclear membrane and are generally hetero-
chromatic, with a high level of DNA methylation. Transcriptionally active
genes have less compact nucleosomes than silent genes, and active and
silent genes are distinguished by differing post-translational modifications
of histones (green represents on and red represents off), as well as increasedDNA methylation (shown in blue) of silent genes. b | In cancer there is a
reduction of LOCKs, as well as general disorganization of the nuclear mem-
brane and hypomethylation of large blocks of DNA corresponding approxi-
mately to the LOCKs and LADs. Chromatin is in a more stem cell-like state
with the ability to differentiate into euchromatin and hypomethylated
genes, or into heterochromatin and hypermethylated genes. Our argument
is that epigenetic dysregulation allows for selection in response to the cell-
ular environment for cellular growth advantage at the expense of the host.
Mechanisms include mutations in epigenetic regulatory genes (for example,
DOT1-like (DOT1L), mixed lineage leukaemia (MLL),EP300(which encodes
p300) and tet methylcytosine dioxygenase 2 (TET2)) and primary epigenetic
modifications with positive feedback. c | Loss of boundary stability of methy-lation at CpG islands includes the encroaching of boundaries, leading to
CpG island hypermethylation, and the shifting out of boundaries, leading
to hypomethylated CpG shores. Both mean shifts in methylation and
hypervariability allow for selection.
P E R S P E C T I V E S
NATURE REVIEWS |CANCER ADVANCE ONLINE PUBLICATION |3
2013 Macmillan Publishers Limited. All rights reserved
-
7/27/2019 Cancer as a Dysregulated Epigenome
4/14
immortalized cell lines, which show markedhypermethylation of CGIs in general50.Furthermore, as Bestor and others5153 haverepeatedly pointed out, most hypermethy-lated tumour suppressor gene-associatedCGIs are not in the promoters of thesegenes, and thus the hypermethylation ofthese sequences is likely to be consequentialrather than causal5153. It has been proposedthat a priori methylation is a mechanism oftumour suppressor gene silencing that cancause cancer predisposition (rather thanbeing a late event in tumorigenesis) in a simi-lar manner to the cancer predisposition thatis caused by germline mutations; however,the data supporting this have been relativelysparse51. The most exciting example isMLH1methylation transmitted as a germline trait,but this report was repudiated by most of itsauthors owing to contamination of the germcells with stroma54,55.
Indeed, we believe the mechanism oftumour suppressor gene silencing to beprimarily driven by chromatin modificationand not by DNA methylation. Vogelstein andcolleagues56 showed in 2003 that tumoursuppressor gene silencing seems to bedriven by histone modifications before DNAmethylation changes56. Recently, Sprouland colleagues57 directly showed that DNAhypermethylation of tumour suppressorgenes in breast cancer occurs at sites that
are already repressed in normal cells of thesame lineage57. The same group extendedthese convincing results regarding the lack ofa role for DNA hypermethylation in cancerdevelopment to 1,154 human cancer samplesfrom seven different tissue types58. Similarfindings at individual loci have also beenshown by others52,53.
Loss of CGI boundary stability in cancer.With the advent of whole-genome epigeneticanalysis, it was possible to broaden the focusof cancer epigenetics to consider regions out-side the relatively limited CGIs. We designeda microarray using an algorithm that isagnostic to genes and CGIs59, and found thatmost methylation differences between tissues(tissue-specific differentially methylated regions(tDMRs)) occurred outside CGIs, oftenwithin 2 kb of CGI boundaries in regionsthat are now commonly referred to as CGI
shores60. A retrospective analysis of previouswork agrees with this result; tDMR loca-tions are more common in low-CpG-densitypromoters, although the connection to CGIshores was not identified24.
Furthermore, compared with the fibro-blasts of origin, 70% of altered methylationregions (reprogramming-specific differentiallymethylated regions (rDMRs)) in iPS cells arelocated in CGI shores. This suggests thatshores are important for differentiation and
reprogramming61. Indeed, colon cancer canin most cases be distinguished from normalcolon tissue using these rDMRs, suggestingthat carcinogenesis may involve a partialreprogramming of the epigenome towards amore stem cell-like state61.
What is the function of CGI shores? Onepossibility is that they are sites of alternativetranscription and enhancer binding regions;this is in contrast to CGIs, which Wolf andMigeon34 and others demonstrated arestrongly protected from DNA methylationto maintain housekeeping gene function34.Indeed, hypomethylated CGI shores wereshown to activate alternative transcriptionalstart sites proximate to cancer-specific differen-tially methylated regions (cDMRs), as shownby 5 rapid amplification of cDNA ends(RACE) experiments60.
Another clue to the function of CGIshores comes from whole-genome bisul-
phite sequencing. Cancers lose the sharplydemarcated boundary between high andlow methylation that is defined by CGIs;that is, at the CGI shores7. Thus, when theboundary between high and low methy-lation shifts inwards towards the CGI, theCGI shore becomes hypermethylated. Whenthe boundary shifts outwards, the CGI shorebecomes hypomethylated (FIG. 1c). The ero-sion of these sharply defined boundariesresults in altered gene expression7.
Glossary
Bivalent modifications
Nucleosomes containing both euchromatic histone H3
lysine 4 trimethylation (H3K4me3) and heterochromatic
H3K27me3 post-translational modifications.
Cancer hallmarks
Ten biological properties of cancer that are said to define
the disease; we argue that they arise by natural selection
for cellular survival at the expense of the host in the setting
of epigenetic dysregulation and random variation.
Cancer-specific differentially methylated regions
(cDMRs). Differentially methylated regions that distinguish
cancer cells from normal cells.
ChemopreventionAdministration of pharmacological compounds to reduce
cancer incidence without certain knowledge of its effect on
a given patient.
CpG islands
(CGIs). Areas of high CpG dinucleotide density in the
genome, typically defined as a region at least 200 bp long
with >50% GC dinucleotides and an observed-to-expected
CpG ratio of >0.6.
CpG island shores
(CGI shores). The region 2 kb on either side of a CpG
island, and the location of most cancer-specific,
tissue-specific and reprogramming-specific differentially
methylated regions.
Epigenetic dysregulation
The loss of normal control of DNA methylation or chromatin
as a result of injury, epigenetic change or mutation, leading
to phenotypic drift.
Epigenetic variability
Increased inter-sample variation in the methylation or
chromatin state. This was recently identified as a common
property of cancer, allowing for more accurate detection
between samples.
Euchromatin
Areas of the genome that are more open to transcription
owing to post-translational modifications of histones and
with less nucleosome density.
Heterochromatin
Areas of the genome that are less open to transcription
owing to post-translational modifications of histones and
with greater nucleosome density. Facultative
heterochromatin can change between the two states.
Large organized chromatin lysine modifications and
lamina-associated domains describe heterochromatin
over relatively large regions and are associated with the
nuclear membrane.
Hypomethylated blocks
Large (mean 144 kb) regions that are broadly
hypomethylated in cancer and that mostly overlap with
large organized chromatin lysine modifications and
lamina-associated domains.
Lamina-associated domains
(LADs). Genomic regions located in the nuclear periphery
that are associated with lamina (an inner nuclear
membrane-associated protein) and usually have low
expression levels.
Large organized chromatin lysine modifications
(LOCKs). Large heterochromatic regions characterized by
low gene expression that are altered between somatic and
stem cells; they are typically lost in cancer cells.
Loss of imprinting
(LOI). Loss of parent of origin-specific expression in cancer
of imprinted genes, first observed for insulin-like growth
factor 2 (IGF2) in Wilms tumour and colorectal cancer.
OrnsteinUhlenbeck process
An overdamped Brownian harmonic oscillator that is,
stochastic variation from a normal state with no persistence
of the rate of change opposed by a stronger restoring
force towards the equilibrium point. We are using this to
model stochastic change in DNA methylation.
Reprogramming-specific differentially methylated
regions
(rDMRs). Differentially methylated regions that distinguish
reprogrammed stem cells from somatic cells.
Tissue-specific differentially methylated regions
(tDMRs). Differentially methylated regions that distinguish
normal tissues from each other.
P E R S P E C T I V E S
4 | ADVANCE ONLINE PUBLICATION www.nature.com/reviews/cancer
2013 Macmillan Publishers Limited. All rights reserved
-
7/27/2019 Cancer as a Dysregulated Epigenome
5/14
Furthermore, a striking hypervariabilityin DNA methylation is found at these CGIshores or boundaries in cancer samples,similar to the hypervariability of DNA methy-lation described above for the large blocks7(FIG. 1c). This same property of hypervariableDNA methylation at the CGI shores orboundaries and blocks is a general propertyof cancer, affecting at least breast, colorectal,kidney, lung and thyroid tumours7. Tissueheterogeneity does not explain this hypervari-ability because normal tissue displays evengreater heterogeneity than cancer samples7. Infact, an increase in methylation hypervariabil-ity in phenotypically normal tissue is predic-tive of future cancer development62. A similarhypervariability is found in gene expression incancer63, and the most hypervariable of thesegenes are found within the large blocks7. Weemphasize that mean changes in gene expres-sion are important but that variance may be
equally important in tumour progression.
Small chromatin domains. Individual nucleo-somes, as opposed to the large domainsdescribed above, are also likely to be involvedin affecting tumour progression. The organi-zation of chromatin into euchromatic andheterochromatic structures is controlled bynucleosome positioning, which functionstogether with post-translational modifica-tions of histone tails (for example, in enhanc-ers)64,65. Physical access to DNA is restrictedby nucleosome positioning and packing chromatin remodelling complexes act toalter this in cancer66. For example, transcrip-tional activity is associated with nucleosomedepletion67, with transcription-activatinghistone modifications such as acetylatedH3K14 (REF. 68) and with the presence ofspecific histone variants such as H3.3 (REF. 69)and H2A.Z70. It is important to consider thecomplex combinatorial nature of the histonecode; different histone modifications oftenact together, meaning that each modificationmust be considered in context with the othermodifications that are present on the nucleo-some71. Some regions are even bivalent,
with nucleosomes having both H3K4me3(a euchromatic, transcriptionally activemodification) and H3K27me3 (a heterochro-matic, repressive modification), implyinga metastable pluripotent state72. Althoughprevious work has demonstrated that bivalentmodifications do not occur on the same his-tone tail73, more recent work has shown thatthese opposing modifications localize to asingle nucleosome, with repressive and acti-
vating marks on different H3 proteins withinthe same nucleosome74. These same regionsare associated with hypermethylated CGIs in
cancer75,76(FIG. 1a,b) and reprogramming61,and in fact a relationship has been shownbetween DNA methylation and nucleosomepositioning77 and histone modification18.
We and others78,79 have identified onemechanism for heterochromatin-inducedsilencing of tumour suppressor genes.Antisense expression of cyclin-dependentkinase inhibitor 2B (Cdkn2b, which encodesp15) generates heterochromatin formation atthe sense promoter, leading to gene silencing,and this mechanism seems to be importantin leukaemia78,79. Hypermethylation ensueson cell differentiation, and the expressionof antisense RNA has been shown to actas a mediator of chromatin remodellingand heterochromatin formation (FIG. 1a,b).Intriguingly, CGI hypermethylation doesoccur but only arises after heterochromatinformation79. Cis-acting non-coding RNAs atpromoters and enhancers can regulate chro-
matin at gene promoters80. The observationof chromatin modifications preceding DNAmethylation during differentiation resonateswith the observations that CGI hypermethy-lation arises secondarily to chromatinmodifications in tumour suppressor genesilencing56, and that the hypermethylatedCGIs in cancer are located at genes that havealready been silenced in normal tissue57,58,presumably through chromatin.
Mutations and the epigenome
The recent discovery of several mutatedepigenetic modifiers in human cancer pro-
vides a potential mechanism by which DNAmutation might lead to epigenetic alterations.Given the apparently universal presence ofDNA methylation and chromatin alterationsin human cancer, we summarize below thefrequency of mutations of epigenetic modify-ing genes, going gene by gene and tumourby tumour, beginning with the Catalogueof Somatic Mutations in Cancer (COSMIC)database81 and then reviewing the originalcitations. The classes of genes include histone
variants (direct substitution of a mutant his-tone isoform); DNA methyltransferases;
histone acetyltransferases; histone deacety-lases; histone methyltransferases; histonedemethylases; and chromatin remodel-ling factors, which can induce changes ineuchromatin and heterochromatin (TABLE 1).Mutations in chromatin readers are also occa-sionally involved in cancer but apparently notas drivers of cancer progression82.
Our analysis of the mutation frequencyof epigenome modifiers in cancer reveals asurprising pattern. Although there is clearlya relationship between mutations and epi-genetic modification in cancer, most of the
epigenetic-associated mutations in solidtumours identified to date involve eitherrare aggressive variants of adult tumours orpaediatric cancers. For example, the com-mon form of pancreatic adenocarcinomashows an 8% mutation frequency in thehistone acetyltransferase p300 (EP300)83, butthe rarer pancreas neuroendocrine cancerhas a 44% frequency of the histone methyl-transferase multiple endocrine neoplasia I(MEN1)84(TABLE 1). Similarly, childhoodglioblastoma, an extremely rare brain cancer,shows frequent (35.6%) mutations in histoneH3 family 3A(H3F3A), but adult glioblas-tomas show a drastically lower frequency ofH3F3A mutations (3.4%)85(TABLE 1).
By contrast, haematological cancersfrequently involve chromosomal rearrange-ments of epigenetic modifiers, and this hasbeen known for many years; for example, inmixed-lineage leukaemias86. Acute myeloid
leukaemia involves mutations in several genesencoding proteins that modify DNA methy-lation, including DNA methyltransferase 3A(DNMT3A), isocitrate dehydrogenase 1(IDH1) and IDH2. These mutations lead toeither decreases in DNA methylation (IDH1and IDH2 mutations) or increases in DNAmethylation (DNMT3A mutations)8789.Chronic myelomonocytic leukaemia involvesmutations in tet methylcytosine dioxyge-nase 2 (TET2), which is involved in DNAdemethylation90. Lymphomas involve fre-quent inactivating mutations in the histoneacetyltransferasesEP300 and CREB bind-ing protein (CBP; also known as CREBBP),leading to increased heterochromatin andresulting in gene silencing91,92. The histonemethyltransferase mixed-lineage leukaemia(MLL), which undergoes translocations asa defining characteristic of mixed-lineageleukaemia, functions through DOT1-like(DOT1L), an H3K79me2 methyltransferase,which leads to specific gene activation93.
Epigenetic mutations also frequentlyoccur in cancers that relapse or that areotherwise resistant to therapy, such as muta-tions ofCBPin relapsed ALL94. A detailed
analysis of these mutations can be found inTABLE 1. Chromatin remodelling proteins for example, AT-rich interactive domain-containing protein 1A (ARID1A) areperhaps the most frequently mutated classof epigenetic modifying proteins in com-mon solid tumours, and their consequentinactivation leads to increased levels ofeuchromatin and gene activation (TABLE 1).A dramatic example of epigenetic genemutation coupled with aggressiveness incancer is the recent finding that the his-tone H3 variant H3.3 is itself mutated in
P E R S P E C T I V E S
NATURE REVIEWS |CANCER ADVANCE ONLINE PUBLICATION |5
2013 Macmillan Publishers Limited. All rights reserved
http://cancer.sanger.ac.uk/cancergenome/projects/cosmic/http://cancer.sanger.ac.uk/cancergenome/projects/cosmic/ -
7/27/2019 Cancer as a Dysregulated Epigenome
6/14
Table 1 | Epigenome-modifying gene mutations in human cancer
Gene Cancer Frequency or stage of cancer
Frequency ofmutation (N)
Effect Refs
Histone variants
HIST1H1B Colorectal cancer Common 4% (24) 149
HIST1H1C Non-Hodgkins lymphoma Common 7% (127) 150
H3F3A Paediatric glioblastoma Rare aggressivepaediatric, high grade
36% (90) Prevents PTMs on H3K27or H3K36
85
Diffuse intrinsic pontine glioma Rare aggressive paediatric 60% (50) Prevents PTMs on H3K27 95
HIST1H3B Diffuse intrinsic pontine glioma Rare aggressive paediatric 18% (50) Prevents PTMs on H3K27 95
DNA methyltransferases
DNMT1 Colorectal cancer 2% (29) Mutation 151
DNMT3A AML Stage M4 13.6% (66) 87
Stage M5 20.5% (112) 87
AML Common 22.1% (281) 88
DNA demethylases
TET2 BCR-ABL-negative myeloproliferativeneoplasms
Rare form 13% (239) 152
CMML Common form 50% (88) 90
MDS Rare 26% (102) 153
IDH1 Anaplastic astrocytoma Rare 73% (52) 154
Diffuse astrocytoma Rare 90% (30) 154
AML Common 6.2% (385) 89
IDH2 AML Common 8.6% (385) 89
Histone acetyltransferases
EP300 (which encodesp300)
Pancreas adenocarcinoma Common form 8% (24) Mutation 83
DLBCL Common form 10% (134) Mutation 91
Follicular lymphoma Uncommon form 8.7% (46) Mutation 91
Head and neck squamous cell cancer Common 11% (74) Mutation 155Transitional cell carcinoma (bladder) Common 13% (97) Mutation 156
CREBBP (whichencodes CBP)
Ovary Common 3% (75) Inactivated 157
Breast adenocarcinoma Common 8% (183) Gain of copy 158
Lung cancer Common 5.3% (95) Mutation 159
DLBCL Common form 22.4% (134) Mutation 91
DLBCL Common form 18% (111) Mutation 92
Follicular lymphoma Uncommon form 32.6% (46) Mutation 91
Relapsed ALL 18.3% (71) Mutation 94
Transitional cell carcinoma (bladder) Common 13% (97) Mutation 156
ELP4 Breast adenocarcinoma Common 4% (183) Gain of copy 158
Histone deacetylases
HDAC4 Breast adenocarcinoma Common 4% (24) Mutation 149
HDAC9 Prostate adenocarcinoma Common 42.9% (7) Mutation 160
Histone methyltransferases
SETD2 Renal clear cell carcinoma Common 3% (407) Mutation 161
MEN1 Pancreas neuroendocrine cancer Rare 44% (68) Mutation 84
Parathyroid cancer Rare 35% (185) Mutation 162
MLL Squamous cell lung cancer Rare 3% (63) Mutation 158
Transitional cell carcinoma (bladder) Common 7% (97) Mutation 156
Mixed lineage leukaemia Common 100% (definition) Fusion 86
P E R S P E C T I V E S
6 | ADVANCE ONLINE PUBLICATION www.nature.com/reviews/cancer
2013 Macmillan Publishers Limited. All rights reserved
-
7/27/2019 Cancer as a Dysregulated Epigenome
7/14
Table 1 (cont.) | Epigenome-modifying gene mutations in human cancer
Gene Cancer Frequency or stage of cancer
Frequency ofmutation (N)
Effect Refs
Histone methyltransferases (cont.)
MLL2 Renal clear cell carcinoma Common 4% (407) Mutation 161
Childhood medulloblastoma Rare 8.7% (92) Mutation 163
Childhood medulloblastoma Rare 13.6% (88) Mutation 164
DLBCL Common form 32% (37) Mutation 150
DLBCL Common form 22.8% (92) Mutation 92
Follicular lymphoma Uncommon form 89% (35) Mutation 150
Head and neck squamous cell cancer Common 11% (74) Mutation 155
MLL3 Childhood medulloblastoma Rare 3.4% (88) Mutation 164
Transitional cell carcinoma (bladder) Common 5% (97) Mutation 156
Colorectal cancer Common 20.8% (24) 149
EZH2 Non-Hodgkins lymphoma Common 7.8% (681) Mutations 165
DLBCL Common form 5.6% (107) Mutation 92
MDS and MPNs Rare 12% (219) Mutations 166
Myelofibrosis Rare 13% (30) Mutations 166
Follicular lymphoma Uncommon form 12% (221) Mutations 167
Histone demethylases
KDM5C (also known asJARID1C)
Renal clear cell carcinoma Common 3% (407) Mutation 161
KDM6A (also knownasUTX)
Transitional cell carcinoma (bladder) Common 20% (97) Mutation 156
Childhood medulloblastoma Rare 3.2% (92) Mutation 163
KDM2B DLBCL Uncommon form 7.4% (54) Mutation 92
Chromatin remodelling factors
ARID1A Pancreas adenocarcinoma Common 8% (24) Mutation 83
Ovarian clear cell carcinoma Rare 57% (42) Mutation 96
Ovarian clear cell carcinoma Rare 46% (119) Mutation 97
Endometrial cancer Common 30% (33) Mutation 97
Transitional cell carcinoma (bladder) Common 13% (97) Mutation 156
Hepatocellular carcinoma Common 16.8% (125) Mutation 168
Colorectal adenocarcinoma Hypermutated 37% (30) Mutation 169
Non-hypermutated 5% (165)
ARID1B Breast adenocarcinoma Common 5% (100) Mutation 170
ARID2 Hepatocellular carcinoma Common 5.6% (125) Mutation 168
Melanoma Common 9% (121) Nonsense mutation 171
CHD1 Prostate adenocarcinoma Common 42.9% (7) Mutation 160
CHD5 Prostate adenocarcinoma Common 42.9% (7) Mutation 160
PBRM1 Clear cell renal carcinoma Common 41% (227) Mutation 172
ATRX Pancreas neuroendocrine cancer Rare 25% (68) Mutation 84
DAXX Pancreas neuroendocrine cancer Rare 17.6% (68) Mutation 84
SMARCD1 Breast adenocarcinoma Common 4% (100) Mutation 170
SMARCB1 (also knownasSNF5and INI1)
Malignant rhabdoid cancer Rare 100% (29) Loss of copy or mutation 173
SMARCA4 Childhood medulloblastoma Rare 4.3% (92) Mutation 163
ALL, acute lymphoblastic leukaemia;AML,acute myeloid leukaemia;ARID,AT-rich interactive domain;ATRX,-thalassaemia/mental retardation syndrome X-linked;CHD, chromodomain helicase DNA binding protein;CMML, chronic myelomonocytic leukaemia;CREBBP,CREB binding protein;DAXX,death-domain associated protein;DLBCL, diffuse large B cell lymphoma; DNMT, DNA methyltransferase;ELP4,elongator acetyltransferase complex subunit 4; EZH2, enhancer of zest homologue 2;H3F3A,histone H3 family 3A;HDAC, histone deacetylase;HIST1H1B,histone cluster 1, H1b;HIST1H1C,histone cluster 1, H1c;HIST1H3B,histone cluster 1, H3b; IDH, isocitratedehydrogenase;KDM, lysine-specific demethylase;MDS, myelodysplastic syndrome;MEN1,multiple endocrine neoplasia I;MLL, mixed lineage leukaemia;MPNs,myeloproliferative neoplasms;PBRM1,polybromo 1;PTMs,post-translational modifications;SETD2,SET domain containing 2; TET2, tet methylcytosine dioxygenase 2.
P E R S P E C T I V E S
NATURE REVIEWS |CANCER ADVANCE ONLINE PUBLICATION |7
2013 Macmillan Publishers Limited. All rights reserved
-
7/27/2019 Cancer as a Dysregulated Epigenome
8/14
paediatric glioblastoma, thus preventingH3.3K27 modifications85,95. Such a mutationwould be expected to substantially affectchromatin structure, causing aberrant geneexpression and potentially allowing for theacquisition of aggressive properties; it shouldbe noted that adult glioblastomas do notfrequently contain this mutation. A similarargument can be made for ovarian clear cellcarcinoma mutations in the chromatinremodeller ARID1A were found to be com-mon in this aggressive subtype in two differ-ent studies (57% and 46% frequency)96,97but were not present in high-grade serousovarian carcinoma97.
A very important possibility in consider-ing the relationship between DNA mutationsand epigenetic modification in cancer isthat altered DNA methylation or chromatinmodifications may change the mutation rateitself. For example, guanine quadruplexes
(G4s) increase the risk of DNA breakage andactivation of the homologous recombinationDNA repair pathway; these breaks are inhib-ited by DNA methylation. Hypomethylatedloci in cancer often coordinate with DNAbreak hotspots, and these are enriched inG4s26. Methylation-mediated mutationthrough spontaneous deamination may alsogive rise to mutation; 18.2% of inherited genemutations occur as CG>TA mutations inCpG dinucleotides98; CG>TA mutations
also make up the bulk of substitutions inmany cancers99, even specifically at CpGdinucleotides in some cases83. More substan-tially, chromatin state has been shown to cor-relate extremely well with somatic mutationrate: H3K9me3 levels alone are predictiveof >40% of somatic mutation loci in humancancer samples100. Common DNA fragilesites, which are implicated in copy number
variation in cancer, also have decreased sta-bility in regions of heterochromatin101. Theorganization of chromatin and genetic archi-tecture of the nucleus have a direct effect onthe rate and effectiveness of copy numberalterations and rearrangements in cancer102.These data represent clear correlationsbetween areas ofepigenetic dysregulation andmutation, suggesting a collaborative effortbetween epimutation and genetic mutationin cancer development.
Abnormal expression of epigenetic modifiergenes. It is also important to note that manyalterations in the expression of epigeneticmodifiers have been reported, and someof these seem to have an important rolein tumour progression (and have beenreviewed extensively elsewhere64,65,103).Perhaps the most notable example is over-expression of enhancer of zeste homo-logue 2 (EZH2), which results in increasedH3K27me3 levels and thus silences tumour
suppressor gene expression and promotesmetastasis104. Therefore, these alterations inthe epigenetic machinery are more complexto understand, and they should be viewednot as equivalent to the mutations sum-marized in TABLE 1 but rather as membersof a positive feedback loop that leads toepigenetic dysregulation. An example of apositive epigenetic feedback loop in malig-nant transformation has recently been dem-onstrated directly in the nuclear factor-B(NF-B) pathway105. Another example is theclass of reprogramming factors that lead tothe generation of iPS cells106. We have alsoobserved a significant overlap of rDMRs andcDMRs61. Furthermore, many of the repro-gramming genes are overexpressed in can-cer107. A non-exhaustive list of examples ofalterations in gene expression of epigeneticmodifiers in cancer is provided in TABLE 2.
One lesson from examining the expression
of epigenetic modifiers in cancer is that thebalance of euchromatic and heterochromatichistone modifications is crucial a modi-fication too far in either direction towardseuchromatin or heterochromatin leads todysregulation of gene expression and isadvantageous for tumour growth (see REF. 108for an example). This imbalance could be atarget for therapy: for example, histone modi-fications could be brought back into balancethrough small-molecule inhibitors of histonedeacetylases, histone acetyltransferases orhistone methyltransferases108.
Cancer as epigenetic dysregulation
In summary, most cancers may share sev-eral common epigenetic modifications:large-scale alterations in chromatin involv-ing LOCKs or LADs and hypomethylatedblocks, and loss of methylated boundarystability at CGIs leading to hypermethy-lated CGIs, hypomethylated CGI shores andaberrant gene expression. This would leadto a drift towards a hybrid stemsomaticcell state, with increased methylation ofPolycomb target regions and loss of methy-lation at pluripotency loci. A simple unifying
explanation of these results is that cancer iscaused by epigenetic dysregulation, whichcould account for the high degree of pheno-typic variability that is observed among indi-
vidual cancers and that leads to selection forcancer cell survival independent of the host.
To visualize this process, consider theclassic epigenetic landscape described byWaddington109, by which the normal epige-netic signature of the cell is represented bya ball trapped in a valley, the walls of whichrepresent a restoring force constraining theball in its normal state (FIG. 2a). Although this
Table 2 | Altered expression of some epigenetic modifying genes in cancerGene Change Cancer Refs
IGF2 Increased LOI in colorectal, gastric and breast cancers 123,174,175
Class I HDACs Increased Gastrointestinal, prostate, breast andcervical cancers
176180
EZH2 Increased Prostate cancer 104
EZH2 Increased Breast cancer 181
HDACs Increased Several 182
HATs Decreased Several 182
HDACs Increased Colon cancer 183,184
HDAC6 Increased Breast cancer 185
SIRT1Increased Prostate cancer 186
SIRT3 Increased Breast cancer 187
KDM5C Increased Breast cancer 188
SMYD3 Increased Liver, colon and breast cancers 189
EHMT1 Decreased Medulloblastoma 190
DNMT1 Increased Pancreas, liver, bladder and breast cancers 191194
DNMT3B Increased Breast cancer 195
AID Increased Leukaemia 196
AID, activation-induced cytidine deaminase;DNMT, DNA methyltransferase;EHMT1, euchromatichistone-lysineN-methyltransferase 1; EZH2, enhancer of zeste homologue 2; HAT, histone acetyltransferase;
HDAC, histone deacetylase; IGF2, insulin-like growth factor 2; KDM5C, lysine-specific demethylase 5C (alsoknown asJARID1B);LOI, loss of imprinting; SIRT, sirtuin; SMYD3, SET and MYND domain-containing 3.
P E R S P E C T I V E S
8 | ADVANCE ONLINE PUBLICATION www.nature.com/reviews/cancer
2013 Macmillan Publishers Limited. All rights reserved
-
7/27/2019 Cancer as a Dysregulated Epigenome
9/14
100
80
60
40
20
0
Methylationle
vel(%)
Time
Normal tissue
Adenoma Carcinoma
100806040200Methylation (%)
Samplehistogram
100806040200
Methylation (%)
Simulatedhistogram
Regulationpotential
Regulationpotential
Methylation level
Methylation level
a c d
eb
signature differs among tissues, it is highlyregulated, invariant among individuals andultimately defined genetically, according tothe classic view109. In order to be dynamicand flexible, the epigenetic signature mustallow for variation, provided in the form ofintrinsic noise: that is, a biochemical char-acteristic of the system that leads to randomdeparture from a set point110. For example,methylation inheritance shows an error ratethat is estimated to be 4% for a given CpGmotif per cell division in a cell population111.This is also consistent with the idea thatepigenetic variability can lead to phenotypicselection on a much shorter timescale thancan mutation112. Furthermore, variation inthe epigenome may be controlled by factorsin the genetic code, providing a potential
mechanism for control of the level of thisvariation (which is represented by the slopeof the valley walls)113.
In cancer, an initial dysregulation of theepigenome would flatten this valley, break-ing the delicate balance of regulation thatmaintained the stable epigenetic signaturein the face of noise (FIG. 2b). This could bethe result of repeated restructuring of theepigenetic landscape through inflammatoryinsult, as in the case of Barretts oesopha-gus114, or it could be through an initial muta-tion in one of the genes directing the fragile
balance of the epigenome (almost any of themutations presented in TABLE 1). A splicing
variation in an epigenetic modifier genecould also cause dysregulation. For example,an isoform of DNMT3B commonly found incancer, DNMT3B7, results in increased vari-ation in the methylation signature115. Eventhe classical gatekeeper mutations mayhave a role in unleashing epigenetic varia-tion, given that cancer-associated mutants ofadenomatous polyposis coli (APC), BRCAand p53 interact with epigenome-controllingproteins. We propose that after regula-tion of the epigenetic signature is relaxed,stochastic variation becomes the drivingforce in patterning the epigenome, allowingDNA methylation or chromatin structure togradually diffuse away from its initial state.
Natural selection within the host then allowseach cancer to move differently, also leadingto substantial epigenetic variation betweena given cancer type across individuals orbetween metastases of a given cancer.
This epigenetic signature is constantlybuffeted by stochastic variation, as if theball is rocking randomly from side to side.The restoring force that is, the network ofgenes that maintain epigenetic homeostasis prevents the epigenetic signature fromwandering too far from its equilibrium pointin normal tissue. We can model this variation
using an OrnsteinUhlenbeck process(FIG. 2c);such processes are already used in biology formodelling selection pressure versus randomgenetic drift116. On the basis of this model,our argument that cancer involves a loss ofregulation of the epigenome means that therestoring force is reduced or lost altogether(the positive feedback referred to above).
We generated simulated methylation lev-els using this process (FIG. 2c). Before carcino-genesis, methylation levels have relatively low
variability, oscillating stochastically aroundtheir equilibrium point. When the simulationreaches the carcinogenesis point, we reducedthe restoring force in the process, flatten-ing the Waddington valley(FIG. 2b), andallowed the simulation to continue. Thesimulated methylation levels subsequently
exhibited a random walk away from theirpreviously well-ordered profile. It is impor-tant to note that the distance from the origi-nal equilibrium point increases over time,but not directionally: instead it increasesby a diffusive spreading of the epigeneticsignature (FIG. 2c).
We then applied our model to existingDNA methylation data from our previ-ous work7, selecting a CpG from within ahypomethylated block as an example. Thedensity plot of methylation values for nor-mal, adenoma and cancer tissue at this CpG
Figure 2 | Modelling epigenetic dysregulation using an Ornstein
Uhlenbeck process. We used an OrnsteinUhlenbeck process to model
stochastic change in DNA methylation opposed by regulatory proteins.
a | In normal tissue, the methylation level can be represented as a ball at thebottom of a valley stochastic noise allows it to vary slightly, but regulatory
forces represented by the walls of the valley keep the levels clustered
around a single point. b | An early carcinogenic event flattens the landscape,
leading to more variable methylation levels. c | Using the EulerMurayama
method, we can model this behaviour as = ( M)dt+dW, with M beingthe methylation value, the equilibrium point, the restoring force, thenoise level (4%) and dWa Wiener process increment. Shown are ten
example traces of simulated methylation levels. Regulatory forces () are sethigh in the normal tissue and low after a carcinogenic event. As time pro-
gresses, samples of the simulation are taken, representing different stages
of cancer progression: that is, normal, adenoma and carcinoma. d | Densityplots of methylation data of a single CpG site fromREF. 7 (Gene Expression
Omnibus number: GSE29381) showing methylation histograms for normal
(green;N= 29), adenoma (blue; N= 31) and carcinoma (red; N= 10) colon
samples. The plots were generated by Gaussian kernel smoothed density
function inR. e | Density plots of simulations for the same CpG showing the
combined results from 100 simulations using the same sampling methods
as in part d. The model provides an excellent fit to the data.
P E R S P E C T I V E S
NATURE REVIEWS |CANCER ADVANCE ONLINE PUBLICATION |9
2013 Macmillan Publishers Limited. All rights reserved
http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE29381http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE29381 -
7/27/2019 Cancer as a Dysregulated Epigenome
10/14
demonstrate the tight distribution in normaltissue, with a progressive relaxation fromnormal to adenoma to carcinoma (FIG. 2d).The simulated values match the actual dataextremely well (FIG. 2d,e). This model isintended to suggest the underlying behaviourthat may explain the increased distance fromthe normal profile we observe in tumourcells over time. We have not studied thisexhaustively, and only one CpG is used asan example of the type of stochastic processthat might apply. There are two interestingimplications of this suggested model. First,the increased variation over time shown indata and matched by the model supportsthe idea that disruption of epigenetic regula-tion occurs at the earliest stages of cancer.Second, it shows that the idea of looking fora defined epigenetic signature for cancer isflawed. Rather, we should be looking for ananti-profile: that is, stochastic departure from
a normal epigenetic signature.
Epigenetics and cancer hallmarks
Cancer is usually viewed as a complex groupof multiple disorders that are mostly driven bysomatic mutation and that involve the accre-tion of ten proposed properties: enhancedproliferation, growth suppressor evasion,anti-apoptosis, replicative immortality, angio-genesis, inflammation, altered metabolism,genomic instability and metastasis signal-ling117. We suggest that epigenetic disruptionlies at the heart of all of these processes andthat mutations enable and collaborate in thesedisruptions.The hallmark properties of can-cer arise not only by mutation but also gener-ally through stochastic epigenetic variation(as described above) and by natural selectionof phenotypes that are advantageous to cancercell survival and growth at the expense of thehost. This is described graphically in FIG. 3, inwhich the epigenome sits at the intersectionof the environment, genetic mutation andtumour cell growth.
Epigenetic damage arises from carcino-gens or diet (for example, methionine), aswell as injury and inflammation (for exam-
ple, altered LOCKs and LADs in EMT, asdescribed above). Errors in the mainte-nance of the epigenome over time thatis, simply ageing may also have a role inaccumulating stochastic epigenetic damage.The mechanism for maintaining epigeneticintegrity is damaged in either of two ways:through mutation in the genes that encodethese mechanisms (TABLE 1) or through epi-genetic modifications themselves in suchgenes (TABLE 2) resulting in positive feedback.Examples of epigenetic modifications thatlead to positive feedback include epigenetic
silencing ofEZH2 and loss of imprinting(LOI) in colorectal cancer. This damagedepigenetic maintenance machinery leads tostochastic drift from a normal epigenetic setpoint (FIG. 2) and selection for cancer hall-marks that give the cell a growth advantageat the expense of the host.
Some epigenetic modifications directlyaffect the hallmark properties of cancer;for example, shifts in the methylationboundaries at CGIs and CGI shores resultin enhanced proliferation and metabolicchange, and hypomethylated blocks lead toincreased invasive potential61. In addition,shifting DNA methylation at CGI bounda-ries leads to hypomethylated shores andthe activation of cell cycle genes that areoverexpressed in cancer7, another cancerhallmark. In this case loss of DNA methy-lation, which seems to be a ground state forembryonic stem cells24, seems to promote
gene expression, normally in developmentand abnormally in cancer. Similarly, LOI candirectly change the balance between apoptosisand proliferation118,119(FIG. 3).
Epigenetic dysregulation occurs veryearly in cancer. For example, DNA methy-lation is altered in the normal tissue of cancerpatients, and these changes increase with age,suggesting a mechanism by which cancerfrequency increases in an age-dependentmanner120122(FIG. 3). A second example isLOI of insulin-like growth factor 2(IGF2) incolorectal cancer, in which the normally silentallele becomes activated, leading to a doubledose of this mitogen. In colorectal cancer,LOI ofIGF2 is found in both the tumoursand the normal tissue of affected patients123,and induced LOI ofIgf2 in mice doubles thefrequency of tumour development124. A thirdexample is that Barretts oesophagus showsepigenetic modifications long before pro-gression to overt malignancy114. Epigeneticmodifications, particularly increased hetero-geneous outlier variability in DNA methy-lation, are also predictive of malignant changein cervical cancer125.
Where do mutations fit into this process?
We have previously suggested that evolutionmay favour mechanisms through geneticselection that allow for epigenetic hetero-geneity by providing a selective advantagein a changing environment113. We suggest asimilar mechanism might occur in cancer:that is, selection for epigenetic plasticityitself. More generally, we suggest that theinitial dysregulation of the epigenome col-laborates with crucial mutations to providethe phenotypic variation that allows theselection of the hallmark properties of can-cer. There is interplay between epigenetic
dysregulation, which provides the phenotypicheterogeneity that generates the hallmarkproperties of cancer and that potentiallyalters the mutation rate, and genetic muta-tions, which directly alter genes and path-ways to confer hallmark properties of cancer,as well as enabling and assisting in epigeneticdysregulation (FIG. 3). Furthermore, the epige-netic changes themselves may contribute toincreased mutation frequency.
We note that these mutations in epi-genetic controlling genes generally occurin two types of tumours: haematopoieticmalignancies and rare solid tumours(childhood solid tumours or aggressivesubtypes of common adult solid tumours).The implication of this observation, whichseems fairly well supported by a great dealof sequencing data, is that primary epige-netic modifications are a more prominentmechanism for cancer progression in com-
mon solid tumours than are mutations inepigenetic modifiers. The converse of thisargument is that mutations in epigeneticmodifiers have extremely strong effectson cellular behaviour, which is consist-ent with the profound aggressiveness ofsuch tumours and their relative rarity.Haematopoietic malignancies are consistentwith this view, as they commonly exhibitmutations in epigenetic modifier genes butarise almost completely progressed (thatis, widely disseminated) compared withsolid tumours. Even lymphomas, which canbe relatively indolent for long periods, aremore aggressive when associated withmutations in epigenetic modifiers.
Note that we do not exclude a primaryrole for mutations in conferring the hall-mark properties of cancer, and in fact webelieve that they do. We simply suggest thatthe epigenome is not relegated to merelya surrogate for mutation but rather thatit has important non-local ramifications.Limiting our understanding of epigeneticsin the context of cancer to the gene-centric
view may neglect valuable insights intohow cancer causes a general disruption of
genetic regulation through the epigenome.With that in mind, even genes that are notnormally thought to have a primary epi-genetic role may have a strong interactionwith the epigenome. For example, BRCA2,which at one time was thought to be a his-tone acetyltransferase126, is actually a bind-ing partner of the histone acetyltransferasePCAF (also known as KAT2B)127. SMAD4is mutated in pancreas adenocarcinoma andhas been shown to be an important driverof this cancer type, but it also interactswith chromatin remodellers83. p53 directly
P E R S P E C T I V E S
10 | ADVANCE ONLINE PUBLICATION www.nature.com/reviews/cancer
2013 Macmillan Publishers Limited. All rights reserved
-
7/27/2019 Cancer as a Dysregulated Epigenome
11/14
Epigenome
Disruptionand positivefeedback (EZH2,MYC and KLF4)
Genomeinstability
Mutation
AgeingEpigeneticregulators (TET2,MLL and ARID1A)
Environmentalfactors and
carcinogens
Inflammation Invasion Metabolism
Apoptosis
Proliferation
LOI andothers
Stochasticageing
Canonical genes
APCp53
Decreased
Increased
Precursorlesion
interacts with DNMT1, which act togetherto control the expression of anti-apoptoticgenes128. The epigenome can also be indi-
rectly affected through pathways that arecommonly mutated in cancer that canaffect the expression level of epigenetic con-trolling proteins. Examples include the reg-ulation of histone deacteylase 2 (HDAC2)129and DNMT1 (REF. 130) by mutations of theWNT-catenin pathway, such as APCmutations in colorectal cancer. In fact, theprotein stability of DNMT1, rather than itsexpression, may be altered by the PI3KAKTpathway so that levels of the DNMT1 pro-tein (and not mRNA) are changed withoutnew expression131.
Implications for diagnosis and therapy
Viewing epigenetic dysregulation itself as acommon driver of cancer progression has
important implications for cancer diagno-sis and therapy. For diagnosis it suggests apromising approach for identifying patients
very early in the course of disease, and fortherapy it suggests novel targets that couldbe the focus of therapeutic interventionearly or even before the development ofovert cancer.
In epigenetic detection, a great deal ofeffort has been invested in identifying CGIhypermethylation, with limited success49.We suggest that this is because an implica-tion of the recent whole-genome epigenetic
work shows that departure from a normalprofile is more pathognomonic than aspecific cancer epigenetic signature. Forexample, septin 9 (SEPT9) was reportedto be a sensitive and specific serum-basedmarker for colorectal cancer132. However,a casecontrol study carried out indepen-dently of the commercial developer showed90% sensitivity and 88% specificity forcolorectal cancer but only 71% sensitivityfor early cancers and 12% sensitivity foradenomas133, and therefore this marker isnot useful for early screening. Similarly,methylation of bone morphogenetic pro-tein 3(BMP3), eyes absent homologue 2(EYA2), ALX homeobox 4(ALX4) or
vimentin (VIM) was found in 66%, 66%,68% and 72%, respectively, of primarycolorectal cancers but in 7%, 5%, 11% and11%, respectively, of normal mucosa134.One notable exception to the disappointing
history of epigenetic detection of canceris glutathione S-transferase pi 1(GSTP1)in prostate cancer135, and this is probablybecause this particular gene has a crucialrole in the early stage of the disease ratherthan being an indicator of epigeneticdysregulation per se.
If a detection scheme was developedon the basis of anti-profiling instead, amuch higher sensitivity and predictive
value might be achieved. Consistent withthis idea, Teschendorff and colleagues125,136recently showed that the DNA methylation
variability was more predictive of cancerprogression than mean changes in DNAmethylation in both cervical125 and breast136cancers. This test can differentiate betweennormal and neoplastic cervical tissue with95% sensitivity and 78% specificity125. Infact, this test is predictive using tissue takenbefore the development of cervical neoplasia;neoplasia development within 3 years ofsample collection is predictable with a 71%sensitivity and 50% specificity125. DNAmethylation instability at specific loci ispredictive of survival in endometrial, ovar-ian, cervical and breast cancers, emphasiz-
ing the usefulness of epigenetic instabilityin chemoresistance136. Such an approachcould even be used in screening patientsfor early cancer: epigenetic anti-profilingof tumour cells using a sophisticated digitalPCR-based approach has already shownconsiderable utility for the early diagnosisof colorectal cancer in stool137.
The methylation signature at specificCGIs can also suggest specific forms oftailored therapy, as in O6-methylguanineDNA alkyltransferase (MGMT) methylation,which can indicate the therapeutic choice
Figure 3 | Collaboration of epigenetic modification and mutation in the hallmarks of cancer.
The epigenome sits at the intersection of the environment, genetic mutation and tumour cell growth.
Environmental factors, such as carcinogens or diet, as well as injury and inflammation, cause epi-
genetic reprogramming. The epigenome also accumulates damage stochastically and through age-
ing. The machinery for maintaining epigenetic integrity can be stably disrupted in either of two ways:
by mutation or by epigenetic change itself with positive feedback. Examples of mutation include
epigenetic regulator mutations (TABLE 1), whereas examples of epigenetic change include loss of
imprinting (LOI) of insulin-like growth factor 2 ( IGF2) in colorectal carcinogenesis, enhancer of zeste
homologue 2 (EZH2) silencing in prostate cancer (TABLE 2) and overexpression of reprogramming
factors. The disruption of epigenetic integrity maintenance leads to the loss of epigenetic regulation
and stochastic drift from a normal set point, followed by selection for cellular growth at the expense
of other cells (FIGS 1,2). Some epigenetic modifications, such as shifting methylation boundaries at
CpG islands and shores, lead to metabolic change and enhanced proliferation. Others, such as hypo-methylated blocks, lead to increased invasion. Still others, such as LOI, directly change the balance
between apoptosis and proliferation. Canonical mutations, such as in adenomatous polyposis coli
(APC) and TP53 (which encodes p53), directly affect cancer hallmarks but can also cause epigenetic
dysregulation. Similarly, epigenetic disruption, such as regional hypomethylation or CpG
hypermethylation, can lead to increased chromosomal rearrangements and mutations, respectively.
Instability of CpG island methylation boundaries also contributes to epigenetic dysregulation, allow-
ing for selection in response to the cellular environment for cellular growth advantage at the expense
of the host. ARID1A, AT-rich interactive domain-containing protein 1A; KLF4, Krppel-like factor 4;MLL, mixed lineage leukaemia; TET2, tet methylcytosine dioxygenase 2.
P E R S P E C T I V E S
NATURE REVIEWS |CANCER ADVANCE ONLINE PUBLICATION |11
2013 Macmillan Publishers Limited. All rights reserved
-
7/27/2019 Cancer as a Dysregulated Epigenome
12/14
for glioma138. However, other hypermethy-lated CGI markers have not shown a highdegree of sensitivity or specificity for canceror for key phenotypes, such as drug resist-ance, although such efforts are still ongo-ing139. In these cases DNA methylation isbeing used as a surrogate measure of geneexpression and/or gene regulation.
In the case of haematological malignan-cies, in which mutations in epigenetic regu-lators are most frequent (TABLE 1), therapytargeted towards these pathways is alreadyquite promising, including EZH2 inhibitorsfor lymphoma140,141, inhibitors of mutantIDH1 for acute myeloid leukaemia andmyelodysplasia142, and HDAC inhibitors forcutaneous T cell lymphoma108. Additionally,a clever approach has been taken to inhib-iting DOT1L, which is not mutated itselfbut which is a common target ofMLLtranslocations143.
In our view, the most exciting poten-tial application of this model of epige-netic dysregulation as a common driverthroughout cancer progression is thepotential for targeted chemoprevention.Currently, chemoprevention involveseither nutritional recommendations orthe use of non-prescription agents suchas cyclooxygenase 2 (COX2; also knownas PTGS2) inhibitors targeted towardsindividuals with a high risk of develop-ing colorectal cancer, such as those witha strong family history of colorectal can-cer 144,145. The approach to chemopreven-tion is far more limited than in cardiology,in which chemoprevention with prescrip-tion medication (statins) is extremelyeffective at preventing heart disease146; thiswas a controversial approach when statinswere first introduced for widespread use inthe population. But what if we could iden-tify the epigenetic disruption in patientsbefore neoplastic growth even begins(as has been shown for cervical cancer 125)?Then we might treat such patients withspecific inhibitors even before theydevelop cancer. Similarly, LOI ofIGF2 is
associated with an increased frequency ofcolorectal cancer123 and may be associatedwith gastric cancer risk147. LOI ofIGF2also substantially increases the frequencyof colon preneoplastic aberrant crypt fociin mice treated with the carcinogen azoxy-methane, and inhibition of signalling atthe IGF2 receptor reduces the incidence ofneoplasms to a level even lower than thatfound in mice with normal Igf2 imprint-ing148. Thus, we might be able to identifydisruption of the epigenome or even therisk of such disruption through epigenetic
or genetic testing before cancer arises, andthen treat patients preventively to reducecancer incidence. This seems to us to be apotentially far more effective mechanismfor reducing cancer mortality than thetreatment of late-stage disease, and it wouldargue strongly for an epigenome-centredapproach.
Winston Timp and Andrew P. Feinberg are at the Center
for Epigenetics, Johns Hopkins University School of
Medicine, 855 N. Wolfe Street, Rangos 570,
Baltimore, Maryland 21205, USA.
Winston Timp is also at the Department of Biomedical
Engineering, Johns Hopkins University School of
Medicine, 720 Rutland Avenue, Baltimore,
Maryland 21205, USA.
Andrew P. Feinberg is also at the Department of
Medicine, Johns Hopkins University School
of Medicine, 855 N. Wolfe Street, Rangos 570,
Baltimore, Maryland 21205, USA.
Correspondence to A.P.F.
e-mail: [email protected]
doi:10.1038/nrc3486
Published online 13 June 2013
1. Weinhouse, S. Isozymes in cancer. Cancer Res. 31,
11661167 (1971).
2. Shih, C. & Weinberg, R. A. Isolation of a transforming
sequence from a human bladder carcinoma cell line.
Cell29, 161169 (1982).
3. Feinberg, A. P. & Vogelstein, B. Hypomethylation
distinguishes genes of some human cancers from their
normal counterparts. Nature 301, 8992 (1983).
4. Knudson, A. G. Mutation and cancer: statistical study
of retinoblastoma. Proc. Natl Acad. Sci. USA 68,
820823 (1971).
5. Greger, V., Passarge, E., Hopping, W., Messmer, E. &
Horsthemke, B. Epigenetic changes may contribute to
the formation and spontaneous regression of
retinoblastoma. Hum. Genet. 83, 155158 (1989).
6. Sakai, T. et al. Allele-specific hypermethylation of the
retinoblastoma tumor-suppressor gene. Amer.J. Hum.
Genet. 48, 880888 (1991).
7. Hansen, K. D. et al. Increased methylation variation in
epigenetic domains across cancer types. Nature
Genet. 43, 768775 (2011).
8. Boveri, T. Concerning the Origin of Malignant Tumors
(Williams and Wilkins, 1929).
9. Zink, D., Fischer, A. H. & Nickerson, J. A. Nuclear
structure in cancer cells. Nature Rev. Cancer4,
677687 (2004).
10. Lever, E. & Sheer, D. The role of nuclear organization
in cancer.J. Pathol. 220, 114125 (2010).
11. Wen, B., Wu, H., Shinkai, Y., Irizarry, R. A. &
Feinberg, A. P. Large histone H3 lysine 9 dimethylated
chromatin blocks distinguish differentiated from
embryonic stem cells. Nature Genet. 41, 246250
(2009).
12. Zullo, J. M. et al. DNA sequence-dependent
compartmentalization and silencing of chromatin at
the nuclear lamina. Cell149, 14741487 (2012).
13. Hu, S., Cheng, L. & Wen, B. Large chromatin domains
in pluripotent and differentiated cells.Acta Biochim.Biophys. Sin. (Shanghai) 44, 4853 (2012).
14. Peric-Hupkes, D. et al. Molecular maps of the
reorganization of genome-nuclear lamina interactions
during differentiation. Mol. Cell38, 603613 (2010).
15. Reddy, K. L., Zullo, J. M., Bertolino, E. & Singh, H.
Transcriptional repression mediated by repositioning
of genes to the nuclear lamina. Nature 452, 243247
(2008).
16. Chow, K. H., Factor, R. E. & Ullman, K. S. The nuclear
envelope environment and its cancer connections.
Nature Rev. Cancer12, 196209 (2012).
17. Hawkins, R. D. et al. Distinct epigenomic landscapes of
pluripotent and lineage-committed human cells. Cell
Stem Cell6, 479491 (2010).
18. Hon, G. C. et al. Global DNA hypomethylation coupled
to repressive chromatin domain formation and gene
silencing in breast cancer. Genome Res. 22, 246258
(2012).
19. McDonald, O. G., Wu, H., Timp, W., Doi, A. &
Feinberg, A. P. Genome-scale epigenetic
reprogramming during epithelial-to-mesenchymal
transition. Nature Struct. Mol. Biol. 18, 867874
(2011).
20. Zhou, V. W., Goren, A. & Bernstein, B. E. Charting
histone modifications and the functional organization
of mammalian genomes. Nature Rev. Genet. 12, 718
(2011).21. Wen, B. et al. Euchromatin islands in large
heterochromatin domains are enriched for CTCF
binding and differentially DNA-methylated regions.BMC Genomics 13, 566 (2012).
22. Berman, B. P.et al. Regions of focal DNA
hypermethylation and long-range hypomethylation in
colorectal cancer coincide with nuclear lamina-
associated domains. Nature Genet. 44, 4046
(2012).
23. Ehrlich, M. DNA hypomethylation in cancer cells.
Epigenomics1, 239259 (2009).
24. Lister, R. et al. Human DNA methylomes at base
resolution show widespread epigenomic differences.
Nature 462, 315322 (2009).
25. Lister, R. et al. Hotspots of aberrant epigenomic
reprogramming in human induced pluripotent stem
cells. Nature 471, 6873 (2011).
26. De, S. & Michor, F. DNA secondary structures and
epigenetic determinants of cancer genome evolution.
Nature Struct. Mol. Biol. 18, 950955 (2011).
27. Nestor, C., Ruzov, A., Meehan, R. & Dunican, D.
Enzymatic approaches and bisulfite sequencing cannot
distinguish between 5-methylcytosine and
5-hydroxymethylcytosine in DNA. Biotechniques 48,
317319 (2010).
28. Jin, S.-G., Kadam, S. & Pfeifer, G. P. Examination of the
specificity of DNA methylation profiling techniques
towards 5-methylcytosine and 5-hydroxymethylcytosine.
Nucleic Acids Res. 38, e125 (2010).
29. Chen, M.-L. et al. Quantification of 5-methylcytosineand 5-hydroxymethylcytosine in genomic DNA from
hepatocellular carcinoma tissues by capillary
hydrophilic-interaction liquid chromatography/
quadrupole TOF mass spectrometry. Clin. Chem.59,824832 (2013).
30. Lian, Christine, G. et al. Loss of
5-hydroxymethylcytosine is an epigenetic hallmark of
melanoma. Cell150, 11351146 (2012).
31. Zhang, L.-T. et al. Quantification of the sixth DNA base
5-hydroxymethylcytosine in colorectal cancer tissue
and C-26 cell line. Bioanalysis5, 839845 (2013).
32. Booth, M. J. et al. Quantitative sequencing of
5-methylcytosine and 5-hydroxymethylcytosine at
single-base resolution. Science 336, 934937 (2012).33. Sun, Z. et al. High-resolution enzymatic mapping of
genomic 5-hydroxymethylcytosine in mouse embryonic
stem cells. Cell Rep. 3, 567576 (2013).
34. Wolf, S. F. & Migeon, B. R. Studies of X chromosome
DNA methylation in normal human cells. Nature 295,
667671 (1982).
35. Bird, A., Taggart, M., Frommer, M., Miller, O. J. &
Macleod, D. A fraction of the mouse genome that is
derived from islands of nonmethylated, CpG-rich DNA.
Cell40, 9199 (1985).
36. Goelz, S. E., Vogelstein, B., Hamilton, S. R. &
Feinberg, A. P. Hypomethylation of DNA from benign
and malignant human colon neoplasms. Science 228,
187190 (1985).
37. Feinberg, A. P., Gehrke, C. W., Kuo, K. C. & Ehrlich, M.
Reduced genomic 5-methylcytosine content in human
colonic neoplasia. Cancer Res. 48, 11591161
(1988).38. Feinberg, A. P. & Vogelstein, B. Hypomethylation of
ras oncogenes in primary human cancers. Biochem.Biophys. Res. Commun. 111, 4754 (1983).
39. De Smet, C. et al. The activation of human gene
MAGE-1 in tumor cells is correlated with genome-widedemethylation. Proc. Natl Acad. Sci. USA 93,
71497153 (1996).
40. Iacobuzio-Donahue, C. A. et al. Exploration of global
gene expression patterns in pancreatic
adenocarcinoma using cDNA microarrays. Amer.
J. Pathol. 162, 11511162 (2003).
41. Oshimo, Y.et al. Promoter methylation of cyclin D2
gene in gastric carcinoma. Int. J. Oncol. 23,
16631670 (2003).
42. Akiyama, Y., Maesawa, C., Ogasawara, S.,
Terashima, M. & Masuda, T. Cell-type-specific
repression of the maspin gene is disrupted frequently
by demethylation at the promoter region in gastric
intestinal metaplasia and cancer cells.Am.J. Pathol.163, 19111919 (2003).
P E R S P E C T I V E S
12 | ADVANCE ONLINE PUBLICATION www.nature.com/reviews/cancer
2013 Macmillan Publishers Limited. All rights reserved
mailto:[email protected]:[email protected] -
7/27/2019 Cancer as a Dysregulated Epigenome
13/14
43. Cho, M. et al. Hypomethylation of the MN/CA9
promoter and upregulated MN/CA9 expression in
human renal cell carcinoma. Br.J. Cancer85,563567 (2001).
44. Nakamura, N. & Takenaga, K. Hypomethylation of the
metastasis-associated S100A4 gene correlates with
gene activation in human colon adenocarcinoma cell
lines. Clin. Exper. Metastasis 16, 471479 (1998).
45. Badal, V. et al. CpG methylation of human
papillomavirus type 16 DNA in cervical cancer cell
lines and in clinical specimens: genomic
hypomethylation correlates with carcinogenicprogression.J. Virol. 77, 62276234 (2003).
46. de Capoa, A. et al. DNA demethylation is directly
related to tumour progression: evidence in normal,
pre-malignant and malignant cells from uterine cervix
samples. Oncol. Rep. 10, 545549 (2003).
47. Sato, N. et al. Frequent hypomethylation of multiple
genes overexpressed in pancreatic ductal
adenocarcinoma. Cancer Res. 63, 41584166 (2003).
48. Piyathilake, C. J. et al. Race- and age-dependent
alterations in global methylation of DNA in squamous
cell carcinoma of the lung (United States). Cancer
Causes Control14, 3742 (2003).
49. Baylin, S. B. & Jones, P. A. A decade of exploring the
cancer epigenome biological and translational
implications. Nature Rev. Cancer11, 726734 (2011).
50. Jones, P. A. et al.De novo methylation of the MyoD1
CpG island during the establishment of immortal cell
lines. Proc. Natl Acad. Sci. USA 87, 61176121 (1990).
51. Bestor, T. H. Unanswered questions about the role of
promoter methylation in carcinogenesis.Ann. NY
Acad. Sci. 983, 2227 (2003).52. Hosoya, K. et al. Adenomatous polyposis coli 1A is
likely to be methylated as a passenger in human
gastric carcinogenesis. Cancer Lett. 285, 182189
(2009).
53. Levanon, D. et al. Absence of Runx3 expression in
normal gastrointestinal epithelium calls into question
its tumour suppressor function. EMBO Mol. Med. 3,
593604 (2011).
54. Hitchins, M. P. et al. Inheritance of a cancer-associated
MLH1 germ-line epimutation. N. Engl. J. Med. 356,697705 (2007).
55. Hitchins, M. P. & Ward, R. L. Erasure of MLH1
methylation in spermatozoa-implications for epigenetic
inheritance. Nature Genet. 39, 1289 (2007).
56. Bachman, K. E. et al. Histone modifications and
silencing prior to DNA methylation of a tumor
suppressor gene. Cancer Cell3, 8995 (2003).
57. Sproul, D. et al. Transcriptionally repressed genes
become aberrantly methylated and distinguish tumors
of different lineages in breast cancer. Proc. Natl Acad.Sci. USA 108, 43644369 (2011).
58. Sproul, D. et al. Tissue of origin determines cancer-
associated CpG island promoter hypermethylation
patterns. Genome Biol. 13, R84 (2012).59. Irizarry, R. A. et al. Comprehensive high-throughput
arrays for relative methylation (CHARM). Genome Res.
18, 780790 (2008).
60. Irizarry, R. A. et al. The human colon cancer
methylome shows similar hypo- and hypermethylation
at conserved tissue-specific CpG island shores. Nature
Genet. 41, 178186 (2009).61. Doi, A. et al. Differential methylation of tissue- and
cancer-specific CpG island shores distinguishes human
induced pluripotent stem cells, embryonic stem cells
and fibroblasts. Nature Genet. 41, 13501353 (2009).
62. Teschendorff, A. et al. Epigenetic variability in cells of
normal cytology is associated with the risk of future
morphological transformation. Genome Med. 4, 24
(2012).
63. Corrada Bravo, H., Pihur, V., McCall, M., Irizarry, R. &Leek, J. Gene expression anti-profiles as a basis for
accurate universal cancer signatures. BMC
Bioinformatics 13, 272 (2012).
64. Wang, G. G., Allis, C. D. & Chi, P. Chromatin
remodeling and cancer, part II: ATP-dependent
chromatin remodeling. Trends Mol. Med. 13,
373380 (2007).
65. Wang, G. G., Allis, C. D. & Chi, P. Chromatin
remodeling and cancer, part I: covalent histone
modifications. Trends Mol. Med. 13, 363372 (2007).
66. Roberts, C. W. & Orkin, S. H. The SWI/SNF complex--
chromatin and cancer. Nature Rev. Cancer4,
133142 (2004).
67. Kaplan, N. et al. The DNA-encoded nucleosome
organization of a eukaryotic genome. Nature 458,
362366 (2009).
68. Strahl, B. D. & Allis, C. D. The language of covalent
histone modifications. Nature 403, 4145 (2000).
69. Mito, Y., Henikoff, J. G. & Henikoff, S. Genome-scale
profiling of histone H3.3 replacement patterns. Nature
Genet. 37, 10901097 (2005).
70. Zofall, M. et al. Histone H2A.Z cooperates with RNAi
and heterochromatin factors to suppress antisense
RNAs. Nature 461, 419422 (2009).
71. Gardner, K. E., Allis, C. D. & Strahl, B. D. Operating on
chromatin, a colorful language where context matters.
J. Mol. Biol. 409, 3646 (2011).72. Mikkelsen, T. S. et al. Genome-wide maps of chromatin
state in pluripotent and lineage-committed cells.
Nature 448, 553560 (2007).73. Young, N. L. et al. High throughput characterization of
combinatorial histone codes. Mol.Cell. Proteomics 8,22662284 (2009).
74. Voigt, P. et al. Asymmetrically modified nucleosomes.
Cell151, 181193 (2012).
75. Ohm, J. E. et al. A stem cell-like chromatin pattern
may predispose tumor suppressor genes to DNA
hypermethylation and heritable silencing. Nature
Genet. 39, 237242 (2007).
76. Widschwendter, M. et al. Epigenetic stem cell
signature in cancer. Nature Genet. 39, 157158
(2007).
77. Chodavarapu, R. K. et al. Relationship between
nucleosome positioning and DNA methylation. Nature
466, 388392 (2010).
78. Gupta, R. A. et al. Long non-coding RNAHOTAIR
reprograms chromatin state to promote cancer
metastasis. Nature 464, 10711076 (2010).
79. Yu, W. et al. Epigenetic silencing of tumour suppressor
gene p15 by its antisense RNA. Nature 451,
202206 (2008).80. Guil, S. & Esteller, M. Cis-acting noncoding RNAs:
friends and foes. Nature Struct. Mol. Biol. 19,
10681075 (2012).
81. Forbes, S. A. et al. COSMIC: mining complete cancer
genomes in the Catalogue of Somatic Mutations in
Cancer. Nucleic Acids Res. 39, D945D950 (2011).
82. Imielinski, M. et al. Mapping the hallmarks of lung
adenocarcinoma with massively parallel sequencing.
Cell150, 11071120 (2012).
83. Jones, S. et al. Core signaling pathways in human
pancreatic cancers revealed by global genomic
analyses. Science 321, 18011806 (2008).
84. Jiao, Y.et al. DAXX/ATRX, MEN1, and mTOR pathway
genes are frequently altered in pancreatic
neuroendocrine tumors. Science 331, 11991203
(2011).
85. Schwartzentruber, J. et al. Driver mutations in histone
H3.3 and chromatin remodelling genes in paediatric
glioblastoma. Nature 482, 226231 (2012).
86. Krivtsov, A. V. & Armstrong, S. A. MLL translocations,histone modifications and leukaemia stem-cell
development. Nature Rev. Cancer7, 823833
(2007).
87. Yan, X. J. et al. Exome sequencing identifies somatic
mutations of DNA methyltransferase gene DNMT3A in
acute monocytic leukemia. Nature Genet. 43,
309315 (2011).
88. Ley, T. J. et al. DNMT3A mutations in acute myeloid
leukemia. N. Engl. J. Med. 363, 24242433 (2010).89. Figueroa, M. E. et al. Leukemic IDH1 and IDH2
mutations result in a hypermethylation phenotype,
disrupt TET2 function, and impair hematopoietic
differentiation. Cancer Cell18, 553567 (2010).
90. Kosmider, O. et al.TET2 gene mutation is a frequent
and adverse event in chronic myelomonocytic
leukemia. Haematologica 94, 16761681 (2009).
91. Pasqualucci, L. et al. Inactivating mutations of
acetyltransferase genes in B-cell lymphoma. Nature471, 189195 (2011).
92. Pasqualucci, L. et al. Analysis of the coding genome ofdiffuse large B-cell lymphoma. Nature Genet. 43,
830837 (2011).
93. Bernt, K. M. et al. MLL-rearranged leukemia is
dependent on aberrant H3K79 methylation by
DOT1L. Cancer Cell20, 6678 (2011).
94. Mullighan, C. G. et al.CREBBPmutations in relapsed
acute lymphoblastic leukaemia. Nature 471,
235239 (2011).95. Wu, G. et al. Somatic histone H3 alterations in
pediatric diffuse intrinsic pontine gliomas and non-
brainstem glioblastomas. Nature Genet. 44, 251253
(2012).
96. Jones, S. et al. Frequent mutations of chromatin
remodeling geneARID1A in ovarian clear cell
carcinoma. Science 330, 228231 (2010).97. Wiegand, K. C. et al.ARID1A mutations in
endometriosis-associated ovarian carcinomas. N. Engl.J. Med. 363, 15321543 (2010).
98. Cooper, D. N., Mort, M., Stenson, P. D., Ball, E. V. &
Chuzhanova, N. A. Methylation-mediated deamination
of 5-methylcytosine appears to give rise to mutations
causing human inherited disease in CpNpG
trinucleotides, as well as in CpG dinucleotides. Hum.
Genomics 4, 406410 (2010).
99. Greenman, C. et al. Patterns of somatic mutation in
human cancer genomes. Nature 446, 153158
(2007).100. Schuster-Bockler, B. & Lehner, B. Chromatin
organization is a major influence on reg